diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/alloc.c | 60 | ||||
| -rw-r--r-- | src/alloc.h | 194 | ||||
| -rw-r--r-- | src/atomic.h | 37 | ||||
| -rw-r--r-- | src/bar.c | 189 | ||||
| -rw-r--r-- | src/bar.h | 4 | ||||
| -rw-r--r-- | src/bfs.h | 241 | ||||
| -rw-r--r-- | src/bfstd.c | 357 | ||||
| -rw-r--r-- | src/bfstd.h | 207 | ||||
| -rw-r--r-- | src/bftw.c | 86 | ||||
| -rw-r--r-- | src/bftw.h | 21 | ||||
| -rw-r--r-- | src/bit.h | 200 | ||||
| -rw-r--r-- | src/color.c | 670 | ||||
| -rw-r--r-- | src/color.h | 29 | ||||
| -rw-r--r-- | src/ctx.c | 135 | ||||
| -rw-r--r-- | src/ctx.h | 37 | ||||
| -rw-r--r-- | src/diag.c | 49 | ||||
| -rw-r--r-- | src/diag.h | 150 | ||||
| -rw-r--r-- | src/dir.c | 20 | ||||
| -rw-r--r-- | src/dir.h | 31 | ||||
| -rw-r--r-- | src/dstring.c | 47 | ||||
| -rw-r--r-- | src/dstring.h | 161 | ||||
| -rw-r--r-- | src/eval.c | 306 | ||||
| -rw-r--r-- | src/eval.h | 8 | ||||
| -rw-r--r-- | src/exec.c | 14 | ||||
| -rw-r--r-- | src/exec.h | 12 | ||||
| -rw-r--r-- | src/expr.c | 10 | ||||
| -rw-r--r-- | src/expr.h | 38 | ||||
| -rw-r--r-- | src/fsade.c | 35 | ||||
| -rw-r--r-- | src/fsade.h | 24 | ||||
| -rw-r--r-- | src/ioq.c | 604 | ||||
| -rw-r--r-- | src/ioq.h | 87 | ||||
| -rw-r--r-- | src/list.h | 150 | ||||
| -rw-r--r-- | src/main.c | 7 | ||||
| -rw-r--r-- | src/mtab.c | 34 | ||||
| -rw-r--r-- | src/mtab.h | 10 | ||||
| -rw-r--r-- | src/opt.c | 398 | ||||
| -rw-r--r-- | src/opt.h | 2 | ||||
| -rw-r--r-- | src/parse.c | 1106 | ||||
| -rw-r--r-- | src/parse.h | 4 | ||||
| -rw-r--r-- | src/prelude.h | 421 | ||||
| -rw-r--r-- | src/printf.c | 57 | ||||
| -rw-r--r-- | src/printf.h | 12 | ||||
| -rw-r--r-- | src/pwcache.c | 5 | ||||
| -rw-r--r-- | src/pwcache.h | 24 | ||||
| -rw-r--r-- | src/sanity.h | 50 | ||||
| -rw-r--r-- | src/sighook.c | 692 | ||||
| -rw-r--r-- | src/sighook.h | 83 | ||||
| -rw-r--r-- | src/stat.c | 74 | ||||
| -rw-r--r-- | src/stat.h | 22 | ||||
| -rw-r--r-- | src/thread.c | 15 | ||||
| -rw-r--r-- | src/thread.h | 14 | ||||
| -rw-r--r-- | src/trie.c | 271 | ||||
| -rw-r--r-- | src/trie.h | 91 | ||||
| -rw-r--r-- | src/typo.c | 1 | ||||
| -rw-r--r-- | src/typo.h | 4 | ||||
| -rw-r--r-- | src/version.c | 32 | ||||
| -rw-r--r-- | src/xregex.c | 67 | ||||
| -rw-r--r-- | src/xregex.h | 20 | ||||
| -rw-r--r-- | src/xspawn.c | 199 | ||||
| -rw-r--r-- | src/xspawn.h | 21 | ||||
| -rw-r--r-- | src/xtime.c | 215 | ||||
| -rw-r--r-- | src/xtime.h | 75 |
62 files changed, 5591 insertions, 2648 deletions
diff --git a/src/alloc.c b/src/alloc.c index ebaff38..f505eda 100644 --- a/src/alloc.c +++ b/src/alloc.c @@ -1,11 +1,13 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "alloc.h" + +#include "bfs.h" #include "bit.h" #include "diag.h" #include "sanity.h" + #include <errno.h> #include <stdlib.h> #include <stdint.h> @@ -18,24 +20,22 @@ # define ALLOC_MAX (SIZE_MAX / 2) #endif -/** Portable aligned_alloc()/posix_memalign(). */ +/** posix_memalign() wrapper. */ static void *xmemalign(size_t align, size_t size) { bfs_assert(has_single_bit(align)); bfs_assert(align >= sizeof(void *)); - bfs_assert(is_aligned(align, size)); -#if BFS_HAS_ALIGNED_ALLOC - return aligned_alloc(align, size); -#else + // Since https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2072.htm, + // aligned_alloc() doesn't require the size to be a multiple of align. + // But the sanitizers don't know about that yet, so always use + // posix_memalign(). void *ptr = NULL; errno = posix_memalign(&ptr, align, size); return ptr; -#endif } void *alloc(size_t align, size_t size) { bfs_assert(has_single_bit(align)); - bfs_assert(is_aligned(align, size)); if (size > ALLOC_MAX) { errno = EOVERFLOW; @@ -51,7 +51,6 @@ void *alloc(size_t align, size_t size) { void *zalloc(size_t align, size_t size) { bfs_assert(has_single_bit(align)); - bfs_assert(is_aligned(align, size)); if (size > ALLOC_MAX) { errno = EOVERFLOW; @@ -71,8 +70,6 @@ void *zalloc(size_t align, size_t size) { void *xrealloc(void *ptr, size_t align, size_t old_size, size_t new_size) { bfs_assert(has_single_bit(align)); - bfs_assert(is_aligned(align, old_size)); - bfs_assert(is_aligned(align, new_size)); if (new_size == 0) { free(ptr); @@ -106,10 +103,10 @@ void *reserve(void *ptr, size_t align, size_t size, size_t count) { size_t old_size = size * count; // Capacity is doubled every power of two, from 0→1, 1→2, 2→4, etc. - // If we stayed within the same size class, re-use ptr. + // If we stayed within the same size class, reuse ptr. if (count & (count - 1)) { // Tell sanitizers about the new array element - sanitize_alloc((char *)ptr + old_size, size); + sanitize_resize(ptr, old_size, old_size + size, bit_ceil(count) * size); errno = 0; return ptr; } @@ -124,7 +121,7 @@ void *reserve(void *ptr, size_t align, size_t size, size_t count) { } // Pretend we only allocated one more element - sanitize_free((char *)ret + old_size + size, new_size - old_size - size); + sanitize_resize(ret, new_size, old_size + size, new_size); errno = 0; return ret; } @@ -176,7 +173,7 @@ void arena_init(struct arena *arena, size_t align, size_t size) { } /** Allocate a new slab. */ -attr(cold) +_cold static int slab_alloc(struct arena *arena) { // Make the initial allocation size ~4K size_t size = 4096; @@ -231,6 +228,7 @@ void arena_free(struct arena *arena, void *ptr) { union chunk *chunk = ptr; chunk_set_next(arena, chunk, arena->chunks); arena->chunks = chunk; + sanitize_uninit(chunk, arena->size); sanitize_free(chunk, arena->size); } @@ -250,7 +248,7 @@ void arena_destroy(struct arena *arena) { sanitize_uninit(arena); } -void varena_init(struct varena *varena, size_t align, size_t min, size_t offset, size_t size) { +void varena_init(struct varena *varena, size_t align, size_t offset, size_t size) { varena->align = align; varena->offset = offset; varena->size = size; @@ -259,7 +257,7 @@ void varena_init(struct varena *varena, size_t align, size_t min, size_t offset, // The smallest size class is at least as many as fit in the smallest // aligned allocation size - size_t min_count = (flex_size(align, min, offset, size, 1) - offset + size - 1) / size; + size_t min_count = (flex_size(align, offset, size, 1) - offset + size - 1) / size; varena->shift = bit_width(min_count - 1); } @@ -272,7 +270,7 @@ static size_t varena_size_class(struct varena *varena, size_t count) { /** Get the exact size of a flexible struct. */ static size_t varena_exact_size(const struct varena *varena, size_t count) { - return flex_size(varena->align, 0, varena->offset, varena->size, count); + return flex_size(varena->align, varena->offset, varena->size, count); } /** Get the arena for the given array length. */ @@ -306,8 +304,7 @@ void *varena_alloc(struct varena *varena, size_t count) { } // Tell the sanitizers the exact size of the allocated struct - sanitize_free(ret, arena->size); - sanitize_alloc(ret, varena_exact_size(varena, count)); + sanitize_resize(ret, arena->size, varena_exact_size(varena, count), arena->size); return ret; } @@ -319,15 +316,14 @@ void *varena_realloc(struct varena *varena, void *ptr, size_t old_count, size_t return NULL; } - size_t new_exact_size = varena_exact_size(varena, new_count); - size_t old_exact_size = varena_exact_size(varena, old_count); + size_t old_size = old_arena->size; + size_t new_size = new_arena->size; if (new_arena == old_arena) { - if (new_count < old_count) { - sanitize_free((char *)ptr + new_exact_size, old_exact_size - new_exact_size); - } else if (new_count > old_count) { - sanitize_alloc((char *)ptr + old_exact_size, new_exact_size - old_exact_size); - } + sanitize_resize(ptr, + varena_exact_size(varena, old_count), + varena_exact_size(varena, new_count), + new_size); return ptr; } @@ -336,16 +332,18 @@ void *varena_realloc(struct varena *varena, void *ptr, size_t old_count, size_t return NULL; } - size_t old_size = old_arena->size; - sanitize_alloc((char *)ptr + old_exact_size, old_size - old_exact_size); + // Non-sanitized builds don't bother computing exact sizes, and just use + // the potentially-larger arena size for each size class instead. To + // allow the below memcpy() to work with the less-precise sizes, expand + // the old allocation to its full capacity. + sanitize_resize(ptr, varena_exact_size(varena, old_count), old_size, old_size); - size_t new_size = new_arena->size; size_t min_size = new_size < old_size ? new_size : old_size; memcpy(ret, ptr, min_size); arena_free(old_arena, ptr); - sanitize_free((char *)ret + new_exact_size, new_size - new_exact_size); + sanitize_resize(ret, new_size, varena_exact_size(varena, new_count), new_size); return ret; } diff --git a/src/alloc.h b/src/alloc.h index 095134a..1fafbab 100644 --- a/src/alloc.h +++ b/src/alloc.h @@ -8,127 +8,145 @@ #ifndef BFS_ALLOC_H #define BFS_ALLOC_H -#include "prelude.h" +#include "bfs.h" + #include <errno.h> #include <stddef.h> #include <stdlib.h> +#define IS_ALIGNED(align, size) \ + (((size) & ((align) - 1)) == 0) + /** Check if a size is properly aligned. */ static inline bool is_aligned(size_t align, size_t size) { - return (size & (align - 1)) == 0; + return IS_ALIGNED(align, size); } +#define ALIGN_FLOOR(align, size) \ + ((size) & ~((align) - 1)) + /** Round down to a multiple of an alignment. */ static inline size_t align_floor(size_t align, size_t size) { - return size & ~(align - 1); + return ALIGN_FLOOR(align, size); } +#define ALIGN_CEIL(align, size) \ + ((((size) - 1) | ((align) - 1)) + 1) + /** Round up to a multiple of an alignment. */ static inline size_t align_ceil(size_t align, size_t size) { - return align_floor(align, size + align - 1); + return ALIGN_CEIL(align, size); } /** - * Saturating array size. - * - * @param align - * Array element alignment. - * @param size - * Array element size. - * @param count - * Array element count. - * @return - * size * count, saturating to the maximum aligned value on overflow. + * Saturating size addition. + */ +static inline size_t size_add(size_t lhs, size_t rhs) { + size_t ret = lhs + rhs; + return ret >= lhs ? ret : (size_t)-1; +} + +/** + * Saturating size multiplication. */ -static inline size_t array_size(size_t align, size_t size, size_t count) { +static inline size_t size_mul(size_t size, size_t count) { size_t ret = size * count; - return ret / size == count ? ret : ~(align - 1); + return ret / size == count ? ret : (size_t)-1; } /** Saturating array sizeof. */ #define sizeof_array(type, count) \ - array_size(alignof(type), sizeof(type), count) + size_mul(sizeof(type), count) /** Size of a struct/union field. */ #define sizeof_member(type, member) \ sizeof(((type *)NULL)->member) /** + * @internal + * Our flexible struct size calculations assume that structs have the minimum + * trailing padding to align the type properly. A pathological ABI that adds + * extra padding would result in us under-allocating space for those structs, + * so we static_assert() that no such padding exists. + */ +#define ASSERT_FLEX_ABI(type, member) \ + ASSERT_FLEX_ABI_( \ + ALIGN_CEIL(alignof(type), offsetof(type, member)) >= sizeof(type), \ + "Unexpected tail padding in " #type) + +/** + * @internal + * The contortions here allow static_assert() to be used in expressions, rather + * than just declarations. + */ +#define ASSERT_FLEX_ABI_(...) \ + ((void)sizeof(struct { char _; static_assert(__VA_ARGS__); })) + +/** * Saturating flexible struct size. * - * @param align + * @align * Struct alignment. - * @param min - * Minimum struct size. - * @param offset + * @offset * Flexible array member offset. - * @param size + * @size * Flexible array element size. - * @param count + * @count * Flexible array element count. * @return * The size of the struct with count flexible array elements. Saturates * to the maximum aligned value on overflow. */ -static inline size_t flex_size(size_t align, size_t min, size_t offset, size_t size, size_t count) { - size_t ret = size * count; - size_t overflow = ret / size != count; - - size_t extra = offset + align - 1; - ret += extra; - overflow |= ret < extra; - ret |= -overflow; +static inline size_t flex_size(size_t align, size_t offset, size_t size, size_t count) { + size_t ret = size_mul(size, count); + ret = size_add(ret, offset + align - 1); ret = align_floor(align, ret); - - // Make sure flex_sizeof(type, member, 0) >= sizeof(type), even if the - // type has more padding than necessary for alignment - if (min > align_ceil(align, offset)) { - ret = ret < min ? min : ret; - } - return ret; } /** * Computes the size of a flexible struct. * - * @param type + * @type * The type of the struct containing the flexible array. - * @param member + * @member * The name of the flexible array member. - * @param count + * @count * The length of the flexible array. * @return * The size of the struct with count flexible array elements. Saturates * to the maximum aligned value on overflow. */ #define sizeof_flex(type, member, count) \ - flex_size(alignof(type), sizeof(type), offsetof(type, member), sizeof_member(type, member[0]), count) + (ASSERT_FLEX_ABI(type, member), flex_size( \ + alignof(type), offsetof(type, member), sizeof_member(type, member[0]), count)) /** * General memory allocator. * - * @param align + * @align * The required alignment. - * @param size + * @size * The size of the allocation. * @return * The allocated memory, or NULL on failure. */ -attr(malloc(free, 1), aligned_alloc(1, 2)) +_malloc(free, 1) +_aligned_alloc(1, 2) void *alloc(size_t align, size_t size); /** * Zero-initialized memory allocator. * - * @param align + * @align * The required alignment. - * @param size + * @size * The size of the allocation. * @return * The allocated memory, or NULL on failure. */ -attr(malloc(free, 1), aligned_alloc(1, 2)) +_malloc(free, 1) +_aligned_alloc(1, 2) void *zalloc(size_t align, size_t size); /** Allocate memory for the given type. */ @@ -158,18 +176,19 @@ void *zalloc(size_t align, size_t size); /** * Alignment-aware realloc(). * - * @param ptr + * @ptr * The pointer to reallocate. - * @param align + * @align * The required alignment. - * @param old_size + * @old_size * The previous allocation size. - * @param new_size + * @new_size * The new allocation size. * @return * The reallocated memory, or NULL on failure. */ -attr(nodiscard, aligned_alloc(2, 4)) +_aligned_alloc(2, 4) +_nodiscard void *xrealloc(void *ptr, size_t align, size_t old_size, size_t new_size); /** Reallocate memory for an array. */ @@ -183,11 +202,11 @@ void *xrealloc(void *ptr, size_t align, size_t old_size, size_t new_size); /** * Reserve space for one more element in a dynamic array. * - * @param ptr + * @ptr * The pointer to reallocate. - * @param align + * @align * The required alignment. - * @param count + * @count * The current size of the array. * @return * The reallocated memory, on both success *and* failure. On success, @@ -195,17 +214,17 @@ void *xrealloc(void *ptr, size_t align, size_t old_size, size_t new_size); * for (count + 1) elements. On failure, errno will be non-zero, and * ptr will returned unchanged. */ -attr(nodiscard) +_nodiscard void *reserve(void *ptr, size_t align, size_t size, size_t count); /** * Convenience macro to grow a dynamic array. * - * @param type + * @type * The array element type. - * @param type **ptr + * @type **ptr * A pointer to the array. - * @param size_t *count + * @size_t *count * A pointer to the array's size. * @return * On success, a pointer to the newly reserved array element, i.e. @@ -253,7 +272,7 @@ void arena_free(struct arena *arena, void *ptr); /** * Allocate an object out of the arena. */ -attr(malloc(arena_free, 2)) +_malloc(arena_free, 2) void *arena_alloc(struct arena *arena); /** @@ -287,40 +306,39 @@ struct varena { /** * Initialize a varena for a struct with the given layout. * - * @param varena + * @varena * The varena to initialize. - * @param align + * @align * alignof(type) - * @param min - * sizeof(type) - * @param offset + * @offset * offsetof(type, flexible_array) - * @param size + * @size * sizeof(flexible_array[i]) */ -void varena_init(struct varena *varena, size_t align, size_t min, size_t offset, size_t size); +void varena_init(struct varena *varena, size_t align, size_t offset, size_t size); /** * Initialize a varena for the given type and flexible array. * - * @param varena + * @varena * The varena to initialize. - * @param type + * @type * A struct type containing a flexible array. - * @param member + * @member * The name of the flexible array member. */ #define VARENA_INIT(varena, type, member) \ - varena_init(varena, alignof(type), sizeof(type), offsetof(type, member), sizeof_member(type, member[0])) + (ASSERT_FLEX_ABI(type, member), varena_init( \ + varena, alignof(type), offsetof(type, member), sizeof_member(type, member[0]))) /** * Free an arena-allocated flexible struct. * - * @param varena + * @varena * The that allocated the object. - * @param ptr + * @ptr * The object to free. - * @param count + * @count * The length of the flexible array. */ void varena_free(struct varena *varena, void *ptr, size_t count); @@ -328,46 +346,46 @@ void varena_free(struct varena *varena, void *ptr, size_t count); /** * Arena-allocate a flexible struct. * - * @param varena + * @varena * The varena to allocate from. - * @param count + * @count * The length of the flexible array. * @return * The allocated struct, or NULL on failure. */ -attr(malloc(varena_free, 2)) +_malloc(varena_free, 2) void *varena_alloc(struct varena *varena, size_t count); /** * Resize a flexible struct. * - * @param varena + * @varena * The varena to allocate from. - * @param ptr + * @ptr * The object to resize. - * @param old_count - * The old array lenth. - * @param new_count + * @old_count + * The old array length. + * @new_count * The new array length. * @return * The resized struct, or NULL on failure. */ -attr(nodiscard) +_nodiscard void *varena_realloc(struct varena *varena, void *ptr, size_t old_count, size_t new_count); /** * Grow a flexible struct by an arbitrary amount. * - * @param varena + * @varena * The varena to allocate from. - * @param ptr + * @ptr * The object to resize. - * @param count + * @count * Pointer to the flexible array length. * @return * The resized struct, or NULL on failure. */ -attr(nodiscard) +_nodiscard void *varena_grow(struct varena *varena, void *ptr, size_t *count); /** diff --git a/src/atomic.h b/src/atomic.h index f1a6bea..5c2826f 100644 --- a/src/atomic.h +++ b/src/atomic.h @@ -8,6 +8,8 @@ #ifndef BFS_ATOMIC_H #define BFS_ATOMIC_H +#include "bfs.h" + #include <stdatomic.h> /** @@ -18,9 +20,9 @@ /** * Shorthand for atomic_load_explicit(). * - * @param obj + * @obj * A pointer to the atomic object. - * @param order + * @order * The memory ordering to use, without the memory_order_ prefix. * @return * The loaded value. @@ -82,4 +84,35 @@ #define fetch_and(obj, arg, order) \ atomic_fetch_and_explicit(obj, arg, memory_order_##order) +/** + * Shorthand for atomic_thread_fence(). + */ +#if __SANITIZE_THREAD__ +// TSan doesn't support fences: https://github.com/google/sanitizers/issues/1415 +# define thread_fence(obj, order) \ + fetch_add(obj, 0, order) +#else +# define thread_fence(obj, order) \ + atomic_thread_fence(memory_order_##order) +#endif + +/** + * Shorthand for atomic_signal_fence(). + */ +#define signal_fence(order) \ + atomic_signal_fence(memory_order_##order) + +/** + * A hint to the CPU to relax while it spins. + */ +#if __has_builtin(__builtin_ia32_pause) +# define spin_loop() __builtin_ia32_pause() +#elif __has_builtin(__builtin_arm_yield) +# define spin_loop() __builtin_arm_yield() +#elif BFS_HAS_BUILTIN_RISCV_PAUSE +# define spin_loop() __builtin_riscv_pause() +#else +# define spin_loop() ((void)0) +#endif + #endif // BFS_ATOMIC_H @@ -1,52 +1,55 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "bar.h" + +#include "alloc.h" #include "atomic.h" +#include "bfs.h" #include "bfstd.h" #include "bit.h" #include "dstring.h" +#include "sighook.h" + #include <errno.h> #include <fcntl.h> #include <signal.h> #include <stdarg.h> #include <stdio.h> +#include <stdlib.h> #include <string.h> -#include <sys/ioctl.h> +#include <termios.h> +#include <unistd.h> struct bfs_bar { int fd; atomic unsigned int width; atomic unsigned int height; -}; -/** The global status bar instance. */ -static struct bfs_bar the_bar = { - .fd = -1, + struct sighook *exit_hook; + struct sighook *winch_hook; }; /** Get the terminal size, if possible. */ static int bfs_bar_getsize(struct bfs_bar *bar) { -#ifdef TIOCGWINSZ struct winsize ws; - if (ioctl(bar->fd, TIOCGWINSZ, &ws) != 0) { + if (xtcgetwinsize(bar->fd, &ws) != 0) { return -1; } store(&bar->width, ws.ws_col, relaxed); store(&bar->height, ws.ws_row, relaxed); return 0; -#else - errno = ENOTSUP; - return -1; -#endif } -/** Async Signal Safe puts(). */ -static int ass_puts(int fd, const char *str) { - size_t len = strlen(str); - return xwrite(fd, str, len) == len ? 0 : -1; +/** Write a string to the status bar (async-signal-safe). */ +static int bfs_bar_write(struct bfs_bar *bar, const char *str, size_t len) { + return xwrite(bar->fd, str, len) == len ? 0 : -1; +} + +/** Write a string to the status bar (async-signal-safe). */ +static int bfs_bar_puts(struct bfs_bar *bar, const char *str) { + return bfs_bar_write(bar, str, strlen(str)); } /** Number of decimal digits needed for terminal sizes. */ @@ -68,66 +71,63 @@ static char *ass_itoa(char *str, unsigned int n) { return str + len; } +/** Reset the scrollable region and hide the bar. */ +static int bfs_bar_reset(struct bfs_bar *bar) { + return bfs_bar_puts(bar, + "\0337" // DECSC: Save cursor + "\033[r" // DECSTBM: Reset scrollable region + "\0338" // DECRC: Restore cursor + "\033[J" // ED: Erase display from cursor to end + ); +} + +/** Hide the bar if the terminal is shorter than this. */ +#define BFS_BAR_MIN_HEIGHT 3 + /** Update the size of the scrollable region. */ static int bfs_bar_resize(struct bfs_bar *bar) { - char esc_seq[12 + ITOA_DIGITS] = + unsigned int height = load(&bar->height, relaxed); + if (height < BFS_BAR_MIN_HEIGHT) { + return bfs_bar_reset(bar); + } + + static const char PREFIX[] = + "\033D" // IND: Line feed, possibly scrolling + "\033[1A" // CUU: Move cursor up 1 row "\0337" // DECSC: Save cursor "\033[;"; // DECSTBM: Set scrollable region + static const char SUFFIX[] = + "r" // (end of DECSTBM) + "\0338" // DECRC: Restore the cursor + "\033[J"; // ED: Erase display from cursor to end - // DECSTBM takes the height as the second argument - unsigned int height = load(&bar->height, relaxed); - char *ptr = esc_seq + strlen(esc_seq); - ptr = ass_itoa(ptr, height - 1); + char esc_seq[sizeof(PREFIX) + ITOA_DIGITS + sizeof(SUFFIX)]; - strcpy(ptr, - "r" // DECSTBM - "\0338" // DECRC: Restore the cursor - "\033[J" // ED: Erase display from cursor to end - ); + // DECSTBM takes the height as the second argument + char *cur = stpcpy(esc_seq, PREFIX); + cur = ass_itoa(cur, height - 1); + cur = stpcpy(cur, SUFFIX); - return ass_puts(bar->fd, esc_seq); + return bfs_bar_write(bar, esc_seq, cur - esc_seq); } #ifdef SIGWINCH /** SIGWINCH handler. */ -static void sighand_winch(int sig) { - int error = errno; - - bfs_bar_getsize(&the_bar); - bfs_bar_resize(&the_bar); - - errno = error; +static void bfs_bar_sigwinch(int sig, siginfo_t *info, void *arg) { + struct bfs_bar *bar = arg; + bfs_bar_getsize(bar); + bfs_bar_resize(bar); } #endif -/** Reset the scrollable region and hide the bar. */ -static int bfs_bar_reset(struct bfs_bar *bar) { - return ass_puts(bar->fd, - "\0337" // DECSC: Save cursor - "\033[r" // DECSTBM: Reset scrollable region - "\0338" // DECRC: Restore cursor - "\033[J" // ED: Erase display from cursor to end - ); -} - /** Signal handler for process-terminating signals. */ -static void sighand_reset(int sig) { - bfs_bar_reset(&the_bar); - raise(sig); -} - -/** Register sighand_reset() for a signal. */ -static void reset_before_death_by(int sig) { - struct sigaction sa = { - .sa_handler = sighand_reset, - .sa_flags = SA_RESETHAND, - }; - sigemptyset(&sa.sa_mask); - sigaction(sig, &sa, NULL); +static void bfs_bar_sigexit(int sig, siginfo_t *info, void *arg) { + struct bfs_bar *bar = arg; + bfs_bar_reset(bar); } /** printf() to the status bar with a single write(). */ -attr(printf(2, 3)) +_printf(2, 3) static int bfs_bar_printf(struct bfs_bar *bar, const char *format, ...) { va_list args; va_start(args, format); @@ -138,64 +138,47 @@ static int bfs_bar_printf(struct bfs_bar *bar, const char *format, ...) { return -1; } - int ret = ass_puts(bar->fd, str); + int ret = bfs_bar_write(bar, str, dstrlen(str)); dstrfree(str); return ret; } struct bfs_bar *bfs_bar_show(void) { - if (the_bar.fd >= 0) { - errno = EBUSY; - goto fail; + struct bfs_bar *bar = ALLOC(struct bfs_bar); + if (!bar) { + return NULL; } - char term[L_ctermid]; - ctermid(term); - if (strlen(term) == 0) { - errno = ENOTTY; + bar->fd = open_cterm(O_RDWR | O_CLOEXEC); + if (bar->fd < 0) { goto fail; } - the_bar.fd = open(term, O_RDWR | O_CLOEXEC); - if (the_bar.fd < 0) { - goto fail; + if (bfs_bar_getsize(bar) != 0) { + goto fail_close; } - if (bfs_bar_getsize(&the_bar) != 0) { + bar->exit_hook = atsigexit(bfs_bar_sigexit, bar); + if (!bar->exit_hook) { goto fail_close; } - reset_before_death_by(SIGABRT); - reset_before_death_by(SIGINT); - reset_before_death_by(SIGPIPE); - reset_before_death_by(SIGQUIT); - reset_before_death_by(SIGTERM); - #ifdef SIGWINCH - struct sigaction sa = { - .sa_handler = sighand_winch, - .sa_flags = SA_RESTART, - }; - sigemptyset(&sa.sa_mask); - sigaction(SIGWINCH, &sa, NULL); + bar->winch_hook = sighook(SIGWINCH, bfs_bar_sigwinch, bar, 0); + if (!bar->winch_hook) { + goto fail_hook; + } #endif - unsigned int height = load(&the_bar.height, relaxed); - bfs_bar_printf(&the_bar, - "\n" // Make space for the bar - "\0337" // DECSC: Save cursor - "\033[;%ur" // DECSTBM: Set scrollable region - "\0338" // DECRC: Restore cursor - "\033[1A", // CUU: Move cursor up 1 row - height - 1 - ); - - return &the_bar; + bfs_bar_resize(bar); + return bar; +fail_hook: + sigunhook(bar->exit_hook); fail_close: - close_quietly(the_bar.fd); - the_bar.fd = -1; + close_quietly(bar->fd); fail: + free(bar); return NULL; } @@ -205,6 +188,10 @@ unsigned int bfs_bar_width(const struct bfs_bar *bar) { int bfs_bar_update(struct bfs_bar *bar, const char *str) { unsigned int height = load(&bar->height, relaxed); + if (height < BFS_BAR_MIN_HEIGHT) { + return 0; + } + return bfs_bar_printf(bar, "\0337" // DECSC: Save cursor "\033[%u;0f" // HVP: Move cursor to row, column @@ -223,17 +210,11 @@ void bfs_bar_hide(struct bfs_bar *bar) { return; } - signal(SIGABRT, SIG_DFL); - signal(SIGINT, SIG_DFL); - signal(SIGPIPE, SIG_DFL); - signal(SIGQUIT, SIG_DFL); - signal(SIGTERM, SIG_DFL); -#ifdef SIGWINCH - signal(SIGWINCH, SIG_DFL); -#endif + sigunhook(bar->winch_hook); + sigunhook(bar->exit_hook); bfs_bar_reset(bar); xclose(bar->fd); - bar->fd = -1; + free(bar); } @@ -27,9 +27,9 @@ unsigned int bfs_bar_width(const struct bfs_bar *bar); /** * Update the status bar message. * - * @param bar + * @bar * The status bar to update. - * @param str + * @str * The string to display. * @return * 0 on success, -1 on failure. diff --git a/src/bfs.h b/src/bfs.h new file mode 100644 index 0000000..3cee727 --- /dev/null +++ b/src/bfs.h @@ -0,0 +1,241 @@ +// Copyright © Tavian Barnes <tavianator@tavianator.com> +// SPDX-License-Identifier: 0BSD + +/** + * Configuration and fundamental utilities. + */ + +#ifndef BFS_H +#define BFS_H + +// Standard versions + +/** Possible __STDC_VERSION__ values. */ +#define C95 199409L +#define C99 199901L +#define C11 201112L +#define C17 201710L +#define C23 202311L + +/** Possible _POSIX_C_SOURCE and _POSIX_<OPTION> values. */ +#define POSIX_1990 1 +#define POSIX_1992 2 +#define POSIX_1993 199309L +#define POSIX_1995 199506L +#define POSIX_2001 200112L +#define POSIX_2008 200809L +#define POSIX_2024 202405L + +// Build configuration + +#include "config.h" + +#ifndef BFS_COMMAND +# define BFS_COMMAND "bfs" +#endif + +#ifndef BFS_HOMEPAGE +# define BFS_HOMEPAGE "https://tavianator.com/projects/bfs.html" +#endif + +#ifndef BFS_LINT +# define BFS_LINT false +#endif + +// This is a symbol instead of a literal so we don't have to rebuild everything +// when the version number changes +extern const char bfs_version[]; + +extern const char bfs_confflags[]; +extern const char bfs_cc[]; +extern const char bfs_cppflags[]; +extern const char bfs_cflags[]; +extern const char bfs_ldflags[]; +extern const char bfs_ldlibs[]; + +// Get __GLIBC__ +#include <assert.h> + +// Fundamental utilities + +/** + * Get the length of an array. + */ +#define countof(...) (sizeof(__VA_ARGS__) / sizeof(0[__VA_ARGS__])) + +/** + * False sharing/destructive interference/largest cache line size. + */ +#ifdef __GCC_DESTRUCTIVE_SIZE +# define FALSE_SHARING_SIZE __GCC_DESTRUCTIVE_SIZE +#else +# define FALSE_SHARING_SIZE 64 +#endif + +/** + * True sharing/constructive interference/smallest cache line size. + */ +#ifdef __GCC_CONSTRUCTIVE_SIZE +# define TRUE_SHARING_SIZE __GCC_CONSTRUCTIVE_SIZE +#else +# define TRUE_SHARING_SIZE 64 +#endif + +/** + * Alignment specifier that avoids false sharing. + */ +#define cache_align alignas(FALSE_SHARING_SIZE) + +// Wrappers for attributes + +/** + * Silence warnings about switch/case fall-throughs. + */ +#if __has_attribute(fallthrough) +# define _fallthrough __attribute__((fallthrough)) +#else +# define _fallthrough ((void)0) +#endif + +/** + * Silence warnings about unused declarations. + */ +#if __has_attribute(unused) +# define _maybe_unused __attribute__((unused)) +#else +# define _maybe_unused +#endif + +/** + * Warn if a value is unused. + */ +#if __has_attribute(warn_unused_result) +# define _nodiscard __attribute__((warn_unused_result)) +#else +# define _nodiscard +#endif + +/** + * Hint to avoid inlining a function. + */ +#if __has_attribute(noinline) +# define _noinline __attribute__((noinline)) +#else +# define _noinline +#endif + +/** + * Marks a non-returning function. + */ +#if __STDC_VERSION__ >= C23 +# define _noreturn [[noreturn]] +#else +# define _noreturn _Noreturn +#endif + +/** + * Hint that a function is unlikely to be called. + */ +#if __has_attribute(cold) +# define _cold _noinline __attribute__((cold)) +#else +# define _cold _noinline +#endif + +/** + * Adds compiler warnings for bad printf()-style function calls, if supported. + */ +#if __has_attribute(format) +# define _printf(fmt, args) __attribute__((format(printf, fmt, args))) +#else +# define _printf(fmt, args) +#endif + +/** + * Annotates functions that potentially modify and return format strings. + */ +#if __has_attribute(format_arg) +# define _format_arg(arg) __attribute__((format_arg(arg))) +#else +# define _format_arg(arg) +#endif + +/** + * Annotates allocator-like functions. + */ +#if __has_attribute(malloc) +# if __GNUC__ >= 11 && !__OPTIMIZE__ // malloc(deallocator) disables inlining on GCC +# define _malloc(...) _nodiscard __attribute__((malloc(__VA_ARGS__))) +# else +# define _malloc(...) _nodiscard __attribute__((malloc)) +# endif +#else +# define _malloc(...) _nodiscard +#endif + +/** + * Specifies that a function returns allocations with a given alignment. + */ +#if __has_attribute(alloc_align) +# define _alloc_align(param) __attribute__((alloc_align(param))) +#else +# define _alloc_align(param) +#endif + +/** + * Specifies that a function returns allocations with a given size. + */ +#if __has_attribute(alloc_size) +# define _alloc_size(...) __attribute__((alloc_size(__VA_ARGS__))) +#else +# define _alloc_size(...) +#endif + +/** + * Shorthand for _alloc_align() and _alloc_size(). + */ +#define _aligned_alloc(align, ...) _alloc_align(align) _alloc_size(__VA_ARGS__) + +/** + * Check if function multiversioning via GNU indirect functions (ifunc) is supported. + * + * Disabled on TSan due to https://github.com/google/sanitizers/issues/342. + */ +#ifndef BFS_USE_TARGET_CLONES +# if __has_attribute(target_clones) && (__GLIBC__ || __FreeBSD__) && !__SANITIZE_THREAD__ +# define BFS_USE_TARGET_CLONES true +# else +# define BFS_USE_TARGET_CLONES false +# endif +#endif + +/** + * Apply the target_clones attribute, if available. + */ +#if BFS_USE_TARGET_CLONES +# define _target_clones(...) __attribute__((target_clones(__VA_ARGS__))) +#else +# define _target_clones(...) +#endif + +/** + * Mark the size of a flexible array member. + */ +#if __has_attribute(counted_by) +# define _counted_by(...) __attribute__((counted_by(__VA_ARGS__))) +#else +# define _counted_by(...) +#endif + +/** + * Optimization hint to not unroll a loop. + */ +#if BFS_HAS_PRAGMA_NOUNROLL +# define _nounroll _Pragma("nounroll") +#elif __GNUC__ && !__clang__ +# define _nounroll _Pragma("GCC unroll 0") +#else +# define _nounroll +#endif + +#endif // BFS_H diff --git a/src/bfstd.c b/src/bfstd.c index f8ce871..b78af7a 100644 --- a/src/bfstd.c +++ b/src/bfstd.c @@ -1,13 +1,15 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "bfstd.h" + +#include "bfs.h" #include "bit.h" #include "diag.h" #include "sanity.h" #include "thread.h" #include "xregex.h" + #include <errno.h> #include <fcntl.h> #include <langinfo.h> @@ -15,25 +17,28 @@ #include <locale.h> #include <nl_types.h> #include <pthread.h> +#include <sched.h> #include <stddef.h> #include <stdint.h> #include <stdio.h> #include <stdlib.h> #include <string.h> +#include <sys/ioctl.h> #include <sys/resource.h> #include <sys/stat.h> #include <sys/types.h> #include <sys/wait.h> +#include <termios.h> #include <unistd.h> #include <wchar.h> -#if BFS_USE_SYS_SYSMACROS_H +#if __has_include(<sys/sysmacros.h>) # include <sys/sysmacros.h> -#elif BFS_USE_SYS_MKDEV_H +#elif __has_include(<sys/mkdev.h>) # include <sys/mkdev.h> #endif -#if BFS_USE_UTIL_H +#if __has_include(<util.h>) # include <util.h> #endif @@ -199,6 +204,171 @@ const char *xgetprogname(void) { return cmd; } +/** Common prologue for xstrto*() wrappers. */ +static int xstrtox_prologue(const char *str) { + // strto*() skips leading spaces, but we want to reject them + if (xisspace(str[0])) { + errno = EINVAL; + return -1; + } + + errno = 0; + return 0; +} + +/** Common epilogue for xstrto*() wrappers. */ +static int xstrtox_epilogue(const char *str, char **end, char *endp) { + if (errno != 0) { + return -1; + } + + if (end) { + *end = endp; + } + + // If end is NULL, make sure the entire string is valid + if (endp == str || (!end && *endp != '\0')) { + errno = EINVAL; + return -1; + } + + return 0; +} + +int xstrtos(const char *str, char **end, int base, short *value) { + long n; + if (xstrtol(str, end, base, &n) != 0) { + return -1; + } + + if (n < SHRT_MIN || n > SHRT_MAX) { + errno = ERANGE; + return -1; + } + + *value = n; + return 0; +} + +int xstrtoi(const char *str, char **end, int base, int *value) { + long n; + if (xstrtol(str, end, base, &n) != 0) { + return -1; + } + + if (n < INT_MIN || n > INT_MAX) { + errno = ERANGE; + return -1; + } + + *value = n; + return 0; +} + +int xstrtol(const char *str, char **end, int base, long *value) { + if (xstrtox_prologue(str) != 0) { + return -1; + } + + char *endp; + *value = strtol(str, &endp, base); + return xstrtox_epilogue(str, end, endp); +} + +int xstrtoll(const char *str, char **end, int base, long long *value) { + if (xstrtox_prologue(str) != 0) { + return -1; + } + + char *endp; + *value = strtoll(str, &endp, base); + return xstrtox_epilogue(str, end, endp); +} + +int xstrtof(const char *str, char **end, float *value) { + if (xstrtox_prologue(str) != 0) { + return -1; + } + + char *endp; + *value = strtof(str, &endp); + return xstrtox_epilogue(str, end, endp); +} + +int xstrtod(const char *str, char **end, double *value) { + if (xstrtox_prologue(str) != 0) { + return -1; + } + + char *endp; + *value = strtod(str, &endp); + return xstrtox_epilogue(str, end, endp); +} + +int xstrtous(const char *str, char **end, int base, unsigned short *value) { + unsigned long n; + if (xstrtoul(str, end, base, &n) != 0) { + return -1; + } + + if (n > USHRT_MAX) { + errno = ERANGE; + return -1; + } + + *value = n; + return 0; +} + +int xstrtoui(const char *str, char **end, int base, unsigned int *value) { + unsigned long n; + if (xstrtoul(str, end, base, &n) != 0) { + return -1; + } + + if (n > UINT_MAX) { + errno = ERANGE; + return -1; + } + + *value = n; + return 0; +} + +/** Common epilogue for xstrtou*() wrappers. */ +static int xstrtoux_epilogue(const char *str, char **end, char *endp) { + if (xstrtox_epilogue(str, end, endp) != 0) { + return -1; + } + + if (str[0] == '-') { + errno = ERANGE; + return -1; + } + + return 0; +} + +int xstrtoul(const char *str, char **end, int base, unsigned long *value) { + if (xstrtox_prologue(str) != 0) { + return -1; + } + + char *endp; + *value = strtoul(str, &endp, base); + return xstrtoux_epilogue(str, end, endp); +} + +int xstrtoull(const char *str, char **end, int base, unsigned long long *value) { + if (xstrtox_prologue(str) != 0) { + return -1; + } + + char *endp; + *value = strtoull(str, &endp, base); + return xstrtoux_epilogue(str, end, endp); +} + /** Compile and execute a regular expression for xrpmatch(). */ static int xrpregex(nl_item item, const char *response) { const char *pattern = nl_langinfo(item); @@ -285,7 +455,7 @@ const char *xstrerror(int errnum) { // On FreeBSD with MemorySanitizer, duplocale() triggers // https://github.com/llvm/llvm-project/issues/65532 -#if BFS_HAS_STRERROR_L && !(__FreeBSD__ && SANITIZE_MEMORY) +#if BFS_HAS_STRERROR_L && !(__FreeBSD__ && __SANITIZE_MEMORY__) # if BFS_HAS_USELOCALE locale_t loc = uselocale((locale_t)0); # else @@ -322,6 +492,10 @@ const char *xstrerror(int errnum) { return ret; } +const char *errstr(void) { + return xstrerror(errno); +} + /** Get the single character describing the given file type. */ static char type_char(mode_t mode) { switch (mode & S_IFMT) { @@ -437,7 +611,9 @@ int rlim_cmp(rlim_t a, rlim_t b) { } dev_t xmakedev(int ma, int mi) { -#ifdef makedev +#if __QNX__ + return makedev(0, ma, mi); +#elif defined(makedev) return makedev(ma, mi); #else return (ma << 8) | mi; @@ -468,6 +644,32 @@ pid_t xwaitpid(pid_t pid, int *status, int flags) { return ret; } +int open_cterm(int flags) { + char path[L_ctermid]; + if (ctermid(path) == NULL || strlen(path) == 0) { + errno = ENOTTY; + return -1; + } + + return open(path, flags); +} + +int xtcgetwinsize(int fd, struct winsize *ws) { +#if BFS_HAS_TCGETWINSIZE + return tcgetwinsize(fd, ws); +#else + return ioctl(fd, TIOCGWINSZ, ws); +#endif +} + +int xtcsetwinsize(int fd, const struct winsize *ws) { +#if BFS_HAS_TCSETWINSIZE + return tcsetwinsize(fd, ws); +#else + return ioctl(fd, TIOCSWINSZ, ws); +#endif +} + int dup_cloexec(int fd) { #ifdef F_DUPFD_CLOEXEC return fcntl(fd, F_DUPFD_CLOEXEC, 0); @@ -637,8 +839,14 @@ error: return NULL; } +#if BFS_HAS_STRTOFFLAGS +# define BFS_STRTOFFLAGS strtofflags +#elif BFS_HAS_STRING_TO_FLAGS +# define BFS_STRTOFFLAGS string_to_flags +#endif + int xstrtofflags(const char **str, unsigned long long *set, unsigned long long *clear) { -#if BSD && !__GNU__ +#ifdef BFS_STRTOFFLAGS char *str_arg = (char *)*str; #if __OpenBSD__ @@ -649,11 +857,7 @@ int xstrtofflags(const char **str, unsigned long long *set, unsigned long long * bfs_fflags_t set_arg = 0; bfs_fflags_t clear_arg = 0; -#if __NetBSD__ - int ret = string_to_flags(&str_arg, &set_arg, &clear_arg); -#else - int ret = strtofflags(&str_arg, &set_arg, &clear_arg); -#endif + int ret = BFS_STRTOFFLAGS(&str_arg, &set_arg, &clear_arg); *str = str_arg; *set = set_arg; @@ -663,47 +867,124 @@ int xstrtofflags(const char **str, unsigned long long *set, unsigned long long * errno = EINVAL; } return ret; -#else // !BSD +#else // !BFS_STRTOFFLAGS errno = ENOTSUP; return -1; #endif } -size_t asciilen(const char *str) { - return asciinlen(str, strlen(str)); -} +long xsysconf(int name) { +#if __FreeBSD__ && __SANITIZE_MEMORY__ + // Work around https://github.com/llvm/llvm-project/issues/88163 + __msan_scoped_disable_interceptor_checks(); +#endif -size_t asciinlen(const char *str, size_t n) { - size_t i = 0; + long ret = sysconf(name); -#if SIZE_WIDTH % 8 == 0 - // Word-at-a-time isascii() - for (size_t word; i + sizeof(word) <= n; i += sizeof(word)) { - memcpy(&word, str + i, sizeof(word)); +#if __FreeBSD__ && __SANITIZE_MEMORY__ + __msan_scoped_enable_interceptor_checks(); +#endif - const size_t mask = (SIZE_MAX / 0xFF) << 7; // 0x808080... - word &= mask; - if (!word) { - continue; + return ret; +} + +#if BFS_HAS_SCHED_GETAFFINITY +/** Get the CPU count in an affinity mask of the given size. */ +static long bfs_sched_getaffinity(size_t size) { + cpu_set_t set, *pset = &set; + + if (size > sizeof(set)) { + pset = malloc(size); + if (!pset) { + return -1; } + } + + long ret = -1; + if (sched_getaffinity(0, size, pset) == 0) { +# ifdef CPU_COUNT_S + ret = CPU_COUNT_S(size, pset); +# else + bfs_assert(size <= sizeof(set)); + ret = CPU_COUNT(pset); +# endif + } + + if (pset != &set) { + free(pset); + } + return ret; +} +#endif + +long nproc(void) { + long ret = 0; -#if ENDIAN_NATIVE == ENDIAN_BIG - word = bswap(word); -#elif ENDIAN_NATIVE != ENDIAN_LITTLE +#if BFS_HAS_SCHED_GETAFFINITY + size_t size = sizeof(cpu_set_t); + do { + ret = bfs_sched_getaffinity(size); + +# ifdef CPU_COUNT_S + // On Linux, sched_getaffinity(2) says: + // + // When working on systems with large kernel CPU affinity masks, one must + // dynamically allocate the mask argument (see CPU_ALLOC(3)). Currently, + // the only way to do this is by probing for the size of the required mask + // using sched_getaffinity() calls with increasing mask sizes (until the + // call does not fail with the error EINVAL). + size *= 2; +# else + // No support for dynamically-sized CPU masks break; +# endif + } while (ret < 0 && errno == EINVAL); #endif - size_t first = trailing_zeros(word) / 8; - return i + first; + if (ret < 1) { + ret = xsysconf(_SC_NPROCESSORS_ONLN); } -#endif - for (; i < n; ++i) { - if (!xisascii(str[i])) { - break; - } + if (ret < 1) { + ret = 1; } + return ret; +} + +size_t asciilen(const char *str) { + return asciinlen(str, strlen(str)); +} + +size_t asciinlen(const char *str, size_t n) { + const unsigned char *ustr = (const unsigned char *)str; + size_t i = 0; + + // Word-at-a-time isascii() +#define CHUNK(n) CHUNK_(uint##n##_t, load8_leu##n) +#define CHUNK_(type, load8) \ + (n - i >= sizeof(type)) { \ + type word = load8(ustr + i); \ + type mask = (((type)-1) / 0xFF) << 7; /* 0x808080.. */ \ + word &= mask; \ + i += trailing_zeros(word) / 8; \ + if (word) { \ + return i; \ + } \ + } + +#if SIZE_WIDTH >= 64 + while CHUNK(64); + if CHUNK(32); +#else + while CHUNK(32); +#endif + if CHUNK(16); + if CHUNK(8); + +#undef CHUNK_ +#undef CHUNK + return i; } @@ -905,14 +1186,14 @@ static char *dollar_quote(char *dest, char *end, const char *str, size_t len, en /** How much of this string is safe as a bare word? */ static size_t bare_len(const char *str, size_t len) { - // https://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_02 + // https://pubs.opengroup.org/onlinepubs/9799919799/utilities/V3_chap02.html#tag_19_02 size_t ret = strcspn(str, "|&;<>()$`\\\"' *?[#~=%!