]> sourceware.org Git - glibc.git/log
glibc.git
5 hours agox86: Add `Avoid_STOSB` tunable to allow NT memset without ERMS
Noah Goldstein [Wed, 14 Aug 2024 06:37:31 +0000 (14:37 +0800)]
x86: Add `Avoid_STOSB` tunable to allow NT memset without ERMS

The goal of this flag is to allow targets which don't prefer/have ERMS
to still access the non-temporal memset implementation.

There are 4 cases for tuning memset:
    1) `Avoid_STOSB && Avoid_Non_Temporal_Memset`
        - Memset with temporal stores
    2) `Avoid_STOSB && !Avoid_Non_Temporal_Memset`
        - Memset with temporal/non-temporal stores. Non-temporal path
          goes through `rep stosb` path. We accomplish this by setting
          `x86_rep_stosb_threshold` to
          `x86_memset_non_temporal_threshold`.
    3) `!Avoid_STOSB && Avoid_Non_Temporal_Memset`
        - Memset with temporal stores/`rep stosb`
    3) `!Avoid_STOSB && !Avoid_Non_Temporal_Memset`
        - Memset with temporal stores/`rep stosb`/non-temporal stores.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
5 hours agox86: Use `Avoid_Non_Temporal_Memset` to control non-temporal path
Noah Goldstein [Wed, 14 Aug 2024 06:37:30 +0000 (14:37 +0800)]
x86: Use `Avoid_Non_Temporal_Memset` to control non-temporal path

This is just a refactor and there should be no behavioral change from
this commit.

The goal is to make `Avoid_Non_Temporal_Memset` a more universal knob
for controlling whether we use non-temporal memset rather than having
extra logic based on vendor.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
6 hours agox86: Fix bug in strchrnul-evex512 [BZ #32078]
Noah Goldstein [Tue, 13 Aug 2024 15:29:14 +0000 (23:29 +0800)]
x86: Fix bug in strchrnul-evex512 [BZ #32078]

Issue was we were expecting not matches with CHAR before the start of
the string in the page cross case.

The check code in the page cross case:
```
    and    $0xffffffffffffffc0,%rax
    vmovdqa64 (%rax),%zmm17
    vpcmpneqb %zmm17,%zmm16,%k1
    vptestmb %zmm17,%zmm17,%k0{%k1}
    kmovq  %k0,%rax
    inc    %rax
    shr    %cl,%rax
    je     L(continue)
```

expects that all characters that neither match null nor CHAR will be
1s in `rax` prior to the `inc`. Then the `inc` will overflow all of
the 1s where no relevant match was found.

This is incorrect in the page-cross case, as the
`vmovdqa64 (%rax),%zmm17` loads from before the start of the input
string.

If there are matches with CHAR before the start of the string, `rax`
won't properly overflow.

The fix is quite simple. Just replace:

```
    inc    %rax
    shr    %cl,%rax
```
With:
```
    sar    %cl,%rax
    inc    %rax
```

The arithmetic shift will clear any matches prior to the start of the
string while maintaining the signbit so the 1s can properly overflow
to zero in the case of no matches.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
28 hours agoTest errno setting on strtod overflow in tst-strtod-round
Joseph Myers [Wed, 14 Aug 2024 17:15:46 +0000 (17:15 +0000)]
Test errno setting on strtod overflow in tst-strtod-round

We have no tests that errno is set to ERANGE on overflow of
strtod-family functions (we do have some tests for underflow, in
tst-strtod-underflow).  Add such tests to tst-strtod-round.

Tested for x86_64.

