36128 commits

This branch

This branch

All branches

Author	SHA1	Message	Date
Alex Rønne Petersen	26751fb31a	compiler-rt: implement __clear_cache() for arm/mips64-openbsd	2025-12-05 12:21:48 +01:00
Alex Rønne Petersen	616488eec7	std.fs: fix build error in selfExePath() on openbsd	2025-12-05 12:21:48 +01:00
Alex Rønne Petersen	7827dc03cb	std.Io.Threaded: apply netbsd todo hacks to openbsd ... https://codeberg.org/ziglang/zig/issues/30063	2025-12-05 12:21:41 +01:00
Alex Rønne Petersen	49a0ffa42f	std.c: openbsd does not have SIGRTMIN/SIGRTMAX	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	2abfbd952a	std.c: remove comptime asserts of siginfo_t size ... These serve no purpose other than to verify that the compiler is doing layout correctly, and this is clearly not the place for that.	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	fc62e6d7ef	compiler-rt: export __stack_chk_fail as __stack_smash_handler on openbsd	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	abb9ec3840	std.zig.target: add openbsd libc support	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	bfe6117b1b	compiler: support building openbsd crt0 and stub shared libraries ... closes #2878	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	7ce50db17d	compiler: don't error on explicit link_libc=false on requiresLibC() targets ... There are various reasons why one might want to still create libc-less compilations on these targets. Case in point: Compiling our bundled crt0 for OpenBSD. We will still default to linking libc on these targets, though.	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	e4f3eedb8a	update_openbsd_libc: add tool for updating openbsd libc startup code	2025-12-05 12:20:05 +01:00
Alex Rønne Petersen	bb7d949100	libc: add openbsd libc startup code for 7.8	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	9a9cc9c96d	compiler: define ___OpenBSD and OpenBSDX_Y when targeting openbsd libc	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	df147734a3	std.zig.target: handle openbsd libc names	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	c3b9f49ac2	std.zig.target: remove two unused consts	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	0ea6bd08d8	std.zig.LibCDirs: add openbsd support	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	6c9cc3450f	std.Target: bump baseline openbsd version from 7.7 to 7.8 ... Since this is the first version we support cross-compiling for.	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	ac86650bf4	std.Target: factor arch/abi into openbsd minimum version logic	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	6921ae8925	std.Target: openbsd's landisk port is only little endian	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	9e3c240fda	std.Target: add isOpenBSDLibC()	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	5f77dde83a	libc: add openbsd 7.8 abilists	2025-12-05 12:20:04 +01:00
Alex Rønne Petersen	1311e997d4	libc: add openbsd 7.8 headers ... This excludes all headers in /usr/include/dev because that directory is bonkers huge (18M). We can add these on an as-needed basis.	2025-12-05 12:19:07 +01:00
Alex Rønne Petersen	154ad83ea4	process_headers: add openbsd support	2025-12-05 12:07:04 +01:00
Alex Rønne Petersen	c166bb36f6	ci: reduce x86_64-linux timeouts ... These excessive timeouts should no longer be necessary with the recent tuning of job capacity and maxrss on these machines.	2025-12-04 20:52:34 +01:00
Alex Rønne Petersen	78cba86928	ci: set maxrss from $ZSF_MAX_RSS if provided by the runner ... All of our runners now define this. When running a CI script locally, this will not be set, so we default to 0, aka "all available system memory".	2025-12-04 20:06:48 +01:00
Alex Rønne Petersen	2728eb5d5e	build: adjust max_rss on a per-CI-host basis ... This avoids pessimizing concurrency on all machines due to e.g. the macOS machine having high memory usage across the board due to 16K page size. This also adds max_rss to test-unit and test-c-abi since those tend to eat a decent chunk of memory too.	2025-12-04 20:06:12 +01:00
Alex Rønne Petersen	44543800a5	std.process.Child: enable rusage collection for dragonfly, netbsd, openbsd	2025-12-04 03:46:36 +01:00
Alex Rønne Petersen	2659fadb95	std.c: add rusage for dragonfly, netbsd, openbsd	2025-12-04 03:46:36 +01:00
Alex Rønne Petersen	9eed87f93e	std.process.Child: enable rusage collection for freebsd, illumos, serenity	2025-12-04 03:46:36 +01:00