{}"); return ret < len ? ret : len; } /** How much of this string is safe to double-quote? */ static size_t quotable_len(const char *str, size_t len) { - // https://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_02_03 + // https://pubs.opengroup.org/onlinepubs/9799919799/utilities/V3_chap02.html#tag_19_02_03 size_t ret = strcspn(str, "`$\\\"!"); return ret < len ? ret : len; } diff --git a/src/bfstd.h b/src/bfstd.h index f91e380..15dd949 100644 --- a/src/bfstd.h +++ b/src/bfstd.h @@ -8,8 +8,8 @@ #ifndef BFS_BFSTD_H #define BFS_BFSTD_H -#include "prelude.h" -#include "sanity.h" +#include "bfs.h" + #include <stddef.h> #include <ctype.h> @@ -18,7 +18,7 @@ * Work around https://github.com/llvm/llvm-project/issues/65532 by forcing a * function, not a macro, to be called. */ -#if __FreeBSD__ && SANITIZE_MEMORY +#if __FreeBSD__ && __SANITIZE_MEMORY__ # define BFS_INTERCEPT(fn) (fn) #else # define BFS_INTERCEPT(fn) fn @@ -48,9 +48,9 @@ * Check if an error code is "like" another one. For example, ENOTDIR is * like ENOENT because they can both be triggered by non-existent paths. * - * @param error + * @error * The error code to check. - * @param category + * @category * The category to test for. Known categories include ENOENT and * ENAMETOOLONG. * @return @@ -66,7 +66,7 @@ bool errno_is_like(int category); /** * Apply the "negative errno" convention. * - * @param ret + * @ret * The return value of the attempted operation. * @return * ret, if non-negative, otherwise -errno. @@ -106,7 +106,7 @@ int try(int ret); /** * Re-entrant dirname() variant that always allocates a copy. * - * @param path + * @path * The path in question. * @return * The parent directory of the path. @@ -116,7 +116,7 @@ char *xdirname(const char *path); /** * Re-entrant basename() variant that always allocates a copy. * - * @param path + * @path * The path in question. * @return * The final component of the path. @@ -126,7 +126,7 @@ char *xbasename(const char *path); /** * Find the offset of the final component of a path. * - * @param path + * @path * The path in question. * @return * The offset of the basename. @@ -138,9 +138,9 @@ size_t xbaseoff(const char *path); /** * fopen() variant that takes open() style flags. * - * @param path + * @path * The path to open. - * @param flags + * @flags * Flags to pass to open(). */ FILE *xfopen(const char *path, int flags); @@ -148,9 +148,9 @@ FILE *xfopen(const char *path, int flags); /** * Convenience wrapper for getdelim(). * - * @param file + * @file * The file to read. - * @param delim + * @delim * The delimiter character to split on. * @return * The read chunk (without the delimiter), allocated with malloc(). @@ -169,6 +169,56 @@ char *xgetdelim(FILE *file, char delim); const char *xgetprogname(void); /** + * Like xstrtol(), but for short. + */ +int xstrtos(const char *str, char **end, int base, short *value); + +/** + * Like xstrtol(), but for int. + */ +int xstrtoi(const char *str, char **end, int base, int *value); + +/** + * Wrapper for strtol() that forbids leading spaces. + */ +int xstrtol(const char *str, char **end, int base, long *value); + +/** + * Wrapper for strtoll() that forbids leading spaces. + */ +int xstrtoll(const char *str, char **end, int base, long long *value); + +/** + * Like xstrtoul(), but for unsigned short. + */ +int xstrtous(const char *str, char **end, int base, unsigned short *value); + +/** + * Like xstrtoul(), but for unsigned int. + */ +int xstrtoui(const char *str, char **end, int base, unsigned int *value); + +/** + * Wrapper for strtoul() that forbids leading spaces, negatives. + */ +int xstrtoul(const char *str, char **end, int base, unsigned long *value); + +/** + * Wrapper for strtoull() that forbids leading spaces, negatives. + */ +int xstrtoull(const char *str, char **end, int base, unsigned long long *value); + +/** + * Wrapper for strtof() that forbids leading spaces. + */ +int xstrtof(const char *str, char **end, float *value); + +/** + * Wrapper for strtod() that forbids leading spaces. + */ +int xstrtod(const char *str, char **end, double *value); + +/** * Process a yes/no prompt. * * @return 1 for yes, 0 for no, and -1 for unknown. @@ -185,9 +235,9 @@ size_t asciilen(const char *str); /** * Get the length of the pure-ASCII prefix of a string. * - * @param str + * @str * The string to check. - * @param n + * @n * The maximum prefix length. */ size_t asciinlen(const char *str, size_t n); @@ -195,9 +245,9 @@ size_t asciinlen(const char *str, size_t n); /** * Allocate a copy of a region of memory. * - * @param src + * @src * The memory region to copy. - * @param size + * @size * The size of the memory region. * @return * A copy of the region, allocated with malloc(), or NULL on failure. @@ -207,12 +257,12 @@ void *xmemdup(const void *src, size_t size); /** * A nice string copying function. * - * @param dest + * @dest * The NUL terminator of the destination string, or `end` if it is * already truncated. - * @param end + * @end * The end of the destination buffer. - * @param src + * @src * The string to copy from. * @return * The new NUL terminator of the destination, or `end` on truncation. @@ -222,14 +272,14 @@ char *xstpecpy(char *dest, char *end, const char *src); /** * A nice string copying function. * - * @param dest + * @dest * The NUL terminator of the destination string, or `end` if it is * already truncated. - * @param end + * @end * The end of the destination buffer. - * @param src + * @src * The string to copy from. - * @param n + * @n * The maximum number of characters to copy. * @return * The new NUL terminator of the destination, or `end` on truncation. @@ -239,7 +289,7 @@ char *xstpencpy(char *dest, char *end, const char *src, size_t n); /** * Thread-safe strerror(). * - * @param errnum + * @errnum * An error number. * @return * A string describing that error, which remains valid until the next @@ -248,11 +298,16 @@ char *xstpencpy(char *dest, char *end, const char *src, size_t n); const char *xstrerror(int errnum); /** + * Shorthand for xstrerror(errno). + */ +const char *errstr(void); + +/** * Format a mode like ls -l (e.g. -rw-r--r--). * - * @param mode + * @mode * The mode to format. - * @param str + * @str * The string to hold the formatted mode. */ void xstrmode(mode_t mode, char str[11]); @@ -307,12 +362,35 @@ int xminor(dev_t dev); */ pid_t xwaitpid(pid_t pid, int *status, int flags); +#include <sys/ioctl.h> // May be necessary for struct winsize +#include <termios.h> + +/** + * Open the controlling terminal. + * + * @flags + * The open() flags. + * @return + * An open file descriptor, or -1 on failure. + */ +int open_cterm(int flags); + +/** + * tcgetwinsize()/ioctl(TIOCGWINSZ) wrapper. + */ +int xtcgetwinsize(int fd, struct winsize *ws); + +/** + * tcsetwinsize()/ioctl(TIOCSWINSZ) wrapper. + */ +int xtcsetwinsize(int fd, const struct winsize *ws); + // #include <unistd.h> /** * Like dup(), but set the FD_CLOEXEC flag. * - * @param fd + * @fd * The file descriptor to duplicate. * @return * A duplicated file descriptor, or -1 on failure. @@ -322,7 +400,7 @@ int dup_cloexec(int fd); /** * Like pipe(), but set the FD_CLOEXEC flag. * - * @param pipefd + * @pipefd * The array to hold the two file descriptors. * @return * 0 on success, -1 on failure. @@ -344,14 +422,14 @@ size_t xread(int fd, void *buf, size_t nbytes); * writes. * * @return - The number of bytes written. A value != nbytes indicates an error. + * The number of bytes written. A value != nbytes indicates an error. */ size_t xwrite(int fd, const void *buf, size_t nbytes); /** * close() variant that preserves errno. * - * @param fd + * @fd * The file descriptor to close. */ void close_quietly(int fd); @@ -359,7 +437,7 @@ void close_quietly(int fd); /** * close() wrapper that asserts the file descriptor is valid. * - * @param fd + * @fd * The file descriptor to close. * @return * 0 on success, or -1 on error. @@ -374,11 +452,11 @@ int xfaccessat(int fd, const char *path, int amode); /** * readlinkat() wrapper that dynamically allocates the result. * - * @param fd + * @fd * The base directory descriptor. - * @param path + * @path * The path to the link, relative to fd. - * @param size + * @size * An estimate for the size of the link name (pass 0 if unknown). * @return * The target of the link, allocated with malloc(), or NULL on failure. @@ -388,7 +466,7 @@ char *xreadlinkat(int fd, const char *path, size_t size); /** * Wrapper for confstr() that allocates with malloc(). * - * @param name + * @name * The ID of the confstr to look up. * @return * The value of the confstr, or NULL on failure. @@ -398,30 +476,53 @@ char *xconfstr(int name); /** * Portability wrapper for strtofflags(). * - * @param str + * @str * The string to parse. The pointee will be advanced to the first * invalid position on error. - * @param set + * @set * The flags that are set in the string. - * @param clear + * @clear * The flags that are cleared in the string. * @return * 0 on success, -1 on failure. */ int xstrtofflags(const char **str, unsigned long long *set, unsigned long long *clear); +/** + * Wrapper for sysconf() that works around an MSan bug. + */ +long xsysconf(int name); + +/** + * Check for a POSIX option[1] at runtime. + * + * [1]: https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/V1_chap02.html#tag_02_01_06 + * + * @name + * The symbolic name of the POSIX option (e.g. SPAWN). + * @return + * The value of the option, either -1 or a date like 202405. + */ +#define sysoption(name) \ + (_POSIX_##name == 0 ? xsysconf(_SC_##name) : _POSIX_##name) + +/** + * Get the number of CPU threads available to the current process. + */ +long nproc(void); + #include <wchar.h> /** * Error-recovering mbrtowc() wrapper. * - * @param str + * @str * The string to convert. - * @param i + * @i * The current index. - * @param len + * @len * The length of the string. - * @param mb + * @mb * The multi-byte decoding state. * @return * The wide character at index *i, or WEOF if decoding fails. In either @@ -432,7 +533,7 @@ wint_t xmbrtowc(const char *str, size_t *i, size_t len, mbstate_t *mb); /** * wcswidth() variant that works on narrow strings. * - * @param str + * @str * The string to measure. * @return * The likely width of that string in a terminal. @@ -484,13 +585,13 @@ enum wesc_flags { /** * Escape a string as a single shell word. * - * @param dest + * @dest * The destination string to fill. - * @param end + * @end * The end of the destination buffer. - * @param src + * @src * The string to escape. - * @param flags + * @flags * Controls which characters to escape. * @return * The new NUL terminator of the destination, or `end` on truncation. @@ -500,15 +601,15 @@ char *wordesc(char *dest, char *end, const char *str, enum wesc_flags flags); /** * Escape a string as a single shell word. * - * @param dest + * @dest * The destination string to fill. - * @param end + * @end * The end of the destination buffer. - * @param src + * @src * The string to escape. - * @param n + * @n * The maximum length of the string. - * @param flags + * @flags * Controls which characters to escape. * @return * The new NUL terminator of the destination, or `end` on truncation. @@ -18,9 +18,10 @@ * various helper functions to take fewer parameters. */ -#include "prelude.h" #include "bftw.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "diag.h" #include "dir.h" @@ -30,6 +31,7 @@ #include "mtab.h" #include "stat.h" #include "trie.h" + #include <errno.h> #include <fcntl.h> #include <stdlib.h> @@ -251,7 +253,7 @@ struct bftw_file { /** The length of the file's name. */ size_t namelen; /** The file's name. */ - char name[]; + char name[]; // _counted_by(namelen + 1) }; /** @@ -446,7 +448,7 @@ static void bftw_queue_rebalance(struct bftw_queue *queue, bool async) { } } -/** Detatch the next waiting file. */ +/** Detach the next waiting file. */ static void bftw_queue_detach(struct bftw_queue *queue, struct bftw_file *file, bool async) { bfs_assert(!file->ioqueued); @@ -913,7 +915,7 @@ static int bftw_state_init(struct bftw_state *state, const struct bftw_args *arg size_t qdepth = 4096; size_t nthreads = args->nthreads; -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING // io_uring uses one fd per ring, ioq uses one ring per thread if (nthreads >= nopenfd - 1) { nthreads = nopenfd - 2; @@ -1006,6 +1008,7 @@ static int bftw_ioq_pop(struct bftw_state *state, bool block) { return -1; } + ioq_submit(ioq); struct ioq_ent *ent = ioq_pop(ioq, block); if (!ent) { return -1; @@ -1049,6 +1052,10 @@ static int bftw_ioq_pop(struct bftw_state *state, bool block) { bftw_queue_attach(&state->fileq, file, true); break; + + default: + bfs_bug("Unexpected ioq op %d", (int)op); + break; } ioq_free(ioq, ent); @@ -1160,12 +1167,13 @@ static int bftw_file_open(struct bftw_state *state, struct bftw_file *file, cons struct bftw_list parents; SLIST_INIT(&parents); - struct bftw_file *cur; - for (cur = file; cur != base; cur = cur->parent) { + // Reverse the chain of parents + for (struct bftw_file *cur = file; cur != base; cur = cur->parent) { SLIST_PREPEND(&parents, cur); } - while ((cur = SLIST_POP(&parents))) { + // Open each component relative to its parent + drain_slist (struct bftw_file, cur, &parents) { if (!cur->parent || cur->parent->fd >= 0) { bftw_file_openat(state, cur, cur->parent, cur->name); } @@ -1281,8 +1289,8 @@ static int bftw_pin_parent(struct bftw_state *state, struct bftw_file *file) { int fd = parent->fd; if (fd < 0) { - bfs_static_assert(AT_FDCWD != -1); - return -1; + // Don't confuse failures with AT_FDCWD + return (int)AT_FDCWD == -1 ? -2 : -1; } bftw_cache_pin(&state->cache, parent); @@ -1298,7 +1306,7 @@ static int bftw_ioq_opendir(struct bftw_state *state, struct bftw_file *file) { } int dfd = bftw_pin_parent(state, file); - if (dfd < 0 && dfd != AT_FDCWD) { + if (dfd < 0 && dfd != (int)AT_FDCWD) { goto fail; } @@ -1431,7 +1439,7 @@ static bool bftw_must_stat(const struct bftw_state *state, size_t depth, enum bf if (!(bftw_stat_flags(state, depth) & BFS_STAT_NOFOLLOW)) { return true; } - fallthru; + _fallthrough; default: #if __linux__ @@ -1450,7 +1458,7 @@ static int bftw_ioq_stat(struct bftw_state *state, struct bftw_file *file) { } int dfd = bftw_pin_parent(state, file); - if (dfd < 0 && dfd != AT_FDCWD) { + if (dfd < 0 && dfd != (int)AT_FDCWD) { goto fail; } @@ -1477,7 +1485,8 @@ fail: /** Check if we should stat() a file asynchronously. */ static bool bftw_should_ioq_stat(struct bftw_state *state, struct bftw_file *file) { - // To avoid surprising users too much, process the roots in order + // POSIX wants the root paths to be processed in order + // See https://www.austingroupbugs.net/view.php?id=1859 if (file->depth == 0) { return false; } @@ -1529,11 +1538,28 @@ static bool bftw_pop_file(struct bftw_state *state) { return bftw_pop(state, &state->fileq); } +/** Add a path component to the path. */ +static void bftw_prepend_path(char *path, size_t nameoff, size_t namelen, const char *name) { + if (nameoff > 0) { + path[nameoff - 1] = '/'; + } + memcpy(path + nameoff, name, namelen); +} + /** Build the path to the current file. */ static int bftw_build_path(struct bftw_state *state, const char *name) { const struct bftw_file *file = state->file; - size_t pathlen = file ? file->nameoff + file->namelen : 0; + size_t nameoff, namelen; + if (name) { + nameoff = file ? bftw_child_nameoff(file) : 0; + namelen = strlen(name); + } else { + nameoff = file->nameoff; + namelen = file->namelen; + } + + size_t pathlen = nameoff + namelen; if (dstresize(&state->path, pathlen) != 0) { state->error = errno; return -1; @@ -1546,11 +1572,11 @@ static int bftw_build_path(struct bftw_state *state, const char *name) { } // Build the path backwards + if (name) { + bftw_prepend_path(state->path, nameoff, namelen, name); + } while (file && file != ancestor) { - if (file->nameoff > 0) { - state->path[file->nameoff - 1] = '/'; - } - memcpy(state->path + file->nameoff, file->name, file->namelen); + bftw_prepend_path(state->path, file->nameoff, file->namelen, file->name); if (ancestor && ancestor->depth == file->depth) { ancestor = ancestor->parent; @@ -1559,20 +1585,6 @@ static int bftw_build_path(struct bftw_state *state, const char *name) { } state->previous = state->file; - - if (name) { - if (pathlen > 0 && state->path[pathlen - 1] != '/') { - if (dstrapp(&state->path, '/') != 0) { - state->error = errno; - return -1; - } - } - if (dstrcat(&state->path, name) != 0) { - state->error = errno; - return -1; - } - } - return 0; } @@ -1676,6 +1688,7 @@ static void bftw_init_ftwbuf(struct bftw_state *state, enum bftw_visit visit) { ftwbuf->visit = visit; ftwbuf->type = BFS_UNKNOWN; ftwbuf->error = state->direrror; + ftwbuf->loopoff = 0; ftwbuf->at_fd = AT_FDCWD; ftwbuf->at_path = ftwbuf->path; bftw_stat_init(&ftwbuf->stat_bufs, &state->stat_buf, &state->lstat_buf); @@ -1733,6 +1746,7 @@ static void bftw_init_ftwbuf(struct bftw_state *state, enum bftw_visit visit) { if (ancestor->dev == statbuf->dev && ancestor->ino == statbuf->ino) { ftwbuf->type = BFS_ERROR; ftwbuf->error = ELOOP; + ftwbuf->loopoff = ancestor->nameoff + ancestor->namelen; return; } } @@ -1863,8 +1877,8 @@ static int bftw_gc(struct bftw_state *state, enum bftw_gc_flags flags) { } state->direrror = 0; - while ((file = SLIST_POP(&state->to_close, ready))) { - bftw_unwrapdir(state, file); + drain_slist (struct bftw_file, dead, &state->to_close, ready) { + bftw_unwrapdir(state, dead); } enum bftw_gc_flags visit = BFTW_VISIT_FILE; @@ -1945,6 +1959,10 @@ static void bftw_flush(struct bftw_state *state) { bftw_queue_flush(&state->dirq); bftw_ioq_opendirs(state); + + if (state->ioq) { + ioq_submit(state->ioq); + } } /** Close the current directory. */ @@ -10,6 +10,7 @@ #include "dir.h" #include "stat.h" + #include <stddef.h> /** @@ -56,6 +57,8 @@ struct BFTW { enum bfs_type type; /** The errno that occurred, if type == BFS_ERROR. */ int error; + /** For filesystem loops, the length of the loop prefix. */ + size_t loopoff; /** A parent file descriptor for the *at() family of calls. */ int at_fd; @@ -72,9 +75,9 @@ struct BFTW { * Get bfs_stat() info for a file encountered during bftw(), caching the result * whenever possible. * - * @param ftwbuf + * @ftwbuf * bftw() data for the file to stat. - * @param flags + * @flags * flags for bfs_stat(). Pass ftwbuf->stat_flags for the default flags. * @return * A pointer to a bfs_stat() buffer, or NULL if the call failed. @@ -85,9 +88,9 @@ const struct bfs_stat *bftw_stat(const struct BFTW *ftwbuf, enum bfs_stat_flags * Get bfs_stat() info for a file encountered during bftw(), if it has already * been cached. * - * @param ftwbuf + * @ftwbuf * bftw() data for the file to stat. - * @param flags + * @flags * flags for bfs_stat(). Pass ftwbuf->stat_flags for the default flags. * @return * A pointer to a bfs_stat() buffer, or NULL if no stat info is cached. @@ -99,9 +102,9 @@ const struct bfs_stat *bftw_cached_stat(const struct BFTW *ftwbuf, enum bfs_stat * whether to follow links. This function will avoid calling bfs_stat() if * possible. * - * @param ftwbuf + * @ftwbuf * bftw() data for the file to check. - * @param flags + * @flags * flags for bfs_stat(). Pass ftwbuf->stat_flags for the default flags. * @return * The type of the file, or BFS_ERROR if an error occurred. @@ -123,9 +126,9 @@ enum bftw_action { /** * Callback function type for bftw(). * - * @param ftwbuf + * @ftwbuf * Data about the current file. - * @param ptr + * @ptr * The pointer passed to bftw(). * @return * An action value. @@ -208,7 +211,7 @@ struct bftw_args { * Like ftw(3) and nftw(3), this function walks a directory tree recursively, * and invokes a callback for each path it encounters. * - * @param args + * @args * The arguments that control the walk. * @return * 0 on success, or -1 on failure. @@ -8,11 +8,12 @@ #ifndef BFS_BIT_H #define BFS_BIT_H -#include "prelude.h" +#include "bfs.h" + #include <limits.h> #include <stdint.h> -#if __STDC_VERSION__ >= C23 +#if __has_include(<stdbit.h>) # include <stdbit.h> #endif @@ -147,7 +148,7 @@ # define INTMAX_WIDTH UINTMAX_WIDTH #endif -// C23 polyfill: byte order +// N3022 polyfill: byte order #ifdef __STDC_ENDIAN_LITTLE__ # define ENDIAN_LITTLE __STDC_ENDIAN_LITTLE__ @@ -173,11 +174,7 @@ # define ENDIAN_NATIVE 0 #endif -#if __STDC_VERSION__ >= C23 -# define bswap_u16 stdc_memreverse8u16 -# define bswap_u32 stdc_memreverse8u32 -# define bswap_u64 stdc_memreverse8u64 -#elif __GNUC__ +#if __GNUC__ # define bswap_u16 __builtin_bswap16 # define bswap_u32 __builtin_bswap32 # define bswap_u64 __builtin_bswap64 @@ -201,15 +198,35 @@ static inline uint8_t bswap_u8(uint8_t n) { return n; } -/** - * Reverse the byte order of an integer. - */ -#define bswap(n) \ - _Generic((n), \ - uint8_t: bswap_u8, \ - uint16_t: bswap_u16, \ - uint32_t: bswap_u32, \ - uint64_t: bswap_u64)(n) +#if UCHAR_WIDTH == 8 +# define bswap_uc bswap_u8 +#endif + +#if USHRT_WIDTH == 16 +# define bswap_us bswap_u16 +#elif USHRT_WIDTH == 32 +# define bswap_us bswap_u32 +#elif USHRT_WIDTH == 64 +# define bswap_us bswap_u64 +#endif + +#if UINT_WIDTH == 16 +# define bswap_ui bswap_u16 +#elif UINT_WIDTH == 32 +# define bswap_ui bswap_u32 +#elif UINT_WIDTH == 64 +# define bswap_ui bswap_u64 +#endif + +#if ULONG_WIDTH == 32 +# define bswap_ul bswap_u32 +#elif ULONG_WIDTH == 64 +# define bswap_ul bswap_u64 +#endif + +#if ULLONG_WIDTH == 64 +# define bswap_ull bswap_u64 +#endif // Define an overload for each unsigned type #define UINT_OVERLOADS(macro) \ @@ -222,25 +239,74 @@ static inline uint8_t bswap_u8(uint8_t n) { // Select an overload based on an unsigned integer type #define UINT_SELECT(n, name) \ _Generic((n), \ - char: name##_uc, \ - signed char: name##_uc, \ unsigned char: name##_uc, \ - signed short: name##_us, \ unsigned short: name##_us, \ - signed int: name##_ui, \ unsigned int: name##_ui, \ - signed long: name##_ul, \ unsigned long: name##_ul, \ - signed long long: name##_ull, \ unsigned long long: name##_ull) +/** + * Reverse the byte order of an integer. + */ +#define bswap(n) UINT_SELECT(n, bswap)(n) + +#define LOAD8_LEU8(ptr, i, n) ((uint##n##_t)((const unsigned char *)ptr)[(i) / 8] << (i)) +#define LOAD8_BEU8(ptr, i, n) ((uint##n##_t)((const unsigned char *)ptr)[(i) / 8] << (n - (i) - 8)) + +/** Load a little-endian 8-bit word. */ +static inline uint8_t load8_leu8(const void *ptr) { + return LOAD8_LEU8(ptr, 0, 8); +} + +/** Load a big-endian 8-bit word. */ +static inline uint8_t load8_beu8(const void *ptr) { + return LOAD8_BEU8(ptr, 0, 8); +} + +#define LOAD8_LEU16(ptr, i, n) (LOAD8_LEU8(ptr, i, n) | LOAD8_LEU8(ptr, i + 8, n)) +#define LOAD8_BEU16(ptr, i, n) (LOAD8_BEU8(ptr, i, n) | LOAD8_BEU8(ptr, i + 8, n)) + +/** Load a little-endian 16-bit word. */ +static inline uint16_t load8_leu16(const void *ptr) { + return LOAD8_LEU16(ptr, 0, 16); +} + +/** Load a big-endian 16-bit word. */ +static inline uint16_t load8_beu16(const void *ptr) { + return LOAD8_BEU16(ptr, 0, 16); +} + +#define LOAD8_LEU32(ptr, i, n) (LOAD8_LEU16(ptr, i, n) | LOAD8_LEU16(ptr, i + 16, n)) +#define LOAD8_BEU32(ptr, i, n) (LOAD8_BEU16(ptr, i, n) | LOAD8_BEU16(ptr, i + 16, n)) + +/** Load a little-endian 32-bit word. */ +static inline uint32_t load8_leu32(const void *ptr) { + return LOAD8_LEU32(ptr, 0, 32); +} + +/** Load a big-endian 32-bit word. */ +static inline uint32_t load8_beu32(const void *ptr) { + return LOAD8_BEU32(ptr, 0, 32); +} + +#define LOAD8_LEU64(ptr, i, n) (LOAD8_LEU32(ptr, i, n) | LOAD8_LEU32(ptr, i + 32, n)) +#define LOAD8_BEU64(ptr, i, n) (LOAD8_BEU32(ptr, i, n) | LOAD8_BEU32(ptr, i + 32, n)) + +/** Load a little-endian 64-bit word. */ +static inline uint64_t load8_leu64(const void *ptr) { + return LOAD8_LEU64(ptr, 0, 64); +} + +/** Load a big-endian 64-bit word. */ +static inline uint64_t load8_beu64(const void *ptr) { + return LOAD8_BEU64(ptr, 0, 64); +} + // C23 polyfill: bit utilities -#if __STDC_VERSION__ >= C23 +#if __STDC_VERSION_STDBIT_H__ >= C23 # define count_ones stdc_count_ones # define count_zeros stdc_count_zeros -# define rotate_left stdc_rotate_left -# define rotate_right stdc_rotate_right # define leading_zeros stdc_leading_zeros # define leading_ones stdc_leading_ones # define trailing_zeros stdc_trailing_zeros @@ -273,31 +339,31 @@ static inline uint8_t bswap_u8(uint8_t n) { #define BUILTIN_WIDTH(suffix) BUILTIN_WIDTH##suffix #define COUNT_ONES(type, suffix, width) \ - static inline int count_ones##suffix(type n) { \ + static inline unsigned int count_ones##suffix(type n) { \ return UINT_BUILTIN(popcount, suffix)(n); \ } #define LEADING_ZEROS(type, suffix, width) \ - static inline int leading_zeros##suffix(type n) { \ + static inline unsigned int leading_zeros##suffix(type n) { \ return n \ ? UINT_BUILTIN(clz, suffix)(n) - (BUILTIN_WIDTH(suffix) - width) \ : width; \ } #define TRAILING_ZEROS(type, suffix, width) \ - static inline int trailing_zeros##suffix(type n) { \ + static inline unsigned int trailing_zeros##suffix(type n) { \ return n ? UINT_BUILTIN(ctz, suffix)(n) : (int)width; \ } #define FIRST_TRAILING_ONE(type, suffix, width) \ - static inline int first_trailing_one##suffix(type n) { \ + static inline unsigned int first_trailing_one##suffix(type n) { \ return UINT_BUILTIN(ffs, suffix)(n); \ } #else // !__GNUC__ #define COUNT_ONES(type, suffix, width) \ - static inline int count_ones##suffix(type n) { \ + static inline unsigned int count_ones##suffix(type n) { \ int ret; \ for (ret = 0; n; ++ret) { \ n &= n - 1; \ @@ -306,7 +372,7 @@ static inline uint8_t bswap_u8(uint8_t n) { } #define LEADING_ZEROS(type, suffix, width) \ - static inline int leading_zeros##suffix(type n) { \ + static inline unsigned int leading_zeros##suffix(type n) { \ type bit = (type)1 << (width - 1); \ int ret; \ for (ret = 0; bit && !(n & bit); ++ret, bit >>= 1); \ @@ -314,7 +380,7 @@ static inline uint8_t bswap_u8(uint8_t n) { } #define TRAILING_ZEROS(type, suffix, width) \ - static inline int trailing_zeros##suffix(type n) { \ + static inline unsigned int trailing_zeros##suffix(type n) { \ type bit = 1; \ int ret; \ for (ret = 0; bit && !(n & bit); ++ret, bit <<= 1); \ @@ -322,7 +388,7 @@ static inline uint8_t bswap_u8(uint8_t n) { } #define FIRST_TRAILING_ONE(type, suffix, width) \ - static inline int first_trailing_one##suffix(type n) { \ + static inline unsigned int first_trailing_one##suffix(type n) { \ return n ? trailing_zeros##suffix(n) + 1 : 0; \ } @@ -333,19 +399,9 @@ UINT_OVERLOADS(LEADING_ZEROS) UINT_OVERLOADS(TRAILING_ZEROS) UINT_OVERLOADS(FIRST_TRAILING_ONE) -#define ROTATE_LEFT(type, suffix, width) \ - static inline type rotate_left##suffix(type n, int c) { \ - return (n << c) | (n >> ((width - c) % width)); \ - } - -#define ROTATE_RIGHT(type, suffix, width) \ - static inline type rotate_right##suffix(type n, int c) { \ - return (n >> c) | (n << ((width - c) % width)); \ - } - #define FIRST_LEADING_ONE(type, suffix, width) \ - static inline int first_leading_one##suffix(type n) { \ - return width - leading_zeros##suffix(n); \ + static inline unsigned int first_leading_one##suffix(type n) { \ + return n ? leading_zeros##suffix(n) + 1 : 0; \ } #define HAS_SINGLE_BIT(type, suffix, width) \ @@ -354,17 +410,30 @@ UINT_OVERLOADS(FIRST_TRAILING_ONE) return n - 1 < (n ^ (n - 1)); \ } -UINT_OVERLOADS(ROTATE_LEFT) -UINT_OVERLOADS(ROTATE_RIGHT) +#define BIT_WIDTH(type, suffix, width) \ + static inline unsigned int bit_width##suffix(type n) { \ + return width - leading_zeros##suffix(n); \ + } + +#define BIT_FLOOR(type, suffix, width) \ + static inline type bit_floor##suffix(type n) { \ + return n ? (type)1 << (bit_width##suffix(n) - 1) : 0; \ + } + +#define BIT_CEIL(type, suffix, width) \ + static inline type bit_ceil##suffix(type n) { \ + return (type)1 << bit_width##suffix(n - !!n); \ + } + UINT_OVERLOADS(FIRST_LEADING_ONE) UINT_OVERLOADS(HAS_SINGLE_BIT) +UINT_OVERLOADS(BIT_WIDTH) +UINT_OVERLOADS(BIT_FLOOR) +UINT_OVERLOADS(BIT_CEIL) #define count_ones(n) UINT_SELECT(n, count_ones)(n) #define count_zeros(n) UINT_SELECT(n, count_ones)(~(n)) -#define rotate_left(n, c) UINT_SELECT(n, rotate_left)(n, c) -#define rotate_right(n, c) UINT_SELECT(n, rotate_right)(n, c) - #define leading_zeros(n) UINT_SELECT(n, leading_zeros)(n) #define leading_ones(n) UINT_SELECT(n, leading_zeros)(~(n)) @@ -379,23 +448,26 @@ UINT_OVERLOADS(HAS_SINGLE_BIT) #define has_single_bit(n) UINT_SELECT(n, has_single_bit)(n) -#define BIT_FLOOR(type, suffix, width) \ - static inline type bit_floor##suffix(type n) { \ - return n ? (type)1 << (first_leading_one##suffix(n) - 1) : 0; \ - } +#define bit_width(n) UINT_SELECT(n, bit_width)(n) +#define bit_floor(n) UINT_SELECT(n, bit_floor)(n) +#define bit_ceil(n) UINT_SELECT(n, bit_ceil)(n) -#define BIT_CEIL(type, suffix, width) \ - static inline type bit_ceil##suffix(type n) { \ - return (type)1 << first_leading_one##suffix(n - !!n); \ +#endif // __STDC_VERSION_STDBIT_H__ < C23 + +#define ROTATE_LEFT(type, suffix, width) \ + static inline type rotate_left##suffix(type n, int c) { \ + return (n << c) | (n >> ((width - c) % width)); \ } -UINT_OVERLOADS(BIT_FLOOR) -UINT_OVERLOADS(BIT_CEIL) +#define ROTATE_RIGHT(type, suffix, width) \ + static inline type rotate_right##suffix(type n, int c) { \ + return (n >> c) | (n << ((width - c) % width)); \ + } -#define bit_width(n) first_leading_one(n) -#define bit_floor(n) UINT_SELECT(n, bit_floor)(n) -#define bit_ceil(n) UINT_SELECT(n, bit_ceil)(n) +UINT_OVERLOADS(ROTATE_LEFT) +UINT_OVERLOADS(ROTATE_RIGHT) -#endif // __STDC_VERSION__ < C23 +#define rotate_left(n, c) UINT_SELECT(n, rotate_left)(n, c) +#define rotate_right(n, c) UINT_SELECT(n, rotate_right)(n, c) #endif // BFS_BIT_H diff --git a/src/color.c b/src/color.c index f004bf2..a026831 100644 --- a/src/color.c +++ b/src/color.c @@ -1,9 +1,10 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "color.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "bftw.h" #include "diag.h" @@ -13,6 +14,7 @@ #include "fsade.h" #include "stat.h" #include "trie.h" + #include <errno.h> #include <fcntl.h> #include <stdarg.h> @@ -29,8 +31,8 @@ struct esc_seq { /** The length of the escape sequence. */ size_t len; - /** The escape sequence iteself, without a terminating NUL. */ - char seq[]; + /** The escape sequence itself, without a terminating NUL. */ + char seq[] _counted_by(len); }; /** @@ -46,7 +48,7 @@ struct ext_color { /** Whether the comparison should be case-sensitive. */ bool case_sensitive; /** The extension to match (NUL-terminated). */ - char ext[]; + char ext[]; // _counted_by(len + 1); }; struct colors { @@ -141,13 +143,7 @@ static int init_esc(struct colors *colors, const char *name, const char *value, *field = esc; - struct trie_leaf *leaf = trie_insert_str(&colors->names, name); - if (!leaf) { - return -1; - } - - leaf->value = field; - return 0; + return trie_set_str(&colors->names, name, field); } /** Check if an escape sequence is equal to a string. */ @@ -157,12 +153,16 @@ static bool esc_eq(const struct esc_seq *esc, const char *str, size_t len) { /** Get an escape sequence from the table. */ static struct esc_seq **get_esc(const struct colors *colors, const char *name) { - const struct trie_leaf *leaf = trie_find_str(&colors->names, name); - return leaf ? leaf->value : NULL; + return trie_get_str(&colors->names, name); +} + +/** Append an escape sequence to a string. */ +static int cat_esc(dchar **dstr, const struct esc_seq *seq) { + return dstrxcat(dstr, seq->seq, seq->len); } /** Set a named escape sequence. */ -static int set_esc(struct colors *colors, const char *name, char *value) { +static int set_esc(struct colors *colors, const char *name, dchar *value) { struct esc_seq **field = get_esc(colors, name); if (!field) { return 0; @@ -209,58 +209,31 @@ static void ext_tolower(char *ext, size_t len) { } } -/** - * The "smart case" algorithm. - * - * @param ext - * The current extension being added. - * @param prev - * The previous case-sensitive match, if any, for the same extension. - * @param iprev - * The previous case-insensitive match, if any, for the same extension. - * @return - * Whether this extension should become case-sensitive. - */ -static bool ext_case_sensitive(struct ext_color *ext, struct ext_color *prev, struct ext_color *iprev) { - // This is the first case-insensitive occurrence of this extension, e.g. - // - // *.gz=01;31:*.tar.gz=01;33 - if (!iprev) { - bfs_assert(!prev); - return false; - } - - // If the last version of this extension is already case-sensitive, - // this one should be too, e.g. - // - // *.tar.gz=01;31:*.TAR.GZ=01;32:*.TAR.GZ=01;33 - if (iprev->case_sensitive) { - return true; - } - - // The case matches the last occurrence exactly, e.g. - // - // *.tar.gz=01;31:*.tar.gz=01;33 - if (iprev == prev) { - return false; - } - - // Different case, but same value, e.g. - // - // *.tar.gz=01;31:*.TAR.GZ=01;31 - if (esc_eq(iprev->esc, ext->esc->seq, ext->esc->len)) { - return false; +/** Insert an extension into a trie. */ +static int insert_ext(struct trie *trie, struct ext_color *ext) { + // A later *.x should override any earlier *.x, *.y.x, etc. + struct trie_leaf *leaf; + while ((leaf = trie_find_postfix(trie, ext->ext))) { + trie_remove(trie, leaf); } - // Different case, different value, e.g. - // - // *.tar.gz=01;31:*.TAR.GZ=01;33 - return true; + size_t len = ext->len + 1; + return trie_set_mem(trie, ext->ext, len, ext); } /** Set the color for an extension. */ -static int set_ext(struct colors *colors, char *key, char *value) { +static int set_ext(struct colors *colors, dchar *key, dchar *value) { size_t len = dstrlen(key); + + // Embedded NUL bytes in extensions can lead to a non-prefix-free + // set of strings, e.g. {".gz", "\0.gz"} would be transformed to + // {"zg.\0", "zg.\0\0"} (showing the implicit terminating NUL). + // Our trie implementation only supports prefix-free key sets, but + // luckily '\0' cannot appear in filenames so we can ignore them. + if (memchr(key, '\0', len)) { + return 0; + } + struct ext_color *ext = varena_alloc(&colors->ext_arena, len + 1); if (!ext) { return -1; @@ -274,45 +247,19 @@ static int set_ext(struct colors *colors, char *key, char *value) { goto fail; } - key = memcpy(ext->ext, key, len + 1); + memcpy(ext->ext, key, len + 1); // Reverse the extension (`*.y.x` -> `x.y.*`) so we can use trie_find_prefix() - ext_reverse(key, len); - - // Find any pre-existing exact match - struct ext_color *prev = NULL; - struct trie_leaf *leaf = trie_find_str(&colors->ext_trie, key); - if (leaf) { - prev = leaf->value; - trie_remove(&colors->ext_trie, leaf); - } - - // A later *.x should override any earlier *.x, *.y.x, etc. - while ((leaf = trie_find_postfix(&colors->ext_trie, key))) { - trie_remove(&colors->ext_trie, leaf); - } + ext_reverse(ext->ext, len); // Insert the extension into the case-sensitive trie - leaf = trie_insert_str(&colors->ext_trie, key); - if (!leaf) { + if (insert_ext(&colors->ext_trie, ext) != 0) { goto fail; } - leaf->value = ext; - // "Smart case": if the same extension is given with two different - // capitalizations (e.g. `*.y.x=31:*.Y.Z=32:`), make it case-sensitive - ext_tolower(key, len); - leaf = trie_insert_str(&colors->iext_trie, key); - if (!leaf) { - goto fail; - } - - struct ext_color *iprev = leaf->value; - if (ext_case_sensitive(ext, prev, iprev)) { - iprev->case_sensitive = true; - ext->case_sensitive = true; + if (colors->ext_len < len) { + colors->ext_len = len; } - leaf->value = ext; return 0; @@ -324,32 +271,83 @@ fail: return -1; } -/** Rebuild the case-insensitive trie after all extensions have been parsed. */ -static int build_iext_trie(struct colors *colors) { - trie_clear(&colors->iext_trie); +/** + * The "smart case" algorithm. + * + * @ext + * The current extension being added. + * @iext + * The previous case-insensitive match, if any, for the same extension. + * @return + * Whether this extension should become case-sensitive. + */ +static bool ext_case_sensitive(struct ext_color *ext, struct ext_color *iext) { + // This is the first case-insensitive occurrence of this extension, e.g. + // + // *.gz=01;31:*.tar.gz=01;33 + if (!iext) { + return false; + } + + // If the last version of this extension is already case-sensitive, + // this one should be too, e.g. + // + // *.tar.gz=01;31:*.TAR.GZ=01;32:*.TAR.GZ=01;33 + if (iext->case_sensitive) { + return true; + } + + // Different case, but same value, e.g. + // + // *.tar.gz=01;31:*.TAR.GZ=01;31 + if (esc_eq(iext->esc, ext->esc->seq, ext->esc->len)) { + return false; + } + // Different case, different value, e.g. + // + // *.tar.gz=01;31:*.TAR.GZ=01;33 + return true; +} + +/** Build the case-insensitive trie, after all extensions have been parsed. */ +static int build_iext_trie(struct colors *colors) { + // Find which extensions should be case-sensitive for_trie (leaf, &colors->ext_trie) { - size_t len = leaf->length - 1; - if (colors->ext_len < len) { - colors->ext_len = len; + struct ext_color *ext = leaf->value; + + // "Smart case": if the same extension is given with two different + // capitalizations (e.g. `*.y.x=31:*.Y.Z=32:`), make it case-sensitive + ext_tolower(ext->ext, ext->len); + + size_t len = ext->len + 1; + struct trie_leaf *ileaf = trie_insert_mem(&colors->iext_trie, ext->ext, len); + if (!ileaf) { + return -1; } + struct ext_color *iext = ileaf->value; + if (ext_case_sensitive(ext, iext)) { + ext->case_sensitive = true; + iext->case_sensitive = true; + } + + ileaf->value = ext; + } + + // Rebuild the trie with only the case-insensitive ones + trie_clear(&colors->iext_trie); + + for_trie (leaf, &colors->ext_trie) { struct ext_color *ext = leaf->value; if (ext->case_sensitive) { continue; } - // set_ext() already reversed and lowercased the extension - struct trie_leaf *ileaf; - while ((ileaf = trie_find_postfix(&colors->iext_trie, ext->ext))) { - trie_remove(&colors->iext_trie, ileaf); - } - - ileaf = trie_insert_str(&colors->iext_trie, ext->ext); - if (!ileaf) { + // We already lowercased the extension above + if (insert_ext(&colors->iext_trie, ext) != 0) { return -1; } - ileaf->value = ext; } return 0; @@ -358,9 +356,8 @@ static int build_iext_trie(struct colors *colors) { /** * Find a color by an extension. */ -static const struct esc_seq *get_ext(const struct colors *colors, const char *filename) { +static const struct esc_seq *get_ext(const struct colors *colors, const char *filename, size_t name_len) { size_t ext_len = colors->ext_len; - size_t name_len = strlen(filename); if (name_len < ext_len) { ext_len = name_len; } @@ -369,7 +366,8 @@ static const struct esc_seq *get_ext(const struct colors *colors, const char *fi char buf[256]; char *copy; if (ext_len < sizeof(buf)) { - copy = memcpy(buf, suffix, ext_len + 1); + copy = memcpy(buf, suffix, ext_len); + copy[ext_len] = '\0'; } else { copy = strndup(suffix, ext_len); if (!copy) { @@ -417,13 +415,13 @@ static const struct esc_seq *get_ext(const struct colors *colors, const char *fi * * See man dir_colors. * - * @param str + * @str * A dstring to fill with the unescaped chunk. - * @param value + * @value * The value to parse. - * @param end + * @end * The character that marks the end of the chunk. - * @param[out] next + * @next[out] * Will be set to the next chunk. * @return * 0 on success, -1 on failure. @@ -578,7 +576,7 @@ static int parse_gnu_ls_colors(struct colors *colors, const char *ls_colors) { break; } - if (dstrncpy(&key, chunk, equals - chunk) != 0) { + if (dstrxcpy(&key, chunk, equals - chunk) != 0) { goto fail; } if (unescape(&value, equals + 1, ':', &next) != 0) { @@ -587,8 +585,8 @@ static int parse_gnu_ls_colors(struct colors *colors, const char *ls_colors) { // All-zero values should be treated like NULL, to fall // back on any other relevant coloring for that file - char *esc = value; - if (strspn(value, "0") == strlen(value) + dchar *esc = value; + if (strspn(value, "0") == dstrlen(value) && strcmp(key, "rs") != 0 && strcmp(key, "lc") != 0 && strcmp(key, "rc") != 0 @@ -693,6 +691,20 @@ struct colors *parse_colors(void) { colors->link->len = 0; } + // Pre-compute the reset escape sequence + if (!colors->endcode) { + dchar *ec = dstralloc(0); + if (!ec + || cat_esc(&ec, colors->leftcode) != 0 + || cat_esc(&ec, colors->reset) != 0 + || cat_esc(&ec, colors->rightcode) != 0 + || set_esc(colors, "ec", ec) != 0) { + dstrfree(ec); + goto fail; + } + dstrfree(ec); + } + return colors; fail: @@ -727,10 +739,11 @@ CFILE *cfwrap(FILE *file, const struct colors *colors, bool close) { } cfile->file = file; + cfile->fd = fileno(file); cfile->need_reset = false; cfile->close = close; - if (isatty(fileno(file))) { + if (isatty(cfile->fd)) { cfile->colors = colors; } else { cfile->colors = NULL; @@ -755,23 +768,196 @@ int cfclose(CFILE *cfile) { return ret; } +bool colors_need_stat(const struct colors *colors) { + return colors->setuid || colors->setgid || colors->executable || colors->multi_hard + || colors->sticky_other_writable || colors->other_writable || colors->sticky; +} + +/** A colorable file path. */ +struct cpath { + /** The full path to color. */ + const char *path; + /** The basename offset of the last valid component. */ + size_t nameoff; + /** The end offset of the last valid component. */ + size_t valid; + /** The total length of the path. */ + size_t len; + + /** The bftw() buffer. */ + const struct BFTW *ftwbuf; + /** bfs_stat() flags for the final component. */ + enum bfs_stat_flags flags; + /** A bfs_stat() buffer, filled in when 0 < valid < len. */ + struct bfs_stat statbuf; +}; + +/** Move the valid range of a path backwards. */ +static void cpath_retreat(struct cpath *cpath) { + const char *path = cpath->path; + size_t nameoff = cpath->nameoff; + size_t valid = cpath->valid; + + if (valid > 0 && path[valid - 1] == '/') { + // Try without trailing slashes, to distinguish "notdir/" from "notdir" + do { + --valid; + } while (valid > 0 && path[valid - 1] == '/'); + + nameoff = valid; + while (nameoff > 0 && path[nameoff - 1] != '/') { + --nameoff; + } + } else { + // Remove the last component and try again + valid = nameoff; + } + + cpath->nameoff = nameoff; + cpath->valid = valid; +} + +/** Initialize a struct cpath. */ +static int cpath_init(struct cpath *cpath, const char *path, const struct BFTW *ftwbuf, enum bfs_stat_flags flags) { + // Normally there are only two components to color: + // + // nameoff valid + // v v + // path/to/filename + // --------+------- + // ${di} ${fi} + // + // Error cases also usually have two components: + // + // valid, + // nameoff + // v + // path/to/nowhere + // --------+------ + // ${di} ${mi} + // + // But with ENOTDIR, there may be three: + // + // nameoff valid + // v v + // path/to/filename/nowhere + // --------+-------+------- + // ${di} ${fi} ${mi} + + cpath->path = path; + cpath->len = strlen(path); + cpath->ftwbuf = ftwbuf; + cpath->flags = flags; + + cpath->valid = cpath->len; + if (path == ftwbuf->path) { + cpath->nameoff = ftwbuf->nameoff; + } else { + cpath->nameoff = xbaseoff(path); + } + + if (bftw_type(ftwbuf, flags) != BFS_ERROR) { + return 0; + } + + cpath_retreat(cpath); + + // Find the base path. For symlinks like + // + // path/to/symlink -> nested/file + // + // this will be something like + // + // path/to/nested/file + int at_fd = AT_FDCWD; + dchar *at_path = NULL; + if (path == ftwbuf->path) { + if (ftwbuf->depth > 0) { + // The parent must have existed to get here + return 0; + } + } else { + // We're in print_link_target(), so resolve relative to the link's parent directory + at_fd = ftwbuf->at_fd; + if (at_fd == (int)AT_FDCWD && path[0] != '/') { + at_path = dstrxdup(ftwbuf->path, ftwbuf->nameoff); + if (!at_path) { + return -1; + } + } + } + + if (!at_path) { + at_path = dstralloc(cpath->valid); + if (!at_path) { + return -1; + } + } + if (dstrxcat(&at_path, path, cpath->valid) != 0) { + dstrfree(at_path); + return -1; + } + + size_t at_off = dstrlen(at_path) - cpath->valid; + + // Find the longest valid path prefix + while (cpath->valid > 0) { + if (bfs_stat(at_fd, at_path, BFS_STAT_FOLLOW, &cpath->statbuf) == 0) { + break; + } + + cpath_retreat(cpath); + dstrshrink(at_path, at_off + cpath->valid); + } + + dstrfree(at_path); + return 0; +} + +/** Get the bfs_stat() buffer for the last valid component. */ +static const struct bfs_stat *cpath_stat(const struct cpath *cpath) { + if (cpath->valid == cpath->len) { + return bftw_stat(cpath->ftwbuf, cpath->flags); + } else { + return &cpath->statbuf; + } +} + +/** Check if a path has non-trivial capabilities. */ +static bool cpath_has_capabilities(const struct cpath *cpath) { + if (cpath->valid == cpath->len) { + return bfs_check_capabilities(cpath->ftwbuf) > 0; + } else { + // TODO: implement capability checks for arbitrary paths + return false; + } +} + /** Check if a symlink is broken. */ -static bool is_link_broken(const struct BFTW *ftwbuf) { +static bool cpath_is_broken(const struct cpath *cpath) { + if (cpath->valid < cpath->len) { + // A valid parent can't be a broken link + return false; + } + + const struct BFTW *ftwbuf = cpath->ftwbuf; if (ftwbuf->stat_flags & BFS_STAT_NOFOLLOW) { return xfaccessat(ftwbuf->at_fd, ftwbuf->at_path, F_OK) != 0; } else { + // A link encountered with BFS_STAT_TRYFOLLOW must be broken return true; } } -bool colors_need_stat(const struct colors *colors) { - return colors->setuid || colors->setgid || colors->executable || colors->multi_hard - || colors->sticky_other_writable || colors->other_writable || colors->sticky; -} - /** Get the color for a file. */ -static const struct esc_seq *file_color(const struct colors *colors, const char *filename, const struct BFTW *ftwbuf, enum bfs_stat_flags flags) { - enum bfs_type type = bftw_type(ftwbuf, flags); +static const struct esc_seq *file_color(const struct colors *colors, const struct cpath *cpath) { + enum bfs_type type; + if (cpath->valid == cpath->len) { + type = bftw_type(cpath->ftwbuf, cpath->flags); + } else { + type = bfs_mode_to_type(cpath->statbuf.mode); + } + if (type == BFS_ERROR) { goto error; } @@ -782,7 +968,7 @@ static const struct esc_seq *file_color(const struct colors *colors, const char switch (type) { case BFS_REG: if (colors->setuid || colors->setgid || colors->executable || colors->multi_hard) { - statbuf = bftw_stat(ftwbuf, flags); + statbuf = cpath_stat(cpath); if (!statbuf) { goto error; } @@ -792,7 +978,7 @@ static const struct esc_seq *file_color(const struct colors *colors, const char color = colors->setuid; } else if (colors->setgid && (statbuf->mode & 02000)) { color = colors->setgid; - } else if (colors->capable && bfs_check_capabilities(ftwbuf) > 0) { + } else if (colors->capable && cpath_has_capabilities(cpath)) { color = colors->capable; } else if (colors->executable && (statbuf->mode & 00111)) { color = colors->executable; @@ -801,7 +987,9 @@ static const struct esc_seq *file_color(const struct colors *colors, const char } if (!color) { - color = get_ext(colors, filename); + const char *name = cpath->path + cpath->nameoff; + size_t namelen = cpath->valid - cpath->nameoff; + color = get_ext(colors, name, namelen); } if (!color) { @@ -812,7 +1000,7 @@ static const struct esc_seq *file_color(const struct colors *colors, const char case BFS_DIR: if (colors->sticky_other_writable || colors->other_writable || colors->sticky) { - statbuf = bftw_stat(ftwbuf, flags); + statbuf = cpath_stat(cpath); if (!statbuf) { goto error; } @@ -831,7 +1019,7 @@ static const struct esc_seq *file_color(const struct colors *colors, const char break; case BFS_LNK: - if (colors->orphan && is_link_broken(ftwbuf)) { + if (colors->orphan && cpath_is_broken(cpath)) { color = colors->orphan; } else { color = colors->link; @@ -874,7 +1062,7 @@ error: /** Print an escape sequence chunk. */ static int print_esc_chunk(CFILE *cfile, const struct esc_seq *esc) { - return dstrxcat(&cfile->buffer, esc->seq, esc->len); + return cat_esc(&cfile->buffer, esc); } /** Print an ANSI escape sequence. */ @@ -908,12 +1096,7 @@ static int print_reset(CFILE *cfile) { } cfile->need_reset = false; - const struct colors *colors = cfile->colors; - if (colors->endcode) { - return print_esc_chunk(cfile, colors->endcode); - } else { - return print_esc(cfile, colors->reset); - } + return print_esc_chunk(cfile, cfile->colors->endcode); } /** Print a shell-escaped string. */ @@ -923,6 +1106,10 @@ static int print_wordesc(CFILE *cfile, const char *str, size_t n, enum wesc_flag /** Print a string with an optional color. */ static int print_colored(CFILE *cfile, const struct esc_seq *esc, const char *str, size_t len) { + if (len == 0) { + return 0; + } + if (print_esc(cfile, esc) != 0) { return -1; } @@ -939,112 +1126,42 @@ static int print_colored(CFILE *cfile, const struct esc_seq *esc, const char *st return 0; } -/** Find the offset of the first broken path component. */ -static ssize_t first_broken_offset(const char *path, const struct BFTW *ftwbuf, enum bfs_stat_flags flags, size_t max) { - ssize_t ret = max; - bfs_assert(ret >= 0); - - if (bftw_type(ftwbuf, flags) != BFS_ERROR) { - goto out; - } - - dchar *at_path; - int at_fd; - if (path == ftwbuf->path) { - if (ftwbuf->depth == 0) { - at_fd = AT_FDCWD; - at_path = dstrndup(path, max); - } else { - // The parent must have existed to get here - goto out; - } - } else { - // We're in print_link_target(), so resolve relative to the link's parent directory - at_fd = ftwbuf->at_fd; - if (at_fd == AT_FDCWD && path[0] != '/') { - at_path = dstrndup(ftwbuf->path, ftwbuf->nameoff); - if (at_path && dstrncat(&at_path, path, max) != 0) { - ret = -1; - goto out_path; - } - } else { - at_path = dstrndup(path, max); - } - } - - if (!at_path) { - ret = -1; - goto out; - } - - while (ret > 0) { - if (xfaccessat(at_fd, at_path, F_OK) == 0) { - break; - } - - size_t len = dstrlen(at_path); - while (ret && at_path[len - 1] == '/') { - --len, --ret; - } - if (errno != ENOTDIR) { - while (ret && at_path[len - 1] != '/') { - --len, --ret; - } - } - - dstresize(&at_path, len); - } - -out_path: - dstrfree(at_path); -out: - return ret; -} - /** Print a path with colors. */ static int print_path_colored(CFILE *cfile, const char *path, const struct BFTW *ftwbuf, enum bfs_stat_flags flags) { - size_t nameoff; - if (path == ftwbuf->path) { - nameoff = ftwbuf->nameoff; - } else { - nameoff = xbaseoff(path); - } - - const char *name = path + nameoff; - size_t pathlen = nameoff + strlen(name); - - ssize_t broken = first_broken_offset(path, ftwbuf, flags, nameoff); - if (broken < 0) { + struct cpath cpath; + if (cpath_init(&cpath, path, ftwbuf, flags) != 0) { return -1; } - size_t split = broken; const struct colors *colors = cfile->colors; const struct esc_seq *dirs_color = colors->directory; - const struct esc_seq *name_color; + const struct esc_seq *name_color = NULL; + const struct esc_seq *err_color = colors->missing; + if (!err_color) { + err_color = colors->orphan; + } - if (split < nameoff) { - name_color = colors->missing; - if (!name_color) { - name_color = colors->orphan; - } - } else { - name_color = file_color(cfile->colors, path + nameoff, ftwbuf, flags); + if (cpath.nameoff < cpath.valid) { + name_color = file_color(colors, &cpath); if (name_color == dirs_color) { - split = pathlen; + cpath.nameoff = cpath.valid; } } - if (split > 0) { - if (print_colored(cfile, dirs_color, path, split) != 0) { - return -1; - } + if (print_colored(cfile, dirs_color, path, cpath.nameoff) != 0) { + return -1; } - if (split < pathlen) { - if (print_colored(cfile, name_color, path + split, pathlen - split) != 0) { - return -1; - } + const char *name = path + cpath.nameoff; + size_t name_len = cpath.valid - cpath.nameoff; + if (print_colored(cfile, name_color, name, name_len) != 0) { + return -1; + } + + const char *tail = path + cpath.valid; + size_t tail_len = cpath.len - cpath.valid; + if (print_colored(cfile, err_color, tail, tail_len) != 0) { + return -1; } return 0; @@ -1052,8 +1169,18 @@ static int print_path_colored(CFILE *cfile, const char *path, const struct BFTW /** Print a file name with colors. */ static int print_name_colored(CFILE *cfile, const char *name, const struct BFTW *ftwbuf, enum bfs_stat_flags flags) { - const struct esc_seq *esc = file_color(cfile->colors, name, ftwbuf, flags); - return print_colored(cfile, esc, name, strlen(name)); + size_t len = strlen(name); + const struct cpath cpath = { + .path = name, + .nameoff = 0, + .valid = len, + .len = len, + .ftwbuf = ftwbuf, + .flags = flags, + }; + + const struct esc_seq *esc = file_color(cfile->colors, &cpath); + return print_colored(cfile, esc, name, cpath.len); } /** Print the name of a file with the appropriate colors. */ @@ -1110,9 +1237,36 @@ static int print_link_target(CFILE *cfile, const struct BFTW *ftwbuf) { } /** Format some colored output to the buffer. */ -attr(printf(2, 3)) +_printf(2, 3) static int cbuff(CFILE *cfile, const char *format, ...); +/** Print an expression's name, for diagnostics. */ +static int print_expr_name(CFILE *cfile, const struct bfs_expr *expr) { + switch (expr->kind) { + case BFS_FLAG: + return cbuff(cfile, "${cyn}%pq${rs}", expr->argv[0]); + case BFS_OPERATOR: + return cbuff(cfile, "${red}%pq${rs}", expr->argv[0]); + default: + return cbuff(cfile, "${blu}%pq${rs}", expr->argv[0]); + } +} + +/** Print an expression's args, for diagnostics. */ +static int print_expr_args(CFILE *cfile, const struct bfs_expr *expr) { + if (print_expr_name(cfile, expr) != 0) { + return -1; + } + + for (size_t i = 1; i < expr->argc; ++i) { + if (cbuff(cfile, " ${bld}%pq${rs}", expr->argv[i]) < 0) { + return -1; + } + } + + return 0; +} + /** Dump a parsed expression tree, for debugging. */ static int print_expr(CFILE *cfile, const struct bfs_expr *expr, bool verbose, int depth) { if (depth >= 2) { @@ -1127,20 +1281,8 @@ static int print_expr(CFILE *cfile, const struct bfs_expr *expr, bool verbose, i return -1; } - if (bfs_expr_is_parent(expr)) { - if (cbuff(cfile, "${red}%pq${rs}", expr->argv[0]) < 0) { - return -1; - } - } else { - if (cbuff(cfile, "${blu}%pq${rs}", expr->argv[0]) < 0) { - return -1; - } - } - - for (size_t i = 1; i < expr->argc; ++i) { - if (cbuff(cfile, " ${bld}%pq${rs}", expr->argv[i]) < 0) { - return -1; - } + if (print_expr_args(cfile, expr) != 0) { + return -1; } if (verbose) { @@ -1156,7 +1298,7 @@ static int print_expr(CFILE *cfile, const struct bfs_expr *expr, bool verbose, i } int count = 0; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (dstrcat(&cfile->buffer, " ") != 0) { return -1; } @@ -1179,10 +1321,9 @@ static int print_expr(CFILE *cfile, const struct bfs_expr *expr, bool verbose, i return 0; } -attr(printf(2, 0)) +_printf(2, 0) static int cvbuff(CFILE *cfile, const char *format, va_list args) { const struct colors *colors = cfile->colors; - int error = errno; // Color specifier (e.g. ${blu}) state struct esc_seq **esc; @@ -1192,7 +1333,7 @@ static int cvbuff(CFILE *cfile, const char *format, va_list args) { for (const char *i = format; *i; ++i) { size_t verbatim = strcspn(i, "%$"); - if (dstrncat(&cfile->buffer, i, verbatim) != 0) { + if (dstrxcat(&cfile->buffer, i, verbatim) != 0) { return -1; } i += verbatim; @@ -1240,12 +1381,6 @@ static int cvbuff(CFILE *cfile, const char *format, va_list args) { } break; - case 'm': - if (dstrcat(&cfile->buffer, xstrerror(error)) != 0) { - return -1; - } - break; - case 'p': switch (*++i) { case 'q': @@ -1287,6 +1422,16 @@ static int cvbuff(CFILE *cfile, const char *format, va_list args) { return -1; } break; + case 'x': + if (print_expr_args(cfile, va_arg(args, const struct bfs_expr *)) != 0) { + return -1; + } + break; + case 'X': + if (print_expr_name(cfile, va_arg(args, const struct bfs_expr *)) != 0) { + return -1; + } + break; default: goto invalid; @@ -1379,7 +1524,7 @@ int cvfprintf(CFILE *cfile, const char *format, va_list args) { } } - dstresize(&cfile->buffer, 0); + dstrshrink(cfile->buffer, 0); return ret; } @@ -1390,3 +1535,14 @@ int cfprintf(CFILE *cfile, const char *format, ...) { va_end(args); return ret; } + +int cfreset(CFILE *cfile) { + const struct colors *colors = cfile->colors; + if (!colors) { + return 0; + } + + const struct esc_seq *esc = colors->endcode; + size_t ret = xwrite(cfile->fd, esc->seq, esc->len); + return ret == esc->len ? 