35 hours agolibio/tst-getdelim: Add new test covering NUL as a delimiter
Frédéric Bérat [Tue, 13 Aug 2024 10:01:26 +0000 (12:01 +0200)]
libio/tst-getdelim: Add new test covering NUL as a delimiter

Add a new test to getdelim to verify that '\0' can be set as a
delimiter.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 days agomanual: Document dprintf and vdprintf
Florian Weimer [Tue, 13 Aug 2024 13:52:34 +0000 (15:52 +0200)]
manual: Document dprintf and vdprintf

Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 days agomanual: Document generic printf error codes
Florian Weimer [Tue, 13 Aug 2024 13:52:34 +0000 (15:52 +0200)]
manual: Document generic printf error codes

Describe EOVERFLOW, ENOMEN, EILSEQ.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>
4 days agoARC: Regenerate ULPs
Pavel Kozlov [Fri, 9 Aug 2024 13:21:31 +0000 (14:21 +0100)]
ARC: Regenerate ULPs

Regenerate fpu and soft-fp ULPs. Based on results from HSDK-4xD board
with GCC 14 build.
Including new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

6 days agosupport: Add options list terminator to the test driver
Florian Weimer [Fri, 9 Aug 2024 15:01:17 +0000 (17:01 +0200)]
support: Add options list terminator to the test driver

This avoids crashes if a test is passed unknown options.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
6 days agoDefine __libc_initial for the static libc
Florian Weimer [Fri, 9 Aug 2024 14:17:14 +0000 (16:17 +0200)]
Define __libc_initial for the static libc

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
6 days agoTurn on -Wimplicit-fallthrough by default if available
Florian Weimer [Fri, 9 Aug 2024 13:34:53 +0000 (15:34 +0200)]
Turn on -Wimplicit-fallthrough by default if available

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
6 days agoLoongArch: Add cfi instructions for _dl_tlsdesc_dynamic
mengqinggang [Fri, 5 Jul 2024 02:40:33 +0000 (10:40 +0800)]
LoongArch: Add cfi instructions for _dl_tlsdesc_dynamic

In _dl_tlsdesc_dynamic, there are three 'addi.d sp, sp, -size'
instructions to allocate stack size for Float/LSX/LASX registers.
Every 'addi.d sp, sp, -size' needs a cfi_adjust_cfa_offset because
of sp is used to compute CFA. But only one 'addi.d sp, sp, -size'
will be run according to HWCAP value. And all cfi_adjust_cfa_offset
will be executed in stack unwinding, it result in incorrect CFA.

Change _dl_tlsdesc_dynamic to _dl_tlsdesc_dynamic,
_dl_tlsdesc_dynamic_lsx and _dl_tlsdesc_dynamic_lasx.
Conflicting cfi instructions can be distributed to the three functions.
And cfi instructions can correspond to stack down instructions.

6 days agoLoongArch: Regenerate ULPs
caiyinyu [Thu, 8 Aug 2024 02:31:03 +0000 (10:31 +0800)]
LoongArch: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

Signed-off-by: caiyinyu <caiyinyu@loongson.cn>
7 days agoRISC-V: Regenerate ULPs
Julian Zhu [Thu, 8 Aug 2024 08:11:32 +0000 (16:11 +0800)]
RISC-V: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
7 days agoMIPS: Regenerate ULPs
Julian Zhu [Thu, 8 Aug 2024 08:11:16 +0000 (16:11 +0800)]
MIPS: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

Signed-off-by: Julian Zhu <jz531210@gmail.com>
Reviewed-by: Florian Weimer <fweimer@redhat.com>
7 days agopowerpc64le: Update ulps
Florian Weimer [Thu, 8 Aug 2024 11:42:12 +0000 (13:42 +0200)]
powerpc64le: Update ulps

Based on results from a POWER8 system with a GCC 8 build.

7 days agoelf: Remove struct dl_init_args from elf/dl-open.c
Florian Weimer [Thu, 8 Aug 2024 11:35:03 +0000 (13:35 +0200)]
elf: Remove struct dl_init_args from elf/dl-open.c

It is completely redundant with struct dl_open_args.

7 days agos390x: Update ulps
Florian Weimer [Thu, 8 Aug 2024 10:57:35 +0000 (12:57 +0200)]
s390x: Update ulps

Based on results from a z16 system with a GCC 8 build.

