summaryrefslogtreecommitdiffstats
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
* ioq: Probe for supported io_uring operationsTavian Barnes2024-02-281-18/+67
|
* ioq: Remove some branches from ioq_slot_{push,pop}()Tavian Barnes2024-02-181-12/+21
|
* ioq: Ensure ioq_ent is sufficiently alignedTavian Barnes2024-02-162-2/+9
| | | | | | | The natural alignment of struct ioq_ent is only 2 on m68k, so over-align it to at least 4 bytes on all platforms. Link: https://buildd.debian.org/status/fetch.php?pkg=bfs&arch=m68k&ver=3.1-1&stamp=1707699583
* Release 3.1.13.1.1Tavian Barnes2024-02-161-1/+1
|
* ioq: Add batched ioqq_push/pop operationsTavian Barnes2024-02-151-49/+127
|
* ioq: Don't push immediately in ioq_check_cancel()Tavian Barnes2024-02-151-10/+6
|
* ioq: Add a missing close() if bfs_opendir() failsTavian Barnes2024-02-151-0/+2
|
* ioq: Don't use the symbolic IO_WQ_[UN]BOUND indicesTavian Barnes2024-02-141-2/+2
| | | | They are only available since liburing 2.2, which is newer than CI.
* ioq: Factor out io_uring initializationTavian Barnes2024-02-141-37/+73
|
* ioq: Make -j also limit the io_uring worker threadsTavian Barnes2024-02-141-0/+9
|
* ioq: Replay IOQ_STOP messages rather than spam themTavian Barnes2024-02-141-4/+5
|
* ioq: Pack ioq_ent args structsTavian Barnes2024-02-141-3/+3
|
* bftw: Document which bftw_file nodes go with which listsTavian Barnes2024-02-141-3/+17
|
* expr: Pack struct bfs_expr a bitTavian Barnes2024-02-141-15/+9
|
* stat: Pack struct bfs_stat a bitTavian Barnes2024-02-142-28/+24
|
* eval: Add thousands separators to -statusTavian Barnes2024-02-121-2/+2
|
* sanity: Don't use self-init for uninit()Tavian Barnes2024-02-123-7/+7
| | | | | | | | | | | Self-initialization like bool ret = ret; is a GCC trick to suppress uninitialized variable warnings, but it's not actually well-defined, and will trip a recent enough MemorySanitizer: src/eval.c:1088:13: runtime error: load of value 128, which is not a valid value for type 'bool'
* ioq: Get rid of IOQ_STRIDETavian Barnes2024-02-121-20/+5
| | | | Benchmarks show it hurts more than it helps.
* ioq: Shrink the io_uringsTavian Barnes2024-02-121-4/+2
|
* Release 3.13.1Tavian Barnes2024-02-061-1/+1
|
* opt: Enable BFTW_STAT when profitableTavian Barnes2024-02-062-0/+93
|
* ctx: Fill in ctx->threads earlierTavian Barnes2024-02-063-23/+20
|
* color: New API to check if stat() is necessaryTavian Barnes2024-02-062-6/+11
|
* bftw: stat() files asynchronouslyTavian Barnes2024-02-063-211/+506
|
* mtab: Take the basename directly in bfs_might_be_mount()Tavian Barnes2024-02-053-6/+5
| | | | This avoids some hot xbaseoff() calls.
* bftw: Always block in bftw_pop_dir() with multiple threadsTavian Barnes2024-02-011-11/+22
|
* bftw: Don't immediately pin open directoriesTavian Barnes2024-02-011-4/+5
| | | | | | | It is undesirable to close a directory that we haven't read yet to free up cache capacity, but it's worse to fail to open the next directory because too many upcoming directories are pinned. This could happen when sorting, because then we can't prioritize the already-opened ones.