0 : -1; +} diff --git a/src/color.h b/src/color.h index 3278cd6..aac8b33 100644 --- a/src/color.h +++ b/src/color.h @@ -8,8 +8,9 @@ #ifndef BFS_COLOR_H #define BFS_COLOR_H -#include "prelude.h" +#include "bfs.h" #include "dstring.h" + #include <stdio.h> /** @@ -42,6 +43,8 @@ typedef struct CFILE { const struct colors *colors; /** A buffer for colored formatting. */ dchar *buffer; + /** Cached file descriptor number. */ + int fd; /** Whether the next ${rs} is actually necessary. */ bool need_reset; /** Whether to close the underlying stream. */ @@ -51,11 +54,11 @@ typedef struct CFILE { /** * Wrap an existing file into a colored stream. * - * @param file + * @file * The underlying file. - * @param colors + * @colors * The color table to use if file is a TTY. - * @param close + * @close * Whether to close the underlying stream when this stream is closed. * @return * A colored wrapper around file. @@ -65,7 +68,7 @@ CFILE *cfwrap(FILE *file, const struct colors *colors, bool close); /** * Close a colored file. * - * @param cfile + * @cfile * The colored file to close. * @return * 0 on success, -1 on failure. @@ -75,9 +78,9 @@ int cfclose(CFILE *cfile); /** * Colored, formatted output. * - * @param cfile + * @cfile * The colored stream to print to. - * @param format + * @format * A printf()-style format string, supporting these format specifiers: * * %c: A single character @@ -85,7 +88,6 @@ int cfclose(CFILE *cfile); * %g: A double * %s: A string * %zu: A size_t - * %m: strerror(errno) * %pq: A shell-escaped string, like bash's printf %q * %pQ: A TTY-escaped string. * %pF: A colored file name, from a const struct BFTW * argument @@ -93,19 +95,26 @@ int cfclose(CFILE *cfile); * %pL: A colored link target, from a const struct BFTW * argument * %pe: Dump a const struct bfs_expr *, for debugging. * %pE: Dump a const struct bfs_expr * in verbose form, for debugging. + * %px: Print a const struct bfs_expr * with syntax highlighting. + * %pX: Print the name of a const struct bfs_expr *, without arguments. * %%: A literal '%' * ${cc}: Change the color to 'cc' * $$: A literal '$' * @return * 0 on success, -1 on failure. */ -attr(printf(2, 3)) +_printf(2, 3) int cfprintf(CFILE *cfile, const char *format, ...); /** * cfprintf() variant that takes a va_list. */ -attr(printf(2, 0)) +_printf(2, 0) int cvfprintf(CFILE *cfile, const char *format, va_list args); +/** + * Reset the TTY state when terminating abnormally (async-signal-safe). + */ +int cfreset(CFILE *cfile); + #endif // BFS_COLOR_H @@ -2,36 +2,28 @@ // SPDX-License-Identifier: 0BSD #include "ctx.h" + #include "alloc.h" +#include "bfstd.h" #include "color.h" #include "diag.h" #include "expr.h" #include "list.h" #include "mtab.h" #include "pwcache.h" +#include "sighook.h" #include "stat.h" #include "trie.h" -#include "xtime.h" + #include <errno.h> #include <limits.h> +#include <signal.h> #include <stdio.h> #include <stdlib.h> +#include <sys/stat.h> +#include <time.h> #include <unistd.h> -/** Get the initial value for ctx->threads (-j). */ -static int bfs_nproc(void) { - long nproc = sysconf(_SC_NPROCESSORS_ONLN); - - if (nproc < 1) { - nproc = 1; - } else if (nproc > 8) { - // Not much speedup after 8 threads - nproc = 8; - } - - return nproc; -} - struct bfs_ctx *bfs_ctx_new(void) { struct bfs_ctx *ctx = ZALLOC(struct bfs_ctx); if (!ctx) { @@ -44,15 +36,24 @@ struct bfs_ctx *bfs_ctx_new(void) { ctx->maxdepth = INT_MAX; ctx->flags = BFTW_RECOVER; ctx->strategy = BFTW_BFS; - ctx->threads = bfs_nproc(); ctx->optlevel = 3; + ctx->threads = nproc(); + if (ctx->threads > 8) { + // Not much speedup after 8 threads + ctx->threads = 8; + } + trie_init(&ctx->files); + ctx->umask = umask(0); + umask(ctx->umask); + if (getrlimit(RLIMIT_NOFILE, &ctx->orig_nofile) != 0) { goto fail; } ctx->cur_nofile = ctx->orig_nofile; + ctx->raise_nofile = true; ctx->users = bfs_users_new(); if (!ctx->users) { @@ -64,7 +65,7 @@ struct bfs_ctx *bfs_ctx_new(void) { goto fail; } - if (xgettime(&ctx->now) != 0) { + if (clock_gettime(CLOCK_REALTIME, &ctx->now) != 0) { goto fail; } @@ -98,13 +99,20 @@ struct bfs_ctx_file { CFILE *cfile; /** The path to the file (for diagnostics). */ const char *path; + /** Signal hook to send a reset escape sequence. */ + struct sighook *hook; /** Remembers I/O errors, to propagate them to the exit status. */ int error; }; +/** Call cfreset() on a tracked file. */ +static void cfreset_hook(int sig, siginfo_t *info, void *arg) { + cfreset(arg); +} + CFILE *bfs_ctx_dedup(struct bfs_ctx *ctx, CFILE *cfile, const char *path) { struct bfs_stat sb; - if (bfs_stat(fileno(cfile->file), NULL, 0, &sb) != 0) { + if (bfs_stat(cfile->fd, NULL, 0, &sb) != 0) { return NULL; } @@ -124,19 +132,31 @@ CFILE *bfs_ctx_dedup(struct bfs_ctx *ctx, CFILE *cfile, const char *path) { leaf->value = ctx_file = ALLOC(struct bfs_ctx_file); if (!ctx_file) { - trie_remove(&ctx->files, leaf); - return NULL; + goto fail; } ctx_file->cfile = cfile; ctx_file->path = path; ctx_file->error = 0; + ctx_file->hook = NULL; + + if (cfile->colors) { + ctx_file->hook = atsigexit(cfreset_hook, cfile); + if (!ctx_file->hook) { + goto fail; + } + } if (cfile != ctx->cout && cfile != ctx->cerr) { ++ctx->nfiles; } return cfile; + +fail: + trie_remove(&ctx->files, leaf); + free(ctx_file); + return NULL; } void bfs_ctx_flush(const struct bfs_ctx *ctx) { @@ -156,9 +176,9 @@ void bfs_ctx_flush(const struct bfs_ctx *ctx) { const char *path = ctx_file->path; if (path) { - bfs_error(ctx, "'%s': %m.\n", path); + bfs_error(ctx, "%pq: %s.\n", path, errstr()); } else if (cfile == ctx->cout) { - bfs_error(ctx, "(standard output): %m.\n"); + bfs_error(ctx, "(standard output): %s.\n", errstr()); } } @@ -188,30 +208,47 @@ static int bfs_ctx_fflush(CFILE *cfile) { static int bfs_ctx_fclose(struct bfs_ctx *ctx, struct bfs_ctx_file *ctx_file) { CFILE *cfile = ctx_file->cfile; - if (cfile == ctx->cout) { - // Will be checked later - return 0; - } else if (cfile == ctx->cerr) { - // Writes to stderr are allowed to fail silently, unless the same file was used by - // -fprint, -fls, etc. - if (ctx_file->path) { - return bfs_ctx_fflush(cfile); - } else { - return 0; - } - } - + // Writes to stderr are allowed to fail silently, unless the same file + // was used by -fprint, -fls, etc. + bool silent = cfile == ctx->cerr && !ctx_file->path; int ret = 0, error = 0; - if (ferror(cfile->file)) { + + if (ctx_file->error) { + // An error was previously reported during bfs_ctx_flush() ret = -1; - error = EIO; + error = ctx_file->error; } - if (cfclose(cfile) != 0) { + + // Flush the file just before we remove the hook, to maximize the chance + // we leave the TTY in a good state + if (bfs_ctx_fflush(cfile) != 0) { ret = -1; error = errno; } - errno = error; + sigunhook(ctx_file->hook); + + // Close the CFILE, except for stdio streams, which are closed later + if (cfile != ctx->cout && cfile != ctx->cerr) { + if (cfclose(cfile) != 0) { + ret = -1; + error = errno; + } + } + + if (silent) { + ret = 0; + } + + if (ret != 0 && ctx->cerr) { + if (ctx_file->path) { + bfs_error(ctx, "%pq: %s.\n", ctx_file->path, xstrerror(error)); + } else if (cfile == ctx->cout) { + bfs_error(ctx, "(standard output): %s.\n", xstrerror(error)); + } + } + + free(ctx_file); return ret; } @@ -229,33 +266,14 @@ int bfs_ctx_free(struct bfs_ctx *ctx) { for_trie (leaf, &ctx->files) { struct bfs_ctx_file *ctx_file = leaf->value; - - if (ctx_file->error) { - // An error was previously reported during bfs_ctx_flush() - ret = -1; - } - if (bfs_ctx_fclose(ctx, ctx_file) != 0) { - if (cerr) { - bfs_error(ctx, "%pq: %m.\n", ctx_file->path); - } ret = -1; } - - free(ctx_file); } trie_destroy(&ctx->files); - if (cout && bfs_ctx_fflush(cout) != 0) { - if (cerr) { - bfs_error(ctx, "(standard output): %m.\n"); - } - ret = -1; - } - cfclose(cout); cfclose(cerr); - free_colors(ctx->colors); for_slist (struct bfs_expr, expr, &ctx->expr_list, freelist) { @@ -268,6 +286,7 @@ int bfs_ctx_free(struct bfs_ctx *ctx) { } free(ctx->paths); + free(ctx->kinds); free(ctx->argv); free(ctx); } @@ -8,14 +8,15 @@ #ifndef BFS_CTX_H #define BFS_CTX_H -#include "prelude.h" #include "alloc.h" #include "bftw.h" #include "diag.h" #include "expr.h" #include "trie.h" + #include <stddef.h> #include <sys/resource.h> +#include <sys/types.h> #include <time.h> struct CFILE; @@ -28,6 +29,8 @@ struct bfs_ctx { size_t argc; /** The unparsed command line arguments. */ char **argv; + /** The argument token kinds. */ + enum bfs_kind *kinds; /** The root paths. */ const char **paths; @@ -67,11 +70,18 @@ struct bfs_ctx { bool status; /** Whether to only return unique files (-unique). */ bool unique; - /** Whether to print warnings (-warn/-nowarn). */ - bool warn; /** Whether to only handle paths with xargs-safe characters (-X). */ bool xargs_safe; + /** Whether bfs was run interactively. */ + bool interactive; + /** Whether to print warnings (-warn/-nowarn). */ + bool warn; + /** Whether to report errors (-noerror). */ + bool ignore_errors; + /** Whether any dangerous actions (-delete/-exec) are present. */ + bool dangerous; + /** Color data. */ struct colors *colors; /** The error that occurred parsing the color table, if any. */ @@ -98,10 +108,15 @@ struct bfs_ctx { /** The number of files owned by the context. */ int nfiles; + /** The current file creation mask. */ + mode_t umask; + /** The initial RLIMIT_NOFILE limits. */ struct rlimit orig_nofile; /** The current RLIMIT_NOFILE limits. */ struct rlimit cur_nofile; + /** Whether the fd limit should be raised. */ + bool raise_nofile; /** The current time. */ struct timespec now; @@ -116,7 +131,7 @@ struct bfs_ctx *bfs_ctx_new(void); /** * Get the mount table. * - * @param ctx + * @ctx * The bfs context. * @return * The cached mount table, or NULL on failure. @@ -126,11 +141,11 @@ const struct bfs_mtab *bfs_ctx_mtab(const struct bfs_ctx *ctx); /** * Deduplicate an opened file. * - * @param ctx + * @ctx * The bfs context. - * @param cfile + * @cfile * The opened file. - * @param path + * @path * The path to the opened file (or NULL for standard streams). * @return * If the same file was opened previously, that file is returned. If cfile is a new file, @@ -141,7 +156,7 @@ struct CFILE *bfs_ctx_dedup(struct bfs_ctx *ctx, struct CFILE *cfile, const char /** * Flush any caches for consistency with external processes. * - * @param ctx + * @ctx * The bfs context. */ void bfs_ctx_flush(const struct bfs_ctx *ctx); @@ -149,9 +164,9 @@ void bfs_ctx_flush(const struct bfs_ctx *ctx); /** * Dump the parsed command line. * - * @param ctx + * @ctx * The bfs context. - * @param flag + * @flag * The -D flag that triggered the dump. */ void bfs_ctx_dump(const struct bfs_ctx *ctx, enum debug_flags flag); @@ -159,7 +174,7 @@ void bfs_ctx_dump(const struct bfs_ctx *ctx, enum debug_flags flag); /** * Free a bfs context. * - * @param ctx + * @ctx * The context to free. * @return * 0 on success, -1 if any errors occurred. @@ -1,38 +1,43 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "diag.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "color.h" #include "ctx.h" #include "dstring.h" #include "expr.h" -#include <errno.h> + #include <stdarg.h> #include <stdio.h> #include <stdlib.h> - -/** bfs_diagf() implementation. */ -attr(printf(2, 0)) -static void bfs_vdiagf(const struct bfs_loc *loc, const char *format, va_list args) { - fprintf(stderr, "%s: %s@%s:%d: ", xgetprogname(), loc->func, loc->file, loc->line); - vfprintf(stderr, format, args); - fprintf(stderr, "\n"); -} - -void bfs_diagf(const struct bfs_loc *loc, const char *format, ...) { +#include <unistd.h> + +/** + * Print an error using dprintf() if possible, because it's more likely to be + * async-signal-safe in practice. + */ +#if BFS_HAS_DPRINTF +# define veprintf(...) vdprintf(STDERR_FILENO, __VA_ARGS__) +#else +# define veprintf(...) vfprintf(stderr, __VA_ARGS__) +#endif + +void bfs_diagf(const char *format, ...) { va_list args; va_start(args, format); - bfs_vdiagf(loc, format, args); + veprintf(format, args); va_end(args); } -noreturn void bfs_abortf(const struct bfs_loc *loc, const char *format, ...) { +_noreturn +void bfs_abortf(const char *format, ...) { va_list args; va_start(args, format); - bfs_vdiagf(loc, format, args); + veprintf(format, args); va_end(args); abort(); @@ -64,7 +69,7 @@ const char *debug_flag_name(enum debug_flags flag) { } void bfs_perror(const struct bfs_ctx *ctx, const char *str) { - bfs_error(ctx, "%s: %m.\n", str); + bfs_error(ctx, "%s: %s.\n", str, errstr()); } void bfs_error(const struct bfs_ctx *ctx, const char *format, ...) { @@ -91,19 +96,12 @@ bool bfs_debug(const struct bfs_ctx *ctx, enum debug_flags flag, const char *for } void bfs_verror(const struct bfs_ctx *ctx, const char *format, va_list args) { - int error = errno; - bfs_error_prefix(ctx); - - errno = error; cvfprintf(ctx->cerr, format, args); } bool bfs_vwarning(const struct bfs_ctx *ctx, const char *format, va_list args) { - int error = errno; - if (bfs_warning_prefix(ctx)) { - errno = error; cvfprintf(ctx->cerr, format, args); return true; } else { @@ -112,10 +110,7 @@ bool bfs_vwarning(const struct bfs_ctx *ctx, const char *format, va_list args) { } bool bfs_vdebug(const struct bfs_ctx *ctx, enum debug_flags flag, const char *format, va_list args) { - int error = errno; - if (bfs_debug_prefix(ctx, flag)) { - errno = error; cvfprintf(ctx->cerr, format, args); return true; } else { @@ -169,7 +164,7 @@ static bool highlight_expr_recursive(const struct bfs_ctx *ctx, const struct bfs } } - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { ret |= highlight_expr_recursive(ctx, child, args); } @@ -8,92 +8,134 @@ #ifndef BFS_DIAG_H #define BFS_DIAG_H -#include "prelude.h" +#include "bfs.h" +#include "bfstd.h" + #include <stdarg.h> /** - * static_assert() with an optional second argument. + * Wrap a diagnostic format string so it looks like + * + * bfs: func@src/file.c:0: Message */ -#if __STDC_VERSION__ >= C23 -# define bfs_static_assert static_assert -#else -# define bfs_static_assert(...) bfs_static_assert_(__VA_ARGS__, #__VA_ARGS__, ) -# define bfs_static_assert_(expr, msg, ...) _Static_assert(expr, msg) -#endif +#define BFS_DIAG_FORMAT_(format) \ + ((format) ? "%s: %s@%s:%d: " format "%s" : "") /** - * A source code location. + * Add arguments to match a BFS_DIAG_FORMAT string. */ -struct bfs_loc { - const char *file; - int line; - const char *func; -}; - -#define BFS_LOC_INIT { .file = __FILE__, .line = __LINE__, .func = __func__ } +#define BFS_DIAG_ARGS_(...) \ + xgetprogname(), __func__, __FILE__, __LINE__, __VA_ARGS__ "\n" /** - * Get the current source code location. + * Print a low-level diagnostic message to standard error. */ -#if __STDC_VERSION__ >= C23 -# define bfs_location() (&(static const struct bfs_loc)BFS_LOC_INIT) -#else -# define bfs_location() (&(const struct bfs_loc)BFS_LOC_INIT) -#endif +_printf(1, 2) +void bfs_diagf(const char *format, ...); /** - * Print a low-level diagnostic message to standard error, formatted like - * - * bfs: func@src/file.c:0: Message + * Unconditional diagnostic message. */ -attr(printf(2, 3)) -void bfs_diagf(const struct bfs_loc *loc, const char *format, ...); +#define bfs_diag(...) \ + bfs_diag_(__VA_ARGS__, ) + +#define bfs_diag_(format, ...) \ + bfs_diagf(BFS_DIAG_FORMAT_(format), BFS_DIAG_ARGS_(__VA_ARGS__)) /** - * Unconditional diagnostic message. + * Print a diagnostic message including the last error. */ -#define bfs_diag(...) bfs_diagf(bfs_location(), __VA_ARGS__) +#define bfs_ediag(...) \ + bfs_ediag_(__VA_ARGS__, ) + +#define bfs_ediag_(format, ...) \ + bfs_diag_(format "%s%s", __VA_ARGS__ (sizeof("" format) > 1 ? ": " : ""), errstr(), ) /** * Print a message to standard error and abort. */ -attr(cold, printf(2, 3)) -noreturn void bfs_abortf(const struct bfs_loc *loc, const char *format, ...); +_cold +_printf(1, 2) +_noreturn +void bfs_abortf(const char *format, ...); /** * Unconditional abort with a message. */ -#define bfs_abort(...) bfs_abortf(bfs_location(), __VA_ARGS__) +#define bfs_abort(...) \ + bfs_abort_(__VA_ARGS__, ) + +#define bfs_abort_(format, ...) \ + bfs_abortf(BFS_DIAG_FORMAT_(format), BFS_DIAG_ARGS_(__VA_ARGS__)) + +/** + * Abort with a message including the last error. + */ +#define bfs_eabort(...) \ + bfs_eabort_(__VA_ARGS__, ) + +#define bfs_eabort_(format, ...) \ + ((format) ? bfs_abort_(format ": %s", __VA_ARGS__ errstr(), ) : (void)0) /** * Abort in debug builds; no-op in release builds. */ #ifdef NDEBUG # define bfs_bug(...) ((void)0) +# define bfs_ebug(...) ((void)0) #else # define bfs_bug bfs_abort +# define bfs_ebug bfs_eabort #endif /** + * Get the default assertion message, if no format string was specified. + */ +#define BFS_DIAG_MSG_(format, str) \ + (sizeof(format) > 1 ? "" : str) + +/** * Unconditional assert. */ #define bfs_verify(...) \ - bfs_verify_(#__VA_ARGS__, __VA_ARGS__, "", "") + bfs_verify_(#__VA_ARGS__, __VA_ARGS__, "", ) #define bfs_verify_(str, cond, format, ...) \ - ((cond) ? (void)0 : bfs_abort( \ + ((cond) ? (void)0 : bfs_verify__(format, BFS_DIAG_MSG_(format, str), __VA_ARGS__)) + +#define bfs_verify__(format, ...) \ + bfs_abortf( \ + sizeof(format) > 1 \ + ? BFS_DIAG_FORMAT_("%s" format "%s") \ + : BFS_DIAG_FORMAT_("Assertion failed: `%s`"), \ + BFS_DIAG_ARGS_(__VA_ARGS__)) + +/** + * Unconditional assert, including the last error. + */ +#define bfs_everify(...) \ + bfs_everify_(#__VA_ARGS__, __VA_ARGS__, "", ) + + +#define bfs_everify_(str, cond, format, ...) \ + ((cond) ? (void)0 : bfs_everify__(format, BFS_DIAG_MSG_(format, str), __VA_ARGS__)) + +#define bfs_everify__(format, ...) \ + bfs_abortf( \ sizeof(format) > 1 \ - ? "%.0s" format "%s%s" \ - : "Assertion failed: `%s`%s", \ - str, __VA_ARGS__)) + ? BFS_DIAG_FORMAT_("%s" format "%s: %s") \ + : BFS_DIAG_FORMAT_("Assertion failed: `%s`: %s"), \ + BFS_DIAG_ARGS_(__VA_ARGS__ errstr(), )) /** * Assert in debug builds; no-op in release builds. */ #ifdef NDEBUG # define bfs_assert(...) ((void)0) +# define bfs_eassert(...) ((void)0) #else # define bfs_assert bfs_verify +# define bfs_eassert bfs_everify #endif struct bfs_ctx; @@ -129,13 +171,14 @@ const char *debug_flag_name(enum debug_flags flag); /** * Like perror(), but decorated like bfs_error(). */ -attr(cold) +_cold void bfs_perror(const struct bfs_ctx *ctx, const char *str); /** * Shorthand for printing error messages. */ -attr(cold, printf(2, 3)) +_cold +_printf(2, 3) void bfs_error(const struct bfs_ctx *ctx, const char *format, ...); /** @@ -143,7 +186,8 @@ void bfs_error(const struct bfs_ctx *ctx, const char *format, ...); * * @return Whether a warning was printed. */ -attr(cold, printf(2, 3)) +_cold +_printf(2, 3) bool bfs_warning(const struct bfs_ctx *ctx, const char *format, ...); /** @@ -151,67 +195,71 @@ bool bfs_warning(const struct bfs_ctx *ctx, const char *format, ...); * * @return Whether a debug message was printed. */ -attr(cold, printf(3, 4)) +_cold +_printf(3, 4) bool bfs_debug(const struct bfs_ctx *ctx, enum debug_flags flag, const char *format, ...); /** * bfs_error() variant that takes a va_list. */ -attr(cold, printf(2, 0)) +_cold +_printf(2, 0) void bfs_verror(const struct bfs_ctx *ctx, const char *format, va_list args); /** * bfs_warning() variant that takes a va_list. */ -attr(cold, printf(2, 0)) +_cold +_printf(2, 0) bool bfs_vwarning(const struct bfs_ctx *ctx, const char *format, va_list args); /** * bfs_debug() variant that takes a va_list. */ -attr(cold, printf(3, 0)) +_cold +_printf(3, 0) bool bfs_vdebug(const struct bfs_ctx *ctx, enum debug_flags flag, const char *format, va_list args); /** * Print the error message prefix. */ -attr(cold) +_cold void bfs_error_prefix(const struct bfs_ctx *ctx); /** * Print the warning message prefix. */ -attr(cold) +_cold bool bfs_warning_prefix(const struct bfs_ctx *ctx); /** * Print the debug message prefix. */ -attr(cold) +_cold bool bfs_debug_prefix(const struct bfs_ctx *ctx, enum debug_flags flag); /** * Highlight parts of the command line in an error message. */ -attr(cold) +_cold void bfs_argv_error(const struct bfs_ctx *ctx, const bool args[]); /** * Highlight parts of an expression in an error message. */ -attr(cold) +_cold void bfs_expr_error(const struct bfs_ctx *ctx, const struct bfs_expr *expr); /** * Highlight parts of the command line in a warning message. */ -attr(cold) +_cold bool bfs_argv_warning(const struct bfs_ctx *ctx, const bool args[]); /** * Highlight parts of an expression in a warning message. */ -attr(cold) +_cold bool bfs_expr_warning(const struct bfs_ctx *ctx, const struct bfs_expr *expr); #endif // BFS_DIAG_H @@ -1,13 +1,15 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "dir.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "diag.h" #include "sanity.h" #include "trie.h" + #include <dirent.h> #include <errno.h> #include <fcntl.h> @@ -25,7 +27,13 @@ static ssize_t bfs_getdents(int fd, void *buf, size_t size) { sanitize_uninit(buf, size); -#if BFS_HAS_GETDENTS +#if BFS_HAS_POSIX_GETDENTS + int flags = 0; +# ifdef DT_FORCE_TYPE + flags |= DT_FORCE_TYPE; +# endif + ssize_t ret = posix_getdents(fd, buf, size, flags); +#elif BFS_HAS_GETDENTS ssize_t ret = getdents(fd, buf, size); #elif BFS_HAS_GETDENTS64 ssize_t ret = getdents64(fd, buf, size); @@ -44,11 +52,13 @@ static ssize_t bfs_getdents(int fd, void *buf, size_t size) { #endif // BFS_USE_GETDENTS -#if BFS_USE_GETDENTS && !BFS_HAS_GETDENTS /** Directory entry type for bfs_getdents() */ -typedef struct dirent64 sys_dirent; -#else +#if !BFS_USE_GETDENTS || BFS_HAS_GETDENTS typedef struct dirent sys_dirent; +#elif BFS_HAS_POSIX_GETDENTS +typedef struct posix_dent sys_dirent; +#else +typedef struct dirent64 sys_dirent; #endif enum bfs_type bfs_mode_to_type(mode_t mode) { @@ -8,15 +8,22 @@ #ifndef BFS_DIR_H #define BFS_DIR_H -#include "prelude.h" +#include "bfs.h" + #include <sys/types.h> /** * Whether the implementation uses the getdents() syscall directly, rather than * libc's readdir(). */ -#if !defined(BFS_USE_GETDENTS) && (__linux__ || __FreeBSD__) -# define BFS_USE_GETDENTS (BFS_HAS_GETDENTS || BFS_HAS_GETDENTS64 | BFS_HAS_GETDENTS64_SYSCALL) +#ifndef BFS_USE_GETDENTS +# if BFS_HAS_POSIX_GETDENTS +# define BFS_USE_GETDENTS true +# elif __linux__ || __FreeBSD__ +# define BFS_USE_GETDENTS (BFS_HAS_GETDENTS || BFS_HAS_GETDENTS64 | BFS_HAS_GETDENTS64_SYSCALL) +# else +# define BFS_USE_GETDENTS false +# endif #endif /** @@ -82,7 +89,7 @@ struct arena; /** * Initialize an arena for directories. * - * @param arena + * @arena * The arena to initialize. */ void bfs_dir_arena(struct arena *arena); @@ -100,14 +107,14 @@ enum bfs_dir_flags { /** * Open a directory. * - * @param dir + * @dir * The allocated directory. - * @param at_fd + * @at_fd * The base directory for path resolution. - * @param at_path + * @at_path * The path of the directory to open, relative to at_fd. Pass NULL to * open at_fd itself. - * @param flags + * @flags * Flags that control which directory entries are listed. * @return * 0 on success, or -1 on failure. @@ -122,7 +129,7 @@ int bfs_dirfd(const struct bfs_dir *dir); /** * Performs any I/O necessary for the next bfs_readdir() call. * - * @param dir + * @dir * The directory to poll. * @return * 1 on success, 0 on EOF, or -1 on failure. @@ -132,9 +139,9 @@ int bfs_polldir(struct bfs_dir *dir); /** * Read a directory entry. * - * @param dir + * @dir * The directory to read. - * @param[out] dirent + * @dirent[out] * The directory entry to populate. * @return * 1 on success, 0 on EOF, or -1 on failure. @@ -160,7 +167,7 @@ int bfs_closedir(struct bfs_dir *dir); /** * Detach the file descriptor from an open directory. * - * @param dir + * @dir * The directory to detach. * @return * The file descriptor of the directory. diff --git a/src/dstring.c b/src/dstring.c index b5bf3d3..678d685 100644 --- a/src/dstring.c +++ b/src/dstring.c @@ -1,11 +1,12 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "dstring.h" + #include "alloc.h" #include "bit.h" #include "diag.h" + #include <stdarg.h> #include <stddef.h> #include <stdint.h> @@ -22,7 +23,7 @@ struct dstring { /** Length of the string, *excluding* the terminating NUL. */ size_t len; /** The string itself. */ - alignas(dchar) char str[]; + alignas(dchar) char str[] _counted_by(cap); }; #define DSTR_OFFSET offsetof(struct dstring, str) @@ -45,6 +46,13 @@ static dchar *dstrdata(struct dstring *header) { return (char *)header + DSTR_OFFSET; } +/** Set the length of a dynamic string. */ +static void dstrsetlen(struct dstring *header, size_t len) { + bfs_assert(len < header->cap); + header->len = len; + header->str[len] = '\0'; +} + /** Allocate a dstring with the given contents. */ static dchar *dstralloc_impl(size_t cap, size_t len, const char *str) { // Avoid reallocations for small strings @@ -58,11 +66,10 @@ static dchar *dstralloc_impl(size_t cap, size_t len, const char *str) { } header->cap = cap; - header->len = len; + dstrsetlen(header, len); - char *ret = dstrdata(header); + dchar *ret = dstrdata(header); memcpy(ret, str, len); - ret[len] = '\0'; return ret; } @@ -120,11 +127,16 @@ int dstresize(dchar **dstr, size_t len) { } struct dstring *header = dstrheader(*dstr); - header->len = len; - header->str[len] = '\0'; + dstrsetlen(header, len); return 0; } +void dstrshrink(dchar *dstr, size_t len) { + struct dstring *header = dstrheader(dstr); + bfs_assert(len <= header->len); + dstrsetlen(header, len); +} + int dstrcat(dchar **dest, const char *src) { return dstrxcat(dest, src, strlen(src)); } @@ -174,7 +186,7 @@ int dstrxcpy(dchar **dest, const char *src, size_t len) { return 0; } -char *dstrprintf(const char *format, ...) { +dchar *dstrprintf(const char *format, ...) { va_list args; va_start(args, format); @@ -184,7 +196,7 @@ char *dstrprintf(const char *format, ...) { return str; } -char *dstrvprintf(const char *format, va_list args) { +dchar *dstrvprintf(const char *format, va_list args) { // Guess a capacity to try to avoid reallocating dchar *str = dstralloc(2 * strlen(format)); if (!str) { @@ -277,3 +289,20 @@ void dstrfree(dchar *dstr) { free(dstrheader(dstr)); } } + +dchar *dstrepeat(const char *str, size_t n) { + size_t len = strlen(str); + dchar *ret = dstralloc(n * len); + if (!ret) { + return NULL; + } + + for (size_t i = 0; i < n; ++i) { + if (dstrxcat(&ret, str, len) < 0) { + dstrfree(ret); + return NULL; + } + } + + return ret; +} diff --git a/src/dstring.h b/src/dstring.h index 9ea7eb9..ce7ef86 100644 --- a/src/dstring.h +++ b/src/dstring.h @@ -8,8 +8,9 @@ #ifndef BFS_DSTRING_H #define BFS_DSTRING_H -#include "prelude.h" +#include "bfs.h" #include "bfstd.h" + #include <stdarg.h> #include <stddef.h> @@ -30,7 +31,7 @@ typedef char dchar; /** * Free a dynamic string. * - * @param dstr + * @dstr * The string to free. */ void dstrfree(dchar *dstr); @@ -38,56 +39,56 @@ void dstrfree(dchar *dstr); /** * Allocate a dynamic string. * - * @param cap + * @cap * The initial capacity of the string. */ -attr(malloc(dstrfree, 1)) +_malloc(dstrfree, 1) dchar *dstralloc(size_t cap); /** * Create a dynamic copy of a string. * - * @param str + * @str * The NUL-terminated string to copy. */ -attr(malloc(dstrfree, 1)) +_malloc(dstrfree, 1) dchar *dstrdup(const char *str); /** * Create a length-limited dynamic copy of a string. * - * @param str + * @str * The string to copy. - * @param n + * @n * The maximum number of characters to copy from str. */ -attr(malloc(dstrfree, 1)) +_malloc(dstrfree, 1) dchar *dstrndup(const char *str, size_t n); /** * Create a dynamic copy of a dynamic string. * - * @param dstr + * @dstr * The dynamic string to copy. */ -attr(malloc(dstrfree, 1)) +_malloc(dstrfree, 1) dchar *dstrddup(const dchar *dstr); /** * Create an exact-sized dynamic copy of a string. * - * @param str + * @str * The string to copy. - * @param len + * @len * The length of the string, which may include internal NUL bytes. */ -attr(malloc(dstrfree, 1)) +_malloc(dstrfree, 1) dchar *dstrxdup(const char *str, size_t len); /** * Get a dynamic string's length. * - * @param dstr + * @dstr * The string to measure. * @return * The length of dstr. @@ -97,9 +98,9 @@ size_t dstrlen(const dchar *dstr); /** * Reserve some capacity in a dynamic string. * - * @param dstr + * @dstr * The dynamic string to preallocate. - * @param cap + * @cap * The new capacity for the string. * @return * 0 on success, -1 on failure. @@ -109,214 +110,246 @@ int dstreserve(dchar **dstr, size_t cap); /** * Resize a dynamic string. * - * @param dstr + * @dstr * The dynamic string to resize. - * @param len + * @len * The new length for the dynamic string. * @return * 0 on success, -1 on failure. */ +_nodiscard int dstresize(dchar **dstr, size_t len); /** + * Shrink a dynamic string. + * + * @dstr + * The dynamic string to shrink. + * @len + * The new length. Must not be greater than the current length. + */ +void dstrshrink(dchar *dstr, size_t len); + +/** * Append to a dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The string to append. * @return 0 on success, -1 on failure. */ +_nodiscard int dstrcat(dchar **dest, const char *src); /** * Append to a dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The string to append. - * @param n + * @n * The maximum number of characters to take from src. * @return * 0 on success, -1 on failure. */ +_nodiscard int dstrncat(dchar **dest, const char *src, size_t n); /** * Append a dynamic string to another dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The dynamic string to append. * @return * 0 on success, -1 on failure. */ +_nodiscard int dstrdcat(dchar **dest, const dchar *src); /** * Append to a dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The string to append. - * @param len + * @len * The exact number of characters to take from src. * @return * 0 on success, -1 on failure. */ +_nodiscard int dstrxcat(dchar **dest, const char *src, size_t len); /** * Append a single character to a dynamic string. * - * @param str + * @str * The string to append to. - * @param c + * @c * The character to append. * @return * 0 on success, -1 on failure. */ +_nodiscard int dstrapp(dchar **str, char c); /** * Copy a string into a dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The string to copy. * @returns * 0 on success, -1 on failure. */ +_nodiscard int dstrcpy(dchar **dest, const char *str); /** * Copy a dynamic string into another one. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The dynamic string to copy. * @returns * 0 on success, -1 on failure. */ +_nodiscard int dstrdcpy(dchar **dest, const dchar *str); /** * Copy a string into a dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The dynamic string to copy. - * @param n + * @n * The maximum number of characters to take from src. * @returns * 0 on success, -1 on failure. */ +_nodiscard int dstrncpy(dchar **dest, const char *str, size_t n); /** * Copy a string into a dynamic string. * - * @param dest + * @dest * The destination dynamic string. - * @param src + * @src * The dynamic string to copy. - * @param len + * @len * The exact number of characters to take from src. * @returns * 0 on success, -1 on failure. */ +_nodiscard int dstrxcpy(dchar **dest, const char *str, size_t len); /** * Create a dynamic string from a format string. * - * @param format + * @format * The format string to fill in. - * @param ... + * @... * Any arguments for the format string. * @return * The created string, or NULL on failure. */ -attr(printf(1, 2)) -char *dstrprintf(const char *format, ...); +_nodiscard +_printf(1, 2) +dchar *dstrprintf(const char *format, ...); /** * Create a dynamic string from a format string and a va_list. * - * @param format + * @format * The format string to fill in. - * @param args + * @args * The arguments for the format string. * @return * The created string, or NULL on failure. */ -attr(printf(1, 0)) -char *dstrvprintf(const char *format, va_list args); +_nodiscard +_printf(1, 0) +dchar *dstrvprintf(const char *format, va_list args); /** * Format some text onto the end of a dynamic string. * - * @param str + * @str * The destination dynamic string. - * @param format + * @format * The format string to fill in. - * @param ... + * @... * Any arguments for the format string. * @return * 0 on success, -1 on failure. */ -attr(printf(2, 3)) +_nodiscard +_printf(2, 3) int dstrcatf(dchar **str, const char *format, ...); /** * Format some text from a va_list onto the end of a dynamic string. * - * @param str + * @str * The destination dynamic string. - * @param format + * @format * The format string to fill in. - * @param args + * @args * The arguments for the format string. * @return * 0 on success, -1 on failure. */ -attr(printf(2, 0)) +_nodiscard +_printf(2, 0) int dstrvcatf(dchar **str, const char *format, va_list args); /** * Concatenate while shell-escaping. * - * @param dest + * @dest * The destination dynamic string. - * @param str + * @str * The string to escape. - * @param flags + * @flags * Flags for wordesc(). * @return * 0 on success, -1 on failure. */ +_nodiscard int dstrescat(dchar **dest, const char *str, enum wesc_flags flags); /** * Concatenate while shell-escaping. * - * @param dest + * @dest * The destination dynamic string. - * @param str + * @str * The string to escape. - * @param n + * @n * The maximum length of the string. - * @param flags + * @flags * Flags for wordesc(). * @return * 0 on success, -1 on failure. */ +_nodiscard int dstrnescat(dchar **dest, const char *str, size_t n, enum wesc_flags flags); +/** + * Repeat a string n times. + */ +_nodiscard +dchar *dstrepeat(const char *str, size_t n); + #endif // BFS_DSTRING_H @@ -5,9 +5,11 @@ * Implementation of all the primary expressions. */ -#include "prelude.h" #include "eval.h" + +#include "atomic.h" #include "bar.h" +#include "bfs.h" #include "bfstd.h" #include "bftw.h" #include "color.h" @@ -22,14 +24,18 @@ #include "printf.h" #include "pwcache.h" #include "sanity.h" +#include "sighook.h" #include "stat.h" #include "trie.h" #include "xregex.h" +#include "xtime.h" + #include <errno.h> #include <fcntl.h> #include <fnmatch.h> #include <grp.h> #include <pwd.h> +#include <signal.h> #include <stdarg.h> #include <stdint.h> #include <stdio.h> @@ -37,6 +43,7 @@ #include <string.h> #include <strings.h> #include <sys/resource.h> +#include <sys/time.h> #include <sys/types.h> #include <time.h> #include <unistd.h> @@ -51,6 +58,8 @@ struct bfs_eval { enum bftw_action action; /** The bfs_eval() return value. */ int *ret; + /** The number of errors that have occurred. */ + size_t *nerrors; /** Whether to quit immediately. */ bool quit; }; @@ -58,20 +67,24 @@ struct bfs_eval { /** * Print an error message. */ -attr(printf(2, 3)) +_printf(2, 3) static void eval_error(struct bfs_eval *state, const char *format, ...) { + const struct bfs_ctx *ctx = state->ctx; + + ++*state->nerrors; + if (ctx->ignore_errors) { + return; + } + // By POSIX, any errors should be accompanied by a non-zero exit status *state->ret = EXIT_FAILURE; - int error = errno; - const struct bfs_ctx *ctx = state->ctx; CFILE *cerr = ctx->cerr; bfs_error(ctx, "%pP: ", state->ftwbuf); va_list args; va_start(args, format); - errno = error; cvfprintf(cerr, format, args); va_end(args); } @@ -90,7 +103,7 @@ static bool eval_should_ignore(const struct bfs_eval *state, int error) { */ static void eval_report_error(struct bfs_eval *state) { if (!eval_should_ignore(state, errno)) { - eval_error(state, "%m.\n"); + eval_error(state, "%s.\n", errstr()); } } @@ -99,9 +112,9 @@ static void eval_report_error(struct bfs_eval *state) { */ static void eval_io_error(const struct bfs_expr *expr, struct bfs_eval *state) { if (expr->path) { - eval_error(state, "'%s': %m.\n", expr->path); + eval_error(state, "'%s': %s.\n", expr->path, errstr()); } else { - eval_error(state, "(standard output): %m.\n"); + eval_error(state, "(standard output): %s.\n", errstr()); } // Don't report the error again in bfs_ctx_free() @@ -124,11 +137,9 @@ static const struct bfs_stat *eval_stat(struct bfs_eval *state) { * Get the difference (in seconds) between two struct timespecs. */ static time_t timespec_diff(const struct timespec *lhs, const struct timespec *rhs) { - time_t ret = lhs->tv_sec - rhs->tv_sec; - if (lhs->tv_nsec < rhs->tv_nsec) { - --ret; - } - return ret; + struct timespec diff = *lhs; + timespec_sub(&diff, rhs); + return diff.tv_sec; } bool bfs_expr_cmp(const struct bfs_expr *expr, long long n) { @@ -228,7 +239,7 @@ bool eval_context(const struct bfs_expr *expr, struct bfs_eval *state) { static const struct timespec *eval_stat_time(const struct bfs_stat *statbuf, enum bfs_stat_field field, struct bfs_eval *state) { const struct timespec *ret = bfs_stat_time(statbuf, field); if (!ret) { - eval_error(state, "Couldn't get file %s: %m.\n", bfs_stat_field_name(field)); + eval_error(state, "Couldn't get file %s: %s.\n", bfs_stat_field_name(field), errstr()); } return ret; } @@ -247,8 +258,7 @@ bool eval_newer(const struct bfs_expr *expr, struct bfs_eval *state) { return false; } - return time->tv_sec > expr->reftime.tv_sec - || (time->tv_sec == expr->reftime.tv_sec && time->tv_nsec > expr->reftime.tv_nsec); + return timespec_cmp(time, &expr->reftime) > 0; } /** @@ -269,10 +279,10 @@ bool eval_time(const struct bfs_expr *expr, struct bfs_eval *state) { switch (expr->time_unit) { case BFS_DAYS: diff /= 60 * 24; - fallthru; + _fallthrough; case BFS_MINUTES: diff /= 60; - fallthru; + _fallthrough; case BFS_SECONDS: break; } @@ -398,13 +408,13 @@ static int eval_exec_finish(const struct bfs_expr *expr, const struct bfs_ctx *c if (expr->eval_fn == eval_exec) { if (bfs_exec_finish(expr->exec) != 0) { if (errno != 0) { - bfs_error(ctx, "%s %s: %m.\n", expr->argv[0], expr->argv[1]); + bfs_error(ctx, "${blu}%pq${rs} ${bld}%pq${rs}: %s.\n", expr->argv[0], expr->argv[1], errstr()); } ret = -1; } } - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (eval_exec_finish(child, ctx) != 0) { ret = -1; } @@ -419,7 +429,7 @@ static int eval_exec_finish(const struct bfs_expr *expr, const struct bfs_ctx *c bool eval_exec(const struct bfs_expr *expr, struct bfs_eval *state) { bool ret = bfs_exec(expr->exec, state->ftwbuf) == 0; if (errno != 0) { - eval_error(state, "%s %s: %m.\n", expr->argv[0], expr->argv[1]); + eval_error(state, "${blu}%pq${rs} ${bld}%pq${rs}: %s.\n", expr->argv[0], expr->argv[1], errstr()); } return ret; } @@ -690,6 +700,34 @@ static int print_owner(FILE *file, const char *name, uintmax_t id, int *width) { } } +/** Print a file's modification time. */ +static int print_time(FILE *file, time_t time, time_t now) { + struct tm tm; + if (!localtime_r(&time, &tm)) { + goto error; + } + + char time_str[256]; + size_t time_ret; + + time_t six_months_ago = now - 6 * 30 * 24 * 60 * 60; + time_t tomorrow = now + 24 * 60 * 60; + if (time <= six_months_ago || time >= tomorrow) { + time_ret = strftime(time_str, sizeof(time_str), "%b %e %Y", &tm); + } else { + time_ret = strftime(time_str, sizeof(time_str), "%b %e %H:%M", &tm); + } + + if (time_ret == 0) { + goto error; + } + + return fprintf(file, " %s", time_str); + +error: + return fprintf(file, " %jd", (intmax_t)time); +} + /** * -f?ls action. */ @@ -746,28 +784,11 @@ bool eval_fls(const struct bfs_expr *expr, struct bfs_eval *state) { time_t time = statbuf->mtime.tv_sec; time_t now = ctx->now.tv_sec; - time_t six_months_ago = now - 6 * 30 * 24 * 60 * 60; - time_t tomorrow = now + 24 * 60 * 60; - struct tm tm; - if (!localtime_r(&time, &tm)) { - goto error; - } - char time_str[256]; - size_t time_ret; - if (time <= six_months_ago || time >= tomorrow) { - time_ret = strftime(time_str, sizeof(time_str), "%b %e %Y", &tm); - } else { - time_ret = strftime(time_str, sizeof(time_str), "%b %e %H:%M", &tm); - } - if (time_ret == 0) { - errno = EOVERFLOW; - goto error; - } - if (cfprintf(cfile, " %s${rs}", time_str) < 0) { + if (print_time(file, time, now) < 0) { goto error; } - if (cfprintf(cfile, " %pP", ftwbuf) < 0) { + if (cfprintf(cfile, "${rs} %pP", ftwbuf) < 0) { goto error; } @@ -902,7 +923,7 @@ bool eval_regex(const struct bfs_expr *expr, struct bfs_eval *state) { eval_error(state, "%s.\n", str); free(str); } else { - eval_error(state, "bfs_regerror(): %m.\n"); + eval_error(state, "bfs_regerror(): %s.\n", errstr()); } } @@ -999,6 +1020,13 @@ bool eval_xtype(const struct bfs_expr *expr, struct bfs_eval *state) { const struct BFTW *ftwbuf = state->ftwbuf; enum bfs_stat_flags flags = ftwbuf->stat_flags ^ (BFS_STAT_NOFOLLOW | BFS_STAT_TRYFOLLOW); enum bfs_type type = bftw_type(ftwbuf, flags); + + // GNU find treats ELOOP as a broken symbolic link for -xtype l + // (but not -L -type l) + if ((flags & BFS_STAT_TRYFOLLOW) && type == BFS_ERROR && errno == ELOOP) { + type = BFS_LNK; + } + if (type == BFS_ERROR) { eval_report_error(state); return false; @@ -1007,40 +1035,23 @@ bool eval_xtype(const struct bfs_expr *expr, struct bfs_eval *state) { } } -#if _POSIX_MONOTONIC_CLOCK > 0 -# define BFS_CLOCK CLOCK_MONOTONIC -#elif _POSIX_TIMERS > 0 -# define BFS_CLOCK CLOCK_REALTIME -#endif - /** - * Call clock_gettime(), if available. + * clock_gettime() wrapper. */ static int eval_gettime(struct bfs_eval *state, struct timespec *ts) { -#ifdef BFS_CLOCK - int ret = clock_gettime(BFS_CLOCK, ts); - if (ret != 0) { - bfs_warning(state->ctx, "%pP: clock_gettime(): %m.\n", state->ftwbuf); + clockid_t clock = CLOCK_REALTIME; + +#if defined(_POSIX_MONOTONIC_CLOCK) && _POSIX_MONOTONIC_CLOCK >= 0 + if (sysoption(MONOTONIC_CLOCK) > 0) { + clock = CLOCK_MONOTONIC; } - return ret; -#else - return -1; #endif -} -/** - * Record an elapsed time. - */ -static void timespec_elapsed(struct timespec *elapsed, const struct timespec *start, const struct timespec *end) { - elapsed->tv_sec += end->tv_sec - start->tv_sec; - elapsed->tv_nsec += end->tv_nsec - start->tv_nsec; - if (elapsed->tv_nsec < 0) { - elapsed->tv_nsec += 1000000000L; - --elapsed->tv_sec; - } else if (elapsed->tv_nsec >= 1000000000L) { - elapsed->tv_nsec -= 1000000000L; - ++elapsed->tv_sec; + int ret = clock_gettime(clock, ts); + if (ret != 0) { + bfs_warning(state->ctx, "%pP: clock_gettime(): %s.\n", state->ftwbuf, errstr()); } + return ret; } /** @@ -1061,7 +1072,8 @@ static bool eval_expr(struct bfs_expr *expr, struct bfs_eval *state) { if (time) { if (eval_gettime(state, &end) == 0) { - timespec_elapsed(&expr->elapsed, &start, &end); + timespec_sub(&end, &start); + timespec_add(&expr->elapsed, &end); } } @@ -1091,7 +1103,7 @@ bool eval_not(const struct bfs_expr *expr, struct bfs_eval *state) { * Evaluate a conjunction. */ bool eval_and(const struct bfs_expr *expr, struct bfs_eval *state) { - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (!eval_expr(child, state) || state->quit) { return false; } @@ -1104,7 +1116,7 @@ bool eval_and(const struct bfs_expr *expr, struct bfs_eval *state) { * Evaluate a disjunction. */ bool eval_or(const struct bfs_expr *expr, struct bfs_eval *state) { - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (eval_expr(child, state) || state->quit) { return true; } @@ -1119,7 +1131,7 @@ bool eval_or(const struct bfs_expr *expr, struct bfs_eval *state) { bool eval_comma(const struct bfs_expr *expr, struct bfs_eval *state) { bool ret uninit(false); - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { ret = eval_expr(child, state); if (state->quit) { break; @@ -1130,20 +1142,7 @@ bool eval_comma(const struct bfs_expr *expr, struct bfs_eval *state) { } /** Update the status bar. */ -static void eval_status(struct bfs_eval *state, struct bfs_bar *bar, struct timespec *last_status, size_t count) { - struct timespec now; - if (eval_gettime(state, &now) == 0) { - struct timespec elapsed = {0}; - timespec_elapsed(&elapsed, last_status, &now); - - // Update every 0.1s - if (elapsed.tv_sec > 0 || elapsed.tv_nsec >= 100000000L) { - *last_status = now; - } else { - return; - } - } - +static void eval_status(struct bfs_eval *state, struct bfs_bar *bar, size_t count) { size_t width = bfs_bar_width(bar); if (width < 3) { return; @@ -1159,7 +1158,7 @@ static void eval_status(struct bfs_eval *state, struct bfs_bar *bar, struct time size_t rhslen = xstrwidth(rhs); if (3 + rhslen > width) { - dstresize(&rhs, 0); + dstrshrink(rhs, 0); rhslen = 0; } @@ -1203,7 +1202,7 @@ static void eval_status(struct bfs_eval *state, struct bfs_bar *bar, struct time } pathwidth += cwidth; } - dstresize(&status, lhslen); + dstrshrink(status, lhslen); if (dstrcat(&status, "...") != 0) { goto out; @@ -1270,7 +1269,7 @@ static void debug_stat(const struct bfs_ctx *ctx, const struct BFTW *ftwbuf, enu bfs_debug_prefix(ctx, DEBUG_STAT); fprintf(stderr, "bfs_stat("); - if (ftwbuf->at_fd == AT_FDCWD) { + if (ftwbuf->at_fd == (int)AT_FDCWD) { fprintf(stderr, "AT_FDCWD"); } else { size_t baselen = strlen(ftwbuf->path) - strlen(ftwbuf->at_path); @@ -1286,7 +1285,7 @@ static void debug_stat(const struct bfs_ctx *ctx, const struct BFTW *ftwbuf, enu DEBUG_FLAG(flags, BFS_STAT_TRYFOLLOW); DEBUG_FLAG(flags, BFS_STAT_NOSYNC); - fprintf(stderr, ") == %d", err ? 