8 days agonptl: Fix stray process left by tst-cancel7 blocking testing
Maciej W. Rozycki [Wed, 7 Aug 2024 18:46:21 +0000 (19:46 +0100)]
nptl: Fix stray process left by tst-cancel7 blocking testing

Fix an issue with commit b74121ae4bc5 ("Update.") and prevent a stray
process from being left behind by tst-cancel7 (and also tst-cancelx7,
which is the same test built with '-fexceptions' additionally supplied
to the compiler), which then blocks remote testing until the process has
been killed by hand.

This test case creates a thread that runs an extra copy of the test via
system(3) and using the '--direct' option so that the test wrapper does
not interfere with this instance.  This extra copy executes its business
and calls sigsuspend(2) and then never terminates by itself.  Instead it
relies on being killed by the main test process directly via a thread
cancellation request or, should that fail, by issuing SIGKILL either at
the conclusion of 'do_test' or by the test driver via 'do_cleanup' where
the test timeout has been hit or the test driver interrupted.

However if the main test process has been instead killed by a signal,
such as due to incorrect execution, before it had a chance to kill the
extra copy of the test case, then the test wrapper will terminate
without running 'do_cleanup' and consequently the extra copy of the test
case will remain forever in its suspended state, and in the remote case
in particular it means that the remote test wrapper will wait forever
for the SSH command to complete.

This has been observed with the 'alpha-linux-gnu' target, where the main
test process triggers SIGSEGV and the test wrapper correctly records:

Didn't expect signal from child: got `Segmentation fault'

in nptl/tst-cancel7.out and terminates, but then the calling SSH command
continues waiting for the remaining process started in the same session
on the remote target to complete.

Address this problem by also registering 'do_cleanup' via atexit(3),
observing that 'support_delete_temp_files' is registered by the test
wrapper before the test initializing function 'do_prepare' is called and
that we call all the functions registered in the reverse of the order in
which they were registered, so it is safe to refer to 'pidfilename' in
'do_cleanup' invoked by exit(3) because by that time temporary files
have not yet been deleted.

A minor inconvenience is that if 'signal_handler' is invoked in the test
wrapper as a result of SIGALRM rather than SIGINT, then 'do_cleanup'
will be called twice, once as a cleanup handler and again by exit(3).
In reality it is harmless though, because issuing SIGKILL is guarded by
a record lock, so if the first call has succeeded in killing the extra
copy of the test case, then the subsequent call will do nothing.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
8 days agonptl: Reorder semaphore release in tst-cancel7
Maciej W. Rozycki [Wed, 7 Aug 2024 18:46:21 +0000 (19:46 +0100)]
nptl: Reorder semaphore release in tst-cancel7

Move the release of the semaphore used to synchronize between an extra
copy of the test run as a separate process and the main test process
until after the PID file has been locked.  It is so that if the cleanup
function gets called by the test driver due to premature termination of
the main test process, then the function does not get at the PID file
before it has been locked and conclude that the extra copy of the test
has already terminated.  This won't usually happen due to a relatively
high amount of time required to elapse before timeout triggers in the
test driver, but it will change with the next change.

There is still a small time window remaining with this change in place
where the main test process gets killed for some reason between the
extra copy of the test has been already started by pthread_create(3) and
a successful return from the call to sem_wait(3), in which case the
cleanup function can be reached before PID has been written to the PID
file and the file locked.  It seems that with the test case structured
as it is now and PID-based process management we have no means to avoid
it.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
8 days agopowerpc: Regenerate ULPs for soft-fp
Adhemerval Zanella [Wed, 7 Aug 2024 13:46:59 +0000 (10:46 -0300)]
powerpc: Regenerate ULPs for soft-fp

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

8 days agopowerpc: Update soft-fp ulps
Adhemerval Zanella [Wed, 7 Aug 2024 13:46:37 +0000 (10:46 -0300)]
powerpc: Update soft-fp ulps