Alex Rønne Petersen	e270c97ed1	ci: don't skip release mode tests on x86_64-windows-debug ... pulsar is much faster than george so we don't need to do this anymore.	2025-12-04 03:46:36 +01:00
Alex Rønne Petersen	c21ce53494	ci: skip spirv and wasm tests on x86_64-freebsd ... These are already built and run on x86_64-linux.	2025-12-04 03:46:36 +01:00
Alex Rønne Petersen	e2a9e568b4	build: add -Dskip-spirv and -Dskip-wasm options	2025-12-04 03:46:35 +01:00
Alex Rønne Petersen	ad9a5187ac	build: add some missing darwin tags	2025-12-04 03:46:35 +01:00
jedisct1	d73fbcc3ae	Merge pull request 'Argon2: use the std.Io interface' (#30084) from jedisct1/zig:argon2 into master ... Reviewed-on: https://codeberg.org/ziglang/zig/pulls/30084	2025-12-03 12:18:09 +01:00
Zihad	cb115cf73a	std.process.ArgIteratorWasi: fix no-args deinit	2025-12-03 08:35:24 +01:00
Alex Rønne Petersen	be9649f4ea	ci: set a sensible maxrss in x86_64-windows scripts	2025-12-03 00:24:58 +01:00
Frank Denis	6fe95c28cf	Argon2: use the std.Io interface ... Also reduce the memory required by tests. 4GB for every test is way too much and doesn't provide much benefits in testing the algorithms.	2025-12-02 23:03:52 +01:00
Andrew Kelley	52ad126bb4	Merge pull request 'std.Io.Threaded: rework cancellation' (#30033) from cancellation into master ... Reviewed-on: https://codeberg.org/ziglang/zig/pulls/30033	2025-12-02 17:58:29 +01:00
Andrew Kelley	bb3f56d5d5	std.Io.Threaded: separate out ECANCELED handling again ... If ECANCELED occurs, it's from pthread_cancel which will *permanently* set that thread to be in a "canceling" state, which means the cancel cannot be ignored. That means it cannot be retried, like EINTR. It must be acknowledged.	2025-12-01 19:17:52 -08:00
Andrew Kelley	cf82064ebc	std.Io.Threaded: don't use pthread_cancel with musl ... It doesn't support setting the "canceled" status to false, so once a thread has been canceled, all operations on the thread start permanently failing.	2025-12-01 19:17:52 -08:00
Andrew Kelley	bf0ffc45b9	std.Io.Threaded: musl: handle ECANCELED same as EINTR ... Otherwise the pthread_cancel can affect unrelated tasks.	2025-12-01 19:17:52 -08:00
Andrew Kelley	54a84964f8	std.os.linux: SIG enum is non-exhaustive	2025-12-01 19:17:52 -08:00
Andrew Kelley	57f5de5b77	std.Io.Threaded: use the correct mmsghdr struct	2025-12-01 19:17:52 -08:00
Andrew Kelley	103467fa6c	std.Io.Threaded: make is_musl linux-only	2025-12-01 19:17:52 -08:00
David Rubin	85053a6a36	link.Elf: implement aarch64 relocation	2025-12-01 19:17:52 -08:00
Andrew Kelley	c4f5dda135	std.Io.Threaded: re-introduce retry logic behind config	2025-12-01 19:17:52 -08:00
Andrew Kelley	de87bad4c3	std.Io.Threaded: don't solve the cancel race after all ... Unfortunately, trying again until the cancellation request is acknowledged has been observed to incur a large amount of overhead, and usually strong cancellation guarantees are not needed, so the race condition is not handled here. Users who want to avoid this have this menu of options instead: * Use no libc, in which case Zig std lib can avoid the race (tracking issue: https://codeberg.org/ziglang/zig/issues/30049) * Use musl libc * Use `std.Io.Evented`. But this is not implemented yet. Tracked by - https://codeberg.org/ziglang/zig/issues/30050 - https://codeberg.org/ziglang/zig/issues/30051 glibc + threaded is the only problematic combination.	2025-12-01 19:17:52 -08:00
Andrew Kelley	144206856e	std.Io.Threaded: fix compilation for riscv32-linux	2025-12-01 19:17:52 -08:00
Andrew Kelley	9e981c3ae5	std.os.linux: delete unnecessary @compileError ... Without this, it already fails to compile with a sufficiently helpful error message.	2025-12-01 19:17:52 -08:00
Andrew Kelley	39ac40209b	std.Io.Threaded: use musl's beautiful pthread_cancel semantics	2025-12-01 19:17:52 -08:00
Andrew Kelley	d60760d61e	std.Io.Threaded: tune requestCancel ... On a heavily loaded Linux 6.17.5, I observed a maximum of 20 attempts not acknowledged before the timeout (including exponential backoff) was sufficient, despite the heavy load. The time wasted here sleeping is mitigated by the fact that, later on, the system will likely wait for the canceled task, causing it to indefinitely yield until the canceled task finishes, and the task must acknowledge the cancel before it proceeds to that point.	2025-12-01 19:17:52 -08:00