0 : -1); + fprintf(stderr, ") == %d", err == 0 ? 0 : -1); if (err) { fprintf(stderr, " [%d]", err); @@ -1373,18 +1372,85 @@ struct callback_args { /** The status bar. */ struct bfs_bar *bar; - /** The time of the last status update. */ - struct timespec last_status; + /** The SIGALRM hook. */ + struct sighook *alrm_hook; + /** The interval timer. */ + struct timer *timer; + /** Flag set by SIGALRM. */ + atomic bool alrm_flag; + /** Flag set by SIGINFO. */ + atomic bool info_flag; + /** The number of files visited so far. */ size_t count; /** The set of seen files. */ struct trie *seen; + /** The number of errors that have occurred. */ + size_t nerrors; /** Eventual return value from bfs_eval(). */ int ret; }; +/** Update the status bar in response to SIGALRM. */ +static void eval_sigalrm(int sig, siginfo_t *info, void *ptr) { + struct callback_args *args = ptr; + store(&args->alrm_flag, true, relaxed); +} + +/** Show/hide the bar in response to SIGINFO. */ +static void eval_siginfo(int sig, siginfo_t *info, void *ptr) { + struct callback_args *args = ptr; + store(&args->info_flag, true, relaxed); +} + +/** Show the status bar. */ +static void eval_show_bar(struct callback_args *args) { + args->alrm_hook = sighook(SIGALRM, eval_sigalrm, args, SH_CONTINUE); + if (!args->alrm_hook) { + goto fail; + } + + args->bar = bfs_bar_show(); + if (!args->bar) { + goto fail; + } + + // Update the bar every 0.1s + struct timespec ival = { .tv_nsec = 100 * 1000 * 1000 }; + args->timer = xtimer_start(&ival); + if (!args->timer) { + goto fail; + } + + // Update the bar immediately + store(&args->alrm_flag, true, relaxed); + + return; + +fail: + bfs_warning(args->ctx, "Couldn't show status bar: %s.\n\n", errstr()); + + bfs_bar_hide(args->bar); + args->bar = NULL; + + sigunhook(args->alrm_hook); + args->alrm_hook = NULL; +} + +/** Hide the status bar. */ +static void eval_hide_bar(struct callback_args *args) { + xtimer_stop(args->timer); + args->timer = NULL; + + sigunhook(args->alrm_hook); + args->alrm_hook = NULL; + + bfs_bar_hide(args->bar); + args->bar = NULL; +} + /** * bftw() callback. */ @@ -1399,17 +1465,37 @@ static enum bftw_action eval_callback(const struct BFTW *ftwbuf, void *ptr) { state.ctx = ctx; state.action = BFTW_CONTINUE; state.ret = &args->ret; + state.nerrors = &args->nerrors; state.quit = false; - if (args->bar) { - eval_status(&state, args->bar, &args->last_status, args->count); + // Check whether SIGINFO was delivered and show/hide the bar + if (exchange(&args->info_flag, false, relaxed)) { + if (args->bar) { + eval_hide_bar(args); + } else { + eval_show_bar(args); + } + } + + if (exchange(&args->alrm_flag, false, relaxed)) { + eval_status(&state, args->bar, args->count); } if (ftwbuf->type == BFS_ERROR) { - if (!eval_should_ignore(&state, ftwbuf->error)) { - eval_error(&state, "%s.\n", xstrerror(ftwbuf->error)); - } state.action = BFTW_PRUNE; + + if (ftwbuf->error == ELOOP && ftwbuf->loopoff > 0) { + char *loop = strndup(ftwbuf->path, ftwbuf->loopoff); + if (loop) { + eval_error(&state, "Filesystem loop back to ${di}%pq${rs}\n", loop); + free(loop); + goto done; + } + } else if (eval_should_ignore(&state, ftwbuf->error)) { + goto done; + } + + eval_error(&state, "%s.\n", xstrerror(ftwbuf->error)); goto done; } @@ -1468,6 +1554,9 @@ done: static int raise_fdlimit(struct bfs_ctx *ctx) { rlim_t cur = ctx->orig_nofile.rlim_cur; rlim_t max = ctx->orig_nofile.rlim_max; + if (!ctx->raise_nofile) { + max = cur; + } rlim_t target = 64 << 10; if (rlim_cmp(target, max) > 0) { @@ -1596,7 +1685,7 @@ static bool eval_must_buffer(const struct bfs_expr *expr) { return true; } - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (eval_must_buffer(child)) { return true; } @@ -1617,12 +1706,16 @@ int bfs_eval(struct bfs_ctx *ctx) { }; if (ctx->status) { - args.bar = bfs_bar_show(); - if (!args.bar) { - bfs_warning(ctx, "Couldn't show status bar: %m.\n\n"); - } + eval_show_bar(&args); } +#ifdef SIGINFO + int siginfo = SIGINFO; +#else + int siginfo = SIGUSR1; +#endif + struct sighook *info_hook = sighook(siginfo, eval_siginfo, &args, SH_CONTINUE); + struct trie seen; if (ctx->unique) { trie_init(&seen); @@ -1690,7 +1783,14 @@ int bfs_eval(struct bfs_ctx *ctx) { trie_destroy(&seen); } - bfs_bar_hide(args.bar); + sigunhook(info_hook); + if (args.bar) { + eval_hide_bar(&args); + } + + if (ctx->ignore_errors && args.nerrors > 0) { + bfs_warning(ctx, "Suppressed errors: %zu\n", args.nerrors); + } return args.ret; } @@ -9,8 +9,6 @@ #ifndef BFS_EVAL_H #define BFS_EVAL_H -#include "prelude.h" - struct bfs_ctx; struct bfs_expr; @@ -22,9 +20,9 @@ struct bfs_eval; /** * Expression evaluation function. * - * @param expr + * @expr * The current expression. - * @param state + * @state * The current evaluation state. * @return * The result of the test. @@ -34,7 +32,7 @@ typedef bool bfs_eval_fn(const struct bfs_expr *expr, struct bfs_eval *state); /** * Evaluate the command line. * - * @param ctx + * @ctx * The bfs context to evaluate. * @return * EXIT_SUCCESS on success, otherwise on failure. @@ -1,9 +1,10 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "exec.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "bftw.h" #include "color.h" @@ -11,6 +12,7 @@ #include "diag.h" #include "dstring.h" #include "xspawn.h" + #include <errno.h> #include <fcntl.h> #include <stdarg.h> @@ -22,7 +24,7 @@ #include <unistd.h> /** Print some debugging info. */ -attr(printf(2, 3)) +_printf(2, 3) static void bfs_exec_debug(const struct bfs_exec *execbuf, const char *format, ...) { const struct bfs_ctx *ctx = execbuf->ctx; @@ -56,7 +58,7 @@ static size_t bfs_exec_arg_size(const char *arg) { /** Determine the maximum argv size. */ static size_t bfs_exec_arg_max(const struct bfs_exec *execbuf) { - long arg_max = sysconf(_SC_ARG_MAX); + long arg_max = xsysconf(_SC_ARG_MAX); bfs_exec_debug(execbuf, "ARG_MAX: %ld according to sysconf()\n", arg_max); if (arg_max < 0) { arg_max = BFS_EXEC_ARG_MAX; @@ -82,7 +84,7 @@ static size_t bfs_exec_arg_max(const struct bfs_exec *execbuf) { // Assume arguments are counted with the granularity of a single page, // so allow a one page cushion to account for rounding up - long page_size = sysconf(_SC_PAGESIZE); + long page_size = xsysconf(_SC_PAGESIZE); if (page_size < 4096) { page_size = 4096; } @@ -234,7 +236,7 @@ static char *bfs_exec_format_arg(char *arg, const char *path) { char *last = arg; do { - if (dstrncat(&ret, last, match - last) != 0) { + if (dstrxcat(&ret, last, match - last) != 0) { goto err; } if (dstrcat(&ret, path) != 0) { @@ -268,7 +270,7 @@ static int bfs_exec_openwd(struct bfs_exec *execbuf, const struct BFTW *ftwbuf) bfs_assert(execbuf->wd_fd < 0); bfs_assert(!execbuf->wd_path); - if (ftwbuf->at_fd != AT_FDCWD) { + if (ftwbuf->at_fd != (int)AT_FDCWD) { // Rely on at_fd being the immediate parent bfs_assert(xbaseoff(ftwbuf->at_path) == 0); @@ -67,11 +67,11 @@ struct bfs_exec { /** * Parse an exec action. * - * @param argv + * @argv * The (bfs) command line argument to parse. - * @param flags + * @flags * Any flags for this exec action. - * @param ctx + * @ctx * The bfs context. * @return * The parsed exec action, or NULL on failure. @@ -81,9 +81,9 @@ struct bfs_exec *bfs_exec_parse(const struct bfs_ctx *ctx, char **argv, enum bfs /** * Execute the command for a file. * - * @param execbuf + * @execbuf * The parsed exec action. - * @param ftwbuf + * @ftwbuf * The bftw() data for the current file. * @return 0 if the command succeeded, -1 if it failed. If the command could * be executed, -1 is returned, and errno will be non-zero. For @@ -94,7 +94,7 @@ int bfs_exec(struct bfs_exec *execbuf, const struct BFTW *ftwbuf); /** * Finish executing any commands. * - * @param execbuf + * @execbuf * The parsed exec action. * @return 0 on success, -1 if any errors were encountered. */ @@ -2,6 +2,7 @@ // SPDX-License-Identifier: 0BSD #include "expr.h" + #include "alloc.h" #include "ctx.h" #include "diag.h" @@ -10,9 +11,12 @@ #include "list.h" #include "printf.h" #include "xregex.h" + #include <string.h> -struct bfs_expr *bfs_expr_new(struct bfs_ctx *ctx, bfs_eval_fn *eval_fn, size_t argc, char **argv) { +struct bfs_expr *bfs_expr_new(struct bfs_ctx *ctx, bfs_eval_fn *eval_fn, size_t argc, char **argv, enum bfs_kind kind) { + bfs_assert(kind != BFS_PATH); + struct bfs_expr *expr = arena_alloc(&ctx->expr_arena); if (!expr) { return NULL; @@ -22,6 +26,7 @@ struct bfs_expr *bfs_expr_new(struct bfs_ctx *ctx, bfs_eval_fn *eval_fn, size_t expr->eval_fn = eval_fn; expr->argc = argc; expr->argv = argv; + expr->kind = kind; expr->probability = 0.5; SLIST_PREPEND(&ctx->expr_list, expr, freelist); @@ -63,8 +68,7 @@ void bfs_expr_append(struct bfs_expr *expr, struct bfs_expr *child) { } void bfs_expr_extend(struct bfs_expr *expr, struct bfs_exprs *children) { - while (!SLIST_EMPTY(children)) { - struct bfs_expr *child = SLIST_POP(children); + drain_slist (struct bfs_expr, child, children) { bfs_expr_append(expr, child); } } @@ -8,14 +8,38 @@ #ifndef BFS_EXPR_H #define BFS_EXPR_H -#include "prelude.h" #include "color.h" #include "eval.h" #include "stat.h" + #include <sys/types.h> #include <time.h> /** + * Argument/token/expression kinds. + */ +enum bfs_kind { + /** A regular argument. */ + BFS_ARG, + + /** A flag (-H, -L, etc.). */ + BFS_FLAG, + + /** A root path. */ + BFS_PATH, + + /** An option (-follow, -mindepth, etc.). */ + BFS_OPTION, + /** A test (-name, -size, etc.). */ + BFS_TEST, + /** An action (-print, -exec, etc.). */ + BFS_ACTION, + + /** An operator (-and, -or, etc.). */ + BFS_OPERATOR, +}; + +/** * Integer comparison modes. */ enum bfs_int_cmp { @@ -97,6 +121,8 @@ struct bfs_expr { size_t argc; /** The command line arguments comprising this expression. */ char **argv; + /** The kind of expression this is. */ + enum bfs_kind kind; /** The number of files this expression keeps open between evaluations. */ int persistent_fds; @@ -123,7 +149,7 @@ struct bfs_expr { /** Total time spent running this predicate. */ struct timespec elapsed; - /** Auxilliary data for the evaluation function. */ + /** Auxiliary data for the evaluation function. */ union { /** Child expressions. */ struct bfs_exprs children; @@ -207,7 +233,7 @@ struct bfs_ctx; /** * Create a new expression. */ -struct bfs_expr *bfs_expr_new(struct bfs_ctx *ctx, bfs_eval_fn *eval, size_t argc, char **argv); +struct bfs_expr *bfs_expr_new(struct bfs_ctx *ctx, bfs_eval_fn *eval, size_t argc, char **argv, enum bfs_kind kind); /** * @return Whether this type of expression has children. @@ -244,4 +270,10 @@ bool bfs_expr_cmp(const struct bfs_expr *expr, long long n); */ void bfs_expr_clear(struct bfs_expr *expr); +/** + * Iterate over the children of an expression. + */ +#define for_expr(child, expr) \ + for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) + #endif // BFS_EXPR_H diff --git a/src/fsade.c b/src/fsade.c index d56fb07..dfdf125 100644 --- a/src/fsade.c +++ b/src/fsade.c @@ -1,14 +1,16 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "fsade.h" + #include "atomic.h" +#include "bfs.h" #include "bfstd.h" #include "bftw.h" #include "dir.h" #include "dstring.h" #include "sanity.h" + #include <errno.h> #include <fcntl.h> #include <stddef.h> @@ -26,22 +28,31 @@ # include <selinux/selinux.h> #endif -#if BFS_USE_SYS_EXTATTR_H +#if __has_include(<sys/extattr.h>) # include <sys/extattr.h> -#elif BFS_USE_SYS_XATTR_H +# define BFS_USE_EXTATTR true +#elif __has_include(<sys/xattr.h>) # include <sys/xattr.h> +# define BFS_USE_XATTR true +#endif + +#ifndef BFS_USE_EXTATTR +# define BFS_USE_EXTATTR false +#endif +#ifndef BFS_USE_XATTR +# define BFS_USE_XATTR false #endif /** * Many of the APIs used here don't have *at() variants, but we can try to * emulate something similar if /proc/self/fd is available. */ -attr(maybe_unused) +_maybe_unused static const char *fake_at(const struct BFTW *ftwbuf) { static atomic int proc_works = -1; dchar *path = NULL; - if (ftwbuf->at_fd == AT_FDCWD || load(&proc_works, relaxed) == 0) { + if (ftwbuf->at_fd == (int)AT_FDCWD || load(&proc_works, relaxed) == 0) { goto fail; } @@ -70,7 +81,7 @@ fail: return ftwbuf->path; } -attr(maybe_unused) +_maybe_unused static void free_fake_at(const struct BFTW *ftwbuf, const char *path) { if (path != ftwbuf->path) { dstrfree((dchar *)path); @@ -80,7 +91,7 @@ static void free_fake_at(const struct BFTW *ftwbuf, const char *path) { /** * Check if an error was caused by the absence of support or data for a feature. */ -attr(maybe_unused) +_maybe_unused static bool is_absence_error(int error) { // If the OS doesn't support the feature, it's obviously not enabled for // any files @@ -160,7 +171,7 @@ static int bfs_acl_entry(acl_t acl, int which, acl_entry_t *entry) { } /** Unified interface for acl_get_tag_type(). */ -attr(maybe_unused) +_maybe_unused static int bfs_acl_tag_type(acl_entry_t entry, acl_tag_t *tag) { #if BFS_HAS_ACL_GET_TAG_TYPE return acl_get_tag_type(entry, tag); @@ -344,7 +355,7 @@ int bfs_check_capabilities(const struct BFTW *ftwbuf) { #if BFS_CAN_CHECK_XATTRS -#if BFS_USE_SYS_EXTATTR_H +#if BFS_USE_EXTATTR /** Wrapper for extattr_list_{file,link}. */ static ssize_t bfs_extattr_list(const char *path, enum bfs_type type, int namespace) { @@ -390,13 +401,13 @@ static ssize_t bfs_extattr_get(const char *path, enum bfs_type type, int namespa #endif } -#endif // BFS_USE_SYS_EXTATTR_H +#endif // BFS_USE_EXTATTR int bfs_check_xattrs(const struct BFTW *ftwbuf) { const char *path = fake_at(ftwbuf); ssize_t len; -#if BFS_USE_SYS_EXTATTR_H +#if BFS_USE_EXTATTR len = bfs_extattr_list(path, ftwbuf->type, EXTATTR_NAMESPACE_SYSTEM); if (len <= 0) { len = bfs_extattr_list(path, ftwbuf->type, EXTATTR_NAMESPACE_USER); @@ -432,7 +443,7 @@ int bfs_check_xattr_named(const struct BFTW *ftwbuf, const char *name) { const char *path = fake_at(ftwbuf); ssize_t len; -#if BFS_USE_SYS_EXTATTR_H +#if BFS_USE_EXTATTR len = bfs_extattr_get(path, ftwbuf->type, EXTATTR_NAMESPACE_SYSTEM, name); if (len < 0) { len = bfs_extattr_get(path, ftwbuf->type, EXTATTR_NAMESPACE_USER, name); diff --git a/src/fsade.h b/src/fsade.h index eefef9f..fbe02d8 100644 --- a/src/fsade.h +++ b/src/fsade.h @@ -9,22 +9,26 @@ #ifndef BFS_FSADE_H #define BFS_FSADE_H -#include "prelude.h" +#include "bfs.h" #define BFS_CAN_CHECK_ACL (BFS_HAS_ACL_GET_FILE || BFS_HAS_ACL_TRIVIAL) -#define BFS_CAN_CHECK_CAPABILITIES BFS_USE_LIBCAP +#define BFS_CAN_CHECK_CAPABILITIES BFS_WITH_LIBCAP -#define BFS_CAN_CHECK_CONTEXT BFS_USE_LIBSELINUX +#define BFS_CAN_CHECK_CONTEXT BFS_WITH_LIBSELINUX -#define BFS_CAN_CHECK_XATTRS (BFS_USE_SYS_EXTATTR_H || BFS_USE_SYS_XATTR_H) +#if __has_include(<sys/extattr.h>) || __has_include(<sys/xattr.h>) +# define BFS_CAN_CHECK_XATTRS true +#else +# define BFS_CAN_CHECK_XATTRS false +#endif struct BFTW; /** * Check if a file has a non-trivial Access Control List. * - * @param ftwbuf + * @ftwbuf * The file to check. * @return * 1 if it does, 0 if it doesn't, or -1 if an error occurred. @@ -34,7 +38,7 @@ int bfs_check_acl(const struct BFTW *ftwbuf); /** * Check if a file has a non-trivial capability set. * - * @param ftwbuf + * @ftwbuf * The file to check. * @return * 1 if it does, 0 if it doesn't, or -1 if an error occurred. @@ -44,7 +48,7 @@ int bfs_check_capabilities(const struct BFTW *ftwbuf); /** * Check if a file has any extended attributes set. * - * @param ftwbuf + * @ftwbuf * The file to check. * @return * 1 if it does, 0 if it doesn't, or -1 if an error occurred. @@ -54,9 +58,9 @@ int bfs_check_xattrs(const struct BFTW *ftwbuf); /** * Check if a file has an extended attribute with the given name. * - * @param ftwbuf + * @ftwbuf * The file to check. - * @param name + * @name * The name of the xattr to check. * @return * 1 if it does, 0 if it doesn't, or -1 if an error occurred. @@ -66,7 +70,7 @@ int bfs_check_xattr_named(const struct BFTW *ftwbuf, const char *name); /** * Get a file's SELinux context * - * @param ftwbuf + * @ftwbuf * The file to check. * @return * The file's SELinux context, or NULL on failure. @@ -118,24 +118,27 @@ * [1]: https://arxiv.org/abs/2201.02179 */ -#include "prelude.h" #include "ioq.h" + #include "alloc.h" #include "atomic.h" +#include "bfs.h" #include "bfstd.h" #include "bit.h" #include "diag.h" #include "dir.h" #include "stat.h" #include "thread.h" + #include <errno.h> #include <fcntl.h> #include <pthread.h> #include <stdint.h> #include <stdlib.h> #include <sys/stat.h> +#include <unistd.h> -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING # include <liburing.h> #endif @@ -180,8 +183,7 @@ typedef atomic uintptr_t ioq_slot; /** Amount to add for an additional skip. */ #define IOQ_SKIP_ONE (~IOQ_BLOCKED) -// Need room for two flag bits -bfs_static_assert(alignof(struct ioq_ent) >= (1 << 2)); +static_assert(alignof(struct ioq_ent) >= (1 << 2), "struct ioq_ent is underaligned"); /** * An MPMC queue of I/O commands. @@ -201,7 +203,7 @@ struct ioqq { cache_align atomic size_t tail; /** The circular buffer itself. */ - cache_align ioq_slot slots[]; + cache_align ioq_slot slots[]; // _counted_by(slot_mask + 1) }; /** Destroy an I/O command queue. */ @@ -258,17 +260,45 @@ static struct ioqq *ioqq_create(size_t size) { /** Get the monitor associated with a slot. */ static struct ioq_monitor *ioq_slot_monitor(struct ioqq *ioqq, ioq_slot *slot) { - size_t i = slot - ioqq->slots; + uint32_t i = slot - ioqq->slots; + + // Hash the index to de-correlate waiters + // https://nullprogram.com/blog/2018/07/31/ + // https://github.com/skeeto/hash-prospector/issues/19#issuecomment-1120105785 + i ^= i >> 16; + i *= UINT32_C(0x21f0aaad); + i ^= i >> 15; + i *= UINT32_C(0x735a2d97); + i ^= i >> 15; + return &ioqq->monitors[i & ioqq->monitor_mask]; } /** Atomically wait for a slot to change. */ -attr(noinline) +_noinline static uintptr_t ioq_slot_wait(struct ioqq *ioqq, ioq_slot *slot, uintptr_t value) { + uintptr_t ret; + + // Try spinning a few times (with exponential backoff) before blocking + _nounroll + for (int i = 1; i < 1024; i *= 2) { + _nounroll + for (int j = 0; j < i; ++j) { + spin_loop(); + } + + // Check if the slot changed + ret = load(slot, relaxed); + if (ret != value) { + return ret; + } + } + + // Nothing changed, start blocking struct ioq_monitor *monitor = ioq_slot_monitor(ioqq, slot); mutex_lock(&monitor->mutex); - uintptr_t ret = load(slot, relaxed); + ret = load(slot, relaxed); if (ret != value) { goto done; } @@ -293,7 +323,7 @@ done: } /** Wake up any threads waiting on a slot. */ -attr(noinline) +_noinline static void ioq_slot_wake(struct ioqq *ioqq, ioq_slot *slot) { struct ioq_monitor *monitor = ioq_slot_monitor(ioqq, slot); @@ -313,9 +343,11 @@ static void ioq_slot_wake(struct ioqq *ioqq, ioq_slot *slot) { cond_broadcast(&monitor->cond); } -/** Branch-free (slot & IOQ_SKIP) ? ~IOQ_BLOCKED : 0 */ -static uintptr_t ioq_skip_mask(uintptr_t slot) { - return -(slot >> IOQ_SKIP_BIT) << 1; +/** Branch-free ((slot & IOQ_SKIP) ? skip : full) & ~IOQ_BLOCKED */ +static uintptr_t ioq_slot_blend(uintptr_t slot, uintptr_t skip, uintptr_t full) { + uintptr_t mask = -(slot >> IOQ_SKIP_BIT); + uintptr_t ret = (skip & mask) | (full & ~mask); + return ret & ~IOQ_BLOCKED; } /** Push an entry into a slot. */ @@ -323,19 +355,18 @@ static bool ioq_slot_push(struct ioqq *ioqq, ioq_slot *slot, struct ioq_ent *ent uintptr_t prev = load(slot, relaxed); while (true) { - size_t skip_mask = ioq_skip_mask(prev); - size_t full_mask = ~skip_mask & ~IOQ_BLOCKED; - if (prev & full_mask) { + uintptr_t full = ioq_slot_blend(prev, 0, prev); + if (full) { // full(ptr) → wait prev = ioq_slot_wait(ioqq, slot, prev); continue; } // empty → full(ptr) - uintptr_t next = ((uintptr_t)ent >> 1) & full_mask; + uintptr_t next = (uintptr_t)ent >> 1; // skip(1) → empty // skip(n) → skip(n - 1) - next |= (prev - IOQ_SKIP_ONE) & skip_mask; + next = ioq_slot_blend(prev, prev - IOQ_SKIP_ONE, next); if (compare_exchange_weak(slot, &prev, next, release, relaxed)) { break; @@ -353,13 +384,20 @@ static bool ioq_slot_push(struct ioqq *ioqq, ioq_slot *slot, struct ioq_ent *ent static struct ioq_ent *ioq_slot_pop(struct ioqq *ioqq, ioq_slot *slot, bool block) { uintptr_t prev = load(slot, relaxed); while (true) { +#if __has_builtin(__builtin_prefetch) + // Optimistically prefetch the pointer in this slot. If this + // slot is not full, this will prefetch an invalid address, but + // experimentally this is worth it on both Intel (Alder Lake) + // and AMD (Zen 2). + __builtin_prefetch((void *)(prev << 1), 1 /* write */); +#endif + // empty → skip(1) // skip(n) → skip(n + 1) // full(ptr) → full(ptr - 1) uintptr_t next = prev + IOQ_SKIP_ONE; - // skip(n) → ~IOQ_BLOCKED // full(ptr) → 0 - next &= ioq_skip_mask(next); + next = ioq_slot_blend(next, next, 0); if (block && next) { prev = ioq_slot_wait(ioqq, slot, prev); @@ -378,7 +416,7 @@ static struct ioq_ent *ioq_slot_pop(struct ioqq *ioqq, ioq_slot *slot, bool bloc // empty → 0 // skip(n) → 0 // full(ptr) → ptr - prev &= ioq_skip_mask(~prev); + prev = ioq_slot_blend(prev, 0, prev); return (struct ioq_ent *)(prev << 1); } @@ -408,13 +446,6 @@ static void ioqq_push_batch(struct ioqq *ioqq, struct ioq_ent *batch[], size_t s } while (size > 0); } -/** Pop an entry from the queue. */ -static struct ioq_ent *ioqq_pop(struct ioqq *ioqq, bool block) { - size_t i = fetch_add(&ioqq->tail, 1, relaxed); - ioq_slot *slot = &ioqq->slots[i & ioqq->slot_mask]; - return ioq_slot_pop(ioqq, slot, block); -} - /** Pop a batch of entries from the queue. */ static void ioqq_pop_batch(struct ioqq *ioqq, struct ioq_ent *batch[], size_t size, bool block) { size_t mask = ioqq->slot_mask; @@ -430,36 +461,83 @@ static void ioqq_pop_batch(struct ioqq *ioqq, struct ioq_ent *batch[], size_t si #define IOQ_BATCH (FALSE_SHARING_SIZE / sizeof(ioq_slot)) /** - * A batch of entries to send all at once. + * A batch of I/O queue entries. */ struct ioq_batch { - /** The current batch size. */ - size_t size; + /** The start of the batch. */ + size_t head; + /** The end of the batch. */ + size_t tail; /** The array of entries. */ struct ioq_ent *entries[IOQ_BATCH]; }; -/** Send the batch to a queue. */ +/** Reset a batch. */ +static void ioq_batch_reset(struct ioq_batch *batch) { + batch->head = batch->tail = 0; +} + +/** Check if a batch is empty. */ +static bool ioq_batch_empty(const struct ioq_batch *batch) { + return batch->head >= batch->tail; +} + +/** Send a batch to a queue. */ static void ioq_batch_flush(struct ioqq *ioqq, struct ioq_batch *batch) { - if (batch->size > 0) { - ioqq_push_batch(ioqq, batch->entries, batch->size); - batch->size = 0; + if (batch->tail > 0) { + ioqq_push_batch(ioqq, batch->entries, batch->tail); + ioq_batch_reset(batch); } } -/** An an entry to a batch, flushing if necessary. */ +/** Push an entry to a batch, flushing if necessary. */ static void ioq_batch_push(struct ioqq *ioqq, struct ioq_batch *batch, struct ioq_ent *ent) { - if (batch->size >= IOQ_BATCH) { + batch->entries[batch->tail++] = ent; + + if (batch->tail >= IOQ_BATCH) { ioq_batch_flush(ioqq, batch); } +} + +/** Fill a batch from a queue. */ +static bool ioq_batch_fill(struct ioqq *ioqq, struct ioq_batch *batch, bool block) { + ioqq_pop_batch(ioqq, batch->entries, IOQ_BATCH, block); + + ioq_batch_reset(batch); + for (size_t i = 0; i < IOQ_BATCH; ++i) { + struct ioq_ent *ent = batch->entries[i]; + if (ent) { + batch->entries[batch->tail++] = ent; + } + } + + return batch->tail > 0; +} + +/** Pop an entry from a batch, filling it first if necessary. */ +static struct ioq_ent *ioq_batch_pop(struct ioqq *ioqq, struct ioq_batch *batch, bool block) { + if (ioq_batch_empty(batch)) { + // For non-blocking pops, make sure that each ioq_batch_pop() + // corresponds to a single (amortized) increment of ioqq->head. + // Otherwise, we start skipping many slots and batching ends up + // degrading performance. + if (!block && batch->head < IOQ_BATCH) { + ++batch->head; + return NULL; + } + + if (!ioq_batch_fill(ioqq, batch, block)) { + return NULL; + } + } - batch->entries[batch->size++] = ent; + return batch->entries[batch->head++]; } /** Sentinel stop command. */ static struct ioq_ent IOQ_STOP; -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING /** * Supported io_uring operations. */ @@ -477,7 +555,7 @@ struct ioq_thread { /** Pointer back to the I/O queue. */ struct ioq *parent; -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING /** io_uring instance. */ struct io_uring ring; /** Any error that occurred initializing the ring. */ @@ -497,20 +575,25 @@ struct ioq { /** ioq_ent arena. */ struct arena ents; -#if BFS_USE_LIBURING && BFS_USE_STATX +#if BFS_WITH_LIBURING && BFS_USE_STATX /** struct statx arena. */ struct arena xbufs; #endif - /** Pending I/O requests. */ + /** Pending I/O request queue. */ struct ioqq *pending; - /** Ready I/O responses. */ + /** Ready I/O response queue. */ struct ioqq *ready; + /** Pending request batch. */ + struct ioq_batch pending_batch; + /** Ready request batch. */ + struct ioq_batch ready_batch; + /** The number of background threads. */ size_t nthreads; /** The background threads themselves. */ - struct ioq_thread threads[]; + struct ioq_thread threads[] _counted_by(nthreads); }; /** Cancel a request if we need to. */ @@ -531,6 +614,14 @@ static bool ioq_check_cancel(struct ioq *ioq, struct ioq_ent *ent) { /** Dispatch a single request synchronously. */ static void ioq_dispatch_sync(struct ioq *ioq, struct ioq_ent *ent) { switch (ent->op) { + case IOQ_NOP: + if (ent->nop.type == IOQ_NOP_HEAVY) { + // A fast, no-op syscall + getppid(); + } + ent->result = 0; + return; + case IOQ_CLOSE: ent->result = try(xclose(ent->close.fd)); return; @@ -559,7 +650,7 @@ static void ioq_dispatch_sync(struct ioq *ioq, struct ioq_ent *ent) { ent->result = -ENOSYS; } -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING /** io_uring worker state. */ struct ioq_ring_state { @@ -579,23 +670,161 @@ struct ioq_ring_state { struct ioq_batch ready; }; +/** Reap a single CQE. */ +static void ioq_reap_cqe(struct ioq_ring_state *state, struct io_uring_cqe *cqe) { + struct ioq *ioq = state->ioq; + + struct ioq_ent *ent = io_uring_cqe_get_data(cqe); + ent->result = cqe->res; + + if (ent->result < 0) { + goto push; + } + + switch (ent->op) { + case IOQ_OPENDIR: { + int fd = ent->result; + if (ioq_check_cancel(ioq, ent)) { + xclose(fd); + goto push; + } + + struct ioq_opendir *args = &ent->opendir; + ent->result = try(bfs_opendir(args->dir, fd, NULL, args->flags)); + if (ent->result >= 0) { + // TODO: io_uring_prep_getdents() + bfs_polldir(args->dir); + } else { + xclose(fd); + } + + break; + } + +#if BFS_USE_STATX + case IOQ_STAT: { + struct ioq_stat *args = &ent->stat; + ent->result = try(bfs_statx_convert(args->buf, args->xbuf)); + break; + } +#endif + + default: + break; + } + +push: + ioq_batch_push(ioq->ready, &state->ready, ent); +} + +/** Wait for submitted requests to complete. */ +static void ioq_ring_drain(struct ioq_ring_state *state, size_t wait_nr) { + struct ioq *ioq = state->ioq; + struct io_uring *ring = state->ring; + + bfs_assert(wait_nr <= state->submitted); + + while (state->submitted > 0) { + struct io_uring_cqe *cqe; + if (wait_nr > 0) { + io_uring_wait_cqes(ring, &cqe, wait_nr, NULL, NULL); + } + + unsigned int head; + size_t seen = 0; + io_uring_for_each_cqe (ring, head, cqe) { + ioq_reap_cqe(state, cqe); + ++seen; + } + + io_uring_cq_advance(ring, seen); + state->submitted -= seen; + + if (seen >= wait_nr) { + break; + } + wait_nr -= seen; + } + + ioq_batch_flush(ioq->ready, &state->ready); +} + +/** Submit prepped SQEs, and wait for some to complete. */ +static void ioq_ring_submit(struct ioq_ring_state *state) { + struct io_uring *ring = state->ring; + + size_t unreaped = state->prepped + state->submitted; + size_t wait_nr = 0; + + if (state->prepped == 0 && unreaped > 0) { + // If we have no new SQEs, wait for at least one old one to + // complete, to avoid livelock + wait_nr = 1; + } + + if (unreaped > ring->sq.ring_entries) { + // Keep the completion queue below half full + wait_nr = unreaped - ring->sq.ring_entries; + } + + // Submit all prepped SQEs + while (state->prepped > 0) { + int ret = io_uring_submit_and_wait(state->ring, wait_nr); + if (ret <= 0) { + continue; + } + + state->submitted += ret; + state->prepped -= ret; + if (state->prepped > 0) { + // In the unlikely event of a short submission, any SQE + // links will be broken. Wait for all SQEs to complete + // to preserve any ordering requirements. + ioq_ring_drain(state, state->submitted); + wait_nr = 0; + } + } + + // Drain all the CQEs we waited for (and any others that are ready) + ioq_ring_drain(state, wait_nr); +} + +/** Reserve space for a number of SQEs, submitting if necessary. */ +static void ioq_reserve_sqes(struct ioq_ring_state *state, unsigned int count) { + while (io_uring_sq_space_left(state->ring) < count) { + ioq_ring_submit(state); + } +} + +/** Get an SQE, submitting if necessary. */ +static struct io_uring_sqe *ioq_get_sqe(struct ioq_ring_state *state) { + ioq_reserve_sqes(state, 1); + return io_uring_get_sqe(state->ring); +} + /** Dispatch a single request asynchronously. */ static struct io_uring_sqe *ioq_dispatch_async(struct ioq_ring_state *state, struct ioq_ent *ent) { - struct io_uring *ring = state->ring; enum ioq_ring_ops ops = state->ops; struct io_uring_sqe *sqe = NULL; switch (ent->op) { + case IOQ_NOP: + if (ent->nop.type == IOQ_NOP_HEAVY) { + sqe = ioq_get_sqe(state); + io_uring_prep_nop(sqe); + } + return sqe; + case IOQ_CLOSE: if (ops & IOQ_RING_CLOSE) { - sqe = io_uring_get_sqe(ring); + sqe = ioq_get_sqe(state); io_uring_prep_close(sqe, ent->close.fd); } return sqe; case IOQ_OPENDIR: if (ops & IOQ_RING_OPENAT) { - sqe = io_uring_get_sqe(ring); + sqe = ioq_get_sqe(state); struct ioq_opendir *args = &ent->opendir; int flags = O_RDONLY | O_CLOEXEC | O_DIRECTORY; io_uring_prep_openat(sqe, args->dfd, args->path, flags, 0); @@ -605,7 +834,7 @@ static struct io_uring_sqe *ioq_dispatch_async(struct ioq_ring_state *state, str case IOQ_CLOSEDIR: #if BFS_USE_UNWRAPDIR if (ops & IOQ_RING_CLOSE) { - sqe = io_uring_get_sqe(ring); + sqe = ioq_get_sqe(state); io_uring_prep_close(sqe, bfs_unwrapdir(ent->closedir.dir)); } #endif @@ -614,10 +843,10 @@ static struct io_uring_sqe *ioq_dispatch_async(struct ioq_ring_state *state, str case IOQ_STAT: #if BFS_USE_STATX if (ops & IOQ_RING_STATX) { - sqe = io_uring_get_sqe(ring); + sqe = ioq_get_sqe(state); struct ioq_stat *args = &ent->stat; int flags = bfs_statx_flags(args->flags); - unsigned int mask = STATX_BASIC_STATS | STATX_BTIME; + unsigned int mask = bfs_statx_mask(); io_uring_prep_statx(sqe, args->dfd, args->path, flags, mask, args->xbuf); } #endif @@ -630,7 +859,7 @@ static struct io_uring_sqe *ioq_dispatch_async(struct ioq_ring_state *state, str /** Check if ioq_ring_reap() has work to do. */ static bool ioq_ring_empty(struct ioq_ring_state *state) { - return !state->prepped && !state->submitted && !state->ready.size; + return !state->prepped && !state->submitted && ioq_batch_empty(&state->ready); } /** Prep a single SQE. */ @@ -658,163 +887,94 @@ static bool ioq_ring_prep(struct ioq_ring_state *state) { } struct ioq *ioq = state->ioq; - struct io_uring *ring = state->ring; - struct ioq_ent *pending[IOQ_BATCH]; - - while (io_uring_sq_space_left(ring) >= IOQ_BATCH) { - bool block = ioq_ring_empty(state); - ioqq_pop_batch(ioq->pending, pending, IOQ_BATCH, block); - - bool any = false; - for (size_t i = 0; i < IOQ_BATCH; ++i) { - struct ioq_ent *ent = pending[i]; - if (ent == &IOQ_STOP) { - ioqq_push(ioq->pending, &IOQ_STOP); - state->stop = true; - goto done; - } else if (ent) { - ioq_prep_sqe(state, ent); - any = true; - } - } - - if (!any) { - break; - } - } - -done: - return !ioq_ring_empty(state); -} - -/** Reap a single CQE. */ -static void ioq_reap_cqe(struct ioq_ring_state *state, struct io_uring_cqe *cqe) { - struct ioq *ioq = state->ioq; - struct io_uring *ring = state->ring; - - struct ioq_ent *ent = io_uring_cqe_get_data(cqe); - ent->result = cqe->res; - io_uring_cqe_seen(ring, cqe); - --state->submitted; - - if (ent->result < 0) { - goto push; - } - switch (ent->op) { - case IOQ_OPENDIR: { - int fd = ent->result; - if (ioq_check_cancel(ioq, ent)) { - xclose(fd); - goto push; - } - - struct ioq_opendir *args = &ent->opendir; - ent->result = try(bfs_opendir(args->dir, fd, NULL, args->flags)); - if (ent->result >= 0) { - // TODO: io_uring_prep_getdents() - bfs_polldir(args->dir); - } else { - xclose(fd); - } + struct ioq_batch pending; + ioq_batch_reset(&pending); + while (true) { + bool block = ioq_ring_empty(state); + struct ioq_ent *ent = ioq_batch_pop(ioq->pending, &pending, block); + if (ent == &IOQ_STOP) { + ioqq_push(ioq->pending, ent); + state->stop = true; break; - } - -#if BFS_USE_STATX - case IOQ_STAT: { - struct ioq_stat *args = &ent->stat; - ent->result = try(bfs_statx_convert(args->buf, args->xbuf)); + } else if (ent) { + ioq_prep_sqe(state, ent); + } else { break; } -#endif - - default: - break; } -push: - ioq_batch_push(ioq->ready, &state->ready, ent); + bfs_assert(ioq_batch_empty(&pending)); + return !ioq_ring_empty(state); } -/** Reap a batch of CQEs. */ -static void ioq_ring_reap(struct ioq_ring_state *state) { - struct ioq *ioq = state->ioq; - struct io_uring *ring = state->ring; +/** io_uring worker loop. */ +static int ioq_ring_work(struct ioq_thread *thread) { + struct io_uring *ring = &thread->ring; - while (state->prepped) { - int ret = io_uring_submit_and_wait(ring, 1); - if (ret > 0) { - state->prepped -= ret; - state->submitted += ret; +#ifdef IORING_SETUP_R_DISABLED + if (ring->flags & IORING_SETUP_R_DISABLED) { + if (io_uring_enable_rings(ring) != 0) { + return -1; } } +#endif - while (state->submitted) { - struct io_uring_cqe *cqe; - if (io_uring_wait_cqe(ring, &cqe) < 0) { - continue; - } - - ioq_reap_cqe(state, cqe); - } - - ioq_batch_flush(ioq->ready, &state->ready); -} - -/** io_uring worker loop. */ -static void ioq_ring_work(struct ioq_thread *thread) { struct ioq_ring_state state = { .ioq = thread->parent, - .ring = &thread->ring, + .ring = ring, .ops = thread->ring_ops, }; while (ioq_ring_prep(&state)) { - ioq_ring_reap(&state); + ioq_ring_submit(&state); } + + ioq_ring_drain(&state, state.submitted); + return 0; } -#endif // BFS_USE_LIBURING +#endif // BFS_WITH_LIBURING /** Synchronous syscall loop. */ static void ioq_sync_work(struct ioq_thread *thread) { struct ioq *ioq = thread->parent; - bool stop = false; - while (!stop) { - struct ioq_ent *pending[IOQ_BATCH]; - ioqq_pop_batch(ioq->pending, pending, IOQ_BATCH, true); - - struct ioq_batch ready; - ready.size = 0; - - for (size_t i = 0; i < IOQ_BATCH; ++i) { - struct ioq_ent *ent = pending[i]; - if (ent == &IOQ_STOP) { - ioqq_push(ioq->pending, &IOQ_STOP); - stop = true; - break; - } else if (ent) { - if (!ioq_check_cancel(ioq, ent)) { - ioq_dispatch_sync(ioq, ent); - } - ioq_batch_push(ioq->ready, &ready, ent); - } + struct ioq_batch pending, ready; + ioq_batch_reset(&pending); + ioq_batch_reset(&ready); + + while (true) { + if (ioq_batch_empty(&pending)) { + ioq_batch_flush(ioq->ready, &ready); } - ioq_batch_flush(ioq->ready, &ready); + struct ioq_ent *ent = ioq_batch_pop(ioq->pending, &pending, true); + if (ent == &IOQ_STOP) { + ioqq_push(ioq->pending, ent); + break; + } + + if (!ioq_check_cancel(ioq, ent)) { + ioq_dispatch_sync(ioq, ent); + } + ioq_batch_push(ioq->ready, &ready, ent); } + + bfs_assert(ioq_batch_empty(&pending)); + ioq_batch_flush(ioq->ready, &ready); } /** Background thread entry point. */ static void *ioq_work(void *ptr) { struct ioq_thread *thread = ptr; -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING if (thread->ring_err == 0) { - ioq_ring_work(thread); - return NULL; + if (ioq_ring_work(thread) == 0) { + return NULL; + } } #endif @@ -822,9 +982,30 @@ static void *ioq_work(void *ptr) { return NULL; } +#if BFS_WITH_LIBURING +/** Test whether some io_uring setup flags are supported. */ +static bool ioq_ring_probe_flags(struct io_uring_params *params, unsigned int flags) { + unsigned int saved = params->flags; + params->flags |= flags; + + struct io_uring ring; + int ret = io_uring_queue_init_params(2, &ring, params); + if (ret == 0) { + io_uring_queue_exit(&ring); + } + + if (ret == -EINVAL) { + params->flags = saved; + return false; + } + + return true; +} +#endif + /** Initialize io_uring thread state. */ static int ioq_ring_init(struct ioq *ioq, struct ioq_thread *thread) { -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING struct ioq_thread *prev = NULL; if (thread > ioq->threads) { prev = thread - 1; @@ -835,11 +1016,31 @@ static int ioq_ring_init(struct ioq *ioq, struct ioq_thread *thread) { return -1; } - // Share io-wq workers between rings struct io_uring_params params = {0}; + if (prev) { - params.flags |= IORING_SETUP_ATTACH_WQ; + // Share io-wq workers between rings + params.flags = prev->ring.flags | IORING_SETUP_ATTACH_WQ; params.wq_fd = prev->ring.ring_fd; + } else { +#ifdef IORING_SETUP_SUBMIT_ALL + // Don't abort submission just because an inline request fails + ioq_ring_probe_flags(¶ms, IORING_SETUP_SUBMIT_ALL); +#endif + +#ifdef IORING_SETUP_R_DISABLED + // Don't enable the ring yet (needed for SINGLE_ISSUER) + if (ioq_ring_probe_flags(¶ms, IORING_SETUP_R_DISABLED)) { +# ifdef IORING_SETUP_SINGLE_ISSUER + // Allow optimizations assuming only one task submits SQEs + ioq_ring_probe_flags(¶ms, IORING_SETUP_SINGLE_ISSUER); +# endif +# ifdef IORING_SETUP_DEFER_TASKRUN + // Don't interrupt us aggressively with completion events + ioq_ring_probe_flags(¶ms, IORING_SETUP_DEFER_TASKRUN); +# endif + } +#endif } // Use a page for each SQE ring @@ -877,6 +1078,7 @@ static int ioq_ring_init(struct ioq *ioq, struct ioq_thread *thread) { return -1; } +#if BFS_HAS_IO_URING_MAX_WORKERS // Limit the number of io_uring workers unsigned int values[] = { ioq->nthreads, // [IO_WQ_BOUND] @@ -885,12 +1087,14 @@ static int ioq_ring_init(struct ioq *ioq, struct ioq_thread *thread) { io_uring_register_iowq_max_workers(&thread->ring, values); #endif +#endif // BFS_WITH_LIBURING + return 0; } /** Destroy an io_uring. */ static void ioq_ring_exit(struct ioq_thread *thread) { -#if BFS_USE_LIBURING +#if BFS_WITH_LIBURING if (thread->ring_err == 0) { io_uring_queue_exit(&thread->ring); } @@ -898,7 +1102,8 @@ static void ioq_ring_exit(struct ioq_thread *thread) { } /** Create an I/O queue thread. */ -static int ioq_thread_create(struct ioq *ioq, struct ioq_thread *thread) { +static int ioq_thread_create(struct ioq *ioq, size_t i) { + struct ioq_thread *thread = &ioq->threads[i]; thread->parent = ioq; ioq_ring_init(ioq, thread); @@ -908,6 +1113,11 @@ static int ioq_thread_create(struct ioq *ioq, struct ioq_thread *thread) { return -1; } + char name[16]; + if (snprintf(name, sizeof(name), "ioq-%zu", i) >= 0) { + thread_setname(thread->id, name); + } + return 0; } @@ -926,7 +1136,7 @@ struct ioq *ioq_create(size_t depth, size_t nthreads) { ioq->depth = depth; ARENA_INIT(&ioq->ents, struct ioq_ent); -#if BFS_USE_LIBURING && BFS_USE_STATX +#if BFS_WITH_LIBURING && BFS_USE_STATX ARENA_INIT(&ioq->xbufs, struct statx); #endif @@ -942,7 +1152,7 @@ struct ioq *ioq_create(size_t depth, size_t nthreads) { ioq->nthreads = nthreads; for (size_t i = 0; i < nthreads; ++i) { - if (ioq_thread_create(ioq, &ioq->threads[i]) != 0) { + if (ioq_thread_create(ioq, i) != 0) { ioq->nthreads = i; goto fail; } @@ -984,6 +1194,18 @@ static struct ioq_ent *ioq_request(struct ioq *ioq, enum ioq_op op, void *ptr) { return ent; } +int ioq_nop(struct ioq *ioq, enum ioq_nop_type type, void *ptr) { + struct ioq_ent *ent = ioq_request(ioq, IOQ_NOP, ptr); + if (!ent) { + return -1; + } + + ent->nop.type = type; + + ioq_batch_push(ioq->pending, &ioq->pending_batch, ent); + return 0; +} + int ioq_close(struct ioq *ioq, int fd, void *ptr) { struct ioq_ent *ent = ioq_request(ioq, IOQ_CLOSE, ptr); if (!ent) { @@ -992,7 +1214,7 @@ int ioq_close(struct ioq *ioq, int fd, void *ptr) { ent->close.fd = fd; - ioqq_push(ioq->pending, ent); + ioq_batch_push(ioq->pending, &ioq->pending_batch, ent); return 0; } @@ -1008,7 +1230,7 @@ int ioq_opendir(struct ioq *ioq, struct bfs_dir *dir, int dfd, const char *path, args->path = path; args->flags = flags; - ioqq_push(ioq->pending, ent); + ioq_batch_push(ioq->pending, &ioq->pending_batch, ent); return 0; } @@ -1020,7 +1242,7 @@ int ioq_closedir(struct ioq *ioq, struct bfs_dir *dir, void *ptr) { ent->closedir.dir = dir; - ioqq_push(ioq->pending, ent); + ioq_batch_push(ioq->pending, &ioq->pending_batch, ent); return 0; } @@ -1036,7 +1258,7 @@ int ioq_stat(struct ioq *ioq, int dfd, const char *path, enum bfs_stat_flags fla args->flags = flags; args->buf = buf; -#if BFS_USE_LIBURING && BFS_USE_STATX +#if BFS_WITH_LIBURING && BFS_USE_STATX args->xbuf = arena_alloc(&ioq->xbufs); if (!args->xbuf) { ioq_free(ioq, ent); @@ -1044,23 +1266,30 @@ int ioq_stat(struct ioq *ioq, int dfd, const char *path, enum bfs_stat_flags fla } #endif - ioqq_push(ioq->pending, ent); + ioq_batch_push(ioq->pending, &ioq->pending_batch, ent); return 0; } +void ioq_submit(struct ioq *ioq) { + ioq_batch_flush(ioq->pending, &ioq->pending_batch); +} + struct ioq_ent *ioq_pop(struct ioq *ioq, bool block) { + // Don't forget to submit before popping + bfs_assert(ioq_batch_empty(&ioq->pending_batch)); + if (ioq->size == 0) { return NULL; } - return ioqq_pop(ioq->ready, block); + return ioq_batch_pop(ioq->ready, &ioq->ready_batch, block); } void ioq_free(struct ioq *ioq, struct ioq_ent *ent) { bfs_assert(ioq->size > 0); --ioq->size; -#if BFS_USE_LIBURING && BFS_USE_STATX +#if BFS_WITH_LIBURING && BFS_USE_STATX if (ent->op == IOQ_STAT && ent->stat.xbuf) { arena_free(&ioq->xbufs, ent->stat.xbuf); } @@ -1071,7 +1300,8 @@ void ioq_free(struct ioq *ioq, struct ioq_ent *ent) { void ioq_cancel(struct ioq *ioq) { if (!exchange(&ioq->cancel, true, relaxed)) { - ioqq_push(ioq->pending, &IOQ_STOP); + ioq_batch_push(ioq->pending, &ioq->pending_batch, &IOQ_STOP); + ioq_submit(ioq); } } @@ -1091,7 +1321,7 @@ void ioq_destroy(struct ioq *ioq) { ioqq_destroy(ioq->ready); ioqq_destroy(ioq->pending); -#if BFS_USE_LIBURING && BFS_USE_STATX +#if BFS_WITH_LIBURING && BFS_USE_STATX arena_destroy(&ioq->xbufs); #endif arena_destroy(&ioq->ents); @@ -8,9 +8,10 @@ #ifndef BFS_IOQ_H #define BFS_IOQ_H -#include "prelude.h" +#include "bfs.h" #include "dir.h" #include "stat.h" + #include <stddef.h> /** @@ -22,6 +23,8 @@ struct ioq; * I/O queue operations. */ enum ioq_op { + /** ioq_nop(). */ + IOQ_NOP, /** ioq_close(). */ IOQ_CLOSE, /** ioq_opendir(). */ @@ -33,18 +36,21 @@ enum ioq_op { }; /** - * The I/O queue implementation needs two tag bits in each pointer to a struct - * ioq_ent, so we need to ensure at least 4-byte alignment. The natural - * alignment is enough on most architectures, but not m68k, so over-align it. + * ioq_nop() types. */ -#define IOQ_ENT_ALIGN alignas(4) +enum ioq_nop_type { + /** A lightweight nop that avoids syscalls. */ + IOQ_NOP_LIGHT, + /** A heavyweight nop that involves a syscall. */ + IOQ_NOP_HEAVY, +}; /** * An I/O queue entry. */ struct ioq_ent { /** The I/O operation. */ - IOQ_ENT_ALIGN enum ioq_op op; + cache_align enum ioq_op op; /** The return value (on success) or negative error code (on failure). */ int result; @@ -54,6 +60,10 @@ struct ioq_ent { /** Operation-specific arguments. */ union { + /** ioq_nop() args. */ + struct ioq_nop { + enum ioq_nop_type type; + } nop; /** ioq_close() args. */ struct ioq_close { int fd; @@ -83,9 +93,9 @@ struct ioq_ent { /** * Create an I/O queue. * - * @param depth + * @depth * The maximum depth of the queue. - * @param nthreads + * @nthreads * The maximum number of background threads. * @return * The new I/O queue, or NULL on failure. @@ -98,13 +108,27 @@ struct ioq *ioq_create(size_t depth, size_t nthreads); size_t ioq_capacity(const struct ioq *ioq); /** + * A no-op, for benchmarking. + * + * @ioq + * The I/O queue. + * @type + * The type of operation to perform. + * @ptr + * An arbitrary pointer to associate with the request. + * @return + * 0 on success, or -1 on failure. + */ +int ioq_nop(struct ioq *ioq, enum ioq_nop_type type, void *ptr); + +/** * Asynchronous close(). * - * @param ioq + * @ioq * The I/O queue. - * @param fd + * @fd * The fd to close. - * @param ptr + * @ptr * An arbitrary pointer to associate with the request. * @return * 0 on success, or -1 on failure. @@ -114,17 +138,17 @@ int ioq_close(struct ioq *ioq, int fd, void *ptr); /** * Asynchronous bfs_opendir(). * - * @param ioq + * @ioq * The I/O queue. - * @param dir + * @dir * The allocated directory. - * @param dfd + * @dfd * The base file descriptor. - * @param path + * @path * The path to open, relative to dfd. - * @param flags + * @flags * Flags that control which directory entries are listed. - * @param ptr + * @ptr * An arbitrary pointer to associate with the request. * @return * 0 on success, or -1 on failure. @@ -134,11 +158,11 @@ int ioq_opendir(struct ioq *ioq, struct bfs_dir *dir, int dfd, const char *path, /** * Asynchronous bfs_closedir(). * - * @param ioq + * @ioq * The I/O queue. - * @param dir + * @dir * The directory to close. - * @param ptr + * @ptr * An arbitrary pointer to associate with the request. * @return * 0 on success, or -1 on failure. @@ -148,17 +172,17 @@ int ioq_closedir(struct ioq *ioq, struct bfs_dir *dir, void *ptr); /** * Asynchronous bfs_stat(). * - * @param ioq + * @ioq * The I/O queue. - * @param dfd + * @dfd * The base file descriptor. - * @param path + * @path * The path to stat, relative to dfd. - * @param flags + * @flags * Flags that affect the lookup. - * @param buf + * @buf * A place to store the stat buffer, if successful. - * @param ptr + * @ptr * An arbitrary pointer to associate with the request. * @return * 0 on success, or -1 on failure. @@ -166,9 +190,14 @@ int ioq_closedir(struct ioq *ioq, struct bfs_dir *dir, void *ptr); int ioq_stat(struct ioq *ioq, int dfd, const char *path, enum bfs_stat_flags flags, struct bfs_stat *buf, void *ptr); /** + * Submit any buffered requests. + */ +void ioq_submit(struct ioq *ioq); + +/** * Pop a response from the queue. * - * @param ioq + * @ioq * The I/O queue. * @return * The next response, or NULL. @@ -178,9 +207,9 @@ struct ioq_ent *ioq_pop(struct ioq *ioq, bool block); /** * Free a queue entry. * - * @param ioq + * @ioq * The I/O queue. - * @param ent + * @ent * The entry to free. */ void ioq_free(struct ioq *ioq, struct ioq_ent *ent); @@ -83,13 +83,14 @@ #define BFS_LIST_H #include "diag.h" + #include <stddef.h> #include <string.h> /** * Initialize a singly-linked list. * - * @param list + * @list * The list to initialize. * * --- @@ -116,9 +117,9 @@ /** * Initialize a singly-linked list item. * - * @param item + * @item * The item to initialize. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. * * --- @@ -200,7 +201,7 @@ /** * Get the head of a singly-linked list. * - * @param list + * @list * The list in question. * @return * The first item in the list. @@ -227,9 +228,9 @@ /** * Get the tail of a singly-linked list. * - * @param list + * @list * The list in question. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. * @return * The last item in the list. @@ -246,11 +247,11 @@ /** * Check if an item is attached to a singly-linked list. * - * @param list + * @list * The list to check. - * @param item + * @item * The item to check. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. * @return * Whether the item is attached to the list. @@ -267,13 +268,13 @@ /** * Insert an item into a singly-linked list. * - * @param list + * @list * The list to modify. - * @param cursor + * @cursor * A pointer to the item to insert after, e.g. &list->head or list->tail. - * @param item + * @item * The item to insert. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. * @return * A cursor for the next item. @@ -294,11 +295,11 @@ /** * Add an item to the tail of a singly-linked list. * - * @param list + * @list * The list to modify. - * @param item + * @item * The item to append. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. */ #define SLIST_APPEND(list, ...) \ @@ -310,11 +311,11 @@ /** * Add an item to the head of a singly-linked list. * - * @param list + * @list * The list to modify. - * @param item + * @item * The item to prepend. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. */ #define SLIST_PREPEND(list, ...) \ @@ -324,27 +325,43 @@ LIST_VOID_(SLIST_INSERT_(list, &(list)->head, item, __VA_ARGS__)) /** + * Splice a singly-linked list into another. + * + * @dest + * The destination list. + * @cursor + * A pointer to the item to splice after, e.g. &list->head or list->tail. + * @src + * The source list. + */ +#define SLIST_SPLICE(dest, cursor, src) \ + LIST_VOID_(SLIST_SPLICE_((dest), (cursor), (src))) + +#define SLIST_SPLICE_(dest, cursor, src) \ + *src->tail = *cursor, \ + *cursor = src->head, \ + dest->tail = *dest->tail ? src->tail : dest->tail, \ + SLIST_INIT(src) + +/** * Add an entire singly-linked list to the tail of another. * - * @param dest + * @dest * The destination list. - * @param src + * @src * The source list. */ #define SLIST_EXTEND(dest, src) \ - SLIST_EXTEND_((dest), (src)) - -#define SLIST_EXTEND_(dest, src) \ - (src->head ? (*dest->tail = src->head, dest->tail = src->tail, SLIST_INIT(src)) : (void)0) + SLIST_SPLICE(dest, (dest)->tail, src) /** * Remove an item from a singly-linked list. * - * @param list + * @list * The list to modify. - * @param cursor + * @cursor * A pointer to the item to remove, either &list->head or &prev->next. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. * @return * The removed item. @@ -357,10 +374,10 @@ #define SLIST_REMOVE__(list, cursor, next) \ (list->tail = (*cursor)->next ? list->tail : cursor, \ - slist_remove_impl(*cursor, cursor, &(*cursor)->next, sizeof(*cursor))) + slist_remove_(*cursor, cursor, &(*cursor)->next, sizeof(*cursor))) // Helper for SLIST_REMOVE() -static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_t size) { +static inline void *slist_remove_(void *ret, void *cursor, void *next, size_t size) { // ret = *cursor; // *cursor = ret->next; memcpy(cursor, next, size); @@ -372,9 +389,9 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Pop the head off a singly-linked list. * - * @param list + * @list * The list to modify. - * @param node (optional) + * @node (optional) * If specified, use head->node.next rather than head->next. * @return * The popped item, or NULL if the list was empty. @@ -391,13 +408,13 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Loop over the items in a singly-linked list. * - * @param type + * @type * The list item type. - * @param item + * @item * The induction variable name. - * @param list + * @list * The list to iterate. - * @param node (optional) + * @node (optional) * If specified, use head->node.next rather than head->next. */ #define for_slist(type, item, ...) \ @@ -412,9 +429,24 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ item = _next) /** + * Loop over a singly-linked list, popping each item. + * + * @type + * The list item type. + * @item + * The induction variable name. + * @list + * The list to drain. + * @node (optional) + * If specified, use head->node.next rather than head->next. + */ +#define drain_slist(type, item, ...) \ + for (type *item; (item = SLIST_POP(__VA_ARGS__));) + +/** * Initialize a doubly-linked list. * - * @param list + * @list * The list to initialize. */ #define LIST_INIT(list) \ @@ -433,9 +465,9 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Initialize a doubly-linked list item. * - * @param item + * @item * The item to initialize. - * @param node (optional) + * @node (optional) * If specified, use item->node.next rather than item->next. */ #define LIST_ITEM_INIT(...) \ @@ -465,11 +497,11 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Add an item to the tail of a doubly-linked list. * - * @param list + * @list * The list to modify. - * @param item + * @item * The item to append. - * @param node (optional) + * @node (optional) * If specified, use item->node.{prev,next} rather than item->{prev,next}. */ #define LIST_APPEND(list, ...) \ @@ -478,11 +510,11 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Add an item to the head of a doubly-linked list. * - * @param list + * @list * The list to modify. - * @param item + * @item * The item to prepend. - * @param node (optional) + * @node (optional) * If specified, use item->node.{prev,next} rather than item->{prev,next}. */ #define LIST_PREPEND(list, ...) \ @@ -491,11 +523,11 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Check if an item is attached to a doubly-linked list. * - * @param list + * @list * The list to check. - * @param item + * @item * The item to check. - * @param node (optional) + * @node (optional) * If specified, use item->node.{prev,next} rather than item->{prev,next}. * @return * Whether the item is attached to the list. @@ -512,13 +544,13 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Insert into a doubly-linked list after the given cursor. * - * @param list + * @list * The list to modify. - * @param cursor + * @cursor * Insert after this element. - * @param item + * @item * The item to insert. - * @param node (optional) + * @node (optional) * If specified, use item->node.{prev,next} rather than item->{prev,next}. */ #define LIST_INSERT(list, cursor, ...) \ @@ -537,11 +569,11 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Remove an item from a doubly-linked list. * - * @param list + * @list * The list to modify. - * @param item + * @item * The item to remove. - * @param node (optional) + * @node (optional) * If specified, use item->node.{prev,next} rather than item->{prev,next}. */ #define LIST_REMOVE(list, ...) \ @@ -558,13 +590,13 @@ static inline void *slist_remove_impl(void *ret, void *cursor, void *next, size_ /** * Loop over the items in a doubly-linked list. * - * @param type + * @type * The list item type. - * @param item + * @item * The induction variable name. - * @param list + * @list * The list to iterate. - * @param node (optional) + * @node (optional) * If specified, use head->node.next rather than head->next. */ #define for_list(type, item, ...) \ @@ -20,13 +20,14 @@ * - bftw.[ch] (an extended version of nftw(3)) * * - Utilities: + * - prelude.h (feature test macros; automatically included) * - alloc.[ch] (memory allocation) * - atomic.h (atomic operations) * - bar.[ch] (a terminal status bar) * - bit.h (bit manipulation) + * - bfs.h (configuration and fundamental utilities) * - bfstd.[ch] (standard library wrappers/polyfills) * - color.[ch] (for pretty terminal colors) - * - prelude.h (configuration and feature/platform detection) * - diag.[ch] (formats diagnostic messages) * - dir.[ch] (a directory API facade) * - dstring.[ch] (a dynamic string library) @@ -36,21 +37,23 @@ * - mtab.[ch] (parses the system's mount table) * - pwcache.[ch] (a cache for the user/group tables) * - sanity.h (sanitizer interfaces) + * - sighook.[ch] (signal hooks) * - stat.[ch] (wraps stat(), or statx() on Linux) * - thread.h (multi-threading) * - trie.[ch] (a trie set/map implementation) * - typo.[ch] (fuzzy matching for typos) + * - version.c (embeds version information) * - xregex.[ch] (regular expression support) * - xspawn.[ch] (spawns processes) * - xtime.[ch] (date/time handling utilities) */ -#include "prelude.h" #include "bfstd.h" #include "ctx.h" #include "diag.h" #include "eval.h" #include "parse.h" + #include <errno.h> #include <fcntl.h> #include <locale.h> @@ -1,24 +1,28 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "mtab.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "stat.h" #include "trie.h" + #include <errno.h> #include <fcntl.h> #include <stdlib.h> #include <string.h> #include <sys/types.h> -#if !defined(BFS_USE_MNTENT) && BFS_USE_MNTENT_H -# define BFS_USE_MNTENT true -#elif !defined(BFS_USE_MNTINFO) && BSD -# define BFS_USE_MNTINFO true -#elif !defined(BFS_USE_MNTTAB) && __SVR4 -# define BFS_USE_MNTTAB true +#ifndef BFS_USE_MNTENT +# define BFS_USE_MNTENT BFS_HAS_GETMNTENT_1 +#endif +#ifndef BFS_USE_MNTINFO +# define BFS_USE_MNTINFO (!BFS_USE_MNTENT && BFS_HAS_GETMNTINFO) +#endif +#ifndef BFS_USE_MNTTAB +# define BFS_USE_MNTTAB (!BFS_USE_MNTINFO && BFS_HAS_GETMNTENT_2) #endif #if BFS_USE_MNTENT @@ -27,7 +31,6 @@ # include <stdio.h> #elif BFS_USE_MNTINFO # include <sys/mount.h> -# include <sys/ucred.h> #elif BFS_USE_MNTTAB # include <stdio.h> # include <sys/mnttab.h> @@ -66,7 +69,7 @@ struct bfs_mtab { /** * Add an entry to the mount table. */ -attr(maybe_unused) +_maybe_unused static int bfs_mtab_add(struct bfs_mtab *mtab, const char *path, const char *type) { size_t path_size = strlen(path) + 1; size_t type_size = strlen(type) + 1; @@ -148,7 +151,7 @@ struct bfs_mtab *bfs_mtab_parse(void) { bfs_statfs *mntbuf; int size = getmntinfo(&mntbuf, MNT_WAIT); - if (size < 0) { + if (size <= 0) { error = errno; goto fail; } @@ -253,10 +256,7 @@ static int bfs_mtab_fill_types(struct bfs_mtab *mtab) { continue; } - struct trie_leaf *leaf = trie_insert_mem(&mtab->types, &sb.dev, sizeof(sb.dev)); - if (leaf) { - leaf->value = mount->type; - } else { + if (trie_set_mem(&mtab->types, &sb.mnt_id, sizeof(sb.mnt_id), mount->type) != 0) { goto fail; } } @@ -279,9 +279,9 @@ const char *bfs_fstype(const struct bfs_mtab *mtab, const struct bfs_stat *statb } } - const struct trie_leaf *leaf = trie_find_mem(&mtab->types, &statbuf->dev, sizeof(statbuf->dev)); - if (leaf) { - return leaf->value; + const char *type = trie_get_mem(&mtab->types, &statbuf->mnt_id, sizeof(statbuf->mnt_id)); + if (type) { + return type; } else { return "unknown"; } @@ -8,8 +8,6 @@ #ifndef BFS_MTAB_H #define BFS_MTAB_H -#include "prelude.h" - struct bfs_stat; /** @@ -28,9 +26,9 @@ struct bfs_mtab *bfs_mtab_parse(void); /** * Determine the file system type that a file is on. * - * @param mtab + * @mtab * The current mount table. - * @param statbuf + * @statbuf * The bfs_stat() buffer for the file in question. * @return * The type of file system containing this file, "unknown" if not known, @@ -41,9 +39,9 @@ const char *bfs_fstype(const struct bfs_mtab *mtab, const struct bfs_stat *statb /** * Check if a file could be a mount point. * - * @param mtab + * @mtab * The current mount table. - * @param name + * @name * The name of the file to check. * @return * Whether the named file could be a mount point. @@ -25,8 +25,10 @@ * effects are reachable at all, skipping the traversal if not. */ -#include "prelude.h" #include "opt.h" + +#include "bfs.h" +#include "bfstd.h" #include "bftw.h" #include "bit.h" #include "color.h" @@ -38,6 +40,8 @@ #include "expr.h" #include "list.h" #include "pwcache.h" +#include "xspawn.h" + #include <errno.h> #include <limits.h> #include <stdarg.h> @@ -102,42 +106,23 @@ enum pred_type { PRED_TYPES, }; -/** Get the name of a predicate type. */ -static const char *pred_type_name(enum pred_type type) { - switch (type) { - case READABLE_PRED: - return "-readable"; - case WRITABLE_PRED: - return "-writable"; - case EXECUTABLE_PRED: - return "-executable"; - case ACL_PRED: - return "-acl"; - case CAPABLE_PRED: - return "-capable"; - case EMPTY_PRED: - return "-empty"; - case HIDDEN_PRED: - return "-hidden"; - case NOGROUP_PRED: - return "-nogroup"; - case NOUSER_PRED: - return "-nouser"; - case SPARSE_PRED: - return "-sparse"; - case XATTR_PRED: - return "-xattr"; - - case PRED_TYPES: - break; - } - - bfs_bug("Unknown predicate %d", (int)type); - return "???"; -} +/** Predicate type names. */ +static const char *const pred_names[] = { + [READABLE_PRED] = "-readable", + [WRITABLE_PRED] = "-writable", + [EXECUTABLE_PRED] = "-executable", + [ACL_PRED] = "-acl", + [CAPABLE_PRED] = "-capable", + [EMPTY_PRED] = "-empty", + [HIDDEN_PRED] = "-hidden", + [NOGROUP_PRED] = "-nogroup", + [NOUSER_PRED] = "-nouser", + [SPARSE_PRED] = "-sparse", + [XATTR_PRED] = "-xattr", +}; /** - * A contrained integer range. + * A constrained integer range. */ struct df_range { /** The (inclusive) minimum value. */ @@ -192,11 +177,17 @@ static void constrain_min(struct df_range *range, long long value) { range->min = max_value(range->min, value); } -/** Contrain the maximum of a range. */ +/** Constrain the maximum of a range. */ static void constrain_max(struct df_range *range, long long value) { range->max = min_value(range->max, value); } +/** Constrain a range to a single value. */ +static void constrain_range(struct df_range *range, long long value) { + constrain_min(range, value); + constrain_max(range, value); +} + /** Remove a single value from a range. */ static void range_remove(struct df_range *range, long long value) { if (range->min == value) { @@ -242,29 +233,15 @@ enum range_type { RANGE_TYPES, }; -/** Get the name of a range type. */ -static const char *range_type_name(enum range_type type) { - switch (type) { - case DEPTH_RANGE: - return "-depth"; - case GID_RANGE: - return "-gid"; - case INUM_RANGE: - return "-inum"; - case LINKS_RANGE: - return "-links"; - case SIZE_RANGE: - return "-size"; - case UID_RANGE: - return "-uid"; - - case RANGE_TYPES: - break; - } - - bfs_bug("Unknown range %d", (int)type); - return "???"; -} +/** Range type names. */ +static const char *const range_names[] = { + [DEPTH_RANGE] = "-depth", + [GID_RANGE] = "-gid", + [INUM_RANGE] = "-inum", + [LINKS_RANGE] = "-links", + [SIZE_RANGE] = "-size", + [UID_RANGE] = "-uid", +}; /** * The data flow analysis domain. @@ -333,27 +310,27 @@ static void df_init_top(struct df_domain *value) { /** Check for the top element. */ static bool df_is_top(const struct df_domain *value) { - for (int i = 0; i < PRED_TYPES; ++i) { - if (value->preds[i] != PRED_TOP) { - return false; - } - } + for (int i = 0; i < PRED_TYPES; ++i) { + if (value->preds[i] != PRED_TOP) { + return false; + } + } - for (int i = 0; i < RANGE_TYPES; ++i) { - if (!range_is_top(&value->ranges[i])) { - return false; - } - } + for (int i = 0; i < RANGE_TYPES; ++i) { + if (!range_is_top(&value->ranges[i])) { + return false; + } + } - if (value->types != ~0U) { - return false; - } + if (value->types != ~0U) { + return false; + } - if (value->xtypes != ~0U) { - return false; - } + if (value->xtypes != ~0U) { + return false; + } - return true; + return true; } /** Compute the union of two fact sets. */ @@ -397,7 +374,7 @@ struct bfs_opt { }; /** Log an optimization. */ -attr(printf(2, 3)) +_printf(2, 3) static bool opt_debug(struct bfs_opt *opt, const char *format, ...) { if (bfs_debug_prefix(opt->ctx, DEBUG_OPT)) { for (int i = 0; i < opt->depth; ++i) { @@ -415,7 +392,7 @@ static bool opt_debug(struct bfs_opt *opt, const char *format, ...) { } /** Log a recursive call. */ -attr(printf(2, 3)) +_printf(2, 3) static bool opt_enter(struct bfs_opt *opt, const char *format, ...) { int depth = opt->depth; if (depth > 0) { @@ -435,7 +412,7 @@ static bool opt_enter(struct bfs_opt *opt, const char *format, ...) { } /** Log a recursive return. */ -attr(printf(2, 3)) +_printf(2, 3) static bool opt_leave(struct bfs_opt *opt, const char *format, ...) { bool debug = false; int depth = opt->depth; @@ -459,7 +436,7 @@ static bool opt_leave(struct bfs_opt *opt, const char *format, ...) { } /** Log a shallow visit. */ -attr(printf(2, 3)) +_printf(2, 3) static bool opt_visit(struct bfs_opt *opt, const char *format, ...) { int depth = opt->depth; if (depth > 0) { @@ -479,7 +456,7 @@ static bool opt_visit(struct bfs_opt *opt, const char *format, ...) { } /** Log the deletion of an expression. */ -attr(printf(2, 3)) +_printf(2, 3) static bool opt_delete(struct bfs_opt *opt, const char *format, ...) { int depth = opt->depth; @@ -503,7 +480,7 @@ typedef bool dump_fn(struct bfs_opt *opt, const char *format, ...); /** Print a df_pred. */ static void pred_dump(dump_fn *dump, struct bfs_opt *opt, const struct df_domain *value, enum pred_type type) { - dump(opt, "${blu}%s${rs}: ", pred_type_name(type)); + dump(opt, "${blu}%s${rs}: ", pred_names[type]); FILE *file = opt->ctx->cerr->file; switch (value->preds[type]) { @@ -524,7 +501,7 @@ static void pred_dump(dump_fn *dump, struct bfs_opt *opt, const struct df_domain /** Print a df_range. */ static void range_dump(dump_fn *dump, struct bfs_opt *opt, const struct df_domain *value, enum range_type type) { - dump(opt, "${blu}%s${rs}: ", range_type_name(type)); + dump(opt, "${blu}%s${rs}: ", range_names[type]); FILE *file = opt->ctx->cerr->file; const struct df_range *range = &value->ranges[type]; @@ -641,22 +618,26 @@ static bool is_const(const struct bfs_expr *expr) { } /** Warn about an expression. */ -attr(printf(3, 4)) -static void opt_warning(const struct bfs_opt *opt, const struct bfs_expr *expr, const char *format, ...) { +_printf(3, 4) +static bool opt_warning(const struct bfs_opt *opt, const struct bfs_expr *expr, const char *format, ...) { if (!opt->warn) { - return; + return false; } if (bfs_expr_is_parent(expr) || is_const(expr)) { - return; + return false; } - if (bfs_expr_warning(opt->ctx, expr)) { - va_list args; - va_start(args, format); - bfs_vwarning(opt->ctx, format, args); - va_end(args); + if (!bfs_expr_warning(opt->ctx, expr)) { + return false; } + + va_list args; + va_start(args, format); + bfs_vwarning(opt->ctx, format, args); + va_end(args); + + return true; } /** Remove and return an expression's children. */ @@ -756,9 +737,7 @@ static struct bfs_expr *visit_and(struct bfs_opt *opt, struct bfs_expr *expr, co df_init_bottom(&opt->after_false); struct bfs_opt nested = *opt; - while (!SLIST_EMPTY(&children)) { - struct bfs_expr *child = SLIST_POP(&children); - + drain_slist (struct bfs_expr, child, &children) { if (SLIST_EMPTY(&children)) { nested.ignore_result = opt->ignore_result; } else { @@ -790,9 +769,7 @@ static struct bfs_expr *visit_or(struct bfs_opt *opt, struct bfs_expr *expr, con df_init_bottom(&opt->after_true); struct bfs_opt nested = *opt; - while (!SLIST_EMPTY(&children)) { - struct bfs_expr *child = SLIST_POP(&children); - + drain_slist (struct bfs_expr, child, &children) { if (SLIST_EMPTY(&children)) { nested.ignore_result = opt->ignore_result; } else { @@ -822,9 +799,7 @@ static struct bfs_expr *visit_comma(struct bfs_opt *opt, struct bfs_expr *expr, struct bfs_opt nested = *opt; - while (!SLIST_EMPTY(&children)) { - struct bfs_expr *child = SLIST_POP(&children); - + drain_slist (struct bfs_expr, child, &children) { if (SLIST_EMPTY(&children)) { nested.ignore_result = opt->ignore_result; } else { @@ -1088,7 +1063,7 @@ static struct bfs_expr *annotate_and(struct bfs_opt *opt, struct bfs_expr *expr, expr->cost = 0.0; expr->probability = 1.0; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { expr->pure &= child->pure; expr->always_true &= child->always_true; expr->always_false |= child->always_false; @@ -1107,7 +1082,7 @@ static struct bfs_expr *annotate_or(struct bfs_opt *opt, struct bfs_expr *expr, expr->cost = 0.0; float false_prob = 1.0; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { expr->pure &= child->pure; expr->always_true |= child->always_true; expr->always_false &= child->always_false; @@ -1124,7 +1099,7 @@ static struct bfs_expr *annotate_comma(struct bfs_opt *opt, struct bfs_expr *exp expr->pure = true; expr->cost = 0.0; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { expr->pure &= child->pure; expr->always_true = child->always_true; expr->always_false = child->always_false; @@ -1360,7 +1335,7 @@ static struct bfs_expr *opt_const(struct bfs_opt *opt, bool value) { static bfs_eval_fn *const fns[] = {eval_false, eval_true}; static char *fake_args[] = {"-false", "-true"}; - struct bfs_expr *expr = bfs_expr_new(opt->ctx, fns[value], 1, &fake_args[value]); + struct bfs_expr *expr = bfs_expr_new(opt->ctx, fns[value], 1, &fake_args[value], BFS_TEST); return visit_shallow(opt, expr, &annotate); } @@ -1374,7 +1349,7 @@ static struct bfs_expr *negate_expr(struct bfs_opt *opt, struct bfs_expr *expr, return opt_const(opt, true); } - struct bfs_expr *ret = bfs_expr_new(opt->ctx, eval_not, 1, argv); + struct bfs_expr *ret = bfs_expr_new(opt->ctx, eval_not, 1, argv, BFS_OPERATOR); if (!ret) { return NULL; } @@ -1403,8 +1378,7 @@ static struct bfs_expr *sink_not_andor(struct bfs_opt *opt, struct bfs_expr *exp struct bfs_exprs children; foster_children(expr, &children); - struct bfs_expr *child; - while ((child = SLIST_POP(&children))) { + drain_slist (struct bfs_expr, child, &children) { opt_enter(opt, "%pe\n", child); child = negate_expr(opt, child, argv); @@ -1422,18 +1396,16 @@ static struct bfs_expr *sink_not_andor(struct bfs_opt *opt, struct bfs_expr *exp /** Sink a negation into a comma expression. */ static struct bfs_expr *sink_not_comma(struct bfs_opt *opt, struct bfs_expr *expr) { - bfs_assert(expr->eval_fn == eval_comma); - - opt_enter(opt, "%pe\n", expr); - char **argv = expr->argv; expr = only_child(expr); + opt_enter(opt, "%pe\n", expr); + + bfs_assert(expr->eval_fn == eval_comma); struct bfs_exprs children; foster_children(expr, &children); - struct bfs_expr *child; - while ((child = SLIST_POP(&children))) { + drain_slist (struct bfs_expr, child, &children) { if (SLIST_EMPTY(&children)) { opt_enter(opt, "%pe\n", child); opt_debug(opt, "sink\n"); @@ -1461,7 +1433,6 @@ static struct bfs_expr *canonicalize_not(struct bfs_opt *opt, struct bfs_expr *e if (rhs->eval_fn == eval_not) { opt_debug(opt, "double negation\n"); - rhs = only_child(expr); return only_child(rhs); } else if (rhs->eval_fn == eval_and || rhs->eval_fn == eval_or) { return sink_not_andor(opt, expr); @@ -1483,8 +1454,7 @@ static struct bfs_expr *canonicalize_assoc(struct bfs_opt *opt, struct bfs_expr struct bfs_exprs flat; SLIST_INIT(&flat); - struct bfs_expr *child; - while ((child = SLIST_POP(&children))) { + drain_slist (struct bfs_expr, child, &children) { if (child->eval_fn == expr->eval_fn) { struct bfs_expr *head = SLIST_HEAD(&child->children); struct bfs_expr *tail = SLIST_TAIL(&child->children); @@ -1592,8 +1562,7 @@ static struct bfs_expr *reorder_andor(struct bfs_opt *opt, struct bfs_expr *expr struct bfs_exprs pure; SLIST_INIT(&pure); - struct bfs_expr *child; - while ((child = SLIST_POP(&children))) { + drain_slist (struct bfs_expr, child, &children) { if (child->pure) { SLIST_APPEND(&pure, child); } else { @@ -1634,8 +1603,7 @@ static void data_flow_icmp(struct bfs_opt *opt, const struct bfs_expr *expr, enu switch (expr->int_cmp) { case BFS_INT_EQUAL: - constrain_min(true_range, value); - constrain_max(true_range, value); + constrain_range(true_range, value); range_remove(false_range, value); break; @@ -1655,14 +1623,31 @@ static void data_flow_icmp(struct bfs_opt *opt, const struct bfs_expr *expr, enu /** Transfer function for -{execut,read,writ}able. */ static struct bfs_expr *data_flow_access(struct bfs_opt *opt, struct bfs_expr *expr, const struct visitor *visitor) { - if (expr->num & R_OK) { + switch (expr->num) { + case R_OK: data_flow_pred(opt, READABLE_PRED, true); - } - if (expr->num & W_OK) { + break; + case W_OK: data_flow_pred(opt, WRITABLE_PRED, true); - } - if (expr->num & X_OK) { + break; + case X_OK: data_flow_pred(opt, EXECUTABLE_PRED, true); + break; + default: + bfs_bug("Unknown access() mode %lld", expr->num); + break; + } + + return expr; +} + +/** Transfer function for -empty. */ +static struct bfs_expr *data_flow_empty(struct bfs_opt *opt, struct bfs_expr *expr, const struct visitor *visitor) { + opt->after_true.types &= (1 << BFS_REG) | (1 << BFS_DIR); + + if (opt->before.types == (1 << BFS_REG)) { + constrain_range(&opt->after_true.ranges[SIZE_RANGE], 0); + range_remove(&opt->after_false.ranges[SIZE_RANGE], 0); } return expr; @@ -1675,7 +1660,7 @@ static struct bfs_expr *data_flow_gid(struct bfs_opt *opt, struct bfs_expr *expr gid_t gid = range->min; bool nogroup = !bfs_getgrgid(opt->ctx->groups, gid); if (errno == 0) { - data_flow_pred(opt, NOGROUP_PRED, nogroup); + constrain_pred(&opt->after_true.preds[NOGROUP_PRED], nogroup); } } @@ -1706,11 +1691,16 @@ static struct bfs_expr *data_flow_links(struct bfs_opt *opt, struct bfs_expr *ex return expr; } +/** Transfer function for -lname. */ +static struct bfs_expr *data_flow_lname(struct bfs_opt *opt, struct bfs_expr *expr, const struct visitor *visitor) { + opt->after_true.types &= 1 << BFS_LNK; + return expr; +} + /** Transfer function for -samefile. */ static struct bfs_expr *data_flow_samefile(struct bfs_opt *opt, struct bfs_expr *expr, const struct visitor *visitor) { struct df_range *true_range = &opt->after_true.ranges[INUM_RANGE]; - constrain_min(true_range, expr->ino); - constrain_max(true_range, expr->ino); + constrain_range(true_range, expr->ino); struct df_range *false_range = &opt->after_false.ranges[INUM_RANGE]; range_remove(false_range, expr->ino); @@ -1744,7 +1734,7 @@ static struct bfs_expr *data_flow_uid(struct bfs_opt *opt, struct bfs_expr *expr uid_t uid = range->min; bool nouser = !bfs_getpwuid(opt->ctx->users, uid); if (errno == 0) { - data_flow_pred(opt, NOUSER_PRED, nouser); + constrain_pred(&opt->after_true.preds[NOUSER_PRED], nouser); } } @@ -1790,7 +1780,7 @@ static struct bfs_expr *data_flow_leave(struct bfs_opt *opt, struct bfs_expr *ex if (df_is_bottom(&opt->after_false)) { if (!expr->pure) { expr->always_true = true; - expr->probability = 0.0; + expr->probability = 1.0; } else if (expr->eval_fn != eval_true) { opt_warning(opt, expr, "This expression is always true.\n\n"); opt_debug(opt, "pure, always true\n"); @@ -1818,12 +1808,45 @@ static struct bfs_expr *data_flow_leave(struct bfs_opt *opt, struct bfs_expr *ex return visit_leave(opt, expr, visitor); } -/** Data flow visitor function. */ -static struct bfs_expr *data_flow_visit(struct bfs_opt *opt, struct bfs_expr *expr, const struct visitor *visitor) { - if (opt->ignore_result && expr->pure) { +/** Ignore an expression (and possibly warn/prompt). */ +static struct bfs_expr *opt_ignore(struct bfs_opt *opt, struct bfs_expr *expr, bool delete) { + if (delete) { + opt_delete(opt, "%pe [ignored result]\n", expr); + } else { opt_debug(opt, "ignored result\n"); - opt_warning(opt, expr, "The result of this expression is ignored.\n\n"); + } + + if (expr->kind != BFS_TEST) { + goto done; + } + + if (!opt_warning(opt, expr, "The result of this expression is ignored.\n")) { + goto done; + } + + struct bfs_ctx *ctx = opt->ctx; + if (ctx->interactive && ctx->dangerous) { + bfs_warning(ctx, "Do you want to continue? "); + if (ynprompt() <= 0) { + errno = 0; + return NULL; + } + } + + fprintf(stderr, "\n"); + +done: + if (!delete && expr->pure) { + // If we're not deleting the expression entirely, replace it with -false expr = opt_const(opt, false); + } + return expr; +} + +/** Data flow visitor function. */ +static struct bfs_expr *data_flow_visit(struct bfs_opt *opt, struct bfs_expr *expr, const struct visitor *visitor) { + if (opt->ignore_result) { + expr = opt_ignore(opt, expr, false); if (!expr) { return NULL; } @@ -1893,9 +1916,11 @@ static const struct visitor data_flow = { .leave = data_flow_leave, .table = (const struct visitor_table[]) { {eval_access, data_flow_access}, + {eval_empty, data_flow_empty}, {eval_gid, data_flow_gid}, {eval_inum, data_flow_inum}, {eval_links, data_flow_links}, + {eval_lname, data_flow_lname}, {eval_samefile, data_flow_samefile}, {eval_size, data_flow_size}, {eval_type, data_flow_type}, @@ -1919,7 +1944,7 @@ static struct bfs_expr *simplify_not(struct bfs_opt *opt, struct bfs_expr *expr, static struct bfs_expr *lift_andor_not(struct bfs_opt *opt, struct bfs_expr *expr) { // Only lift negations if it would reduce the number of (-not) expressions size_t added = 0, removed = 0; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (child->eval_fn == eval_not) { ++removed; } else { @@ -1944,8 +1969,7 @@ static struct bfs_expr *lift_andor_not(struct bfs_opt *opt, struct bfs_expr *exp struct bfs_exprs children; foster_children(expr, &children); - struct bfs_expr *child; - while ((child = SLIST_POP(&children))) { + drain_slist (struct bfs_expr, child, &children) { opt_enter(opt, "%pe\n", child); child = negate_expr(opt, child, &fake_not_arg); @@ -1958,6 +1982,10 @@ static struct bfs_expr *lift_andor_not(struct bfs_opt *opt, struct bfs_expr *exp } expr = visit_shallow(opt, expr, &annotate); + if (!expr) { + return NULL; + } + return negate_expr(opt, expr, &fake_not_arg); } @@ -1968,7 +1996,7 @@ static struct bfs_expr *first_ignorable(struct bfs_opt *opt, struct bfs_expr *ex } struct bfs_expr *ret = NULL; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { if (!child->pure) { ret = NULL; } else if (!ret) { @@ -1987,16 +2015,15 @@ static struct bfs_expr *simplify_and(struct bfs_opt *opt, struct bfs_expr *expr, struct bfs_exprs children; foster_children(expr, &children); - while (!SLIST_EMPTY(&children)) { - struct bfs_expr *child = SLIST_POP(&children); - + drain_slist (struct bfs_expr, child, &children) { if (child == ignorable) { ignore = true; } if (ignore) { - opt_delete(opt, "%pe [ignored result]\n", child); - opt_warning(opt, child, "The result of this expression is ignored.\n\n"); + if (!opt_ignore(opt, child, true)) { + return NULL; + } continue; } @@ -2009,8 +2036,8 @@ static struct bfs_expr *simplify_and(struct bfs_opt *opt, struct bfs_expr *expr, bfs_expr_append(expr, child); if (child->always_false) { - while ((child = SLIST_POP(&children))) { - opt_delete(opt, "%pe [short-circuit]\n", child); + drain_slist (struct bfs_expr, dead, &children) { + opt_delete(opt, "%pe [short-circuit]\n", dead); } } } @@ -2035,16 +2062,15 @@ static struct bfs_expr *simplify_or(struct bfs_opt *opt, struct bfs_expr *expr, struct bfs_exprs children; foster_children(expr, &children); - while (!SLIST_EMPTY(&children)) { - struct bfs_expr *child = SLIST_POP(&children); - + drain_slist (struct bfs_expr, child, &children) { if (child == ignorable) { ignore = true; } if (ignore) { - opt_delete(opt, "%pe [ignored result]\n", child); - opt_warning(opt, child, "The result of this expression is ignored.\n\n"); + if (!opt_ignore(opt, child, true)) { + return NULL; + } continue; } @@ -2057,8 +2083,8 @@ static struct bfs_expr *simplify_or(struct bfs_opt *opt, struct bfs_expr *expr, bfs_expr_append(expr, child); if (child->always_true) { - while ((child = SLIST_POP(&children))) { - opt_delete(opt, "%pe [short-circuit]\n", child); + drain_slist (struct bfs_expr, dead, &children) { + opt_delete(opt, "%pe [short-circuit]\n", dead); } } } @@ -2080,12 +2106,11 @@ static struct bfs_expr *simplify_comma(struct bfs_opt *opt, struct bfs_expr *exp struct bfs_exprs children; foster_children(expr, &children); - while (!SLIST_EMPTY(&children)) { - struct bfs_expr *child = SLIST_POP(&children); - + drain_slist (struct bfs_expr, child, &children) { if (opt->level >= 2 && child->pure && !SLIST_EMPTY(&children)) { - opt_delete(opt, "%pe [ignored result]\n", child); - opt_warning(opt, child, "The result of this expression is ignored.\n\n"); + if (!opt_ignore(opt, child, true)) { + return NULL; + } continue; } @@ -2136,6 +2161,8 @@ static struct bfs_expr *optimize(struct bfs_opt *opt, struct bfs_expr *expr) { }; struct df_domain impure; + df_init_top(&opt->after_true); + df_init_top(&opt->after_false); for (int i = 0; i < 3; ++i) { struct bfs_opt nested = *opt; @@ -2149,9 +2176,11 @@ static struct bfs_expr *optimize(struct bfs_opt *opt, struct bfs_expr *expr) { continue; } + const struct visitor *visitor = passes[j].visitor; + // Skip reordering the first time through the passes, to // make warnings more understandable - if (passes[j].visitor == &reorder) { + if (visitor == &reorder) { if (i == 0) { continue; } else { @@ -2159,10 +2188,15 @@ static struct bfs_expr *optimize(struct bfs_opt *opt, struct bfs_expr *expr) { } } - expr = visit(&nested, expr, passes[j].visitor); + expr = visit(&nested, expr, visitor); if (!expr) { return NULL; } + + if (visitor == &data_flow) { + opt->after_true = nested.after_true; + opt->after_false = nested.after_false; + } } opt_leave(&nested, NULL); @@ -2176,17 +2210,20 @@ static struct bfs_expr *optimize(struct bfs_opt *opt, struct bfs_expr *expr) { return expr; } -/** Estimate the odds of an expression calling stat(). */ -static float expr_stat_odds(struct bfs_expr *expr) { - if (expr->calls_stat) { +/** An expression predicate. */ +typedef bool expr_pred(const struct bfs_expr *expr); + +/** Estimate the odds that a matching expression will be evaluated. */ +static float estimate_odds(const struct bfs_expr *expr, expr_pred *pred) { + if (pred(expr)) { return 1.0; } - float nostat_odds = 1.0; + float nonmatch_odds = 1.0; float reached_odds = 1.0; - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { - float child_odds = expr_stat_odds(child); - nostat_odds *= 1.0 - reached_odds * child_odds; + for_expr (child, expr) { + float child_odds = estimate_odds(child, pred); + nonmatch_odds *= 1.0 - reached_odds * child_odds; if (expr->eval_fn == eval_and) { reached_odds *= child->probability; @@ -2195,7 +2232,12 @@ static float expr_stat_odds(struct bfs_expr *expr) { } } - return 1.0 - nostat_odds; + return 1.0 - nonmatch_odds; +} + +/** Whether an expression calls stat(). */ +static bool calls_stat(const struct bfs_expr *expr) { + return expr->calls_stat; } /** Estimate the odds of calling stat(). */ @@ -2204,15 +2246,20 @@ static float estimate_stat_odds(struct bfs_ctx *ctx) { return 1.0; } - float nostat_odds = 1.0 - expr_stat_odds(ctx->exclude); + float nostat_odds = 1.0 - estimate_odds(ctx->exclude, calls_stat); float reached_odds = 1.0 - ctx->exclude->probability; - float expr_odds = expr_stat_odds(ctx->expr); + float expr_odds = estimate_odds(ctx->expr, calls_stat); nostat_odds *= 1.0 - reached_odds * expr_odds; return 1.0 - nostat_odds; } +/** Matches -(exec|ok) ... \; */ +static bool single_exec(const struct bfs_expr *expr) { + return expr->eval_fn == eval_exec && !(expr->exec->flags & BFS_EXEC_MULTI); +} + int bfs_optimize(struct bfs_ctx *ctx) { bfs_ctx_dump(ctx, DEBUG_OPT); @@ -2291,6 +2338,17 @@ int bfs_optimize(struct bfs_ctx *ctx) { opt_leave(&opt, "eager stat cost: ${ylw}%g${rs}\n", eager_cost); } +#ifndef POSIX_SPAWN_SETRLIMIT + // If bfs_spawn_setrlimit() would force us to use fork() over + // posix_spawn(), the extra cost may outweigh the benefit of a + // higher RLIMIT_NOFILE + float single_exec_odds = estimate_odds(ctx->expr, single_exec); + if (single_exec_odds >= 0.5) { + opt_enter(&opt, "single ${blu}-exec${rs} odds: ${ylw}%g${rs}\n", single_exec_odds); + ctx->raise_nofile = false; + opt_leave(&opt, "not raising RLIMIT_NOFILE\n"); + } +#endif } opt_leave(&opt, NULL); @@ -13,7 +13,7 @@ struct bfs_ctx; /** * Apply optimizations to the command line. * - * @param ctx + * @ctx * The bfs context to optimize. * @return * 0 if successful, -1 on error. diff --git a/src/parse.c b/src/parse.c index a1155c0..5ec4c0e 100644 --- a/src/parse.c +++ b/src/parse.c @@ -8,9 +8,10 @@ * flags like always-true options, and skipping over paths wherever they appear. */ -#include "prelude.h" #include "parse.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "bftw.h" #include "color.h" @@ -31,6 +32,7 @@ #include "xregex.h" #include "xspawn.h" #include "xtime.h" + #include <errno.h> #include <fcntl.h> #include <fnmatch.h> @@ -78,14 +80,10 @@ struct bfs_parser { /** Whether stdout is a terminal. */ bool stdout_tty; - /** Whether this session is interactive (stdin and stderr are each a terminal). */ - bool interactive; /** Whether -color or -nocolor has been passed. */ enum use_color use_color; /** Whether a -print action is implied. */ bool implicit_print; - /** Whether the default root "." should be used. */ - bool implicit_root; /** Whether the expression has started. */ bool expr_started; /** Whether an information option like -help or -version was passed. */ @@ -95,44 +93,26 @@ struct bfs_parser { /** The last non-path argument. */ char **last_arg; - /** A "-depth"-type argument, if any. */ - char **depth_arg; - /** A "-limit" argument, if any. */ - char **limit_arg; - /** A "-prune" argument, if any. */ - char **prune_arg; - /** A "-mount" argument, if any. */ - char **mount_arg; - /** An "-xdev" argument, if any. */ - char **xdev_arg; - /** A "-files0-from -" argument, if any. */ - char **files0_stdin_arg; - /** An "-ok"-type expression, if any. */ - const struct bfs_expr *ok_expr; + /** A "-depth"-type expression, if any. */ + const struct bfs_expr *depth_expr; + /** A "-limit" expression, if any. */ + const struct bfs_expr *limit_expr; + /** A "-prune" expression, if any. */ + const struct bfs_expr *prune_expr; + /** A "-mount" expression, if any. */ + const struct bfs_expr *mount_expr; + /** An "-xdev" expression, if any. */ + const struct bfs_expr *xdev_expr; + /** A "-files0-from" expression, if any. */ + const struct bfs_expr *files0_expr; + /** An expression that consumes stdin, if any. */ + const struct bfs_expr *stdin_expr; /** The current time (maybe modified by -daystart). */ struct timespec now; }; /** - * Possible token types. - */ -enum token_type { - /** A flag. */ - T_FLAG, - /** A root path. */ - T_PATH, - /** An option. */ - T_OPTION, - /** A test. */ - T_TEST, - /** An action. */ - T_ACTION, - /** An operator. */ - T_OPERATOR, -}; - -/** * Print a low-level error message during parsing. */ static void parse_perror(const struct bfs_parser *parser, const char *str) { @@ -158,9 +138,8 @@ static void highlight_args(const struct bfs_ctx *ctx, char **argv, size_t argc, /** * Print an error message during parsing. */ -attr(printf(2, 3)) +_printf(2, 3) static void parse_error(const struct bfs_parser *parser, const char *format, ...) { - int error = errno; const struct bfs_ctx *ctx = parser->ctx; bool highlight[ctx->argc]; @@ -170,7 +149,6 @@ static void parse_error(const struct bfs_parser *parser, const char *format, ... va_list args; va_start(args, format); - errno = error; bfs_verror(parser->ctx, format, args); va_end(args); } @@ -178,9 +156,8 @@ static void parse_error(const struct bfs_parser *parser, const char *format, ... /** * Print an error about some command line arguments. */ -attr(printf(4, 5)) +_printf(4, 5) static void parse_argv_error(const struct bfs_parser *parser, char **argv, size_t argc, const char *format, ...) { - int error = errno; const struct bfs_ctx *ctx = parser->ctx; bool highlight[ctx->argc]; @@ -190,7 +167,6 @@ static void parse_argv_error(const struct bfs_parser *parser, char **argv, size_ va_list args; va_start(args, format); - errno = error; bfs_verror(ctx, format, args); va_end(args); } @@ -198,20 +174,18 @@ static void parse_argv_error(const struct bfs_parser *parser, char **argv, size_ /** * Print an error about conflicting command line arguments. */ -attr(printf(6, 7)) -static void parse_conflict_error(const struct bfs_parser *parser, char **argv1, size_t argc1, char **argv2, size_t argc2, const char *format, ...) { - int error = errno; +_printf(4, 5) +static void parse_conflict_error(const struct bfs_parser *parser, const struct bfs_expr *expr1, const struct bfs_expr *expr2, const char *format, ...) { const struct bfs_ctx *ctx = parser->ctx; bool highlight[ctx->argc]; init_highlight(ctx, highlight); - highlight_args(ctx, argv1, argc1, highlight); - highlight_args(ctx, argv2, argc2, highlight); + highlight_args(ctx, expr1->argv, expr1->argc, highlight); + highlight_args(ctx, expr2->argv, expr2->argc, highlight); bfs_argv_error(ctx, highlight); va_list args; va_start(args, format); - errno = error; bfs_verror(ctx, format, args); va_end(args); } @@ -219,16 +193,14 @@ static void parse_conflict_error(const struct bfs_parser *parser, char **argv1, /** * Print an error about an expression. */ -attr(printf(3, 4)) +_printf(3, 4) static void parse_expr_error(const struct bfs_parser *parser, const struct bfs_expr *expr, const char *format, ...) { - int error = errno; const struct bfs_ctx *ctx = parser->ctx; bfs_expr_error(ctx, expr); va_list args; va_start(args, format); - errno = error; bfs_verror(ctx, format, args); va_end(args); } @@ -236,9 +208,8 @@ static void parse_expr_error(const struct bfs_parser *parser, const struct bfs_e /** * Print a warning message during parsing. */ -attr(printf(2, 3)) +_printf(2, 3) static bool parse_warning(const struct bfs_parser *parser, const char *format, ...) { - int error = errno; const struct bfs_ctx *ctx = parser->ctx; bool highlight[ctx->argc]; @@ -250,7 +221,6 @@ static bool parse_warning(const struct bfs_parser *parser, const char *format, . va_list args; va_start(args, format); - errno = error; bool ret = bfs_vwarning(parser->ctx, format, args); va_end(args); return ret; @@ -259,22 +229,20 @@ static bool parse_warning(const struct bfs_parser *parser, const char *format, . /** * Print a warning about conflicting command line arguments. */ -attr(printf(6, 7)) -static bool parse_conflict_warning(const struct bfs_parser *parser, char **argv1, size_t argc1, char **argv2, size_t argc2, const char *format, ...) { - int error = errno; +_printf(4, 5) +static bool parse_conflict_warning(const struct bfs_parser *parser, const struct bfs_expr *expr1, const struct bfs_expr *expr2, const char *format, ...) { const struct bfs_ctx *ctx = parser->ctx; bool highlight[ctx->argc]; init_highlight(ctx, highlight); - highlight_args(ctx, argv1, argc1, highlight); - highlight_args(ctx, argv2, argc2, highlight); + highlight_args(ctx, expr1->argv, expr1->argc, highlight); + highlight_args(ctx, expr2->argv, expr2->argc, highlight); if (!bfs_argv_warning(ctx, highlight)) { return false; } va_list args; va_start(args, format); - errno = error; bool ret = bfs_vwarning(ctx, format, args); va_end(args); return ret; @@ -283,9 +251,8 @@ static bool parse_conflict_warning(const struct bfs_parser *parser, char **argv1 /** * Print a warning about an expression. */ -attr(printf(3, 4)) +_printf(3, 4) static bool parse_expr_warning(const struct bfs_parser *parser, const struct bfs_expr *expr, const char *format, ...) { - int error = errno; const struct bfs_ctx *ctx = parser->ctx; if (!bfs_expr_warning(ctx, expr)) { @@ -294,17 +261,31 @@ static bool parse_expr_warning(const struct bfs_parser *parser, const struct bfs va_list args; va_start(args, format); - errno = error; bool ret = bfs_vwarning(ctx, format, args); va_end(args); return ret; } /** + * Report an error if stdin is already consumed, then consume it. + */ +static bool consume_stdin(struct bfs_parser *parser, const struct bfs_expr *expr) { + if (parser->stdin_expr) { + parse_conflict_error(parser, parser->stdin_expr, expr, + "%pX and %pX can't both use standard input.\n", + parser->stdin_expr, expr); + return false; + } + + parser->stdin_expr = expr; + return true; +} + +/** * Allocate a new expression. */ -static struct bfs_expr *parse_new_expr(const struct bfs_parser *parser, bfs_eval_fn *eval_fn, size_t argc, char **argv) { - struct bfs_expr *expr = bfs_expr_new(parser->ctx, eval_fn, argc, argv); +static struct bfs_expr *parse_new_expr(const struct bfs_parser *parser, bfs_eval_fn *eval_fn, size_t argc, char **argv, enum bfs_kind kind) { + struct bfs_expr *expr = bfs_expr_new(parser->ctx, eval_fn, argc, argv, kind); if (!expr) { parse_perror(parser, "bfs_expr_new()"); } @@ -315,7 +296,7 @@ static struct bfs_expr *parse_new_expr(const struct bfs_parser *parser, bfs_eval * Create a new unary expression. */ static struct bfs_expr *new_unary_expr(const struct bfs_parser *parser, bfs_eval_fn *eval_fn, struct bfs_expr *rhs, char **argv) { - struct bfs_expr *expr = parse_new_expr(parser, eval_fn, 1, argv); + struct bfs_expr *expr = parse_new_expr(parser, eval_fn, 1, argv, BFS_OPERATOR); if (!expr) { return NULL; } @@ -329,7 +310,7 @@ static struct bfs_expr *new_unary_expr(const struct bfs_parser *parser, bfs_eval * Create a new binary expression. */ static struct bfs_expr *new_binary_expr(const struct bfs_parser *parser, bfs_eval_fn *eval_fn, struct bfs_expr *lhs, struct bfs_expr *rhs, char **argv) { - struct bfs_expr *expr = parse_new_expr(parser, eval_fn, 1, argv); + struct bfs_expr *expr = parse_new_expr(parser, eval_fn, 1, argv, BFS_OPERATOR); if (!expr) { return NULL; } @@ -381,7 +362,7 @@ static int expr_open(struct bfs_parser *parser, struct bfs_expr *expr, const cha return 0; fail: - parse_expr_error(parser, expr, "%m.\n"); + parse_expr_error(parser, expr, "%s.\n", errstr()); if (cfile) { cfclose(cfile); } else if (file) { @@ -401,7 +382,7 @@ static int stat_arg(const struct bfs_parser *parser, char **arg, struct bfs_stat int ret = bfs_stat(AT_FDCWD, *arg, flags, sb); if (ret != 0) { - parse_argv_error(parser, arg, 1, "%m.\n"); + parse_argv_error(parser, arg, 1, "%s.\n", errstr()); } return ret; } @@ -414,15 +395,20 @@ static struct bfs_expr *parse_expr(struct bfs_parser *parser); /** * Advance by a single token. */ -static char **parser_advance(struct bfs_parser *parser, enum token_type type, size_t argc) { - if (type != T_FLAG && type != T_PATH) { +static char **parser_advance(struct bfs_parser *parser, enum bfs_kind kind, size_t argc) { + struct bfs_ctx *ctx = parser->ctx; + + if (kind != BFS_FLAG && kind != BFS_PATH) { parser->expr_started = true; } - if (type != T_PATH) { + if (kind != BFS_PATH) { parser->last_arg = parser->argv; } + size_t i = parser->argv - ctx->argv; + ctx->kinds[i] = kind; + char **argv = parser->argv; parser->argv += argc; return argv; @@ -446,7 +432,6 @@ static int parse_root(struct bfs_parser *parser, const char *path) { return -1; } - parser->implicit_root = false; return 0; } @@ -465,7 +450,7 @@ static int skip_paths(struct bfs_parser *parser) { // find uses -- to separate flags from the rest // of the command line. We allow mixing flags // and paths/predicates, so we just ignore --. - parser_advance(parser, T_FLAG, 1); + parser_advance(parser, BFS_FLAG, 1); continue; } if (strcmp(arg, "-") != 0) { @@ -497,7 +482,7 @@ static int skip_paths(struct bfs_parser *parser) { return -1; } - parser_advance(parser, T_PATH, 1); + parser_advance(parser, BFS_PATH, 1); } } @@ -517,20 +502,14 @@ enum int_flags { * Parse an integer. */ static const char *parse_int(const struct bfs_parser *parser, char **arg, const char *str, void *result, enum int_flags flags) { - // strtoll() skips leading spaces, but we want to reject them - if (xisspace(str[0])) { - goto bad; - } - int base = flags & IF_BASE_MASK; if (base == 0) { base = 10; } char *endptr; - errno = 0; - long long value = strtoll(str, &endptr, base); - if (errno != 0) { + long long value; + if (xstrtoll(str, &endptr, base, &value) != 0) { if (errno == ERANGE) { goto range; } else { @@ -538,13 +517,6 @@ static const char *parse_int(const struct bfs_parser *parser, char **arg, const } } - // https://github.com/llvm/llvm-project/issues/64946 - sanitize_init(&endptr); - - if (endptr == str) { - goto bad; - } - if (!