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

8 days agosparc: Regenerate ULPs
Adhemerval Zanella [Wed, 7 Aug 2024 13:38:26 +0000 (10:38 -0300)]
sparc: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

8 days agoi386: Regenerate ULPs
Adhemerval Zanella [Wed, 7 Aug 2024 13:34:06 +0000 (10:34 -0300)]
i386: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

8 days agoarm: Regenerate ULPs
Adhemerval Zanella [Wed, 7 Aug 2024 13:27:38 +0000 (10:27 -0300)]
arm: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

8 days agoaarch64: Regenerate ULPs
Adhemerval Zanella [Wed, 7 Aug 2024 13:27:04 +0000 (10:27 -0300)]
aarch64: Regenerate ULPs

From new tests added by 07972839108495245d8b93ca546462b3f4dad47f.

8 days agosysdeps: Re-flow and sort multiline gnu/Makefile definitions
Adhemerval Zanella [Wed, 8 Nov 2023 16:49:52 +0000 (13:49 -0300)]
sysdeps: Re-flow and sort multiline gnu/Makefile definitions

8 days agologin: Re-flow and sort multiline Makefile definitions
Adhemerval Zanella [Wed, 8 Nov 2023 16:49:48 +0000 (13:49 -0300)]
login: Re-flow and sort multiline Makefile definitions

8 days agobenchtests: Add random memset benchmark
Wilco Dijkstra [Wed, 7 Aug 2024 13:49:33 +0000 (14:49 +0100)]
benchtests: Add random memset benchmark

Add a new randomized memset test similar to bench-random-memcpy.  Instead of
repeating the same call to memset over and over again, it times a large number
of different inputs.  The distribution of memset length and alignment is based
on SPEC2017 (length up to 4096 and alignment up to 64).

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
8 days agoAArch64: Improve generic strlen
Wilco Dijkstra [Wed, 7 Aug 2024 13:43:47 +0000 (14:43 +0100)]
AArch64: Improve generic strlen

Improve performance by handling another 16 bytes before entering the loop.
Use ADDHN in the loop to avoid SHRN+FMOV when it terminates.  Change final
size computation to avoid increasing latency.  On Neoverse V1 performance
of the random strlen benchmark improves by 4.6%.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
8 days agonss: Fix incorrect switch fall-through in tst-nss-gai-actions
Florian Weimer [Wed, 7 Aug 2024 12:59:08 +0000 (14:59 +0200)]
nss: Fix incorrect switch fall-through in tst-nss-gai-actions

This only happened on test failure.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
8 days agoadded inputs giving large errors on x86_64 for new C23 functions
Paul Zimmermann [Thu, 25 Jul 2024 14:38:08 +0000 (16:38 +0200)]
added inputs giving large errors on x86_64 for new C23 functions

These functions are exp10m1, exp2m1, log10p1, log2p1.
Also regenerated ulps on x86_64.

For each format, there are 4 values, one for each rounding mode.
(For the intel96 format, there are 8 values, 4 for Intel hardware,
and 4 for AMD hardware. However, regen-ulps was only run on Intel.
It should be run in a separate patch on a AMD x86_64.)
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
9 days agostdlib: Link tst-concurrent-quick_exit with $(shared-thread-library)
Adhemerval Zanella [Tue, 6 Aug 2024 16:53:52 +0000 (13:53 -0300)]
stdlib: Link tst-concurrent-quick_exit with $(shared-thread-library)

This avoids a Hurd build failure.  Fixes commit c6af8a9a3c
("stdlib: Allow concurrent quick_exit (BZ 31997)").

9 days agoUse binutils 2.43 branch in build-many-glibcs.py
Joseph Myers [Tue, 6 Aug 2024 17:00:50 +0000 (17:00 +0000)]
Use binutils 2.43 branch in build-many-glibcs.py

This patch makes build-many-glibcs.py use binutils 2.43 branch.

Tested with build-many-glibcs.py (host-libraries, compilers and glibcs
builds).