* bftw: Allow forcing bfs_dir allocation from the main threadTavian Barnes2024-02-011-12/+35
| | | | | | | | | When sorting, we can be forced to pop an unopened directory. If enough other directories are already open, that can lead to ENOMEM when we try to open it synchronously. To avoid this, force allocations from the main thread to be attempted even if they would go over the limit. Also, fix the accounting in bftw_allocdir() if allocation fails.
* bftw: Kill trivial bftw_queue_balance() helperTavian Barnes2024-02-011-7/+2
|
* bfstd: Don't shadow FreeBSD's fflags_tTavian Barnes2024-02-011-4/+4
|
* bftw: Actually stop if the callback returns BFTW_STOPTavian Barnes2024-01-311-1/+1
| | | | | | Otherwise, bftw_ids() or bftw_eds() might keep going! Fixes: 5f16169 ("bftw: Share the bftw_state between iterations of ids/eds")
* opt: Charge eval_flags() for a stat() callTavian Barnes2024-01-311-0/+1
|
* bftw: Optimize -s -j2 searchesTavian Barnes2024-01-311-2/+1
| | | | | Maintaining balance and strict ordering at the same time forces too much work onto the main thread.
* bftw: Use a bftw_queue for files tooTavian Barnes2024-01-311-26/+31
|
* bftw: New bftw_queue abstractionTavian Barnes2024-01-311-74/+292
|
* list: Return the next cursor from SLIST_INSERT()Tavian Barnes2024-01-301-7/+10
|
* eval: Squelch an uninitialized variable warningTavian Barnes2024-01-221-1/+3
|
* ioq: Implement ioq_stat()Tavian Barnes2024-01-183-10/+124
|
* ioq: Refactor to take advantage of -WswitchTavian Barnes2024-01-181-71/+91
|
* stat: Expose bfs_{fstatat,statx}_flags()Tavian Barnes2024-01-182-21/+47
|
* stat: Expose bfs_stat{,x}_convert()Tavian Barnes2024-01-182-103/+121
|
* xspawn: Check for _POSIX_SPAWN in bfs_spawn_addfchdir()Tavian Barnes2024-01-171-1/+1
|
* fsade: Mark bfs_acl_tag_type() as maybe_unusedTavian Barnes2024-01-171-0/+1
| | | | | It's not used on at least macOS, which doesn't have ACL_{USER,GROUP}_OBJ or ACL_OTHER.
* ioq: Use the negative errno conventionTavian Barnes2024-01-135-26/+33
|
* bfstd: New {error,errno}_is_like() functionsTavian Barnes2024-01-135-21/+42
| | | | | | | We used to have is_nonexistence_error() to consistently treat ENOENT and ENOTDIR the same. Recently, we started considering EFAULT the same as ENAMETOOLONG on DragonFly BSD to work around a kernel bug. Unify both of these behind a more generic interface.
* config: Disable sys/acl.h on illumosTavian Barnes2024-01-091-1/+1
| | | | | | Their ACL API is quite a bit different from most. Link: https://illumos.org/man/2/acl
* eval: Check for xbasename() allocation failureTavian Barnes2024-01-071-1/+8
|
* expr: Make expressions variadicTavian Barnes2024-01-078-890/+1719
| | | | | | | | | Rather than only unary/binary expressions, we now support an arbitrary number of children. The optimizer has been re-written almost completely and now supports optimal reordering of longer expression chains, rather than just arm-swapping. Fixes #85.
* list: New SLIST_HEAD() and SLIST_TAIL() macrosTavian Barnes2024-01-071-3/+40
|
* Work around DragonFly BSD kernel bugTavian Barnes2024-01-041-1/+13
| | | | | | | | | DragonFly's x86_64 assembly implementation of copyinstr() checks the wrong pointer when deciding whether to return EFAULT or ENAMETOOLONG, causing it to always return EFAULT for overlong paths. Work around it by treating EFAULT the same as ENAMETOOLONG on DragonFly. Link: https://twitter.com/tavianator/status/1742991411203485713