(flags & IF_PARTIAL_OK) && *endptr != '\0') { goto bad; } @@ -641,8 +613,8 @@ static bool looks_like_icmp(const char *str) { * Parse a single flag. */ static struct bfs_expr *parse_flag(struct bfs_parser *parser, size_t argc) { - char **argv = parser_advance(parser, T_FLAG, argc); - return parse_new_expr(parser, eval_true, argc, argv); + char **argv = parser_advance(parser, BFS_FLAG, argc); + return parse_new_expr(parser, eval_true, argc, argv, BFS_FLAG); } /** @@ -657,9 +629,11 @@ static struct bfs_expr *parse_nullary_flag(struct bfs_parser *parser) { */ static struct bfs_expr *parse_unary_flag(struct bfs_parser *parser) { const char *arg = parser->argv[0]; + char flag = arg[strlen(arg) - 1]; + const char *value = parser->argv[1]; if (!value) { - parse_error(parser, "${cyn}%s${rs} needs a value.\n", arg); + parse_error(parser, "${cyn}-%c${rs} needs a value.\n", flag); return NULL; } @@ -667,11 +641,34 @@ static struct bfs_expr *parse_unary_flag(struct bfs_parser *parser) { } /** + * Parse a prefix flag like -O3, -j8, etc. + */ +static struct bfs_expr *parse_prefix_flag(struct bfs_parser *parser, char flag, bool allow_separate, const char **value) { + const char *arg = parser->argv[0]; + + const char *suffix = strchr(arg, flag) + 1; + if (*suffix) { + *value = suffix; + return parse_nullary_flag(parser); + } + + suffix = parser->argv[1]; + if (allow_separate && suffix) { + *value = suffix; + } else { + parse_error(parser, "${cyn}-%c${rs} needs a value.\n", flag); + return NULL; + } + + return parse_unary_flag(parser); +} + +/** * Parse a single option. */ static struct bfs_expr *parse_option(struct bfs_parser *parser, size_t argc) { - char **argv = parser_advance(parser, T_OPTION, argc); - return parse_new_expr(parser, eval_true, argc, argv); + char **argv = parser_advance(parser, BFS_OPTION, argc); + return parse_new_expr(parser, eval_true, argc, argv, BFS_OPTION); } /** @@ -699,8 +696,8 @@ static struct bfs_expr *parse_unary_option(struct bfs_parser *parser) { * Parse a single test. */ static struct bfs_expr *parse_test(struct bfs_parser *parser, bfs_eval_fn *eval_fn, size_t argc) { - char **argv = parser_advance(parser, T_TEST, argc); - return parse_new_expr(parser, eval_fn, argc, argv); + char **argv = parser_advance(parser, BFS_TEST, argc); + return parse_new_expr(parser, eval_fn, argc, argv, BFS_TEST); } /** @@ -728,7 +725,7 @@ static struct bfs_expr *parse_unary_test(struct bfs_parser *parser, bfs_eval_fn * Parse a single action. */ static struct bfs_expr *parse_action(struct bfs_parser *parser, bfs_eval_fn *eval_fn, size_t argc) { - char **argv = parser_advance(parser, T_ACTION, argc); + char **argv = parser_advance(parser, BFS_ACTION, argc); if (parser->excluding) { parse_argv_error(parser, argv, argc, "This action is not supported within ${red}-exclude${rs}.\n"); @@ -739,7 +736,7 @@ static struct bfs_expr *parse_action(struct bfs_parser *parser, bfs_eval_fn *eva parser->implicit_print = false; } - return parse_new_expr(parser, eval_fn, argc, argv); + return parse_new_expr(parser, eval_fn, argc, argv, BFS_ACTION); } /** @@ -811,7 +808,8 @@ static bool parse_debug_flag(const char *flag, size_t len, const char *expected) static struct bfs_expr *parse_debug(struct bfs_parser *parser, int arg1, int arg2) { struct bfs_ctx *ctx = parser->ctx; - struct bfs_expr *expr = parse_unary_flag(parser); + const char *flags; + struct bfs_expr *expr = parse_prefix_flag(parser, 'D', true, &flags); if (!expr) { cfprintf(ctx->cerr, "\n"); debug_help(ctx->cerr); @@ -820,7 +818,7 @@ static struct bfs_expr *parse_debug(struct bfs_parser *parser, int arg1, int arg bool unrecognized = false; - for (const char *flag = expr->argv[1], *next; flag; flag = next) { + for (const char *flag = flags, *next; flag; flag = next) { size_t len = strcspn(flag, ","); if (flag[len]) { next = flag + len + 1; @@ -868,21 +866,22 @@ static struct bfs_expr *parse_debug(struct bfs_parser *parser, int arg1, int arg * Parse -On. */ static struct bfs_expr *parse_optlevel(struct bfs_parser *parser, int arg1, int arg2) { - struct bfs_expr *expr = parse_nullary_flag(parser); + const char *arg; + struct bfs_expr *expr = parse_prefix_flag(parser, 'O', false, &arg); if (!expr) { return NULL; } int *optlevel = &parser->ctx->optlevel; - if (strcmp(expr->argv[0], "-Ofast") == 0) { + if (strcmp(arg, "fast") == 0) { *optlevel = 4; - } else if (!parse_int(parser, expr->argv, expr->argv[0] + 2, optlevel, IF_INT | IF_UNSIGNED)) { + } else if (!parse_int(parser, expr->argv, arg, optlevel, IF_INT | IF_UNSIGNED)) { return NULL; } if (*optlevel > 4) { - parse_expr_warning(parser, expr, "${cyn}-O${bld}%s${rs} is the same as ${cyn}-O${bld}4${rs}.\n\n", expr->argv[0] + 2); + parse_expr_warning(parser, expr, "${cyn}-O${bld}%s${rs} is the same as ${cyn}-O${bld}4${rs}.\n\n", arg); } return expr; @@ -996,16 +995,16 @@ static struct bfs_expr *parse_time(struct bfs_parser *parser, int field, int arg switch (*tail) { case 'w': time *= 7; - fallthru; + _fallthrough; case 'd': time *= 24; - fallthru; + _fallthrough; case 'h': time *= 60; - fallthru; + _fallthrough; case 'm': time *= 60; - fallthru; + _fallthrough; case 's': break; default: @@ -1117,7 +1116,7 @@ static struct bfs_expr *parse_fnmatch(const struct bfs_parser *parser, struct bf // strcmp() can be much faster than fnmatch() since it doesn't have to // parse the pattern, so special-case patterns with no wildcards. // - // https://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_13_01 + // https://pubs.opengroup.org/onlinepubs/9799919799/utilities/V3_chap02.html#tag_19_14_01 expr->literal = strcspn(expr->pattern, "?*\\[") == len; return expr; @@ -1176,18 +1175,33 @@ static struct bfs_expr *parse_daystart(struct bfs_parser *parser, int arg1, int * Parse -delete. */ static struct bfs_expr *parse_delete(struct bfs_parser *parser, int arg1, int arg2) { - parser->ctx->flags |= BFTW_POST_ORDER; - parser->depth_arg = parser->argv; - return parse_nullary_action(parser, eval_delete); + struct bfs_expr *expr = parse_nullary_action(parser, eval_delete); + if (!expr) { + return NULL; + } + + struct bfs_ctx *ctx = parser->ctx; + ctx->flags |= BFTW_POST_ORDER; + ctx->dangerous = true; + + parser->depth_expr = expr; + return expr; } /** * Parse -d. */ -static struct bfs_expr *parse_depth(struct bfs_parser *parser, int arg1, int arg2) { +static struct bfs_expr *parse_depth(struct bfs_parser *parser, int flag, int arg2) { + struct bfs_expr *expr = flag + ? parse_nullary_flag(parser) + : parse_nullary_option(parser); + if (!expr) { + return NULL; + } + parser->ctx->flags |= BFTW_POST_ORDER; - parser->depth_arg = parser->argv; - return parse_nullary_flag(parser); + parser->depth_expr = expr; + return expr; } /** @@ -1233,11 +1247,48 @@ static struct bfs_expr *parse_empty(struct bfs_parser *parser, int arg1, int arg return expr; } +/** Check for unsafe relative paths in $PATH. */ +static const char *unsafe_path(const struct bfs_exec *execbuf) { + if (!(execbuf->flags & BFS_EXEC_CHDIR)) { + // Not -execdir or -okdir + return NULL; + } + + const char *exe = execbuf->tmpl_argv[0]; + if (strchr(exe, '/')) { + // No $PATH lookups for /foo or foo/bar + return NULL; + } + + if (strstr(exe, "{}")) { + // Substituted paths always contain a / + return NULL; + } + + const char *path = getenv("PATH"); + while (path) { + if (path[0] != '/') { + // Relative $PATH component! + return path; + } + + path = strchr(path, ':'); + if (path) { + ++path; + } + } + + // No relative components in $PATH + return NULL; +} + /** * Parse -exec(dir)?/-ok(dir)?. */ static struct bfs_expr *parse_exec(struct bfs_parser *parser, int flags, int arg2) { - struct bfs_exec *execbuf = bfs_exec_parse(parser->ctx, parser->argv, flags); + struct bfs_ctx *ctx = parser->ctx; + + struct bfs_exec *execbuf = bfs_exec_parse(ctx, parser->argv, flags); if (!execbuf) { return NULL; } @@ -1253,29 +1304,21 @@ static struct bfs_expr *parse_exec(struct bfs_parser *parser, int flags, int arg // For pipe() in bfs_spawn() expr->ephemeral_fds = 2; - if (execbuf->flags & BFS_EXEC_CHDIR) { - // Check for relative paths in $PATH - const char *path = getenv("PATH"); - while (path) { - if (*path != '/') { - size_t len = strcspn(path, ":"); - char *comp = strndup(path, len); - if (comp) { - parse_expr_error(parser, expr, - "This action would be unsafe, since ${bld}$$PATH${rs} contains the relative path ${bld}%pq${rs}\n", comp); - free(comp); - } else { - parse_perror(parser, "strndup()"); - } - return NULL; - } - - path = strchr(path, ':'); - if (path) { - ++path; - } + const char *unsafe = unsafe_path(execbuf); + if (unsafe) { + size_t len = strcspn(unsafe, ":"); + char *comp = strndup(unsafe, len); + if (comp) { + parse_expr_error(parser, expr, + "This action would be unsafe, since ${bld}$$PATH${rs} contains the relative path ${bld}%pq${rs}\n", comp); + free(comp); + } else { + parse_perror(parser, "strndup()"); } + return NULL; + } + if (execbuf->flags & BFS_EXEC_CHDIR) { // To dup() the parent directory if (execbuf->flags & BFS_EXEC_MULTI) { ++expr->persistent_fds; @@ -1285,7 +1328,11 @@ static struct bfs_expr *parse_exec(struct bfs_parser *parser, int flags, int arg } if (execbuf->flags & BFS_EXEC_CONFIRM) { - parser->ok_expr = expr; + if (!consume_stdin(parser, expr)) { + return NULL; + } + } else { + ctx->dangerous = true; } return expr; @@ -1314,11 +1361,17 @@ static struct bfs_expr *parse_exit(struct bfs_parser *parser, int arg1, int arg2 * Parse -f PATH. */ static struct bfs_expr *parse_f(struct bfs_parser *parser, int arg1, int arg2) { + struct bfs_ctx *ctx = parser->ctx; + struct bfs_expr *expr = parse_unary_flag(parser); if (!expr) { return NULL; } + // Mark the path as a path, not a regular argument + size_t i = expr->argv - ctx->argv; + ctx->kinds[i + 1] = BFS_PATH; + if (parse_root(parser, expr->argv[1]) != 0) { return NULL; } @@ -1335,50 +1388,14 @@ static struct bfs_expr *parse_files0_from(struct bfs_parser *parser, int arg1, i return NULL; } - const char *from = expr->argv[1]; - - FILE *file; - if (strcmp(from, "-") == 0) { - file = stdin; - } else { - file = xfopen(from, O_RDONLY | O_CLOEXEC); - } - if (!file) { - parse_expr_error(parser, expr, "%m.\n"); - return NULL; - } - - while (true) { - char *path = xgetdelim(file, '\0'); - if (!path) { - if (errno) { - goto fail; - } else { - break; - } - } - - int ret = parse_root(parser, path); - free(path); - if (ret != 0) { - goto fail; - } - } - - if (file == stdin) { - parser->files0_stdin_arg = expr->argv; - } else { - fclose(file); - } - - parser->implicit_root = false; + // For compatibility with GNU find, + // + // bfs -files0-from a -files0-from b + // + // should *only* use b, not a. So stash the expression here and only + // process the last one at the end of parsing. + parser->files0_expr = expr; return expr; - -fail: - if (file != stdin) { - fclose(file); - } - return NULL; } /** @@ -1509,7 +1526,7 @@ static struct bfs_expr *parse_fstype(struct bfs_parser *parser, int arg1, int ar } if (!bfs_ctx_mtab(parser->ctx)) { - parse_expr_error(parser, expr, "Couldn't parse the mount table: %m.\n"); + parse_expr_error(parser, expr, "Couldn't parse the mount table: %s.\n", errstr()); return NULL; } @@ -1534,7 +1551,7 @@ static struct bfs_expr *parse_group(struct bfs_parser *parser, int arg1, int arg return NULL; } } else if (errno) { - parse_expr_error(parser, expr, "%m.\n"); + parse_expr_error(parser, expr, "%s.\n", errstr()); return NULL; } else { parse_expr_error(parser, expr, "No such group.\n"); @@ -1577,7 +1594,7 @@ static struct bfs_expr *parse_user(struct bfs_parser *parser, int arg1, int arg2 return NULL; } } else if (errno) { - parse_expr_error(parser, expr, "%m.\n"); + parse_expr_error(parser, expr, "%s.\n", errstr()); return NULL; } else { parse_expr_error(parser, expr, "No such user.\n"); @@ -1613,13 +1630,14 @@ static struct bfs_expr *parse_inum(struct bfs_parser *parser, int arg1, int arg2 * Parse -j<n>. */ static struct bfs_expr *parse_jobs(struct bfs_parser *parser, int arg1, int arg2) { - struct bfs_expr *expr = parse_nullary_flag(parser); + const char *arg; + struct bfs_expr *expr = parse_prefix_flag(parser, 'j', false, &arg); if (!expr) { return NULL; } unsigned int n; - if (!parse_int(parser, expr->argv, expr->argv[0] + 2, &n, IF_INT | IF_UNSIGNED)) { + if (!parse_int(parser, expr->argv, arg, &n, IF_INT | IF_UNSIGNED)) { return NULL; } @@ -1647,11 +1665,11 @@ static struct bfs_expr *parse_limit(struct bfs_parser *parser, int arg1, int arg } if (expr->num <= 0) { - parse_expr_error(parser, expr, "The ${blu}%s${rs} must be at least ${bld}1${rs}.\n", expr->argv[0]); + parse_expr_error(parser, expr, "The %pX must be at least ${bld}1${rs}.\n", expr); return NULL; } - parser->limit_arg = expr->argv; + parser->limit_expr = expr; return expr; } @@ -1684,11 +1702,8 @@ static struct bfs_expr *parse_mount(struct bfs_parser *parser, int arg1, int arg return NULL; } - parse_expr_warning(parser, expr, "In the future, ${blu}%s${rs} will skip mount points entirely, unlike\n", expr->argv[0]); - bfs_warning(parser->ctx, "${blu}-xdev${rs}, due to http://austingroupbugs.net/view.php?id=1133.\n\n"); - - parser->ctx->flags |= BFTW_PRUNE_MOUNTS; - parser->mount_arg = expr->argv; + parser->ctx->flags |= BFTW_SKIP_MOUNTS; + parser->mount_expr = expr; return expr; } @@ -1737,7 +1752,7 @@ static int parse_reftime(const struct bfs_parser *parser, struct bfs_expr *expr) if (xgetdate(expr->argv[1], &expr->reftime) == 0) { return 0; } else if (errno != EINVAL) { - parse_expr_error(parser, expr, "%m.\n"); + parse_expr_error(parser, expr, "%s.\n", errstr()); return -1; } @@ -1831,6 +1846,14 @@ static struct bfs_expr *parse_newerxy(struct bfs_parser *parser, int arg1, int a } /** + * Parse -noerror. + */ +static struct bfs_expr *parse_noerror(struct bfs_parser *parser, int arg1, int arg2) { + parser->ctx->ignore_errors = true; + return parse_nullary_option(parser); +} + +/** * Parse -nogroup. */ static struct bfs_expr *parse_nogroup(struct bfs_parser *parser, int arg1, int arg2) { @@ -1846,7 +1869,7 @@ static struct bfs_expr *parse_nogroup(struct bfs_parser *parser, int arg1, int a * Parse -nohidden. */ static struct bfs_expr *parse_nohidden(struct bfs_parser *parser, int arg1, int arg2) { - struct bfs_expr *hidden = parse_new_expr(parser, eval_hidden, 1, &fake_hidden_arg); + struct bfs_expr *hidden = parse_new_expr(parser, eval_hidden, 1, &fake_hidden_arg, BFS_TEST); if (!hidden) { return NULL; } @@ -1859,9 +1882,15 @@ static struct bfs_expr *parse_nohidden(struct bfs_parser *parser, int arg1, int * Parse -noleaf. */ static struct bfs_expr *parse_noleaf(struct bfs_parser *parser, int arg1, int arg2) { - parse_warning(parser, "${ex}%s${rs} does not apply the optimization that ${blu}%s${rs} inhibits.\n\n", - BFS_COMMAND, parser->argv[0]); - return parse_nullary_option(parser); + struct bfs_expr *expr = parse_nullary_option(parser); + if (!expr) { + return NULL; + } + + parse_expr_warning(parser, expr, + "${ex}%s${rs} does not apply the optimization that %px inhibits.\n\n", + BFS_COMMAND, expr); + return expr; } /** @@ -1894,6 +1923,8 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct return 0; } + mode_t umask = parser->ctx->umask; + expr->file_mode = 0; expr->dir_mode = 0; @@ -1914,7 +1945,7 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct // // PERMCOPY : "u" | "g" | "o" - // Parser machine parser + // State machine state enum { MODE_CLAUSE, MODE_WHO, @@ -1922,7 +1953,7 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct MODE_ACTION_APPLY, MODE_OP, MODE_PERM, - } mparser = MODE_CLAUSE; + } state = MODE_CLAUSE; enum { MODE_PLUS, @@ -1931,16 +1962,18 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct } op uninit(MODE_EQUALS); mode_t who uninit(0); + mode_t mask uninit(0); mode_t file_change uninit(0); mode_t dir_change uninit(0); const char *i = mode; while (true) { - switch (mparser) { + switch (state) { case MODE_CLAUSE: who = 0; - mparser = MODE_WHO; - fallthru; + mask = 0777; + state = MODE_WHO; + _fallthrough; case MODE_WHO: switch (*i) { @@ -1957,7 +1990,7 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct who |= 0777; break; default: - mparser = MODE_ACTION; + state = MODE_ACTION; continue; } break; @@ -1967,7 +2000,7 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct case MODE_EQUALS: expr->file_mode &= ~who; expr->dir_mode &= ~who; - fallthru; + _fallthrough; case MODE_PLUS: expr->file_mode |= file_change; expr->dir_mode |= dir_change; @@ -1977,37 +2010,40 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct expr->dir_mode &= ~dir_change; break; } - fallthru; + _fallthrough; case MODE_ACTION: if (who == 0) { who = 0777; + mask = who & ~umask; + } else { + mask = who; } switch (*i) { case '+': op = MODE_PLUS; - mparser = MODE_OP; + state = MODE_OP; break; case '-': op = MODE_MINUS; - mparser = MODE_OP; + state = MODE_OP; break; case '=': op = MODE_EQUALS; - mparser = MODE_OP; + state = MODE_OP; break; case ',': - if (mparser == MODE_ACTION_APPLY) { - mparser = MODE_CLAUSE; + if (state == MODE_ACTION_APPLY) { + state = MODE_CLAUSE; } else { goto fail; } break; case '\0': - if (mparser == MODE_ACTION_APPLY) { + if (state == MODE_ACTION_APPLY) { goto done; } else { goto fail; @@ -2036,32 +2072,32 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct default: file_change = 0; dir_change = 0; - mparser = MODE_PERM; + state = MODE_PERM; continue; } file_change |= (file_change << 6) | (file_change << 3); - file_change &= who; + file_change &= mask; dir_change |= (dir_change << 6) | (dir_change << 3); - dir_change &= who; - mparser = MODE_ACTION_APPLY; + dir_change &= mask; + state = MODE_ACTION_APPLY; break; case MODE_PERM: switch (*i) { case 'r': - file_change |= who & 0444; - dir_change |= who & 0444; + file_change |= mask & 0444; + dir_change |= mask & 0444; break; case 'w': - file_change |= who & 0222; - dir_change |= who & 0222; + file_change |= mask & 0222; + dir_change |= mask & 0222; break; case 'x': - file_change |= who & 0111; - fallthru; + file_change |= mask & 0111; + _fallthrough; case 'X': - dir_change |= who & 0111; + dir_change |= mask & 0111; break; case 's': if (who & 0700) { @@ -2080,7 +2116,7 @@ static int parse_mode(const struct bfs_parser *parser, const char *mode, struct } break; default: - mparser = MODE_ACTION_APPLY; + state = MODE_ACTION_APPLY; continue; } break; @@ -2122,7 +2158,7 @@ static struct bfs_expr *parse_perm(struct bfs_parser *parser, int field, int arg ++mode; break; } - fallthru; + _fallthrough; default: expr->mode_cmp = BFS_MODE_EQUAL; break; @@ -2190,8 +2226,13 @@ static struct bfs_expr *parse_printx(struct bfs_parser *parser, int arg1, int ar * Parse -prune. */ static struct bfs_expr *parse_prune(struct bfs_parser *parser, int arg1, int arg2) { - parser->prune_arg = parser->argv; - return parse_nullary_action(parser, eval_prune); + struct bfs_expr *expr = parse_nullary_action(parser, eval_prune); + if (!expr) { + return NULL; + } + + parser->prune_expr = expr; + return expr; } /** @@ -2253,16 +2294,27 @@ static struct bfs_expr *parse_regextype(struct bfs_parser *parser, int arg1, int // See https://www.gnu.org/software/gnulib/manual/html_node/Predefined-Syntaxes.html const char *type = expr->argv[1]; if (strcmp(type, "posix-basic") == 0 + || strcmp(type, "posix-minimal-basic") == 0 || strcmp(type, "ed") == 0 || strcmp(type, "sed") == 0) { parser->regex_type = BFS_REGEX_POSIX_BASIC; } else if (strcmp(type, "posix-extended") == 0) { parser->regex_type = BFS_REGEX_POSIX_EXTENDED; -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA + } else if (strcmp(type, "awk") == 0 + || strcmp(type, "posix-awk") == 0) { + parser->regex_type = BFS_REGEX_AWK; + } else if (strcmp(type, "gnu-awk") == 0) { + parser->regex_type = BFS_REGEX_GNU_AWK; } else if (strcmp(type, "emacs") == 0) { parser->regex_type = BFS_REGEX_EMACS; } else if (strcmp(type, "grep") == 0) { parser->regex_type = BFS_REGEX_GREP; + } else if (strcmp(type, "egrep") == 0 + || strcmp(type, "posix-egrep") == 0) { + parser->regex_type = BFS_REGEX_EGREP; + } else if (strcmp(type, "findutils-default") == 0) { + parser->regex_type = BFS_REGEX_GNU_FIND; #endif } else if (strcmp(type, "help") == 0) { parser->just_info = true; @@ -2277,14 +2329,23 @@ static struct bfs_expr *parse_regextype(struct bfs_parser *parser, int arg1, int list_types: cfprintf(cfile, "Supported types are:\n\n"); - cfprintf(cfile, " ${bld}posix-basic${rs}: POSIX basic regular expressions (BRE)\n"); - cfprintf(cfile, " ${bld}posix-extended${rs}: POSIX extended regular expressions (ERE)\n"); - cfprintf(cfile, " ${bld}ed${rs}: Like ${grn}ed${rs} (same as ${bld}posix-basic${rs})\n"); -#if BFS_USE_ONIGURUMA - cfprintf(cfile, " ${bld}emacs${rs}: Like ${grn}emacs${rs}\n"); - cfprintf(cfile, " ${bld}grep${rs}: Like ${grn}grep${rs}\n"); + cfprintf(cfile, " ${bld}posix-basic${rs}: POSIX basic regular expressions (BRE)\n"); + cfprintf(cfile, " ${bld}ed${rs}: Like ${grn}ed${rs} (same as ${bld}posix-basic${rs})\n"); + cfprintf(cfile, " ${bld}sed${rs}: Like ${grn}sed${rs} (same as ${bld}posix-basic${rs})\n\n"); + + cfprintf(cfile, " ${bld}posix-extended${rs}: POSIX extended regular expressions (ERE)\n\n"); + +#if BFS_WITH_ONIGURUMA + cfprintf(cfile, " [${bld}posix-${rs}]${bld}awk${rs}: Like ${grn}awk${rs}\n"); + cfprintf(cfile, " ${bld}gnu-awk${rs}: Like GNU ${grn}awk${rs}\n\n"); + + cfprintf(cfile, " ${bld}emacs${rs}: Like ${grn}emacs${rs}\n\n"); + + cfprintf(cfile, " ${bld}grep${rs}: Like ${grn}grep${rs}\n"); + cfprintf(cfile, " [${bld}posix-${rs}]${bld}egrep${rs}: Like ${grn}grep${rs} ${cyn}-E${rs}\n\n"); + + cfprintf(cfile, " ${bld}findutils-default${rs}: Like GNU ${grn}find${rs}\n"); #endif - cfprintf(cfile, " ${bld}sed${rs}: Like ${grn}sed${rs} (same as ${bld}posix-basic${rs})\n"); return NULL; } @@ -2322,13 +2383,13 @@ static struct bfs_expr *parse_search_strategy(struct bfs_parser *parser, int arg struct bfs_ctx *ctx = parser->ctx; CFILE *cfile = ctx->cerr; - struct bfs_expr *expr = parse_unary_flag(parser); + const char *arg; + struct bfs_expr *expr = parse_prefix_flag(parser, 'S', true, &arg); if (!expr) { cfprintf(cfile, "\n"); goto list_strategies; } - const char *arg = expr->argv[1]; if (strcmp(arg, "bfs") == 0) { ctx->strategy = BFTW_BFS; } else if (strcmp(arg, "dfs") == 0) { @@ -2549,9 +2610,14 @@ static struct bfs_expr *parse_xattrname(struct bfs_parser *parser, int arg1, int * Parse -xdev. */ static struct bfs_expr *parse_xdev(struct bfs_parser *parser, int arg1, int arg2) { + struct bfs_expr *expr = parse_nullary_option(parser); + if (!expr) { + return NULL; + } + parser->ctx->flags |= BFTW_PRUNE_MOUNTS; - parser->xdev_arg = parser->argv; - return parse_nullary_option(parser); + parser->xdev_expr = expr; + return expr; } /** @@ -2744,8 +2810,9 @@ static struct bfs_expr *parse_help(struct bfs_parser *parser, int arg1, int arg2 cfprintf(cout, " ${blu}-mindepth${rs} ${bld}N${rs}\n"); cfprintf(cout, " Ignore files deeper/shallower than ${bld}N${rs}\n"); cfprintf(cout, " ${blu}-mount${rs}\n"); - cfprintf(cout, " Don't descend into other mount points (same as ${blu}-xdev${rs} for now, but will\n"); - cfprintf(cout, " skip mount points entirely in the future)\n"); + cfprintf(cout, " Exclude mount points entirely from the results\n"); + cfprintf(cout, " ${blu}-noerror${rs}\n"); + cfprintf(cout, " Ignore any errors that occur during traversal\n"); cfprintf(cout, " ${blu}-nohidden${rs}\n"); cfprintf(cout, " Exclude hidden files\n"); cfprintf(cout, " ${blu}-noleaf${rs}\n"); @@ -2922,18 +2989,60 @@ static struct bfs_expr *parse_help(struct bfs_parser *parser, int arg1, int arg2 return NULL; } +/** Print the bfs "logo". */ +static void print_logo(CFILE *cout) { + if (!cout->colors) { + goto boring; + } + + size_t vwidth = xstrwidth(bfs_version); + dchar *spaces = dstrepeat(" ", vwidth); + dchar *lines = dstrepeat("─", vwidth); + if (!spaces || !lines) { + dstrfree(lines); + dstrfree(spaces); + goto boring; + } + + // We do ----\r<emoji> rather than <emoji>--- so we don't have to assume + // anything about the width of the emoji + cfprintf(cout, "╭─────%s╮\r📂\n", lines); + cfprintf(cout, "├${ex}b${rs} %s │\n", spaces); + cfprintf(cout, "╰├${ex}f${rs} ${bld}%s${rs} │\n", bfs_version); + cfprintf(cout, " ╰├${ex}s${rs} %s │\n", spaces); + cfprintf(cout, " ╰──%s─╯\n\n", lines); + + dstrfree(lines); + dstrfree(spaces); + return; + +boring: + printf("%s %s\n\n", BFS_COMMAND, bfs_version); +} + /** * "Parse" -version. */ static struct bfs_expr *parse_version(struct bfs_parser *parser, int arg1, int arg2) { - cfprintf(parser->ctx->cout, "${ex}%s${rs} ${bld}%s${rs}\n\n", BFS_COMMAND, bfs_version); + print_logo(parser->ctx->cout); + + printf("Copyright © Tavian Barnes and the bfs contributors\n"); + printf("No rights reserved (https://opensource.org/license/0BSD)\n\n"); + + printf("CONFFLAGS := %s\n", bfs_confflags); + printf("CC := %s\n", bfs_cc); + printf("CPPFLAGS := %s\n", bfs_cppflags); + printf("CFLAGS := %s\n", bfs_cflags); + printf("LDFLAGS := %s\n", bfs_ldflags); + printf("LDLIBS := %s\n", bfs_ldlibs); - printf("%s\n", BFS_HOMEPAGE); + printf("\n%s\n", BFS_HOMEPAGE); parser->just_info = true; return NULL; } +/** Parser callback function type. */ typedef struct bfs_expr *parse_fn(struct bfs_parser *parser, int arg1, int arg2); /** @@ -2941,137 +3050,139 @@ typedef struct bfs_expr *parse_fn(struct bfs_parser *parser, int arg1, int arg2) */ struct table_entry { char *arg; - enum token_type type; + enum bfs_kind kind; parse_fn *parse; int arg1; int arg2; bool prefix; + bool needs_arg; }; /** * The parse table for primary expressions. */ static const struct table_entry parse_table[] = { - {"--", T_FLAG}, - {"--help", T_ACTION, parse_help}, - {"--version", T_ACTION, parse_version}, - {"-Bmin", T_TEST, parse_min, BFS_STAT_BTIME}, - {"-Bnewer", T_TEST, parse_newer, BFS_STAT_BTIME}, - {"-Bsince", T_TEST, parse_since, BFS_STAT_BTIME}, - {"-Btime", T_TEST, parse_time, BFS_STAT_BTIME}, - {"-D", T_FLAG, parse_debug}, - {"-E", T_FLAG, parse_regex_extended}, - {"-H", T_FLAG, parse_follow, BFTW_FOLLOW_ROOTS, false}, - {"-L", T_FLAG, parse_follow, BFTW_FOLLOW_ALL, false}, - {"-O", T_FLAG, parse_optlevel, 0, 0, true}, - {"-P", T_FLAG, parse_follow, 0, false}, - {"-S", T_FLAG, parse_search_strategy}, - {"-X", T_FLAG, parse_xargs_safe}, - {"-a", T_OPERATOR}, - {"-acl", T_TEST, parse_acl}, - {"-amin", T_TEST, parse_min, BFS_STAT_ATIME}, - {"-and", T_OPERATOR}, - {"-anewer", T_TEST, parse_newer, BFS_STAT_ATIME}, - {"-asince", T_TEST, parse_since, BFS_STAT_ATIME}, - {"-atime", T_TEST, parse_time, BFS_STAT_ATIME}, - {"-capable", T_TEST, parse_capable}, - {"-cmin", T_TEST, parse_min, BFS_STAT_CTIME}, - {"-cnewer", T_TEST, parse_newer, BFS_STAT_CTIME}, - {"-color", T_OPTION, parse_color, true}, - {"-context", T_TEST, parse_context, true}, - {"-csince", T_TEST, parse_since, BFS_STAT_CTIME}, - {"-ctime", T_TEST, parse_time, BFS_STAT_CTIME}, - {"-d", T_FLAG, parse_depth}, - {"-daystart", T_OPTION, parse_daystart}, - {"-delete", T_ACTION, parse_delete}, - {"-depth", T_OPTION, parse_depth_n}, - {"-empty", T_TEST, parse_empty}, - {"-exclude", T_OPERATOR}, - {"-exec", T_ACTION, parse_exec, 0}, - {"-execdir", T_ACTION, parse_exec, BFS_EXEC_CHDIR}, - {"-executable", T_TEST, parse_access, X_OK}, - {"-exit", T_ACTION, parse_exit}, - {"-f", T_FLAG, parse_f}, - {"-false", T_TEST, parse_const, false}, - {"-files0-from", T_OPTION, parse_files0_from}, - {"-flags", T_TEST, parse_flags}, - {"-fls", T_ACTION, parse_fls}, - {"-follow", T_OPTION, parse_follow, BFTW_FOLLOW_ALL, true}, - {"-fprint", T_ACTION, parse_fprint}, - {"-fprint0", T_ACTION, parse_fprint0}, - {"-fprintf", T_ACTION, parse_fprintf}, - {"-fstype", T_TEST, parse_fstype}, - {"-gid", T_TEST, parse_group}, - {"-group", T_TEST, parse_group}, - {"-help", T_ACTION, parse_help}, - {"-hidden", T_TEST, parse_hidden}, - {"-ignore_readdir_race", T_OPTION, parse_ignore_races, true}, - {"-ilname", T_TEST, parse_lname, true}, - {"-iname", T_TEST, parse_name, true}, - {"-inum", T_TEST, parse_inum}, - {"-ipath", T_TEST, parse_path, true}, - {"-iregex", T_TEST, parse_regex, BFS_REGEX_ICASE}, - {"-iwholename", T_TEST, parse_path, true}, - {"-j", T_FLAG, parse_jobs, 0, 0, true}, - {"-limit", T_ACTION, parse_limit}, - {"-links", T_TEST, parse_links}, - {"-lname", T_TEST, parse_lname, false}, - {"-ls", T_ACTION, parse_ls}, - {"-maxdepth", T_OPTION, parse_depth_limit, false}, - {"-mindepth", T_OPTION, parse_depth_limit, true}, - {"-mmin", T_TEST, parse_min, BFS_STAT_MTIME}, - {"-mnewer", T_TEST, parse_newer, BFS_STAT_MTIME}, - {"-mount", T_OPTION, parse_mount}, - {"-msince", T_TEST, parse_since, BFS_STAT_MTIME}, - {"-mtime", T_TEST, parse_time, BFS_STAT_MTIME}, - {"-name", T_TEST, parse_name, false}, - {"-newer", T_TEST, parse_newer, BFS_STAT_MTIME}, - {"-newer", T_TEST, parse_newerxy, 0, 0, true}, - {"-nocolor", T_OPTION, parse_color, false}, - {"-nogroup", T_TEST, parse_nogroup}, - {"-nohidden", T_TEST, parse_nohidden}, - {"-noignore_readdir_race", T_OPTION, parse_ignore_races, false}, - {"-noleaf", T_OPTION, parse_noleaf}, - {"-not", T_OPERATOR}, - {"-nouser", T_TEST, parse_nouser}, - {"-nowarn", T_OPTION, parse_warn, false}, - {"-o", T_OPERATOR}, - {"-ok", T_ACTION, parse_exec, BFS_EXEC_CONFIRM}, - {"-okdir", T_ACTION, parse_exec, BFS_EXEC_CONFIRM | BFS_EXEC_CHDIR}, - {"-or", T_OPERATOR}, - {"-path", T_TEST, parse_path, false}, - {"-perm", T_TEST, parse_perm}, - {"-print", T_ACTION, parse_print}, - {"-print0", T_ACTION, parse_print0}, - {"-printf", T_ACTION, parse_printf}, - {"-printx", T_ACTION, parse_printx}, - {"-prune", T_ACTION, parse_prune}, - {"-quit", T_ACTION, parse_quit}, - {"-readable", T_TEST, parse_access, R_OK}, - {"-regex", T_TEST, parse_regex, 0}, - {"-regextype", T_OPTION, parse_regextype}, - {"-rm", T_ACTION, parse_delete}, - {"-s", T_FLAG, parse_s}, - {"-samefile", T_TEST, parse_samefile}, - {"-since", T_TEST, parse_since, BFS_STAT_MTIME}, - {"-size", T_TEST, parse_size}, - {"-sparse", T_TEST, parse_sparse}, - {"-status", T_OPTION, parse_status}, - {"-true", T_TEST, parse_const, true}, - {"-type", T_TEST, parse_type, false}, - {"-uid", T_TEST, parse_user}, - {"-unique", T_OPTION, parse_unique}, - {"-used", T_TEST, parse_used}, - {"-user", T_TEST, parse_user}, - {"-version", T_ACTION, parse_version}, - {"-warn", T_OPTION, parse_warn, true}, - {"-wholename", T_TEST, parse_path, false}, - {"-writable", T_TEST, parse_access, W_OK}, - {"-x", T_FLAG, parse_xdev}, - {"-xattr", T_TEST, parse_xattr}, - {"-xattrname", T_TEST, parse_xattrname}, - {"-xdev", T_OPTION, parse_xdev}, - {"-xtype", T_TEST, parse_type, true}, + {"--", BFS_FLAG}, + {"--help", BFS_ACTION, parse_help}, + {"--version", BFS_ACTION, parse_version}, + {"-Bmin", BFS_TEST, parse_min, BFS_STAT_BTIME}, + {"-Bnewer", BFS_TEST, parse_newer, BFS_STAT_BTIME}, + {"-Bsince", BFS_TEST, parse_since, BFS_STAT_BTIME}, + {"-Btime", BFS_TEST, parse_time, BFS_STAT_BTIME}, + {"-D", BFS_FLAG, parse_debug, .prefix = true}, + {"-E", BFS_FLAG, parse_regex_extended}, + {"-H", BFS_FLAG, parse_follow, BFTW_FOLLOW_ROOTS, false}, + {"-L", BFS_FLAG, parse_follow, BFTW_FOLLOW_ALL, false}, + {"-O", BFS_FLAG, parse_optlevel, .prefix = true}, + {"-P", BFS_FLAG, parse_follow, 0, false}, + {"-S", BFS_FLAG, parse_search_strategy, .prefix = true}, + {"-X", BFS_FLAG, parse_xargs_safe}, + {"-a", BFS_OPERATOR}, + {"-acl", BFS_TEST, parse_acl}, + {"-amin", BFS_TEST, parse_min, BFS_STAT_ATIME}, + {"-and", BFS_OPERATOR}, + {"-anewer", BFS_TEST, parse_newer, BFS_STAT_ATIME}, + {"-asince", BFS_TEST, parse_since, BFS_STAT_ATIME}, + {"-atime", BFS_TEST, parse_time, BFS_STAT_ATIME}, + {"-capable", BFS_TEST, parse_capable}, + {"-cmin", BFS_TEST, parse_min, BFS_STAT_CTIME}, + {"-cnewer", BFS_TEST, parse_newer, BFS_STAT_CTIME}, + {"-color", BFS_OPTION, parse_color, true}, + {"-context", BFS_TEST, parse_context, true}, + {"-csince", BFS_TEST, parse_since, BFS_STAT_CTIME}, + {"-ctime", BFS_TEST, parse_time, BFS_STAT_CTIME}, + {"-d", BFS_FLAG, parse_depth, true}, + {"-daystart", BFS_OPTION, parse_daystart}, + {"-delete", BFS_ACTION, parse_delete}, + {"-depth", BFS_OPTION, parse_depth_n, false}, + {"-empty", BFS_TEST, parse_empty}, + {"-exclude", BFS_OPERATOR}, + {"-exec", BFS_ACTION, parse_exec, 0}, + {"-execdir", BFS_ACTION, parse_exec, BFS_EXEC_CHDIR}, + {"-executable", BFS_TEST, parse_access, X_OK}, + {"-exit", BFS_ACTION, parse_exit}, + {"-f", BFS_FLAG, parse_f, .needs_arg = true}, + {"-false", BFS_TEST, parse_const, false}, + {"-files0-from", BFS_OPTION, parse_files0_from}, + {"-flags", BFS_TEST, parse_flags}, + {"-fls", BFS_ACTION, parse_fls}, + {"-follow", BFS_OPTION, parse_follow, BFTW_FOLLOW_ALL, true}, + {"-fprint", BFS_ACTION, parse_fprint}, + {"-fprint0", BFS_ACTION, parse_fprint0}, + {"-fprintf", BFS_ACTION, parse_fprintf}, + {"-fstype", BFS_TEST, parse_fstype}, + {"-gid", BFS_TEST, parse_group}, + {"-group", BFS_TEST, parse_group}, + {"-help", BFS_ACTION, parse_help}, + {"-hidden", BFS_TEST, parse_hidden}, + {"-ignore_readdir_race", BFS_OPTION, parse_ignore_races, true}, + {"-ilname", BFS_TEST, parse_lname, true}, + {"-iname", BFS_TEST, parse_name, true}, + {"-inum", BFS_TEST, parse_inum}, + {"-ipath", BFS_TEST, parse_path, true}, + {"-iregex", BFS_TEST, parse_regex, BFS_REGEX_ICASE}, + {"-iwholename", BFS_TEST, parse_path, true}, + {"-j", BFS_FLAG, parse_jobs, .prefix = true}, + {"-limit", BFS_ACTION, parse_limit}, + {"-links", BFS_TEST, parse_links}, + {"-lname", BFS_TEST, parse_lname, false}, + {"-ls", BFS_ACTION, parse_ls}, + {"-maxdepth", BFS_OPTION, parse_depth_limit, false}, + {"-mindepth", BFS_OPTION, parse_depth_limit, true}, + {"-mmin", BFS_TEST, parse_min, BFS_STAT_MTIME}, + {"-mnewer", BFS_TEST, parse_newer, BFS_STAT_MTIME}, + {"-mount", BFS_OPTION, parse_mount}, + {"-msince", BFS_TEST, parse_since, BFS_STAT_MTIME}, + {"-mtime", BFS_TEST, parse_time, BFS_STAT_MTIME}, + {"-name", BFS_TEST, parse_name, false}, + {"-newer", BFS_TEST, parse_newer, BFS_STAT_MTIME}, + {"-newer", BFS_TEST, parse_newerxy, .prefix = true}, + {"-nocolor", BFS_OPTION, parse_color, false}, + {"-noerror", BFS_OPTION, parse_noerror}, + {"-nogroup", BFS_TEST, parse_nogroup}, + {"-nohidden", BFS_TEST, parse_nohidden}, + {"-noignore_readdir_race", BFS_OPTION, parse_ignore_races, false}, + {"-noleaf", BFS_OPTION, parse_noleaf}, + {"-not", BFS_OPERATOR}, + {"-nouser", BFS_TEST, parse_nouser}, + {"-nowarn", BFS_OPTION, parse_warn, false}, + {"-o", BFS_OPERATOR}, + {"-ok", BFS_ACTION, parse_exec, BFS_EXEC_CONFIRM}, + {"-okdir", BFS_ACTION, parse_exec, BFS_EXEC_CONFIRM | BFS_EXEC_CHDIR}, + {"-or", BFS_OPERATOR}, + {"-path", BFS_TEST, parse_path, false}, + {"-perm", BFS_TEST, parse_perm}, + {"-print", BFS_ACTION, parse_print}, + {"-print0", BFS_ACTION, parse_print0}, + {"-printf", BFS_ACTION, parse_printf}, + {"-printx", BFS_ACTION, parse_printx}, + {"-prune", BFS_ACTION, parse_prune}, + {"-quit", BFS_ACTION, parse_quit}, + {"-readable", BFS_TEST, parse_access, R_OK}, + {"-regex", BFS_TEST, parse_regex, 0}, + {"-regextype", BFS_OPTION, parse_regextype}, + {"-rm", BFS_ACTION, parse_delete}, + {"-s", BFS_FLAG, parse_s}, + {"-samefile", BFS_TEST, parse_samefile}, + {"-since", BFS_TEST, parse_since, BFS_STAT_MTIME}, + {"-size", BFS_TEST, parse_size}, + {"-sparse", BFS_TEST, parse_sparse}, + {"-status", BFS_OPTION, parse_status}, + {"-true", BFS_TEST, parse_const, true}, + {"-type", BFS_TEST, parse_type, false}, + {"-uid", BFS_TEST, parse_user}, + {"-unique", BFS_OPTION, parse_unique}, + {"-used", BFS_TEST, parse_used}, + {"-user", BFS_TEST, parse_user}, + {"-version", BFS_ACTION, parse_version}, + {"-warn", BFS_OPTION, parse_warn, true}, + {"-wholename", BFS_TEST, parse_path, false}, + {"-writable", BFS_TEST, parse_access, W_OK}, + {"-x", BFS_FLAG, parse_xdev}, + {"-xattr", BFS_TEST, parse_xattr}, + {"-xattrname", BFS_TEST, parse_xattrname}, + {"-xdev", BFS_OPTION, parse_xdev}, + {"-xtype", BFS_TEST, parse_type, true}, {0}, }; @@ -3092,6 +3203,83 @@ static const struct table_entry *table_lookup(const char *arg) { return NULL; } +/** Look up a single-character flag in the parse table. */ +static const struct table_entry *flag_lookup(char flag) { + for (const struct table_entry *entry = parse_table; entry->arg; ++entry) { + enum bfs_kind kind = entry->kind; + if (kind == BFS_FLAG && entry->arg[1] == flag && !entry->arg[2]) { + return entry; + } + } + + return NULL; +} + +/** Check for a multi-flag argument like -LEXO2. */ +static bool is_flag_group(const char *arg) { + // We enforce that at least one flag in a flag group must be a capital + // letter, to avoid ambiguity with primary expressions + bool has_upper = false; + + // Flags that take an argument must appear last + bool needs_arg = false; + + for (size_t i = 1; arg[i]; ++i) { + char c = arg[i]; + if (c >= 'A' && c <= 'Z') { + has_upper = true; + } + + if (needs_arg) { + return false; + } + + const struct table_entry *entry = flag_lookup(c); + if (!entry || !entry->parse) { + return false; + } + + if (entry->prefix) { + // The rest is the flag's argument + break; + } + + needs_arg |= entry->needs_arg; + } + + return has_upper; +} + +/** Parse a multi-flag argument. */ +static struct bfs_expr *parse_flag_group(struct bfs_parser *parser) { + struct bfs_expr *expr = NULL; + + char **start = parser->argv; + char **end = start; + const char *arg = start[0]; + + for (size_t i = 1; arg[i]; ++i) { + parser->argv = start; + + const struct table_entry *entry = flag_lookup(arg[i]); + expr = entry->parse(parser, entry->arg1, entry->arg2); + + if (parser->argv > end) { + end = parser->argv; + } + + if (!expr || entry->prefix) { + break; + } + } + + if (expr) { + bfs_assert(parser->argv == end, "Didn't eat enough tokens"); + } + + return expr; +} + /** Search for a fuzzy match in the parse table. */ static const struct table_entry *table_lookup_fuzzy(const char *arg) { const struct table_entry *best = NULL; @@ -3114,6 +3302,8 @@ static const struct table_entry *table_lookup_fuzzy(const char *arg) { * | ACTION */ static struct bfs_expr *parse_primary(struct bfs_parser *parser) { + struct bfs_ctx *ctx = parser->ctx; + // Paths are already skipped at this point const char *arg = parser->argv[0]; @@ -3130,15 +3320,19 @@ static struct bfs_expr *parse_primary(struct bfs_parser *parser) { } } + if (is_flag_group(arg)) { + return parse_flag_group(parser); + } + match = table_lookup_fuzzy(arg); - CFILE *cerr = parser->ctx->cerr; + CFILE *cerr = ctx->cerr; parse_error(parser, "Unknown argument; did you mean "); - switch (match->type) { - case T_FLAG: + switch (match->kind) { + case BFS_FLAG: cfprintf(cerr, "${cyn}%s${rs}?", match->arg); break; - case T_OPERATOR: + case BFS_OPERATOR: cfprintf(cerr, "${red}%s${rs}?", match->arg); break; default: @@ -3146,7 +3340,7 @@ static struct bfs_expr *parse_primary(struct bfs_parser *parser) { break; } - if (!parser->interactive || !match->parse) { + if (!ctx->interactive || !match->parse) { fprintf(stderr, "\n"); goto unmatched; } @@ -3188,7 +3382,7 @@ static struct bfs_expr *parse_factor(struct bfs_parser *parser) { } if (strcmp(arg, "(") == 0) { - parser_advance(parser, T_OPERATOR, 1); + parser_advance(parser, BFS_OPERATOR, 1); struct bfs_expr *expr = parse_expr(parser); if (!expr) { @@ -3205,7 +3399,7 @@ static struct bfs_expr *parse_factor(struct bfs_parser *parser) { return NULL; } - parser_advance(parser, T_OPERATOR, 1); + parser_advance(parser, BFS_OPERATOR, 1); return expr; } else if (strcmp(arg, "-exclude") == 0) { if (parser->excluding) { @@ -3213,7 +3407,7 @@ static struct bfs_expr *parse_factor(struct bfs_parser *parser) { return NULL; } - char **argv = parser_advance(parser, T_OPERATOR, 1); + char **argv = parser_advance(parser, BFS_OPERATOR, 1); parser->excluding = true; struct bfs_expr *factor = parse_factor(parser); @@ -3224,9 +3418,9 @@ static struct bfs_expr *parse_factor(struct bfs_parser *parser) { parser->excluding = false; bfs_expr_append(parser->ctx->exclude, factor); - return parse_new_expr(parser, eval_true, parser->argv - argv, argv); + return parse_new_expr(parser, eval_true, parser->argv - argv, argv, BFS_OPERATOR); } else if (strcmp(arg, "!") == 0 || strcmp(arg, "-not") == 0) { - char **argv = parser_advance(parser, T_OPERATOR, 1); + char **argv = parser_advance(parser, BFS_OPERATOR, 1); struct bfs_expr *factor = parse_factor(parser); if (!factor) { @@ -3266,7 +3460,7 @@ static struct bfs_expr *parse_term(struct bfs_parser *parser) { char **argv = &fake_and_arg; if (strcmp(arg, "-a") == 0 || strcmp(arg, "-and") == 0) { - argv = parser_advance(parser, T_OPERATOR, 1); + argv = parser_advance(parser, BFS_OPERATOR, 1); } struct bfs_expr *lhs = term; @@ -3303,7 +3497,7 @@ static struct bfs_expr *parse_clause(struct bfs_parser *parser) { break; } - char **argv = parser_advance(parser, T_OPERATOR, 1); + char **argv = parser_advance(parser, BFS_OPERATOR, 1); struct bfs_expr *lhs = clause; struct bfs_expr *rhs = parse_term(parser); @@ -3338,7 +3532,7 @@ static struct bfs_expr *parse_expr(struct bfs_parser *parser) { break; } - char **argv = parser_advance(parser, T_OPERATOR, 1); + char **argv = parser_advance(parser, BFS_OPERATOR, 1); struct bfs_expr *lhs = expr; struct bfs_expr *rhs = parse_clause(parser); @@ -3352,10 +3546,79 @@ static struct bfs_expr *parse_expr(struct bfs_parser *parser) { return expr; } +/** Handle -files0-from after parsing. */ +static int parse_files0_roots(struct bfs_parser *parser) { + const struct bfs_ctx *ctx = parser->ctx; + const struct bfs_expr *expr = parser->files0_expr; + + if (ctx->npaths > 0) { + bool highlight[ctx->argc]; + init_highlight(ctx, highlight); + highlight_args(ctx, expr->argv, expr->argc, highlight); + + for (size_t i = 0; i < ctx->argc; ++i) { + if (ctx->kinds[i] == BFS_PATH) { + highlight[i] = true; + } + } + + bfs_argv_error(ctx, highlight); + bfs_error(ctx, "Cannot combine %pX with explicit root paths.\n", expr); + return -1; + } + + const char *from = expr->argv[1]; + + FILE *file; + if (strcmp(from, "-") == 0) { + if (!consume_stdin(parser, expr)) { + return -1; + } + file = stdin; + } else { + file = xfopen(from, O_RDONLY | O_CLOEXEC); + } + if (!file) { + parse_expr_error(parser, expr, "%s.\n", errstr()); + return -1; + } + + while (true) { + char *path = xgetdelim(file, '\0'); + if (!path) { + if (errno) { + goto fail; + } else { + break; + } + } + + int ret = parse_root(parser, path); + free(path); + if (ret != 0) { + goto fail; + } + } + + if (file != stdin) { + fclose(file); + } + + return 0; + +fail: + if (file != stdin) { + fclose(file); + } + return -1; +} + /** * Parse the top-level expression. */ static struct bfs_expr *parse_whole_expr(struct bfs_parser *parser) { + struct bfs_ctx *ctx = parser->ctx; + if (skip_paths(parser) != 0) { return NULL; } @@ -3364,7 +3627,7 @@ static struct bfs_expr *parse_whole_expr(struct bfs_parser *parser) { if (parser->argv[0]) { expr = parse_expr(parser); } else { - expr = parse_new_expr(parser, eval_true, 1, &fake_true_arg); + expr = parse_new_expr(parser, eval_true, 1, &fake_true_arg, BFS_TEST); } if (!expr) { return NULL; @@ -3375,16 +3638,26 @@ static struct bfs_expr *parse_whole_expr(struct bfs_parser *parser) { return NULL; } + if (parser->files0_expr) { + if (parse_files0_roots(parser) != 0) { + return NULL; + } + } else if (ctx->npaths == 0) { + if (parse_root(parser, ".") != 0) { + return NULL; + } + } + if (parser->implicit_print) { - char **limit = parser->limit_arg; + const struct bfs_expr *limit = parser->limit_expr; if (limit) { - parse_argv_error(parser, parser->limit_arg, 2, - "With ${blu}%s${rs}, you must specify an action explicitly; for example, ${blu}-print${rs} ${blu}%s${rs} ${bld}%s${rs}.\n", - limit[0], limit[0], limit[1]); + parse_expr_error(parser, limit, + "With %pX, you must specify an action explicitly; for example, ${blu}-print${rs} %px.\n", + limit, limit); return NULL; } - struct bfs_expr *print = parse_new_expr(parser, eval_fprint, 1, &fake_print_arg); + struct bfs_expr *print = parse_new_expr(parser, eval_fprint, 1, &fake_print_arg, BFS_ACTION); if (!print) { return NULL; } @@ -3396,20 +3669,20 @@ static struct bfs_expr *parse_whole_expr(struct bfs_parser *parser) { } } - if (parser->mount_arg && parser->xdev_arg) { - parse_conflict_warning(parser, parser->mount_arg, 1, parser->xdev_arg, 1, - "${blu}%s${rs} is redundant in the presence of ${blu}%s${rs}.\n\n", - parser->xdev_arg[0], parser->mount_arg[0]); + if (parser->mount_expr && parser->xdev_expr) { + parse_conflict_warning(parser, parser->mount_expr, parser->xdev_expr, + "%px is redundant in the presence of %px.\n\n", + parser->xdev_expr, parser->mount_expr); } - if (parser->ctx->warn && parser->depth_arg && parser->prune_arg) { - parse_conflict_warning(parser, parser->depth_arg, 1, parser->prune_arg, 1, - "${blu}%s${rs} does not work in the presence of ${blu}%s${rs}.\n", - parser->prune_arg[0], parser->depth_arg[0]); + if (ctx->warn && parser->depth_expr && parser->prune_expr) { + parse_conflict_warning(parser, parser->depth_expr, parser->prune_expr, + "%px does not work in the presence of %px.\n", + parser->prune_expr, parser->depth_expr); - if (parser->interactive) { - bfs_warning(parser->ctx, "Do you want to continue? "); - if (ynprompt() == 0) { + if (ctx->interactive) { + bfs_warning(ctx, "Do you want to continue? "); + if (ynprompt() <= 0) { return NULL; } } @@ -3417,13 +3690,6 @@ static struct bfs_expr *parse_whole_expr(struct bfs_parser *parser) { fprintf(stderr, "\n"); } - if (parser->ok_expr && parser->files0_stdin_arg) { - parse_conflict_error(parser, parser->ok_expr->argv, parser->ok_expr->argc, parser->files0_stdin_arg, 2, - "${blu}%s${rs} conflicts with ${blu}%s${rs} ${bld}%s${rs}.\n", - parser->ok_expr->argv[0], parser->files0_stdin_arg[0], parser->files0_stdin_arg[1]); - return NULL; - } - return expr; } @@ -3458,7 +3724,7 @@ static void dump_expr_multiline(const struct bfs_ctx *ctx, enum debug_flags flag ++rparens; } else { cfprintf(ctx->cerr, "(${red}%s${rs}\n", expr->argv[0]); - for (struct bfs_expr *child = bfs_expr_children(expr); child; child = child->next) { + for_expr (child, expr) { int parens = child->next ? 0 : rparens + 1; dump_expr_multiline(ctx, flag, child, indent + 1, parens); } @@ -3605,6 +3871,12 @@ struct bfs_ctx *bfs_parse_cmdline(int argc, char *argv[]) { goto fail; } + ctx->kinds = ZALLOC_ARRAY(enum bfs_kind, argc); + if (!ctx->kinds) { + perror("zalloc()"); + goto fail; + } + enum use_color use_color = COLOR_AUTO; const char *no_color = getenv("NO_COLOR"); if (no_color && *no_color) { @@ -3643,6 +3915,7 @@ struct bfs_ctx *bfs_parse_cmdline(int argc, char *argv[]) { } else { ctx->warn = stdin_tty; } + ctx->interactive = stdin_tty && stderr_tty; struct bfs_parser parser = { .ctx = ctx, @@ -3650,23 +3923,20 @@ struct bfs_ctx *bfs_parse_cmdline(int argc, char *argv[]) { .command = ctx->argv[0], .regex_type = BFS_REGEX_POSIX_BASIC, .stdout_tty = stdout_tty, - .interactive = stdin_tty && stderr_tty, .use_color = use_color, .implicit_print = true, - .implicit_root = true, .just_info = false, .excluding = false, .last_arg = NULL, - .depth_arg = NULL, - .prune_arg = NULL, - .mount_arg = NULL, - .xdev_arg = NULL, - .files0_stdin_arg = NULL, - .ok_expr = NULL, + .depth_expr = NULL, + .prune_expr = NULL, + .mount_expr = NULL, + .xdev_expr = NULL, + .stdin_expr = NULL, .now = ctx->now, }; - ctx->exclude = parse_new_expr(&parser, eval_or, 1, &fake_or_arg); + ctx->exclude = parse_new_expr(&parser, eval_or, 1, &fake_or_arg, BFS_OPERATOR); if (!ctx->exclude) { goto fail; } @@ -3685,14 +3955,10 @@ struct bfs_ctx *bfs_parse_cmdline(int argc, char *argv[]) { } if (bfs_optimize(ctx) != 0) { - bfs_perror(ctx, "bfs_optimize()"); - goto fail; - } - - if (ctx->npaths == 0 && parser.implicit_root) { - if (parse_root(&parser, ".") != 0) { - goto fail; + if (errno != 0) { + bfs_perror(ctx, "bfs_optimize()"); } + goto fail; } if ((ctx->flags & BFTW_FOLLOW_ALL) && !ctx->unique) { diff --git a/src/parse.h b/src/parse.h index 6895c9f..fcc8234 100644 --- a/src/parse.h +++ b/src/parse.h @@ -11,9 +11,9 @@ /** * Parse the command line. * - * @param argc + * @argc * The number of arguments. - * @param argv + * @argv * The arguments to parse. * @return * A new bfs context, or NULL on failure. diff --git a/src/prelude.h b/src/prelude.h index ddeacbd..de89a6c 100644 --- a/src/prelude.h +++ b/src/prelude.h @@ -2,384 +2,129 @@ // SPDX-License-Identifier: 0BSD /** - * Configuration and feature/platform detection. + * Praeludium. + * + * This header is automatically included in every translation unit, before any + * other headers, so it can set feature test macros[1][2]. This sets up our own + * mini-dialect of C, which includes + * + * - Standard C17 and POSIX.1 2024 features + * - Portable and platform-specific extensions + * - Convenience macros like `bool`, `alignof`, etc. + * - Common compiler extensions like __has_include() + * + * Further bfs-specific utilities are defined in "bfs.h". + * + * [1]: https://www.gnu.org/software/libc/manual/html_node/Feature-Test-Macros.html + * [2]: https://pubs.opengroup.org/onlinepubs/9799919799/functions/V2_chap02.html */ #ifndef BFS_PRELUDE_H #define BFS_PRELUDE_H -// Possible __STDC_VERSION__ values - -#define C95 199409L -#define C99 199901L -#define C11 201112L -#define C17 201710L -#define C23 202311L - -#include <stddef.h> - -#if __STDC_VERSION__ < C23 -# include <stdalign.h> -# include <stdbool.h> -# include <stdnoreturn.h> -#endif - -// bfs packaging configuration - -#include "config.h" - -#ifndef BFS_COMMAND -# define BFS_COMMAND "bfs" -#endif -#ifndef BFS_HOMEPAGE -# define BFS_HOMEPAGE "https://tavianator.com/projects/bfs.html" -#endif - -// This is a symbol instead of a literal so we don't have to rebuild everything -// when the version number changes -extern const char bfs_version[]; - -// Check for system headers +// Feature test macros -#ifdef __has_include - -#if __has_include(<mntent.h>) -# define BFS_HAS_MNTENT_H true -#endif -#if __has_include(<paths.h>) -# define BFS_HAS_PATHS_H true -#endif -#if __has_include(<sys/extattr.h>) -# define BFS_HAS_SYS_EXTATTR_H true -#endif -#if __has_include(<sys/mkdev.h>) -# define BFS_HAS_SYS_MKDEV_H true -#endif -#if __has_include(<sys/param.h>) -# define BFS_HAS_SYS_PARAM_H true -#endif -#if __has_include(<sys/sysmacros.h>) -# define BFS_HAS_SYS_SYSMACROS_H true -#endif -#if __has_include(<sys/xattr.h>) -# define BFS_HAS_SYS_XATTR_H true -#endif -#if __has_include(<threads.h>) -# define BFS_HAS_THREADS_H true -#endif -#if __has_include(<util.h>) -# define BFS_HAS_UTIL_H true -#endif - -#else // !__has_include - -#define BFS_HAS_MNTENT_H __GLIBC__ -#define BFS_HAS_PATHS_H true -#define BFS_HAS_SYS_EXTATTR_H __FreeBSD__ -#define BFS_HAS_SYS_MKDEV_H false -#define BFS_HAS_SYS_PARAM_H true -#define BFS_HAS_SYS_SYSMACROS_H __GLIBC__ -#define BFS_HAS_SYS_XATTR_H __linux__ -#define BFS_HAS_THREADS_H (!__STDC_NO_THREADS__) -#define BFS_HAS_UTIL_H __NetBSD__ - -#endif // !__has_include - -#ifndef BFS_USE_MNTENT_H -# define BFS_USE_MNTENT_H BFS_HAS_MNTENT_H -#endif -#ifndef BFS_USE_PATHS_H -# define BFS_USE_PATHS_H BFS_HAS_PATHS_H -#endif -#ifndef BFS_USE_SYS_EXTATTR_H -# define BFS_USE_SYS_EXTATTR_H BFS_HAS_SYS_EXTATTR_H -#endif -#ifndef BFS_USE_SYS_MKDEV_H -# define BFS_USE_SYS_MKDEV_H BFS_HAS_SYS_MKDEV_H -#endif -#ifndef BFS_USE_SYS_PARAM_H -# define BFS_USE_SYS_PARAM_H BFS_HAS_SYS_PARAM_H -#endif -#ifndef BFS_USE_SYS_SYSMACROS_H -# define BFS_USE_SYS_SYSMACROS_H BFS_HAS_SYS_SYSMACROS_H -#endif -#ifndef BFS_USE_SYS_XATTR_H -# define BFS_USE_SYS_XATTR_H BFS_HAS_SYS_XATTR_H -#endif -#ifndef BFS_USE_THREADS_H -# define BFS_USE_THREADS_H BFS_HAS_THREADS_H -#endif -#ifndef BFS_USE_UTIL_H -# define BFS_USE_UTIL_H BFS_HAS_UTIL_H -#endif +/** + * Linux and BSD handle _POSIX_C_SOURCE differently: on Linux, it enables POSIX + * interfaces that are not visible by default. On BSD, it also *disables* most + * extensions, giving a strict POSIX environment. Since we want the extensions, + * we don't set _POSIX_C_SOURCE. + */ +// #define _POSIX_C_SOURCE 202405L -// Stub out feature detection on old/incompatible compilers +/** openat() etc. */ +#define _ATFILE_SOURCE 1 -#ifndef __has_feature -# define __has_feature(feat) false -#endif +/** BSD-derived extensions. */ +#define _BSD_SOURCE 1 -#ifndef __has_c_attribute -# define __has_c_attribute(attr) false -#endif +/** glibc successor to _BSD_SOURCE. */ +#define _DEFAULT_SOURCE 1 -#ifndef __has_attribute -# define __has_attribute(attr) false -#endif +/** GNU extensions. */ +#define _GNU_SOURCE 1 -// Platform detection +/** Use 64-bit off_t. */ +#define _FILE_OFFSET_BITS 64 -// Get the definition of BSD if available -#if BFS_USE_SYS_PARAM_H -# include <sys/param.h> -#endif +/** Use 64-bit time_t. */ +#define _TIME_BITS 64 -#ifndef __GLIBC_PREREQ -# define __GLIBC_PREREQ(maj, min) false +/** macOS extensions. */ +#if __APPLE__ +# define _DARWIN_C_SOURCE 1 #endif -#ifndef __NetBSD_Prereq__ -# define __NetBSD_Prereq__(maj, min, patch) false +/** Solaris extensions. */ +#if __sun +# define __EXTENSIONS__ 1 +// https://illumos.org/man/3C/getpwnam#standard-conforming +# define _POSIX_PTHREAD_SEMANTICS 1 #endif -// Fundamental utilities - -/** - * Get the length of an array. - */ -#define countof(array) (sizeof(array) / sizeof(0[array])) - -/** - * False sharing/destructive interference/largest cache line size. - */ -#ifdef __GCC_DESTRUCTIVE_SIZE -# define FALSE_SHARING_SIZE __GCC_DESTRUCTIVE_SIZE -#else -# define FALSE_SHARING_SIZE 64 +/** QNX extensions. */ +#if __QNX__ +# define _QNX_SOURCE 1 #endif -/** - * True sharing/constructive interference/smallest cache line size. - */ -#ifdef __GCC_CONSTRUCTIVE_SIZE -# define TRUE_SHARING_SIZE __GCC_CONSTRUCTIVE_SIZE -#else -# define TRUE_SHARING_SIZE 64 -#endif +// Get the convenience macros that became standard spellings in C23 +#if __STDC_VERSION__ < 202311L -/** - * Polyfill max_align_t if we don't already have it. - */ -#if !BFS_HAS_MAX_ALIGN_T -typedef union { -# ifdef __BIGGEST_ALIGNMENT__ - alignas(__BIGGEST_ALIGNMENT__) char c; -# else - long double ld; - long long ll; - void *ptr; -# endif -} max_align_t; -#endif +/** _Static_assert() => static_assert() */ +#include <assert.h> +/** _Alignas(), _Alignof() => alignas(), alignof() */ +#include <stdalign.h> +/** _Bool => bool, true, false */ +#include <stdbool.h> /** - * Alignment specifier that avoids false sharing. + * C23 deprecates `noreturn void` in favour of `[[noreturn]] void`, so we expose + * _noreturn instead with the other attributes in "bfs.h". */ -#define cache_align alignas(FALSE_SHARING_SIZE) +// #include <stdnoreturn.h> -// Wrappers for attributes +/** Part of <threads.h>, but we don't use anything else from it. */ +#define thread_local _Thread_local -/** - * Silence warnings about switch/case fall-throughs. - */ -#if __has_attribute(fallthrough) -# define fallthru __attribute__((fallthrough)) -#else -# define fallthru ((void)0) -#endif +#endif // !C23 -/** - * Silence warnings about unused declarations. - */ -#if __has_attribute(unused) -# define attr_maybe_unused __attribute__((unused)) -#else -# define attr_maybe_unused -#endif +// Feature detection -/** - * Warn if a value is unused. - */ -#if __has_attribute(warn_unused_result) -# define attr_nodiscard __attribute__((warn_unused_result)) -#else -# define attr_nodiscard -#endif - -/** - * Hint to avoid inlining a function. - */ -#if __has_attribute(noinline) -# define attr_noinline __attribute__((noinline)) -#else -# define attr_noinline +// https://clang.llvm.org/docs/LanguageExtensions.html#has-attribute +#ifndef __has_attribute +# define __has_attribute(attr) false #endif -/** - * Hint that a function is unlikely to be called. - */ -#if __has_attribute(cold) -# define attr_cold attr_noinline __attribute__((cold)) -#else -# define attr_cold attr_noinline +// https://clang.llvm.org/docs/LanguageExtensions.html#has-builtin +#ifndef __has_builtin +# define __has_builtin(builtin) false #endif -/** - * Adds compiler warnings for bad printf()-style function calls, if supported. - */ -#if __has_attribute(format) -# define attr_printf(fmt, args) __attribute__((format(printf, fmt, args))) -#else -# define attr_printf(fmt, args) +// https://en.cppreference.com/w/c/language/attributes#Attribute_testing +#ifndef __has_c_attribute +# define __has_c_attribute(attr) false #endif -/** - * Annotates functions that potentially modify and return format strings. - */ -#if __has_attribute(format_arg) -# define attr_format_arg(arg) __attribute__((format_arg(arg))) -#else -# define attr_format_arg(args) +// https://clang.llvm.org/docs/LanguageExtensions.html#has-feature-and-has-extension +#ifndef __has_feature +# define __has_feature(feat) false #endif -/** - * Annotates allocator-like functions. - */ -#if __has_attribute(malloc) -# if __GNUC__ >= 11 && !__OPTIMIZE__ // malloc(deallocator) disables inlining on GCC -# define attr_malloc(...) attr_nodiscard __attribute__((malloc(__VA_ARGS__))) -# else -# define attr_malloc(...) attr_nodiscard __attribute__((malloc)) -# endif -#else -# define attr_malloc(...) attr_nodiscard +// https://en.cppreference.com/w/c/preprocessor/include +#ifndef __has_include +# define __has_include(header) false #endif -/** - * Specifies that a function returns allocations with a given alignment. - */ -#if __has_attribute(alloc_align) -# define attr_alloc_align(param) __attribute__((alloc_align(param))) -#else -# define attr_alloc_align(param) -#endif +// Sanitizer macros (GCC defines these but Clang does not) -/** - * Specifies that a function returns allocations with a given size. - */ -#if __has_attribute(alloc_size) -# define attr_alloc_size(...) __attribute__((alloc_size(__VA_ARGS__))) -#else -# define attr_alloc_size(...) +#if __has_feature(address_sanitizer) && !defined(__SANITIZE_ADDRESS__) +# define __SANITIZE_ADDRESS__ true #endif - -/** - * Shorthand for attr_alloc_align() and attr_alloc_size(). - */ -#define attr_aligned_alloc(align, ...) \ - attr_alloc_align(align) \ - attr_alloc_size(__VA_ARGS__) - -/** - * Check if function multiversioning via GNU indirect functions (ifunc) is supported. - */ -#ifndef BFS_USE_TARGET_CLONES -# if __has_attribute(target_clones) && (__GLIBC__ || __FreeBSD__) -# define BFS_USE_TARGET_CLONES true -# endif +#if __has_feature(memory_sanitizer) && !defined(__SANITIZE_MEMORY__) +# define __SANITIZE_MEMORY__ true #endif - -/** - * Apply the target_clones attribute, if available. - */ -#if BFS_USE_TARGET_CLONES -# define attr_target_clones(...) __attribute__((target_clones(__VA_ARGS__))) -#else -# define attr_target_clones(...) +#if __has_feature(thread_sanitizer) && !defined(__SANITIZE_THREAD__) +# define __SANITIZE_THREAD__ true #endif -/** - * Shorthand for multiple attributes at once. attr(a, b(c), d) is equivalent to - * - * attr_a - * attr_b(c) - * attr_d - */ -#define attr(...) \ - attr__(attr_##__VA_ARGS__, none, none, none, none, none, none, none, none, none, ) - -/** - * attr() helper. For exposition, pretend we support only 2 args, instead of 9. - * There are a few cases: - * - * attr() - * => attr__(attr_, none, none) - * => attr_ => - * attr_none => - * attr_too_many_none() => - * - * attr(a) - * => attr__(attr_a, none, none) - * => attr_a => __attribute__((a)) - * attr_none => - * attr_too_many_none() => - * - * attr(a, b(c)) - * => attr__(attr_a, b(c), none, none) - * => attr_a => __attribute__((a)) - * attr_b(c) => __attribute__((b(c))) - * attr_too_many_none(none) => - * - * attr(a, b(c), d) - * => attr__(attr_a, b(c), d, none, none) - * => attr_a => __attribute__((a)) - * attr_b(c) => __attribute__((b(c))) - * attr_too_many_d(none, none) => error - * - * Some attribute names are the same as standard library functions, e.g. printf. - * Standard libraries are permitted to define these functions as macros, like - * - * #define printf(...) __builtin_printf(__VA_ARGS__) - * - * The token paste in - * - * #define attr(...) attr__(attr_##__VA_ARGS__, none, none) - * - * is necessary to prevent macro expansion before evaluating attr__(). - * Otherwise, we could get - * - * attr(printf(1, 2)) - * => attr__(__builtin_printf(1, 2), none, none) - * => attr____builtin_printf(1, 2) - * => error - */ -#define attr__(a1, a2, a3, a4, a5, a6, a7, a8, a9, none, ...) \ - a1 \ - attr_##a2 \ - attr_##a3 \ - attr_##a4 \ - attr_##a5 \ - attr_##a6 \ - attr_##a7 \ - attr_##a8 \ - attr_##a9 \ - attr_too_many_##none(__VA_ARGS__) - -// Ignore `attr_none` from expanding 1-9 argument attr(a1, a2, ...) -#define attr_none -// Ignore `attr_` from expanding 0-argument attr() -#define attr_ -// Only trigger an error on more than 9 arguments -#define attr_too_many_none(...) - #endif // BFS_PRELUDE_H diff --git a/src/printf.c b/src/printf.c index f8428f7..30ec201 100644 --- a/src/printf.c +++ b/src/printf.c @@ -1,9 +1,10 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "printf.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "bftw.h" #include "color.h" @@ -16,6 +17,7 @@ #include "mtab.h" #include "pwcache.h" #include "stat.h" + #include <errno.h> #include <grp.h> #include <pwd.h> @@ -89,7 +91,7 @@ static bool should_color(CFILE *cfile, const struct bfs_fmt *fmt) { (void)ret /** Return a dynamic format string. */ -attr(format_arg(2)) +_format_arg(2) static const char *dyn_fmt(const char *str, const char *fake) { bfs_assert(strcmp(str + strlen(str) - strlen(fake) + 1, fake + 1) == 0, "Mismatched format specifiers: '%s' vs. '%s'", str, fake); @@ -97,7 +99,7 @@ static const char *dyn_fmt(const char *str, const char *fake) { } /** Wrapper for fprintf(). */ -attr(printf(3, 4)) +_printf(3, 4) static int bfs_fprintf(CFILE *cfile, const struct bfs_fmt *fmt, const char *fake, ...) { va_list args; va_start(args, fake); @@ -505,30 +507,25 @@ static int bfs_printf_u(CFILE *cfile, const struct bfs_fmt *fmt, const struct BF } static const char *bfs_printf_type(enum bfs_type type) { - switch (type) { - case BFS_BLK: - return "b"; - case BFS_CHR: - return "c"; - case BFS_DIR: - return "d"; - case BFS_DOOR: - return "D"; - case BFS_FIFO: - return "p"; - case BFS_LNK: - return "l"; - case BFS_PORT: - return "P"; - case BFS_REG: - return "f"; - case BFS_SOCK: - return "s"; - case BFS_WHT: - return "w"; - default: - return "U"; + const char *const names[] = { + [BFS_BLK] = "b", + [BFS_CHR] = "c", + [BFS_DIR] = "d", + [BFS_DOOR] = "D", + [BFS_FIFO] = "p", + [BFS_LNK] = "l", + [BFS_PORT] = "P", + [BFS_REG] = "f", + [BFS_SOCK] = "s", + [BFS_WHT] = "w", + }; + + const char *name = NULL; + if ((size_t)type < countof(names)) { + name = names[type]; } + + return name ? name : "U"; } /** %y: type */ @@ -544,7 +541,7 @@ static int bfs_printf_Y(CFILE *cfile, const struct bfs_fmt *fmt, const struct BF int error = 0; if (type == BFS_ERROR) { - if (errno_is_like(ELOOP)) { + if (errno == ELOOP) { str = "L"; } else if (errno_is_like(ENOENT)) { str = "N"; @@ -565,7 +562,7 @@ static int bfs_printf_Y(CFILE *cfile, const struct bfs_fmt *fmt, const struct BF } /** %Z: SELinux context */ -attr(maybe_unused) +_maybe_unused static int bfs_printf_Z(CFILE *cfile, const struct bfs_fmt *fmt, const struct BFTW *ftwbuf) { char *con = bfs_getfilecon(ftwbuf); if (!con) { @@ -709,9 +706,9 @@ int bfs_printf_parse(const struct bfs_ctx *ctx, struct bfs_expr *expr, const cha case '#': case '0': case '+': - must_be_numeric = true; - fallthru; case ' ': + must_be_numeric = true; + _fallthrough; case '-': if (strchr(fmt.str, c)) { bfs_expr_error(ctx, expr); diff --git a/src/printf.h b/src/printf.h index 2bff087..e8d862e 100644 --- a/src/printf.h +++ b/src/printf.h @@ -22,11 +22,11 @@ struct bfs_printf; /** * Parse a -printf format string. * - * @param ctx + * @ctx * The bfs context. - * @param expr + * @expr * The expression to fill in. - * @param format + * @format * The format string to parse. * @return * 0 on success, -1 on failure. @@ -36,11 +36,11 @@ int bfs_printf_parse(const struct bfs_ctx *ctx, struct bfs_expr *expr, const cha /** * Evaluate a parsed format string. * - * @param cfile + * @cfile * The CFILE to print to. - * @param format + * @format * The parsed printf format. - * @param ftwbuf + * @ftwbuf * The bftw() data for the current file. * @return * 0 on success, -1 on failure. diff --git a/src/pwcache.c b/src/pwcache.c index af8c237..fa19dad 100644 --- a/src/pwcache.c +++ b/src/pwcache.c @@ -1,16 +1,15 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "pwcache.h" + #include "alloc.h" #include "trie.h" + #include <errno.h> #include <grp.h> #include <pwd.h> #include <stdlib.h> -#include <string.h> -#include <unistd.h> /** Represents cache hits for negative results. */ static void *MISSING = &MISSING; diff --git a/src/pwcache.h b/src/pwcache.h index b6c0b67..d7c602d 100644 --- a/src/pwcache.h +++ b/src/pwcache.h @@ -27,9 +27,9 @@ struct bfs_users *bfs_users_new(void); /** * Get a user entry by name. * - * @param users + * @users * The user cache. - * @param name + * @name * The username to look up. * @return * The matching user, or NULL if not found (errno == 0) or an error @@ -40,9 +40,9 @@ const struct passwd *bfs_getpwnam(struct bfs_users *users, const char *name); /** * Get a user entry by ID. * - * @param users + * @users * The user cache. - * @param uid + * @uid * The ID to look up. * @return * The matching user, or NULL if not found (errno == 0) or an error @@ -53,7 +53,7 @@ const struct passwd *bfs_getpwuid(struct bfs_users *users, uid_t uid); /** * Flush a user cache. * - * @param users + * @users * The cache to flush. */ void bfs_users_flush(struct bfs_users *users); @@ -61,7 +61,7 @@ void bfs_users_flush(struct bfs_users *users); /** * Free a user cache. * - * @param users + * @users * The user cache to free. */ void bfs_users_free(struct bfs_users *users); @@ -82,9 +82,9 @@ struct bfs_groups *bfs_groups_new(void); /** * Get a group entry by name. * - * @param groups + * @groups * The group cache. - * @param name + * @name * The group name to look up. * @return * The matching group, or NULL if not found (errno == 0) or an error @@ -95,9 +95,9 @@ const struct group *bfs_getgrnam(struct bfs_groups *groups, const char *name); /** * Get a group entry by ID. * - * @param groups + * @groups * The group cache. - * @param uid + * @uid * The ID to look up. * @return * The matching group, or NULL if not found (errno == 0) or an error @@ -108,7 +108,7 @@ const struct group *bfs_getgrgid(struct bfs_groups *groups, gid_t gid); /** * Flush a group cache. * - * @param groups + * @groups * The cache to flush. */ void bfs_groups_flush(struct bfs_groups *groups); @@ -116,7 +116,7 @@ void bfs_groups_flush(struct bfs_groups *groups); /** * Free a group cache. * - * @param groups + * @groups * The group cache to free. */ void bfs_groups_free(struct bfs_groups *groups); diff --git a/src/sanity.h b/src/sanity.h index e168b8f..be77eef 100644 --- a/src/sanity.h +++ b/src/sanity.h @@ -8,21 +8,8 @@ #ifndef BFS_SANITY_H #define BFS_SANITY_H -#include "prelude.h" #include <stddef.h> -#if __has_feature(address_sanitizer) || defined(__SANITIZE_ADDRESS__) -# define SANITIZE_ADDRESS true -#endif - -#if __has_feature(memory_sanitizer) || defined(__SANITIZE_MEMORY__) -# define SANITIZE_MEMORY true -#endif - -#if __has_feature(thread_sanitizer) || defined(__SANITIZE_THREAD__) -# define SANITIZE_THREAD true -#endif - // Call macro(ptr, size) or macro(ptr, sizeof(*ptr)) #define SANITIZE_CALL(...) \ SANITIZE_CALL_(__VA_ARGS__, ) @@ -33,7 +20,7 @@ #define SANITIZE_CALL__(macro, ptr, size, ...) \ macro(ptr, size) -#if SANITIZE_ADDRESS +#if __SANITIZE_ADDRESS__ # include <sanitizer/asan_interface.h> /** @@ -50,12 +37,30 @@ */ #define sanitize_free(...) SANITIZE_CALL(__asan_poison_memory_region, __VA_ARGS__) +/** + * Adjust the size of an allocated region, for things like dynamic arrays. + * + * @ptr + * The memory region. + * @old + * The previous usable size of the region. + * @new + * The new usable size of the region. + * @cap + * The total allocated capacity of the region. + */ +static inline void sanitize_resize(const void *ptr, size_t old, size_t new, size_t cap) { + const char *beg = ptr; + __sanitizer_annotate_contiguous_container(beg, beg + cap, beg + old, beg + new); +} + #else -# define sanitize_alloc sanitize_uninit -# define sanitize_free sanitize_uninit +# define sanitize_alloc(...) ((void)0) +# define sanitize_free(...) ((void)0) +# define sanitize_resize(ptr, old, new, cap) ((void)0) #endif -#if SANITIZE_MEMORY +#if __SANITIZE_MEMORY__ # include <sanitizer/msan_interface.h> /** @@ -73,19 +78,14 @@ #define sanitize_uninit(...) SANITIZE_CALL(__msan_allocated_memory, __VA_ARGS__) #else -# define sanitize_init(...) SANITIZE_CALL(sanitize_ignore, __VA_ARGS__) -# define sanitize_uninit(...) SANITIZE_CALL(sanitize_ignore, __VA_ARGS__) +# define sanitize_init(...) ((void)0) +# define sanitize_uninit(...) ((void)0) #endif /** - * Squelch unused variable warnings when not sanitizing. - */ -#define sanitize_ignore(ptr, size) ((void)(ptr), (void)(size)) - -/** * Initialize a variable, unless sanitizers would detect uninitialized uses. */ -#if SANITIZE_MEMORY +#if __SANITIZE_MEMORY__ # define uninit(value) #else # define uninit(value) = value diff --git a/src/sighook.c b/src/sighook.c new file mode 100644 index 0000000..a87bed5 --- /dev/null +++ b/src/sighook.c @@ -0,0 +1,692 @@ +// Copyright © Tavian Barnes <tavianator@tavianator.com> +// SPDX-License-Identifier: 0BSD + +/** + * Dynamic (un)registration of signal handlers. + * + * Because signal handlers can interrupt any thread at an arbitrary point, they + * must be lock-free or risk deadlock. Therefore, we implement the global table + * of signal "hooks" with a simple read-copy-update (RCU) scheme. Readers get a + * reference-counted pointer (struct arc) to the table in a lock-free way, and + * release the reference count when finished. + * + * Updates are managed by struct rcu, which has two slots: one active and one + * inactive. Readers acquire a reference to the active slot. A single writer + * can safely update it by initializing the inactive slot, atomically swapping + * the slots, and waiting for the reference count of the newly inactive slot to + * drop to zero. Once it does, the old pointer can be safely freed. + */ + +#include "sighook.h" + +#include "alloc.h" +#include "atomic.h" +#include "bfs.h" +#include "bfstd.h" +#include "diag.h" +#include "thread.h" + +#include <errno.h> +#include <pthread.h> +#include <signal.h> +#include <stdlib.h> +#include <unistd.h> + +#if __linux__ +# include <sys/syscall.h> +#endif + +// NetBSD opens a file descriptor for each sem_init() +#if defined(_POSIX_SEMAPHORES) && !__NetBSD__ +# define BFS_POSIX_SEMAPHORES _POSIX_SEMAPHORES +#else +# define BFS_POSIX_SEMAPHORES (-1) +#endif + +#if BFS_POSIX_SEMAPHORES >= 0 +# include <semaphore.h> +#endif + +/** + * An atomically reference-counted pointer. + */ +struct arc { + /** The current reference count (0 means empty). */ + atomic size_t refs; + /** The reference itself. */ + void *ptr; + +#if BFS_POSIX_SEMAPHORES >= 0 + /** A semaphore for arc_wait(). */ + sem_t sem; + /** sem_init() result. */ + int sem_status; +#endif +}; + +/** Initialize an arc. */ +static void arc_init(struct arc *arc) { + bfs_verify(atomic_is_lock_free(&arc->refs)); + + atomic_init(&arc->refs, 0); + arc->ptr = NULL; + +#if BFS_POSIX_SEMAPHORES >= 0 + if (sysoption(SEMAPHORES) > 0) { + arc->sem_status = sem_init(&arc->sem, false, 0); + } else { + arc->sem_status = -1; + } +#endif +} + +/** Get the current refcount. */ +static size_t arc_refs(const struct arc *arc) { + return load(&arc->refs, relaxed); +} + +/** Set the pointer in an empty arc. */ +static void arc_set(struct arc *arc, void *ptr) { + bfs_assert(arc_refs(arc) == 0); + bfs_assert(ptr); + + arc->ptr = ptr; + store(&arc->refs, 1, release); +} + +/** Acquire a reference. */ +static void *arc_get(struct arc *arc) { + size_t refs = arc_refs(arc); + do { + if (refs < 1) { + return NULL; + } + } while (!compare_exchange_weak(&arc->refs, &refs, refs + 1, acquire, relaxed)); + + return arc->ptr; +} + +/** Release a reference. */ +static void arc_put(struct arc *arc) { + size_t refs = fetch_sub(&arc->refs, 1, release); + + if (refs == 1) { +#if BFS_POSIX_SEMAPHORES >= 0 + if (arc->sem_status == 0 && sem_post(&arc->sem) != 0) { + abort(); + } +#endif + } +} + +/** Wait on the semaphore. */ +static int arc_sem_wait(struct arc *arc) { +#if BFS_POSIX_SEMAPHORES >= 0 + if (arc->sem_status == 0) { + while (sem_wait(&arc->sem) != 0) { + bfs_everify(errno == EINTR, "sem_wait()"); + } + return 0; + } +#endif + + return -1; +} + +/** Wait for all references to be released. */ +static void *arc_wait(struct arc *arc) { + size_t refs = fetch_sub(&arc->refs, 1, relaxed); + bfs_assert(refs > 0); + + --refs; + while (refs > 0) { + if (arc_sem_wait(arc) == 0) { + bfs_assert(arc_refs(arc) == 0); + // sem_wait() provides enough ordering, so we can skip the fence + goto done; + } + + // Some platforms (like macOS) don't support unnamed semaphores, + // but we can always busy-wait + spin_loop(); + refs = arc_refs(arc); + } + + thread_fence(&arc->refs, acquire); + +done:; + void *ptr = arc->ptr; + arc->ptr = NULL; + return ptr; +} + +/** Destroy an arc. */ +static void arc_destroy(struct arc *arc) { + bfs_assert(arc_refs(arc) == 0); + +#if BFS_POSIX_SEMAPHORES >= 0 + if (arc->sem_status == 0) { + bfs_everify(sem_destroy(&arc->sem) == 0, "sem_destroy()"); + } +#endif +} + +/** + * A simple read-copy-update memory reclamation scheme. + */ +struct rcu { + /** The currently active slot. */ + atomic size_t active; + /** The two slots. */ + struct arc slots[2]; +}; + +/** Sentinel value for RCU, since arc uses NULL already. */ +static void *RCU_NULL = &RCU_NULL; + +/** Map NULL -> RCU_NULL. */ +static void *rcu_encode(void *ptr) { + return ptr ? ptr : RCU_NULL; +} + +/** Map RCU_NULL -> NULL. */ +static void *rcu_decode(void *ptr) { + bfs_assert(ptr != NULL); + return ptr == RCU_NULL ? NULL : ptr; +} + +/** Initialize an RCU block. */ +static void rcu_init(struct rcu *rcu, void *ptr) { + bfs_verify(atomic_is_lock_free(&rcu->active)); + + atomic_init(&rcu->active, 0); + arc_init(&rcu->slots[0]); + arc_init(&rcu->slots[1]); + arc_set(&rcu->slots[0], rcu_encode(ptr)); +} + +/** Get the active slot. */ +static struct arc *rcu_active(struct rcu *rcu) { + size_t i = load(&rcu->active, relaxed); + return &rcu->slots[i]; +} + +/** Destroy an RCU block. */ +static void rcu_destroy(struct rcu *rcu) { + arc_wait(rcu_active(rcu)); + arc_destroy(&rcu->slots[1]); + arc_destroy(&rcu->slots[0]); +} + +/** Read an RCU-protected pointer. */ +static void *rcu_read(struct rcu *rcu, struct arc **slot) { + while (true) { + *slot = rcu_active(rcu); + void *ptr = arc_get(*slot); + if (ptr) { + return rcu_decode(ptr); + } + // Otherwise, the other slot became active; retry + } +} + +/** Get the RCU-protected pointer without acquiring a reference. */ +static void *rcu_peek(struct rcu *rcu) { + struct arc *arc = rcu_active(rcu); + return rcu_decode(arc->ptr); +} + +/** Update an RCU-protected pointer, and return the old one. */ +static void *rcu_update(struct rcu *rcu, void *ptr) { + size_t i = load(&rcu->active, relaxed); + struct arc *prev = &rcu->slots[i]; + + size_t j = i ^ 1; + struct arc *next = &rcu->slots[j]; + + arc_set(next, rcu_encode(ptr)); + store(&rcu->active, j, relaxed); + return rcu_decode(arc_wait(prev)); +} + +/** + * An RCU-protected linked list. + */ +struct rcu_list { + /** The first node in the list. */ + struct rcu head; + /** &last->next */ + struct rcu *tail; +}; + +/** + * An rcu_list node. + */ +struct rcu_node { + /** The RCU pointer to this node. */ + struct rcu *self; + /** The next node in the list. */ + struct rcu next; +}; + +/** Initialize an rcu_list. */ +static void rcu_list_init(struct rcu_list *list) { + rcu_init(&list->head, NULL); + list->tail = &list->head; +} + +/** Append to an rcu_list. */ +static void rcu_list_append(struct rcu_list *list, struct rcu_node *node) { + node->self = list->tail; + list->tail = &node->next; + rcu_init(&node->next, NULL); + rcu_update(node->self, node); +} + +/** Remove from an rcu_list. */ +static void rcu_list_remove(struct rcu_list *list, struct rcu_node *node) { + struct rcu_node *next = rcu_peek(&node->next); + rcu_update(node->self, next); + if (next) { + next->self = node->self; + } else { + list->tail = &list->head; + } + rcu_destroy(&node->next); +} + +/** + * Iterate over an rcu_list. + * + * It is save to `break` out of this loop, but `return` or `goto` will lead to + * a missed arc_put(). + */ +#define for_rcu(type, node, list) \ + for_rcu_(type, node, (list), node##_slot_, node##_prev_, node##_done_) + +#define for_rcu_(type, node, list, slot, prev, done) \ + for (struct arc *slot, *prev, **done = NULL; !done; arc_put(slot), done = &slot) \ + for (type *node = rcu_read(&list->head, &slot); \ + node; \ + prev = slot, \ + node = rcu_read(&((struct rcu_node *)node)->next, &slot), \ + arc_put(prev)) + +struct sighook { + /** The RCU list node (must be the first field). */ + struct rcu_node node; + + /** The signal being hooked, or 0 for atsigexit(). */ + int sig; + /** Signal hook flags. */ + enum sigflags flags; + /** The function to call. */ + sighook_fn *fn; + /** An argument to pass to the function. */ + void *arg; + /** Flag for SH_ONESHOT. */ + atomic bool armed; +}; + +/** The lists of signal hooks. */ +static struct rcu_list sighooks[64]; + +/** Get the hook list for a particular signal. */ +static struct rcu_list *siglist(int sig) { + return &sighooks[sig % countof(sighooks)]; +} + +/** Mutex for initialization and RCU writer exclusion. */ +static pthread_mutex_t sigmutex = PTHREAD_MUTEX_INITIALIZER; + +/** Check if a signal was generated by userspace. */ +static bool is_user_generated(const siginfo_t *info) { + // https://pubs.opengroup.org/onlinepubs/9799919799/functions/V2_chap02.html#tag_16_04_03_03 + // + // If si_code is SI_USER or SI_QUEUE, or any value less than or + // equal to 0, then the signal was generated by a process ... + int code = info->si_code; + return code == SI_USER || code == SI_QUEUE || code <= 0; +} + +/** Check if a signal is caused by a fault. */ +static bool is_fault(const siginfo_t *info) { + int sig = info->si_signo; + if (sig == SIGBUS || sig == SIGFPE || sig == SIGILL || sig == SIGSEGV) { + return !is_user_generated(info); + } else { + return false; + } +} + +// https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/signal.h.html +static const int FATAL_SIGNALS[] = { + SIGABRT, + SIGALRM, + SIGBUS, + SIGFPE, + SIGHUP, + SIGILL, + SIGINT, +#ifdef SIGIO + SIGIO, +#endif + SIGPIPE, +#ifdef SIGPOLL + SIGPOLL, +#endif +#ifdef SIGPROF + SIGPROF, +#endif +#ifdef SIGPWR + SIGPWR, +#endif + SIGQUIT, + SIGSEGV, +#ifdef SIGSTKFLT + SIGSTKFLT, +#endif +#ifdef SIGSYS + SIGSYS, +#endif + SIGTERM, + SIGTRAP, + SIGUSR1, + SIGUSR2, +#ifdef SIGVTALRM + SIGVTALRM, +#endif + SIGXCPU, + SIGXFSZ, +}; + +/** Check if a signal's default action is to terminate the process. */ +static bool is_fatal(int sig) { + for (size_t i = 0; i < countof(FATAL_SIGNALS); ++i) { + if (sig == FATAL_SIGNALS[i]) { + return true; + } + } + +#ifdef SIGRTMIN + // https://pubs.opengroup.org/onlinepubs/9799919799/functions/V2_chap02.html#tag_16_04_03_01 + // + // The default actions for the realtime signals in the range + // SIGRTMIN to SIGRTMAX shall be to terminate the process + // abnormally. + if (sig >= SIGRTMIN && sig <= SIGRTMAX) { + return true; + } +#endif + + return false; +} + +/** Reraise a fatal signal. */ +_noreturn +static void reraise(siginfo_t *info) { + int sig = info->si_signo; + + // Restore the default signal action + if (signal(sig, SIG_DFL) == SIG_ERR) { + goto fail; + } + + // Unblock the signal, since we didn't set SA_NODEFER + sigset_t mask; + if (sigemptyset(&mask) != 0 + || sigaddset(&mask, sig) != 0 + || pthread_sigmask(SIG_UNBLOCK, &mask, NULL) != 0) { + goto fail; + } + +#if __linux__ + // On Linux, try to re-raise the exact siginfo_t (since 3.9, a process can + // signal itself with any siginfo_t) + pid_t tid = syscall(SYS_gettid); + syscall(SYS_rt_tgsigqueueinfo, getpid(), tid, sig, info); +#endif + + raise(sig); +fail: + abort(); +} + +/** Check whether we should run a hook. */ +static bool should_run(int sig, struct sighook *hook) { + if (hook->sig != sig && hook->sig != 0) { + return false; + } + + if (hook->flags & SH_ONESHOT) { + if (!exchange(&hook->armed, false, relaxed)) { + return false; + } + } + + return true; +} + +/** Find any matching hooks and run them. */ +static enum sigflags run_hooks(struct rcu_list *list, int sig, siginfo_t *info) { + enum sigflags ret = 0; + + for_rcu (struct sighook, hook, list) { + if (should_run(sig, hook)) { + hook->fn(sig, info, hook->arg); + ret |= hook->flags; + } + } + + return ret; +} + +/** Dispatches a signal to the registered handlers. */ +static void sigdispatch(int sig, siginfo_t *info, void *context) { + // If we get a fault (e.g. a "real" SIGSEGV, not something like + // kill(..., SIGSEGV)), don't try to run signal hooks, since we could be + // in an arbitrarily corrupted state. + // + // POSIX says that returning normally from a signal handler for a fault + // is undefined. But in practice, it's better to uninstall the handler + // and return, which will re-run the faulting instruction and cause us + // to die "correctly" (e.g. with a core dump pointing at the faulting + // instruction, not reraise()). + if (is_fault(info)) { + // On macOS, we cannot reliably distinguish between faults and + // asynchronous signals. For example, pkill -SEGV bfs will + // result in si_code == SEGV_ACCERR. So we always re-raise the + // signal, because just returning would cause us to ignore + // asynchronous SIG{BUS,ILL,SEGV}. +#if !__APPLE__ + if (signal(sig, SIG_DFL) != SIG_ERR) { + return; + } +#endif + reraise(info); + } + + // https://pubs.opengroup.org/onlinepubs/9799919799/functions/V2_chap02.html#tag_16_04_04 + // + // After returning from a signal-catching function, the value of + // errno is unspecified if the signal-catching function or any + // function it called assigned a value to errno and the signal- + // catching function did not save and restore the original value of + // errno. + int error = errno; + + // Run the normal hooks + struct rcu_list *list = siglist(sig); + enum sigflags flags = run_hooks(list, sig, info); + + // Run the atsigexit() hooks, if we're exiting + if (!(flags & SH_CONTINUE) && is_fatal(sig)) { + list = siglist(0); + run_hooks(list, sig, info); + reraise(info); + } + + errno = error; +} + +/** A saved signal handler, for sigreset() to restore. */ +struct sigsave { + struct rcu_node node; + int sig; + struct sigaction action; +}; + +/** The list of saved signal handlers. */ +static struct rcu_list saved; +/** `saved` initialization status (since rcu_list_init() isn't atomic). */ +static atomic bool initialized = false; + +/** Make sure our signal handler is installed for a given signal. */ +static int siginit(int sig) { +#ifdef SA_RESTART +# define BFS_SA_RESTART SA_RESTART +#else +# define BFS_SA_RESTART 0 +#endif + + static struct sigaction action = { + .sa_sigaction = sigdispatch, + .sa_flags = BFS_SA_RESTART | SA_SIGINFO, + }; + + static sigset_t signals; + + if (!load(&initialized, relaxed)) { + if (sigemptyset(&signals) != 0 + || sigemptyset(&action.sa_mask) != 0) { + return -1; + } + + for (size_t i = 0; i < countof(sighooks); ++i) { + rcu_list_init(&sighooks[i]); + } + + rcu_list_init(&saved); + store(&initialized, true, release); + } + + int installed = sigismember(&signals, sig); + if (installed < 0) { + return -1; + } else if (installed) { + return 0; + } + + sigset_t updated = signals; + if (sigaddset(&updated, sig) != 0) { + return -1; + } + + struct sigaction original; + if (sigaction(sig, NULL, &original) != 0) { + return -1; + } + + struct sigsave *save = ALLOC(struct sigsave); + if (!save) { + return -1; + } + + save->sig = sig; + save->action = original; + rcu_list_append(&saved, &save->node); + + if (sigaction(sig, &action, NULL) != 0) { + rcu_list_remove(&saved, &save->node); + free(save); + return -1; + } + + signals = updated; + return 0; +} + +/** Shared sighook()/atsigexit() implementation. */ +static struct sighook *sighook_impl(int sig, sighook_fn *fn, void *arg, enum sigflags flags) { + struct sighook *hook = ALLOC(struct sighook); + if (!hook) { + return NULL; + } + + hook->sig = sig; + hook->flags = flags; + hook->fn = fn; + hook->arg = arg; + atomic_init(&hook->armed, true); + + struct rcu_list *list = siglist(sig); + rcu_list_append(list, &hook->node); + return hook; +} + +struct sighook *sighook(int sig, sighook_fn *fn, void *arg, enum sigflags flags) { + bfs_assert(sig > 0); + + mutex_lock(&sigmutex); + + struct sighook *ret = NULL; + if (siginit(sig) == 0) { + ret = sighook_impl(sig, fn, arg, flags); + } + + mutex_unlock(&sigmutex); + return ret; +} + +struct sighook *atsigexit(sighook_fn *fn, void *arg) { + mutex_lock(&sigmutex); + + for (size_t i = 0; i < countof(FATAL_SIGNALS); ++i) { + // Ignore errors; atsigexit() is best-effort anyway and things + // like sanitizer runtimes or valgrind may reserve signals for + // their own use + siginit(FATAL_SIGNALS[i]); + } + +#ifdef SIGRTMIN + for (int i = SIGRTMIN; i <= SIGRTMAX; ++i) { + siginit(i); + } +#endif + + struct sighook *ret = sighook_impl(0, fn, arg, 0); + mutex_unlock(&sigmutex); + return ret; +} + +void sigunhook(struct sighook *hook) { + if (!hook) { + return; + } + + mutex_lock(&sigmutex); + + struct rcu_list *list = siglist(hook->sig); + rcu_list_remove(list, &hook->node); + + mutex_unlock(&sigmutex); + + free(hook); +} + +int sigreset(void) { + if (!load(&initialized, acquire)) { + return 0; + } + + int ret = 0; + + for_rcu (struct sigsave, save, &saved) { + if (sigaction(save->sig, &save->action, NULL) != 0) { + ret = -1; + break; + } + } + + return ret; +} diff --git a/src/sighook.h b/src/sighook.h new file mode 100644 index 0000000..7149229 --- /dev/null +++ b/src/sighook.h @@ -0,0 +1,83 @@ +// Copyright © Tavian Barnes <tavianator@tavianator.com> +// SPDX-License-Identifier: 0BSD + +/** + * Signal hooks. + */ + +#ifndef BFS_SIGHOOK_H +#define BFS_SIGHOOK_H + +#include <signal.h> + +/** + * A dynamic signal hook. + */ +struct sighook; + +/** + * Signal hook flags. + */ +enum sigflags { + /** Suppress the default action for this signal. */ + SH_CONTINUE = 1 << 0, + /** Only run this hook once. */ + SH_ONESHOT = 1 << 1, +}; + +/** + * A signal hook callback. Hooks are executed from a signal handler, so must + * only call async-signal-safe functions. + * + * @sig + * The signal number. + * @info + * Additional information about the signal. + * @arg + * An arbitrary pointer passed to the hook. + */ +typedef void sighook_fn(int sig, siginfo_t *info, void *arg); + +/** + * Install a hook for a signal. + * + * @sig + * The signal to hook. + * @fn + * The function to call. + * @arg + * An argument passed to the function. + * @flags + * Flags for the new hook. + * @return + * The installed hook, or NULL on failure. + */ +struct sighook *sighook(int sig, sighook_fn *fn, void *arg, enum sigflags flags); + +/** + * On a best-effort basis, invoke the given hook just before the program is + * abnormally terminated by a signal. + * + * @fn + * The function to call. + * @arg + * An argument passed to the function. + * @return + * The installed hook, or NULL on failure. + */ +struct sighook *atsigexit(sighook_fn *fn, void *arg); + +/** + * Remove a signal hook. + */ +void sigunhook(struct sighook *hook); + +/** + * Restore all signal handlers to their original dispositions (e.g. after fork()). + * + * @return + * 0 on success, -1 on failure. + */ +int sigreset(void); + +#endif // BFS_SIGHOOK_H @@ -1,12 +1,14 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "stat.h" + #include "atomic.h" +#include "bfs.h" #include "bfstd.h" #include "diag.h" #include "sanity.h" + #include <errno.h> #include <fcntl.h> #include <string.h> @@ -49,9 +51,11 @@ const char *bfs_stat_field_name(enum bfs_stat_field field) { return "change time"; case BFS_STAT_MTIME: return "modification time"; + case BFS_STAT_MNT_ID: + return "mount ID"; } - bfs_bug("Unrecognized stat field"); + bfs_bug("Unrecognized stat field %d", (int)field); return "???"; } @@ -62,7 +66,7 @@ int bfs_fstatat_flags(enum bfs_stat_flags flags) { ret |= AT_SYMLINK_NOFOLLOW; } -#if defined(AT_NO_AUTOMOUNT) && (!__GNU__ || __GLIBC_PREREQ(2, 35)) +#ifdef AT_NO_AUTOMOUNT ret |= AT_NO_AUTOMOUNT; #endif @@ -99,6 +103,10 @@ void bfs_stat_convert(struct bfs_stat *dest, const struct stat *src) { dest->rdev = src->st_rdev; dest->mask |= BFS_STAT_RDEV; + // No mount IDs in regular stat(), so use the dev_t as an approximation + dest->mnt_id = dest->dev; + dest->mask |= BFS_STAT_MNT_ID; + #if BFS_HAS_ST_FLAGS dest->attrs = src->st_flags; dest->mask |= BFS_STAT_ATTRS; @@ -116,6 +124,9 @@ void bfs_stat_convert(struct bfs_stat *dest, const struct stat *src) { #if BFS_HAS_ST_BIRTHTIM dest->btime = src->st_birthtim; dest->mask |= BFS_STAT_BTIME; +#elif BFS_HAS___ST_BIRTHTIM + dest->btime = src->__st_birthtim; + dest->mask |= BFS_STAT_BTIME; #elif BFS_HAS_ST_BIRTHTIMESPEC dest->btime = src->st_birthtimespec; dest->mask |= BFS_STAT_BTIME; @@ -164,6 +175,17 @@ int bfs_statx_flags(enum bfs_stat_flags flags) { return ret; } +unsigned int bfs_statx_mask(void) { + unsigned int mask = STATX_BASIC_STATS | STATX_BTIME; +#ifdef STATX_MNT_ID + mask |= STATX_MNT_ID; +#endif +#ifdef STATX_MNT_ID_UNIQUE + mask |= STATX_MNT_ID_UNIQUE; +#endif + return mask; +} + int bfs_statx_convert(struct bfs_stat *dest, const struct statx *src) { // Callers shouldn't have to check anything except the times const unsigned int guaranteed = STATX_BASIC_STATS & ~(STATX_ATIME | STATX_CTIME | STATX_MTIME); @@ -204,6 +226,18 @@ int bfs_statx_convert(struct bfs_stat *dest, const struct statx *src) { dest->attrs = src->stx_attributes; dest->mask |= BFS_STAT_ATTRS; + dest->mnt_id = dest->dev; +#ifdef STATX_MNT_ID + unsigned int mnt_mask = STATX_MNT_ID; +# ifdef STATX_MNT_ID_UNIQUE + mnt_mask |= STATX_MNT_ID_UNIQUE; +# endif + if (src->stx_mask & mnt_mask) { + dest->mnt_id = src->stx_mnt_id; + } +#endif + dest->mask |= BFS_STAT_MNT_ID; + if (src->stx_mask & STATX_ATIME) { dest->atime.tv_sec = src->stx_atime.tv_sec; dest->atime.tv_nsec = src->stx_atime.tv_nsec; @@ -235,7 +269,7 @@ int bfs_statx_convert(struct bfs_stat *dest, const struct statx *src) { * bfs_stat() implementation backed by statx(). */ static int bfs_statx_impl(int at_fd, const char *at_path, int at_flags, struct bfs_stat *buf) { - unsigned int mask = STATX_BASIC_STATS | STATX_BTIME; + unsigned int mask = bfs_statx_mask(); struct statx xbuf; int ret = bfs_statx(at_fd, at_path, at_flags, mask, &xbuf); if (ret != 0) { @@ -297,27 +331,21 @@ int bfs_stat(int at_fd, const char *at_path, enum bfs_stat_flags flags, struct b return bfs_stat_tryfollow(at_fd, at_path, at_flags, flags, buf); } - // Check __GNU__ to work around https://lists.gnu.org/archive/html/bug-hurd/2021-12/msg00001.html -#if defined(AT_EMPTY_PATH) && !__GNU__ - static atomic bool has_at_ep = true; - if (load(&has_at_ep, relaxed)) { - at_flags |= AT_EMPTY_PATH; - int ret = bfs_stat_explicit(at_fd, "", at_flags, buf); - if (ret != 0 && errno == EINVAL) { - store(&has_at_ep, false, relaxed); - } else { - return ret; - } - } -#endif - - struct stat statbuf; - if (fstat(at_fd, &statbuf) == 0) { - bfs_stat_convert(buf, &statbuf); - return 0; - } else { +#if BFS_USE_STATX + // If we have statx(), use it with AT_EMPTY_PATH for its extra features + at_flags |= AT_EMPTY_PATH; + return bfs_stat_explicit(at_fd, "", at_flags, buf); +#else + // Otherwise, just use fstat() rather than fstatat(at_fd, ""), to save + // the kernel the trouble of copying in the empty string + struct stat sb; + if (fstat(at_fd, &sb) != 0) { return -1; } + + bfs_stat_convert(buf, &sb); + return 0; +#endif } const struct timespec *bfs_stat_time(const struct bfs_stat *buf, enum bfs_stat_field field) { @@ -12,7 +12,9 @@ #ifndef BFS_STAT_H #define BFS_STAT_H -#include "prelude.h" +#include "bfs.h" + +#include <stdint.h> #include <sys/stat.h> #include <sys/types.h> #include <time.h> @@ -25,7 +27,7 @@ # define BFS_USE_STATX (BFS_HAS_STATX || BFS_HAS_STATX_SYSCALL) #endif -#if BFS_USE_SYS_PARAM_H +#if __has_include(<sys/param.h>) # include <sys/param.h> #endif @@ -55,6 +57,7 @@ enum bfs_stat_field { BFS_STAT_BTIME = 1 << 11, BFS_STAT_CTIME = 1 << 12, BFS_STAT_MTIME = 1 << 13, + BFS_STAT_MNT_ID = 1 << 14, }; /** @@ -101,6 +104,8 @@ struct bfs_stat { blkcnt_t blocks; /** The device ID represented by this file. */ dev_t rdev; + /** The ID of the mount point containing this file. */ + uint64_t mnt_id; /** Attributes/flags set on the file. */ unsigned long long attrs; @@ -118,14 +123,14 @@ struct bfs_stat { /** * Facade over fstatat(). * - * @param at_fd + * @at_fd * The base file descriptor for the lookup. - * @param at_path + * @at_path * The path to stat, relative to at_fd. Pass NULL to fstat() at_fd * itself. - * @param flags + * @flags * Flags that affect the lookup. - * @param[out] buf + * @buf[out] * A place to store the stat buffer, if successful. * @return * 0 on success, -1 on error. @@ -149,6 +154,11 @@ void bfs_stat_convert(struct bfs_stat *dest, const struct stat *src); int bfs_statx_flags(enum bfs_stat_flags flags); /** + * Get the default statx() mask. + */ +unsigned int bfs_statx_mask(void); + +/** * Convert struct statx to struct bfs_stat. */ int bfs_statx_convert(struct bfs_stat *dest, const struct statx *src); diff --git a/src/thread.c b/src/thread.c index 3793896..b3604f8 100644 --- a/src/thread.c +++ b/src/thread.c @@ -1,13 +1,18 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "thread.h" + #include "bfstd.h" #include "diag.h" + #include <errno.h> #include <pthread.h> +#if __has_include(<pthread_np.h>) +# include <pthread_np.h> +#endif + #define THREAD_FALLIBLE(expr) \ do { \ int err = expr; \ @@ -31,6 +36,14 @@ int thread_create(pthread_t *thread, const pthread_attr_t *attr, thread_fn *fn, THREAD_FALLIBLE(pthread_create(thread, attr, fn, arg)); } +void thread_setname(pthread_t thread, const char *name) { +#if BFS_HAS_PTHREAD_SETNAME_NP + pthread_setname_np(thread, name); +#elif BFS_HAS_PTHREAD_SET_NAME_NP + pthread_set_name_np(thread, name); +#endif +} + void thread_join(pthread_t thread, void **ret) { THREAD_INFALLIBLE(pthread_join(thread, ret)); } diff --git a/src/thread.h b/src/thread.h index db11bd8..3dd8422 100644 --- a/src/thread.h +++ b/src/thread.h @@ -8,17 +8,8 @@ #ifndef BFS_THREAD_H #define BFS_THREAD_H -#include "prelude.h" #include <pthread.h> -#if __STDC_VERSION__ < C23 && !defined(thread_local) -# if BFS_USE_THREADS_H -# include <threads.h> -# else -# define thread_local _Thread_local -# endif -#endif - /** Thread entry point type. */ typedef void *thread_fn(void *arg); @@ -31,6 +22,11 @@ typedef void *thread_fn(void *arg); int thread_create(pthread_t *thread, const pthread_attr_t *attr, thread_fn *fn, void *arg); /** + * Set the name of a thread. + */ +void thread_setname(pthread_t thread, const char *name); + +/** * Wrapper for pthread_join(). */ void thread_join(pthread_t thread, void **ret); @@ -81,21 +81,23 @@ * and insert intermediate singleton "jump" nodes when necessary. */ -#include "prelude.h" #include "trie.h" + #include "alloc.h" +#include "bfs.h" #include "bit.h" #include "diag.h" #include "list.h" + #include <stdint.h> #include <string.h> -bfs_static_assert(CHAR_WIDTH == 8); +static_assert(CHAR_WIDTH == 8, "This trie implementation assumes 8-bit bytes."); #if __i386__ || __x86_64__ -# define trie_clones attr(target_clones("popcnt", "default")) +# define _trie_clones _target_clones("popcnt", "default") #else -# define trie_clones +# define _trie_clones #endif /** Number of bits for the sparse array bitmap, aka the range of a nibble. */ @@ -127,37 +129,37 @@ struct trie_node { * tag to distinguish internal nodes from leaves. This is safe as long * as all dynamic allocations are aligned to more than a single byte. */ - uintptr_t children[]; + uintptr_t children[]; // _counted_by(count_ones(bitmap)) }; -/** Check if an encoded pointer is to a leaf. */ -static bool trie_is_leaf(uintptr_t ptr) { +/** Check if an encoded pointer is to an internal node. */ +static bool trie_is_node(uintptr_t ptr) { return ptr & 1; } -/** Decode a pointer to a leaf. */ -static struct trie_leaf *trie_decode_leaf(uintptr_t ptr) { - bfs_assert(trie_is_leaf(ptr)); - return (struct trie_leaf *)(ptr ^ 1); +/** Decode a pointer to an internal node. */ +static struct trie_node *trie_decode_node(uintptr_t ptr) { + bfs_assert(trie_is_node(ptr)); + return (struct trie_node *)(ptr - 1); } -/** Encode a pointer to a leaf. */ -static uintptr_t trie_encode_leaf(const struct trie_leaf *leaf) { - uintptr_t ptr = (uintptr_t)leaf ^ 1; - bfs_assert(trie_is_leaf(ptr)); +/** Encode a pointer to an internal node. */ +static uintptr_t trie_encode_node(const struct trie_node *node) { + uintptr_t ptr = (uintptr_t)node + 1; + bfs_assert(trie_is_node(ptr)); return ptr; } -/** Decode a pointer to an internal node. */ -static struct trie_node *trie_decode_node(uintptr_t ptr) { - bfs_assert(!trie_is_leaf(ptr)); - return (struct trie_node *)ptr; +/** Decode a pointer to a leaf. */ +static struct trie_leaf *trie_decode_leaf(uintptr_t ptr) { + bfs_assert(!trie_is_node(ptr)); + return (struct trie_leaf *)ptr; } -/** Encode a pointer to an internal node. */ -static uintptr_t trie_encode_node(const struct trie_node *node) { - uintptr_t ptr = (uintptr_t)node; - bfs_assert(!trie_is_leaf(ptr)); +/** Encode a pointer to a leaf. */ +static uintptr_t trie_encode_leaf(const struct trie_leaf *leaf) { + uintptr_t ptr = (uintptr_t)leaf; + bfs_assert(!trie_is_node(ptr)); return ptr; } @@ -169,20 +171,32 @@ void trie_init(struct trie *trie) { } /** Extract the nibble at a certain offset from a byte sequence. */ -static unsigned char trie_key_nibble(const void *key, size_t offset) { +static unsigned char trie_key_nibble(const void *key, size_t length, size_t offset) { const unsigned char *bytes = key; - size_t byte = offset >> 1; + size_t byte = offset / 2; + bfs_assert(byte < length); // A branchless version of // if (offset & 1) { - // return bytes[byte] >> 4; - // } else { // return bytes[byte] & 0xF; + // } else { + // return bytes[byte] >> 4; // } - unsigned int shift = (offset & 1) << 2; + unsigned int shift = 4 * ((offset + 1) % 2); return (bytes[byte] >> shift) & 0xF; } +/** Extract the nibble at a certain offset from a leaf. */ +static unsigned char trie_leaf_nibble(const struct trie_leaf *leaf, size_t offset) { + return trie_key_nibble(leaf->key, leaf->length, offset); +} + +/** Get the number of children of an internal node. */ +_trie_clones +static unsigned int trie_node_size(const struct trie_node *node) { + return count_ones((unsigned int)node->bitmap); +} + /** * Finds a leaf in the trie that matches the key at every branch. If the key * exists in the trie, the representative will match the searched key. But @@ -190,26 +204,24 @@ static unsigned char trie_key_nibble(const void *key, size_t offset) { * that case, the first mismatch between the key and the representative will be * the depth at which to make a new branch to insert the key. */ -trie_clones +_trie_clones static struct trie_leaf *trie_representative(const struct trie *trie, const void *key, size_t length) { uintptr_t ptr = trie->root; - if (!ptr) { - return NULL; - } - size_t offset = 0; - while (!trie_is_leaf(ptr)) { + size_t offset = 0, limit = 2 * length; + while (trie_is_node(ptr)) { struct trie_node *node = trie_decode_node(ptr); offset += node->offset; unsigned int index = 0; - if ((offset >> 1) < length) { - unsigned char nibble = trie_key_nibble(key, offset); + if (offset < limit) { + unsigned char nibble = trie_key_nibble(key, length, offset); unsigned int bit = 1U << nibble; - // bits = bitmap & bit ? bitmap & (bit - 1) : 0 - unsigned int mask = -!!(node->bitmap & bit); - unsigned int bits = node->bitmap & (bit - 1) & mask; - index = count_ones(bits); + unsigned int map = node->bitmap; + unsigned int bits = map & (bit - 1); + unsigned int mask = -!!(map & bit); + // index = (map & bit) ? count_ones(bits) : 0; + index = count_ones(bits) & mask; } ptr = node->children[index]; } @@ -221,7 +233,8 @@ struct trie_leaf *trie_find_str(const struct trie *trie, const char *key) { return trie_find_mem(trie, key, strlen(key) + 1); } -struct trie_leaf *trie_find_mem(const struct trie *trie, const void *key, size_t length) { +_trie_clones +static struct trie_leaf *trie_find_mem_impl(const struct trie *trie, const void *key, size_t length) { struct trie_leaf *rep = trie_representative(trie, key, length); if (rep && rep->length == length && memcmp(rep->key, key, length) == 0) { return rep; @@ -230,7 +243,22 @@ struct trie_leaf *trie_find_mem(const struct trie *trie, const void *key, size_t } } -struct trie_leaf *trie_find_postfix(const struct trie *trie, const char *key) { +struct trie_leaf *trie_find_mem(const struct trie *trie, const void *key, size_t length) { + return trie_find_mem_impl(trie, key, length); +} + +void *trie_get_str(const struct trie *trie, const char *key) { + const struct trie_leaf *leaf = trie_find_str(trie, key); + return leaf ? leaf->value : NULL; +} + +void *trie_get_mem(const struct trie *trie, const void *key, size_t length) { + const struct trie_leaf *leaf = trie_find_mem(trie, key, length); + return leaf ? leaf->value : NULL; +} + +_trie_clones +static struct trie_leaf *trie_find_postfix_impl(const struct trie *trie, const char *key) { size_t length = strlen(key); struct trie_leaf *rep = trie_representative(trie, key, length + 1); if (rep && rep->length >= length && memcmp(rep->key, key, length) == 0) { @@ -240,6 +268,10 @@ struct trie_leaf *trie_find_postfix(const struct trie *trie, const char *key) { } } +struct trie_leaf *trie_find_postfix(const struct trie *trie, const char *key) { + return trie_find_postfix_impl(trie, key); +} + /** * Find a leaf that may end at the current node. */ @@ -251,10 +283,10 @@ static struct trie_leaf *trie_terminal_leaf(const struct trie_node *node) { } uintptr_t ptr = node->children[0]; - if (trie_is_leaf(ptr)) { - return trie_decode_leaf(ptr); - } else { + if (trie_is_node(ptr)) { node = trie_decode_node(ptr); + } else { + return trie_decode_leaf(ptr); } } @@ -270,7 +302,7 @@ static bool trie_check_prefix(struct trie_leaf *leaf, size_t skip, const char *k } } -trie_clones +_trie_clones static struct trie_leaf *trie_find_prefix_impl(const struct trie *trie, const char *key) { uintptr_t ptr = trie->root; if (!ptr) { @@ -281,21 +313,21 @@ static struct trie_leaf *trie_find_prefix_impl(const struct trie *trie, const ch size_t skip = 0; size_t length = strlen(key) + 1; - size_t offset = 0; - while (!trie_is_leaf(ptr)) { + size_t offset = 0, limit = 2 * length; + while (trie_is_node(ptr)) { struct trie_node *node = trie_decode_node(ptr); offset += node->offset; - if ((offset >> 1) >= length) { + if (offset >= limit) { return best; } struct trie_leaf *leaf = trie_terminal_leaf(node); if (trie_check_prefix(leaf, skip, key, length)) { best = leaf; - skip = offset >> 1; + skip = offset / 2; } - unsigned char nibble = trie_key_nibble(key, offset); + unsigned char nibble = trie_key_nibble(key, length, offset); unsigned int bit = 1U << nibble; if (node->bitmap & bit) { unsigned int index = count_ones(node->bitmap & (bit - 1)); @@ -355,16 +387,10 @@ static struct trie_node *trie_node_realloc(struct trie *trie, struct trie_node * /** Free a node. */ static void trie_node_free(struct trie *trie, struct trie_node *node, size_t size) { - bfs_assert(size == (size_t)count_ones(node->bitmap)); + bfs_assert(size == trie_node_size(node)); varena_free(&trie->nodes, node, size); } -#if ENDIAN_NATIVE == ENDIAN_LITTLE -# define TRIE_BSWAP(n) (n) -#elif ENDIAN_NATIVE == ENDIAN_BIG -# define TRIE_BSWAP(n) bswap(n) -#endif - /** Find the offset of the first nibble that differs between two keys. */ static size_t trie_mismatch(const struct trie_leaf *rep, const void *key, size_t length) { if (!rep) { @@ -378,32 +404,34 @@ static size_t trie_mismatch(const struct trie_leaf *rep, const void *key, size_t const char *rep_bytes = rep->key; const char *key_bytes = key; - size_t i = 0; - for (size_t chunk = sizeof(chunk); i + chunk <= length; i += chunk) { - size_t rep_chunk, key_chunk; - memcpy(&rep_chunk, rep_bytes + i, sizeof(rep_chunk)); - memcpy(&key_chunk, key_bytes + i, sizeof(key_chunk)); - - if (rep_chunk != key_chunk) { -#ifdef TRIE_BSWAP - size_t diff = TRIE_BSWAP(rep_chunk ^ key_chunk); - i *= 2; - i += trailing_zeros(diff) / 4; - return i; + size_t ret = 0, i = 0; + +#define CHUNK(n) CHUNK_(uint##n##_t, load8_beu##n) +#define CHUNK_(type, load8) \ + (length - i >= sizeof(type)) { \ + type rep_chunk = load8(rep_bytes + i); \ + type key_chunk = load8(key_bytes + i); \ + type diff = rep_chunk ^ key_chunk; \ + ret += leading_zeros(diff) / 4; \ + if (diff) { \ + return ret; \ + } \ + i += sizeof(type); \ + } + +#if SIZE_WIDTH >= 64 + while CHUNK(64); + if CHUNK(32); #else - break; + while CHUNK(32); #endif - } - } + if CHUNK(16); + if CHUNK(8); - for (; i < length; ++i) { - unsigned char diff = rep_bytes[i] ^ key_bytes[i]; - if (diff) { - return 2 * i + !(diff & 0xF); - } - } +#undef CHUNK_ +#undef CHUNK - return 2 * i; + return ret; } /** @@ -428,10 +456,10 @@ static size_t trie_mismatch(const struct trie_leaf *rep, const void *key, size_t * | Z * +--->... */ -trie_clones +_trie_clones static struct trie_leaf *trie_node_insert(struct trie *trie, uintptr_t *ptr, struct trie_leaf *leaf, unsigned char nibble) { struct trie_node *node = trie_decode_node(*ptr); - unsigned int size = count_ones(node->bitmap); + unsigned int size = trie_node_size(node); // Double the capacity every power of two if (has_single_bit(size)) { @@ -482,10 +510,10 @@ static struct trie_leaf *trie_node_insert(struct trie *trie, uintptr_t *ptr, str * | Y * +--->key */ -static uintptr_t *trie_jump(struct trie *trie, uintptr_t *ptr, const char *key, size_t *offset) { +static uintptr_t *trie_jump(struct trie *trie, uintptr_t *ptr, size_t *offset) { // We only ever need to jump to leaf nodes, since internal nodes are // guaranteed to be within OFFSET_MAX anyway - bfs_assert(trie_is_leaf(*ptr)); + struct trie_leaf *leaf = trie_decode_leaf(*ptr); struct trie_node *node = trie_node_alloc(trie, 1); if (!node) { @@ -495,7 +523,7 @@ static uintptr_t *trie_jump(struct trie *trie, uintptr_t *ptr, const char *key, *offset += OFFSET_MAX; node->offset = OFFSET_MAX; - unsigned char nibble = trie_key_nibble(key, *offset); + unsigned char nibble = trie_leaf_nibble(leaf, *offset); node->bitmap = 1 << nibble; node->children[0] = *ptr; @@ -521,8 +549,8 @@ static uintptr_t *trie_jump(struct trie *trie, uintptr_t *ptr, const char *key, * +--->leaf */ static struct trie_leaf *trie_split(struct trie *trie, uintptr_t *ptr, struct trie_leaf *leaf, struct trie_leaf *rep, size_t offset, size_t mismatch) { - unsigned char key_nibble = trie_key_nibble(leaf->key, mismatch); - unsigned char rep_nibble = trie_key_nibble(rep->key, mismatch); + unsigned char key_nibble = trie_leaf_nibble(leaf, mismatch); + unsigned char rep_nibble = trie_leaf_nibble(rep, mismatch); bfs_assert(key_nibble != rep_nibble); struct trie_node *node = trie_node_alloc(trie, 2); @@ -534,7 +562,7 @@ static struct trie_leaf *trie_split(struct trie *trie, uintptr_t *ptr, struct tr node->bitmap = (1 << key_nibble) | (1 << rep_nibble); size_t delta = mismatch - offset; - if (!trie_is_leaf(*ptr)) { + if (trie_is_node(*ptr)) { struct trie_node *child = trie_decode_node(*ptr); child->offset -= delta; } @@ -551,12 +579,18 @@ struct trie_leaf *trie_insert_str(struct trie *trie, const char *key) { return trie_insert_mem(trie, key, strlen(key) + 1); } -trie_clones +_trie_clones static struct trie_leaf *trie_insert_mem_impl(struct trie *trie, const void *key, size_t length) { struct trie_leaf *rep = trie_representative(trie, key, length); size_t mismatch = trie_mismatch(rep, key, length); - if (mismatch >= (length << 1)) { + size_t misbyte = mismatch / 2; + if (misbyte >= length) { + bfs_assert(misbyte == length); return rep; + } else if (rep && misbyte >= rep->length) { + bfs_bug("trie keys must be prefix-free"); + errno = EINVAL; + return NULL; } struct trie_leaf *leaf = trie_leaf_alloc(trie, key, length); @@ -571,14 +605,14 @@ static struct trie_leaf *trie_insert_mem_impl(struct trie *trie, const void *key size_t offset = 0; uintptr_t *ptr = &trie->root; - while (!trie_is_leaf(*ptr)) { + while (trie_is_node(*ptr)) { struct trie_node *node = trie_decode_node(*ptr); if (offset + node->offset > mismatch) { break; } offset += node->offset; - unsigned char nibble = trie_key_nibble(key, offset); + unsigned char nibble = trie_leaf_nibble(leaf, offset); unsigned int bit = 1U << nibble; if (node->bitmap & bit) { bfs_assert(offset < mismatch); @@ -591,7 +625,7 @@ static struct trie_leaf *trie_insert_mem_impl(struct trie *trie, const void *key } while (mismatch - offset > OFFSET_MAX) { - ptr = trie_jump(trie, ptr, key, &offset); + ptr = trie_jump(trie, ptr, &offset); if (!ptr) { trie_leaf_free(trie, leaf); return NULL; @@ -605,13 +639,33 @@ struct trie_leaf *trie_insert_mem(struct trie *trie, const void *key, size_t len return trie_insert_mem_impl(trie, key, length); } +int trie_set_str(struct trie *trie, const char *key, const void *value) { + struct trie_leaf *leaf = trie_insert_str(trie, key); + if (leaf) { + leaf->value = (void *)value; + return 0; + } else { + return -1; + } +} + +int trie_set_mem(struct trie *trie, const void *key, size_t length, const void *value) { + struct trie_leaf *leaf = trie_insert_mem(trie, key, length); + if (leaf) { + leaf->value = (void *)value; + return 0; + } else { + return -1; + } +} + /** Free a chain of singleton nodes. */ static void trie_free_singletons(struct trie *trie, uintptr_t ptr) { - while (!trie_is_leaf(ptr)) { + while (trie_is_node(ptr)) { struct trie_node *node = trie_decode_node(ptr); // Make sure the bitmap is a power of two, i.e. it has just one child - bfs_assert(has_single_bit(node->bitmap)); + bfs_assert(has_single_bit((size_t)node->bitmap)); ptr = node->children[0]; trie_node_free(trie, node, 1); @@ -639,7 +693,7 @@ static void trie_free_singletons(struct trie *trie, uintptr_t ptr) { */ static int trie_collapse_node(struct trie *trie, uintptr_t *parent, struct trie_node *parent_node, unsigned int child_index) { uintptr_t other = parent_node->children[child_index ^ 1]; - if (!trie_is_leaf(other)) { + if (trie_is_node(other)) { struct trie_node *other_node = trie_decode_node(other); if (other_node->offset + parent_node->offset <= OFFSET_MAX) { other_node->offset += parent_node->offset; @@ -649,22 +703,21 @@ static int trie_collapse_node(struct trie *trie, uintptr_t *parent, struct trie_ } *parent = other; - trie_node_free(trie, parent_node, 1); + trie_node_free(trie, parent_node, 2); return 0; } -trie_clones +_trie_clones static void trie_remove_impl(struct trie *trie, struct trie_leaf *leaf) { uintptr_t *child = &trie->root; uintptr_t *parent = NULL; unsigned int child_bit = 0, child_index = 0; size_t offset = 0; - while (!trie_is_leaf(*child)) { + while (trie_is_node(*child)) { struct trie_node *node = trie_decode_node(*child); offset += node->offset; - bfs_assert((offset >> 1) < leaf->length); - unsigned char nibble = trie_key_nibble(leaf->key, offset); + unsigned char nibble = trie_leaf_nibble(leaf, offset); unsigned int bit = 1U << nibble; unsigned int bitmap = node->bitmap; bfs_assert(bitmap & bit); @@ -689,19 +742,19 @@ static void trie_remove_impl(struct trie *trie, struct trie_leaf *leaf) { } struct trie_node *node = trie_decode_node(*parent); - child = node->children + child_index; - trie_free_singletons(trie, *child); + trie_free_singletons(trie, node->children[child_index]); - node->bitmap ^= child_bit; - unsigned int parent_size = count_ones(node->bitmap); - bfs_assert(parent_size > 0); - if (parent_size == 1 && trie_collapse_node(trie, parent, node, child_index) == 0) { + unsigned int parent_size = trie_node_size(node); + bfs_assert(parent_size > 1); + if (parent_size == 2 && trie_collapse_node(trie, parent, node, child_index) == 0) { return; } - if (child_index < parent_size) { - memmove(child, child + 1, (parent_size - child_index) * sizeof(*child)); + for (size_t i = child_index; i + 1 < parent_size; ++i) { + node->children[i] = node->children[i + 1]; } + node->bitmap &= ~child_bit; + --parent_size; if (has_single_bit(parent_size)) { node = trie_node_realloc(trie, node, 2 * parent_size, parent_size); @@ -6,6 +6,7 @@ #include "alloc.h" #include "list.h" + #include <stddef.h> #include <stdint.h> @@ -20,7 +21,7 @@ struct trie_leaf { /** The length of the key in bytes. */ size_t length; /** The key itself, stored inline. */ - char key[]; + char key[] _counted_by(length); }; /** @@ -45,9 +46,9 @@ void trie_init(struct trie *trie); /** * Find the leaf for a string key. * - * @param trie + * @trie * The trie to search. - * @param key + * @key * The key to look up. * @return * The found leaf, or NULL if the key is not present. @@ -57,11 +58,11 @@ struct trie_leaf *trie_find_str(const struct trie *trie, const char *key); /** * Find the leaf for a fixed-size key. * - * @param trie + * @trie * The trie to search. - * @param key + * @key * The key to look up. - * @param length + * @length * The length of the key in bytes. * @return * The found leaf, or NULL if the key is not present. @@ -69,11 +70,37 @@ struct trie_leaf *trie_find_str(const struct trie *trie, const char *key); struct trie_leaf *trie_find_mem(const struct trie *trie, const void *key, size_t length); /** + * Get the value associated with a string key. + * + * @trie + * The trie to search. + * @key + * The key to look up. + * @return + * The found value, or NULL if the key is not present. + */ +void *trie_get_str(const struct trie *trie, const char *key); + +/** + * Get the value associated with a fixed-size key. + * + * @trie + * The trie to search. + * @key + * The key to look up. + * @length + * The length of the key in bytes. + * @return + * The found value, or NULL if the key is not present. + */ +void *trie_get_mem(const struct trie *trie, const void *key, size_t length); + +/** * Find the shortest leaf that starts with a given key. * - * @param trie + * @trie * The trie to search. - * @param key + * @key * The key to look up. * @return * A leaf that starts with the given key, or NULL. @@ -83,9 +110,9 @@ struct trie_leaf *trie_find_postfix(const struct trie *trie, const char *key); /** * Find the leaf that is the longest prefix of the given key. * - * @param trie + * @trie * The trie to search. - * @param key + * @key * The key to look up. * @return * The longest prefix match for the given key, or NULL. @@ -95,9 +122,9 @@ struct trie_leaf *trie_find_prefix(const struct trie *trie, const char *key); /** * Insert a string key into the trie. * - * @param trie + * @trie * The trie to modify. - * @param key + * @key * The key to insert. * @return * The inserted leaf, or NULL on failure. @@ -107,11 +134,11 @@ struct trie_leaf *trie_insert_str(struct trie *trie, const char *key); /** * Insert a fixed-size key into the trie. * - * @param trie + * @trie * The trie to modify. - * @param key + * @key * The key to insert. - * @param length + * @length * The length of the key in bytes. * @return * The inserted leaf, or NULL on failure. @@ -119,11 +146,41 @@ struct trie_leaf *trie_insert_str(struct trie *trie, const char *key); struct trie_leaf *trie_insert_mem(struct trie *trie, const void *key, size_t length); /** + * Set the value for a string key. + * + * @trie + * The trie to modify. + * @key + * The key to insert. + * @value + * The value to set. + * @return + * 0 on success, -1 on error. + */ +int trie_set_str(struct trie *trie, const char *key, const void *value); + +/** + * Set the value for a fixed-size key. + * + * @trie + * The trie to modify. + * @key + * The key to insert. + * @length + * The length of the key in bytes. + * @value + * The value to set. + * @return + * 0 on success, -1 on error. + */ +int trie_set_mem(struct trie *trie, const void *key, size_t length, const void *value); + +/** * Remove a leaf from a trie. * - * @param trie + * @trie * The trie to modify. - * @param leaf + * @leaf * The leaf to remove. */ void trie_remove(struct trie *trie, struct trie_leaf *leaf); @@ -2,6 +2,7 @@ // SPDX-License-Identifier: 0BSD #include "typo.h" + #include <limits.h> #include <stdint.h> #include <stdlib.h> @@ -7,9 +7,9 @@ /** * Find the "typo" distance between two strings. * - * @param actual + * @actual * The actual string typed by the user. - * @param expected + * @expected * The expected valid string. * @return The distance between the two strings. */ diff --git a/src/version.c b/src/version.c new file mode 100644 index 0000000..7479a9f --- /dev/null +++ b/src/version.c @@ -0,0 +1,32 @@ +// Copyright © Tavian Barnes <tavianator@tavianator.com> +// SPDX-License-Identifier: 0BSD + +#include "bfs.h" + +const char bfs_version[] = { +#include "version.i" +}; + +const char bfs_confflags[] = { +#include "confflags.i" +}; + +const char bfs_cc[] = { +#include "cc.i" +}; + +const char bfs_cppflags[] = { +#include "cppflags.i" +}; + +const char bfs_cflags[] = { +#include "cflags.i" +}; + +const char bfs_ldflags[] = { +#include "ldflags.i" +}; + +const char bfs_ldlibs[] = { +#include "ldlibs.i" +}; diff --git a/src/xregex.c b/src/xregex.c index c2711bc..796544e 100644 --- a/src/xregex.c +++ b/src/xregex.c @@ -1,19 +1,21 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "xregex.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "diag.h" #include "sanity.h" #include "thread.h" + #include <errno.h> #include <pthread.h> #include <stdlib.h> #include <string.h> -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA # include <langinfo.h> # include <oniguruma.h> #else @@ -21,7 +23,7 @@ #endif struct bfs_regex { -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA unsigned char *pattern; OnigRegex impl; int err; @@ -32,11 +34,17 @@ struct bfs_regex { #endif }; -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA static int bfs_onig_status; static OnigEncoding bfs_onig_enc; +static OnigSyntaxType bfs_onig_syntax_awk; +static OnigSyntaxType bfs_onig_syntax_gnu_awk; +static OnigSyntaxType bfs_onig_syntax_emacs; +static OnigSyntaxType bfs_onig_syntax_egrep; +static OnigSyntaxType bfs_onig_syntax_gnu_find; + /** pthread_once() callback. */ static void bfs_onig_once(void) { // Fall back to ASCII by default @@ -103,6 +111,35 @@ static void bfs_onig_once(void) { if (bfs_onig_status != ONIG_NORMAL) { bfs_onig_enc = NULL; } + + // Compute the GNU extensions + OnigSyntaxType *ere = ONIG_SYNTAX_POSIX_EXTENDED; + OnigSyntaxType *gnu = ONIG_SYNTAX_GNU_REGEX; + unsigned int gnu_op = gnu->op & ~ere->op; + unsigned int gnu_op2 = gnu->op2 & ~ere->op2; + unsigned int gnu_behavior = gnu->behavior & ~ere->behavior; + + onig_copy_syntax(&bfs_onig_syntax_awk, ONIG_SYNTAX_POSIX_EXTENDED); + bfs_onig_syntax_awk.behavior |= ONIG_SYN_ALLOW_INVALID_INTERVAL; + bfs_onig_syntax_awk.behavior |= ONIG_SYN_BACKSLASH_ESCAPE_IN_CC; + + onig_copy_syntax(&bfs_onig_syntax_gnu_awk, &bfs_onig_syntax_awk); + bfs_onig_syntax_gnu_awk.op |= gnu_op; + bfs_onig_syntax_gnu_awk.op2 |= gnu_op2; + bfs_onig_syntax_gnu_awk.behavior |= gnu_behavior; + bfs_onig_syntax_gnu_awk.behavior &= ~ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS; + bfs_onig_syntax_gnu_awk.behavior &= ~ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS; + + // https://github.com/kkos/oniguruma/issues/296 + onig_copy_syntax(&bfs_onig_syntax_emacs, ONIG_SYNTAX_EMACS); + bfs_onig_syntax_emacs.op2 |= ONIG_SYN_OP2_QMARK_GROUP_EFFECT; + + onig_copy_syntax(&bfs_onig_syntax_egrep, ONIG_SYNTAX_POSIX_EXTENDED); + bfs_onig_syntax_egrep.behavior |= ONIG_SYN_ALLOW_INVALID_INTERVAL; + bfs_onig_syntax_egrep.behavior &= ~ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS; + + onig_copy_syntax(&bfs_onig_syntax_gnu_find, &bfs_onig_syntax_emacs); + bfs_onig_syntax_gnu_find.options |= ONIG_OPTION_MULTILINE; } /** Initialize Oniguruma. */ @@ -121,7 +158,7 @@ int bfs_regcomp(struct bfs_regex **preg, const char *pattern, enum bfs_regex_typ return -1; } -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA // onig_error_code_to_str() says // // don't call this after the pattern argument of onig_new() is freed @@ -143,12 +180,24 @@ int bfs_regcomp(struct bfs_regex **preg, const char *pattern, enum bfs_regex_typ case BFS_REGEX_POSIX_EXTENDED: syntax = ONIG_SYNTAX_POSIX_EXTENDED; break; + case BFS_REGEX_AWK: + syntax = &bfs_onig_syntax_awk; + break; + case BFS_REGEX_GNU_AWK: + syntax = &bfs_onig_syntax_gnu_awk; + break; case BFS_REGEX_EMACS: - syntax = ONIG_SYNTAX_EMACS; + syntax = &bfs_onig_syntax_emacs; break; case BFS_REGEX_GREP: syntax = ONIG_SYNTAX_GREP; break; + case BFS_REGEX_EGREP: + syntax = &bfs_onig_syntax_egrep; + break; + case BFS_REGEX_GNU_FIND: + syntax = &bfs_onig_syntax_gnu_find; + break; } bfs_assert(syntax, "Invalid regex type"); @@ -204,7 +253,7 @@ fail: int bfs_regexec(struct bfs_regex *regex, const char *str, enum bfs_regexec_flags flags) { size_t len = strlen(str); -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA const unsigned char *ustr = (const unsigned char *)str; const unsigned char *end = ustr + len; @@ -263,7 +312,7 @@ int bfs_regexec(struct bfs_regex *regex, const char *str, enum bfs_regexec_flags void bfs_regfree(struct bfs_regex *regex) { if (regex) { -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA onig_free(regex->impl); free(regex->pattern); #else @@ -278,7 +327,7 @@ char *bfs_regerror(const struct bfs_regex *regex) { return strdup(xstrerror(ENOMEM)); } -#if BFS_USE_ONIGURUMA +#if BFS_WITH_ONIGURUMA unsigned char *str = malloc(ONIG_MAX_ERROR_MESSAGE_LEN); if (str) { onig_error_code_to_str(str, regex->err, ®ex->einfo); diff --git a/src/xregex.h b/src/xregex.h index 998a2b0..c4504ee 100644 --- a/src/xregex.h +++ b/src/xregex.h @@ -15,8 +15,12 @@ struct bfs_regex; enum bfs_regex_type { BFS_REGEX_POSIX_BASIC, BFS_REGEX_POSIX_EXTENDED, + BFS_REGEX_AWK, + BFS_REGEX_GNU_AWK, BFS_REGEX_EMACS, BFS_REGEX_GREP, + BFS_REGEX_EGREP, + BFS_REGEX_GNU_FIND, }; /** @@ -38,13 +42,13 @@ enum bfs_regexec_flags { /** * Wrapper for regcomp() that supports additional regex types. * - * @param[out] preg + * @preg[out] * Will hold the compiled regex. - * @param pattern + * @pattern * The regular expression to compile. - * @param type + * @type * The regular expression syntax to use. - * @param flags + * @flags * Regex compilation flags. * @return * 0 on success, -1 on failure. @@ -54,11 +58,11 @@ int bfs_regcomp(struct bfs_regex **preg, const char *pattern, enum bfs_regex_typ /** * Wrapper for regexec(). * - * @param regex + * @regex * The regular expression to execute. - * @param str + * @str * The string to match against. - * @param flags + * @flags * Regex execution flags. * @return * 1 for a match, 0 for no match, -1 on failure. @@ -73,7 +77,7 @@ void bfs_regfree(struct bfs_regex *regex); /** * Get a human-readable regex error message. * - * @param regex + * @regex * The compiled regex. * @return * A human-readable description of the error, which should be free()'d. diff --git a/src/xspawn.c b/src/xspawn.c index 0b0cea4..ee62c05 100644 --- a/src/xspawn.c +++ b/src/xspawn.c @@ -1,24 +1,29 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "xspawn.h" + #include "alloc.h" +#include "bfs.h" #include "bfstd.h" +#include "diag.h" #include "list.h" +#include "sighook.h" + #include <errno.h> #include <fcntl.h> +#include <signal.h> #include <stdlib.h> #include <string.h> #include <sys/resource.h> #include <sys/types.h> #include <unistd.h> -#if BFS_USE_PATHS_H +#if __has_include(<paths.h>) # include <paths.h> #endif -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 # include <spawn.h> #endif @@ -68,29 +73,42 @@ int bfs_spawn_init(struct bfs_spawn *ctx) { ctx->flags = 0; SLIST_INIT(ctx); -#if _POSIX_SPAWN > 0 - ctx->flags |= BFS_SPAWN_USE_POSIX; +#if BFS_POSIX_SPAWN >= 0 + if (sysoption(SPAWN) > 0) { + ctx->flags |= BFS_SPAWN_USE_POSIX; - errno = posix_spawn_file_actions_init(&ctx->actions); - if (errno != 0) { - return -1; - } + errno = posix_spawn_file_actions_init(&ctx->actions); + if (errno != 0) { + return -1; + } - errno = posix_spawnattr_init(&ctx->attr); - if (errno != 0) { - posix_spawn_file_actions_destroy(&ctx->actions); - return -1; + errno = posix_spawnattr_init(&ctx->attr); + if (errno != 0) { + posix_spawn_file_actions_destroy(&ctx->actions); + return -1; + } } #endif return 0; } -int bfs_spawn_destroy(struct bfs_spawn *ctx) { -#if _POSIX_SPAWN > 0 - posix_spawnattr_destroy(&ctx->attr); - posix_spawn_file_actions_destroy(&ctx->actions); +/** + * Clear the BFS_SPAWN_USE_POSIX flag and free the attributes. + */ +static void bfs_spawn_clear_posix(struct bfs_spawn *ctx) { + if (ctx->flags & BFS_SPAWN_USE_POSIX) { + ctx->flags &= ~BFS_SPAWN_USE_POSIX; + +#if BFS_POSIX_SPAWN >= 0 + posix_spawnattr_destroy(&ctx->attr); + posix_spawn_file_actions_destroy(&ctx->actions); #endif + } +} + +int bfs_spawn_destroy(struct bfs_spawn *ctx) { + bfs_spawn_clear_posix(ctx); for_slist (struct bfs_spawn_action, action, ctx) { free(action); @@ -99,9 +117,9 @@ int bfs_spawn_destroy(struct bfs_spawn *ctx) { return 0; } -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 /** Set some posix_spawnattr flags. */ -attr(maybe_unused) +_maybe_unused static int bfs_spawn_addflags(struct bfs_spawn *ctx, short flags) { short prev; errno = posix_spawnattr_getflags(&ctx->attr, &prev); @@ -119,7 +137,7 @@ static int bfs_spawn_addflags(struct bfs_spawn *ctx, short flags) { return 0; } -#endif // _POSIX_SPAWN > 0 +#endif /** Allocate a spawn action. */ static struct bfs_spawn_action *bfs_spawn_action(enum bfs_spawn_op op) { @@ -141,7 +159,7 @@ int bfs_spawn_addopen(struct bfs_spawn *ctx, int fd, const char *path, int flags return -1; } -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 if (ctx->flags & BFS_SPAWN_USE_POSIX) { errno = posix_spawn_file_actions_addopen(&ctx->actions, fd, path, flags, mode); if (errno != 0) { @@ -165,7 +183,7 @@ int bfs_spawn_addclose(struct bfs_spawn *ctx, int fd) { return -1; } -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 if (ctx->flags & BFS_SPAWN_USE_POSIX) { errno = posix_spawn_file_actions_addclose(&ctx->actions, fd); if (errno != 0) { @@ -186,7 +204,7 @@ int bfs_spawn_adddup2(struct bfs_spawn *ctx, int oldfd, int newfd) { return -1; } -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 if (ctx->flags & BFS_SPAWN_USE_POSIX) { errno = posix_spawn_file_actions_adddup2(&ctx->actions, oldfd, newfd); if (errno != 0) { @@ -214,19 +232,35 @@ int bfs_spawn_adddup2(struct bfs_spawn *ctx, int oldfd, int newfd) { */ #define BFS_POSIX_SPAWNP_AFTER_FCHDIR !(__APPLE__ || __NetBSD__) +/** + * NetBSD even resolves the executable before file actions with posix_spawn()! + */ +#define BFS_POSIX_SPAWN_AFTER_FCHDIR !__NetBSD__ + int bfs_spawn_addfchdir(struct bfs_spawn *ctx, int fd) { struct bfs_spawn_action *action = bfs_spawn_action(BFS_SPAWN_FCHDIR); if (!action) { return -1; } +#if __APPLE__ + // macOS has a bug that causes EBADF when an fchdir() action refers to a + // file opened by the file actions + for_slist (struct bfs_spawn_action, prev, ctx) { + if (fd == prev->out_fd) { + bfs_spawn_clear_posix(ctx); + break; + } + } +#endif + #if BFS_HAS_POSIX_SPAWN_ADDFCHDIR # define BFS_POSIX_SPAWN_ADDFCHDIR posix_spawn_file_actions_addfchdir #elif BFS_HAS_POSIX_SPAWN_ADDFCHDIR_NP # define BFS_POSIX_SPAWN_ADDFCHDIR posix_spawn_file_actions_addfchdir_np #endif -#if _POSIX_SPAWN > 0 && defined(BFS_POSIX_SPAWN_FCHDIR) +#if BFS_POSIX_SPAWN >= 0 && defined(BFS_POSIX_SPAWN_ADDFCHDIR) if (ctx->flags & BFS_SPAWN_USE_POSIX) { errno = BFS_POSIX_SPAWN_ADDFCHDIR(&ctx->actions, fd); if (errno != 0) { @@ -235,7 +269,7 @@ int bfs_spawn_addfchdir(struct bfs_spawn *ctx, int fd) { } } #else - ctx->flags &= ~BFS_SPAWN_USE_POSIX; + bfs_spawn_clear_posix(ctx); #endif action->in_fd = fd; @@ -259,7 +293,7 @@ int bfs_spawn_setrlimit(struct bfs_spawn *ctx, int resource, const struct rlimit goto fail; } #else - ctx->flags &= ~BFS_SPAWN_USE_POSIX; + bfs_spawn_clear_posix(ctx); #endif action->resource = resource; @@ -383,18 +417,40 @@ static bool bfs_resolve_relative(const struct bfs_resolver *res) { return false; } +/** Check if the actions include fchdir(). */ +static bool bfs_spawn_will_chdir(const struct bfs_spawn *ctx) { + if (ctx) { + for_slist (const struct bfs_spawn_action, action, ctx) { + if (action->op == BFS_SPAWN_FCHDIR) { + return true; + } + } + } + + return false; +} + +/** Check if we can call xfaccessat() before file actions. */ +static bool bfs_can_access_early(const struct bfs_resolver *res, const struct bfs_spawn *ctx) { + if (res->exe[0] == '/') { + return true; + } + + if (bfs_spawn_will_chdir(ctx)) { + return false; + } + + return true; +} + /** Check if we can resolve the executable before file actions. */ static bool bfs_can_resolve_early(const struct bfs_resolver *res, const struct bfs_spawn *ctx) { if (!bfs_resolve_relative(res)) { return true; } - if (ctx) { - for_slist (const struct bfs_spawn_action, action, ctx) { - if (action->op == BFS_SPAWN_FCHDIR) { - return false; - } - } + if (bfs_spawn_will_chdir(ctx)) { + return false; } return true; @@ -424,6 +480,17 @@ static int bfs_resolve_early(struct bfs_resolver *res, const char *exe, const st }; if (bfs_can_skip_resolve(res, ctx)) { + if (bfs_can_access_early(res, ctx)) { + // Do this check eagerly, even though posix_spawn()/execv() also + // would, because: + // + // - faccessat() is faster than fork()/clone() + execv() + // - posix_spawn() is not guaranteed to report ENOENT + if (xfaccessat(AT_FDCWD, exe, X_OK) != 0) { + return -1; + } + } + res->done = true; return 0; } @@ -471,7 +538,7 @@ fail: return -1; } -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 /** bfs_spawn() implementation using posix_spawn(). */ static pid_t bfs_posix_spawn(struct bfs_resolver *res, const struct bfs_spawn *ctx, char **argv, char **envp) { @@ -502,13 +569,20 @@ static bool bfs_use_posix_spawn(const struct bfs_resolver *res, const struct bfs } #endif +#if !BFS_POSIX_SPAWN_AFTER_FCHDIR + if (res->exe[0] != '/' && bfs_spawn_will_chdir(ctx)) { + return false; + } +#endif + return true; } -#endif // _POSIX_SPAWN > 0 +#endif // BFS_POSIX_SPAWN >= 0 /** Actually exec() the new process. */ -static noreturn void bfs_spawn_exec(struct bfs_resolver *res, const struct bfs_spawn *ctx, char **argv, char **envp, int pipefd[2]) { +_noreturn +static void bfs_spawn_exec(struct bfs_resolver *res, const struct bfs_spawn *ctx, char **argv, char **envp, const sigset_t *mask, int pipefd[2]) { xclose(pipefd[0]); for_slist (const struct bfs_spawn_action, action, ctx) { @@ -569,6 +643,18 @@ static noreturn void bfs_spawn_exec(struct bfs_resolver *res, const struct bfs_s goto fail; } + // Reset signal handlers to their original values before we unblock + // signals, so that handlers don't run in both the parent and the child + if (sigreset() != 0) { + goto fail; + } + + // Restore the original signal mask for the child process + errno = pthread_sigmask(SIG_SETMASK, mask, NULL); + if (errno != 0) { + goto fail; + } + execve(res->exe, argv, envp); fail:; @@ -590,35 +676,58 @@ static pid_t bfs_fork_spawn(struct bfs_resolver *res, const struct bfs_spawn *ct return -1; } + // Block signals before fork() so handlers don't run in the child + sigset_t new_mask; + if (sigfillset(&new_mask) != 0) { + goto fail; + } + sigset_t old_mask; + errno = pthread_sigmask(SIG_BLOCK, &new_mask, &old_mask); + if (errno != 0) { + goto fail; + } + +#if BFS_HAS__FORK + pid_t pid = _Fork(); +#else pid_t pid = fork(); - if (pid < 0) { - close_quietly(pipefd[1]); - close_quietly(pipefd[0]); - return -1; - } else if (pid == 0) { +#endif + if (pid == 0) { // Child - bfs_spawn_exec(res, ctx, argv, envp, pipefd); + bfs_spawn_exec(res, ctx, argv, envp, &old_mask, pipefd); + } + + // Restore the original signal mask + errno = pthread_sigmask(SIG_SETMASK, &old_mask, NULL); + bfs_everify(errno == 0, "pthread_sigmask()"); + + if (pid < 0) { + // fork() failed + goto fail; } - // Parent xclose(pipefd[1]); int error; ssize_t nbytes = xread(pipefd[0], &error, sizeof(error)); xclose(pipefd[0]); if (nbytes == sizeof(error)) { - int wstatus; - xwaitpid(pid, &wstatus, 0); + xwaitpid(pid, NULL, 0); errno = error; return -1; } return pid; + +fail: + close_quietly(pipefd[1]); + close_quietly(pipefd[0]); + return -1; } /** Call the right bfs_spawn() implementation. */ static pid_t bfs_spawn_impl(struct bfs_resolver *res, const struct bfs_spawn *ctx, char **argv, char **envp) { -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 if (bfs_use_posix_spawn(res, ctx)) { return bfs_posix_spawn(res, ctx, argv, envp); } diff --git a/src/xspawn.h b/src/xspawn.h index 6a8f54a..3c74ccd 100644 --- a/src/xspawn.h +++ b/src/xspawn.h @@ -8,12 +8,17 @@ #ifndef BFS_XSPAWN_H #define BFS_XSPAWN_H -#include "prelude.h" #include <sys/resource.h> #include <sys/types.h> #include <unistd.h> -#if _POSIX_SPAWN > 0 +#ifdef _POSIX_SPAWN +# define BFS_POSIX_SPAWN _POSIX_SPAWN +#else +# define BFS_POSIX_SPAWN (-1) +#endif + +#if BFS_POSIX_SPAWN >= 0 # include <spawn.h> #endif @@ -38,7 +43,7 @@ struct bfs_spawn { struct bfs_spawn_action *head; struct bfs_spawn_action **tail; -#if _POSIX_SPAWN > 0 +#if BFS_POSIX_SPAWN >= 0 /** posix_spawn() context, for when we can use it. */ posix_spawn_file_actions_t actions; posix_spawnattr_t attr; @@ -104,13 +109,13 @@ int bfs_spawn_setrlimit(struct bfs_spawn *ctx, int resource, const struct rlimit /** * Spawn a new process. * - * @param exe + * @exe * The executable to run. - * @param ctx + * @ctx * The context for the new process. - * @param argv + * @argv * The arguments for the new process. - * @param envp + * @envp * The environment variables for the new process (NULL for the current * environment). * @return @@ -122,7 +127,7 @@ pid_t bfs_spawn(const char *exe, const struct bfs_spawn *ctx, char **argv, char * Look up an executable in the current PATH, as BFS_SPAWN_USE_PATH or execvp() * would do. * - * @param exe + * @exe * The name of the binary to execute. Bare names without a '/' will be * searched on the provided PATH. * @return diff --git a/src/xtime.c b/src/xtime.c index 91ed915..6b8a141 100644 --- a/src/xtime.c +++ b/src/xtime.c @@ -1,10 +1,14 @@ // Copyright © Tavian Barnes <tavianator@tavianator.com> // SPDX-License-Identifier: 0BSD -#include "prelude.h" #include "xtime.h" + +#include "alloc.h" +#include "bfs.h" #include "bfstd.h" #include "diag.h" +#include "sanity.h" + #include <errno.h> #include <limits.h> #include <sys/time.h> @@ -12,14 +16,14 @@ #include <unistd.h> int xmktime(struct tm *tm, time_t *timep) { - *timep = mktime(tm); + time_t time = mktime(tm); - if (*timep == -1) { + if (time == -1) { int error = errno; struct tm tmp; - if (!localtime_r(timep, &tmp)) { - bfs_bug("localtime_r(-1): %s", xstrerror(errno)); + if (!localtime_r(&time, &tmp)) { + bfs_ebug("localtime_r(-1)"); return -1; } @@ -30,9 +34,38 @@ int xmktime(struct tm *tm, time_t *timep) { } } + *timep = time; + return 0; +} + +// FreeBSD is missing an interceptor +#if BFS_HAS_TIMEGM && !(__FreeBSD__ && __SANITIZE_MEMORY__) + +int xtimegm(struct tm *tm, time_t *timep) { + time_t time = timegm(tm); + + if (time == -1) { + int error = errno; + + struct tm tmp; + if (!gmtime_r(&time, &tmp)) { + bfs_ebug("gmtime_r(-1)"); + return -1; + } + + if (tm->tm_year != tmp.tm_year || tm->tm_yday != tmp.tm_yday + || tm->tm_hour != tmp.tm_hour || tm->tm_min != tmp.tm_min || tm->tm_sec != tmp.tm_sec) { + errno = error; + return -1; + } + } + + *timep = time; return 0; } +#else + static int safe_add(int *value, int delta) { if (*value >= 0) { if (delta > INT_MAX - *value) { @@ -147,6 +180,8 @@ overflow: return -1; } +#endif // !BFS_HAS_TIMEGM + /** Parse a decimal digit. */ static int xgetdigit(char c) { int ret = c - '0'; @@ -174,6 +209,23 @@ static int xgetpart(const char **str, size_t n, int *result) { } int xgetdate(const char *str, struct timespec *result) { + // Handle @epochseconds + if (str[0] == '@') { + long long value; + if (xstrtoll(str + 1, NULL, 10, &value) != 0) { + goto error; + } + + time_t time = (time_t)value; + if ((long long)time != value) { + errno = ERANGE; + goto error; + } + + result->tv_sec = time; + goto done; + } + struct tm tm = { .tm_isdst = -1, }; @@ -292,6 +344,7 @@ end: } } +done: result->tv_nsec = 0; return 0; @@ -301,16 +354,150 @@ error: return -1; } -int xgettime(struct timespec *result) { -#if _POSIX_TIMERS > 0 - return clock_gettime(CLOCK_REALTIME, result); +/** One nanosecond. */ +static const long NS = 1000L * 1000 * 1000; + +void timespec_add(struct timespec *lhs, const struct timespec *rhs) { + lhs->tv_sec += rhs->tv_sec; + lhs->tv_nsec += rhs->tv_nsec; + if (lhs->tv_nsec >= NS) { + lhs->tv_nsec -= NS; + lhs->tv_sec += 1; + } +} + +void timespec_sub(struct timespec *lhs, const struct timespec *rhs) { + lhs->tv_sec -= rhs->tv_sec; + lhs->tv_nsec -= rhs->tv_nsec; + if (lhs->tv_nsec < 0) { + lhs->tv_nsec += NS; + lhs->tv_sec -= 1; + } +} + +int timespec_cmp(const struct timespec *lhs, const struct timespec *rhs) { + if (lhs->tv_sec < rhs->tv_sec) { + return -1; + } else if (lhs->tv_sec > rhs->tv_sec) { + return 1; + } + + if (lhs->tv_nsec < rhs->tv_nsec) { + return -1; + } else if (lhs->tv_nsec > rhs->tv_nsec) { + return 1; + } + + return 0; +} + +void timespec_min(struct timespec *dest, const struct timespec *src) { + if (timespec_cmp(src, dest) < 0) { + *dest = *src; + } +} + +void timespec_max(struct timespec *dest, const struct timespec *src) { + if (timespec_cmp(src, dest) > 0) { + *dest = *src; + } +} + +double timespec_ns(const struct timespec *ts) { + return 1.0e9 * ts->tv_sec + ts->tv_nsec; +} + +#if defined(_POSIX_TIMERS) && BFS_HAS_TIMER_CREATE +# define BFS_POSIX_TIMERS _POSIX_TIMERS #else - struct timeval tv; - int ret = gettimeofday(&tv, NULL); - if (ret == 0) { - result->tv_sec = tv.tv_sec; - result->tv_nsec = tv.tv_usec * 1000L; +# define BFS_POSIX_TIMERS (-1) +#endif + +struct timer { +#if BFS_POSIX_TIMERS >= 0 + /** The POSIX timer. */ + timer_t timer; +#endif + /** Whether to use timer_create() or setitimer(). */ + bool legacy; +}; + +struct timer *xtimer_start(const struct timespec *interval) { + struct timer *timer = ALLOC(struct timer); + if (!timer) { + return NULL; } - return ret; + +#if BFS_POSIX_TIMERS >= 0 + if (sysoption(TIMERS)) { + clockid_t clock = CLOCK_REALTIME; + +#if defined(_POSIX_MONOTONIC_CLOCK) && _POSIX_MONOTONIC_CLOCK >= 0 + if (sysoption(MONOTONIC_CLOCK) > 0) { + clock = CLOCK_MONOTONIC; + } +#endif + + if (timer_create(clock, NULL, &timer->timer) != 0) { + goto fail; + } + + // https://github.com/llvm/llvm-project/issues/111847 + sanitize_init(&timer->timer); + + struct itimerspec spec = { + .it_value = *interval, + .it_interval = *interval, + }; + if (timer_settime(timer->timer, 0, &spec, NULL) != 0) { + timer_delete(timer->timer); + goto fail; + } + + timer->legacy = false; + return timer; + } +#endif + +#if BFS_POSIX_TIMERS <= 0 + struct timeval tv = { + .tv_sec = interval->tv_sec, + .tv_usec = (interval->tv_nsec + 999) / 1000, + }; + struct itimerval ival = { + .it_value = tv, + .it_interval = tv, + }; + if (setitimer(ITIMER_REAL, &ival, NULL) != 0) { + goto fail; + } + + timer->legacy = true; + return timer; #endif + +fail: + free(timer); + return NULL; +} + +void xtimer_stop(struct timer *timer) { + if (!timer) { + return; + } + + if (timer->legacy) { +#if BFS_POSIX_TIMERS <= 0 + struct itimerval ival = {0}; + int ret = setitimer(ITIMER_REAL, &ival, NULL); + bfs_everify(ret == 0, "setitimer()"); +#endif + } else { +#if BFS_POSIX_TIMERS >= 0 + int ret = timer_delete(timer->timer); + bfs_everify(ret == 0, "timer_delete()"); +#endif + } + + free(timer); } diff --git a/src/xtime.h b/src/xtime.h index fb60ae4..b76fef2 100644 --- a/src/xtime.h +++ b/src/xtime.h @@ -13,9 +13,9 @@ /** * mktime() wrapper that reports errors more reliably. * - * @param[in,out] tm - * The struct tm to convert. - * @param[out] timep + * @tm[in,out] + * The struct tm to convert and normalize. + * @timep[out] * A pointer to the result. * @return * 0 on success, -1 on failure. @@ -25,9 +25,9 @@ int xmktime(struct tm *tm, time_t *timep); /** * A portable timegm(), the inverse of gmtime(). * - * @param[in,out] tm - * The struct tm to convert. - * @param[out] timep + * @tm[in,out] + * The struct tm to convert and normalize. + * @timep[out] * A pointer to the result. * @return * 0 on success, -1 on failure. @@ -37,9 +37,9 @@ int xtimegm(struct tm *tm, time_t *timep); /** * Parse an ISO 8601-style timestamp. * - * @param[in] str + * @str * The string to parse. - * @param[out] result + * @result[out] * A pointer to the result. * @return * 0 on success, -1 on failure. @@ -47,13 +47,62 @@ int xtimegm(struct tm *tm, time_t *timep); int xgetdate(const char *str, struct timespec *result); /** - * Get the current time. + * Add to a timespec. + */ +void timespec_add(struct timespec *lhs, const struct timespec *rhs); + +/** + * Subtract from a timespec. + */ +void timespec_sub(struct timespec *lhs, const struct timespec *rhs); + +/** + * Compare two timespecs. * - * @param[out] result - * A pointer to the result. * @return - * 0 on success, -1 on failure. + * An integer with the sign of (*lhs - *rhs). + */ +int timespec_cmp(const struct timespec *lhs, const struct timespec *rhs); + +/** + * Update a minimum timespec. + */ +void timespec_min(struct timespec *dest, const struct timespec *src); + +/** + * Update a maximum timespec. + */ +void timespec_max(struct timespec *dest, const struct timespec *src); + +/** + * Convert a timespec to floating point. + * + * @return + * The value in nanoseconds. + */ +double timespec_ns(const struct timespec *ts); + +/** + * A timer. + */ +struct timer; + +/** + * Start a timer. + * + * @interval + * The regular interval at which to send SIGALRM. + * @return + * The new timer on success, otherwise NULL. + */ +struct timer *xtimer_start(const struct timespec *interval); + +/** + * Stop a timer. + * + * @timer + * The timer to stop. */ -int xgettime(struct timespec *result); +void xtimer_stop(struct timer *timer); #endif // BFS_XTIME_H |