9 days agortld: Add test case for '--' option
Henrik Lindström [Fri, 26 Jul 2024 10:33:51 +0000 (12:33 +0200)]
rtld: Add test case for '--' option

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
9 days agortld: Fix handling of '--' option
Henrik Lindström [Fri, 26 Jul 2024 10:33:50 +0000 (12:33 +0200)]
rtld: Fix handling of '--' option

It always resulted in the error `unrecognized option '--'` previously.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
9 days agomanual/stdio: Further clarify putc, putwc, getc, and getwc
Arjun Shankar [Tue, 30 Jul 2024 09:37:57 +0000 (11:37 +0200)]
manual/stdio: Further clarify putc, putwc, getc, and getwc

This is a follow-up to 10de4a47ef3f481592e3c62eb07bcda23e9fde4d that
reworded the manual entries for putc and putwc and removed any
performance claims.

This commit further clarifies these entries and brings getc and getwc in
line with the descriptions of putc and putwc, removing any performance
claims from them as well.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
9 days agoLoongArch: Update Ulps.
caiyinyu [Fri, 2 Aug 2024 07:37:42 +0000 (15:37 +0800)]
LoongArch: Update Ulps.

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

Signed-off-by: caiyinyu <caiyinyu@loongson.cn>
10 days agostdlib: Allow concurrent quick_exit (BZ 31997)
Adhemerval Zanella [Mon, 5 Aug 2024 14:27:35 +0000 (11:27 -0300)]
stdlib: Allow concurrent quick_exit (BZ 31997)

As for exit, also allows concurrent quick_exit to avoid race
conditions when it is called concurrently.  Since it uses the same
internal function as exit, the __exit_lock lock is moved to
__run_exit_handlers.  It also solved a potential concurrent when
calling exit and quick_exit concurrently.

The test case 'expected' is expanded to a value larger than the
minimum required by C/POSIX (32 entries) so at_quick_exit() will
require libc to allocate a new block.  This makes the test mre likely to
trigger concurrent issues (through free() at __run_exit_handlers)
if quick_exit() interacts with the at_quick_exit list concurrently.

This is also the latest interpretation of the Austin Ticket [1].

Checked on x86_64-linux-gnu.

[1] https://austingroupbugs.net/view.php?id=1845
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
10 days agoelf: Avoid re-initializing already allocated TLS in dlopen (bug 31717)
Florian Weimer [Thu, 1 Aug 2024 21:31:30 +0000 (23:31 +0200)]
elf: Avoid re-initializing already allocated TLS in dlopen (bug 31717)

The old code used l_init_called as an indicator for whether TLS
initialization was complete.  However, it is possible that
TLS for an object is initialized, written to, and then dlopen
for this object is called again, and l_init_called is not true at
this point.  Previously, this resulted in TLS being initialized
twice, discarding any interim writes (technically introducing a
use-after-free bug even).

This commit introduces an explicit per-object flag, l_tls_in_slotinfo.
It indicates whether _dl_add_to_slotinfo has been called for this
object.  This flag is used to avoid double-initialization of TLS.
In update_tls_slotinfo, the first_static_tls micro-optimization
is removed because preserving the initalization flag for subsequent
use by the second loop for static TLS is a bit complicated, and
another per-object flag does not seem to be worth it.  Furthermore,
the l_init_called flag is dropped from the second loop (for static
TLS initialization) because l_need_tls_init on its own prevents
double-initialization.

The remaining l_init_called usage in resize_scopes and update_scopes
is just an optimization due to the use of scope_has_map, so it is
not changed in this commit.

The isupper check ensures that libc.so.6 is TLS is not reverted.
Such a revert happens if l_need_tls_init is not cleared in
_dl_allocate_tls_init for the main_thread case, now that
l_init_called is not checked anymore in update_tls_slotinfo
in elf/dl-open.c.

Reported-by: Jonathon Anderson <janderson@rice.edu>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
10 days agoelf: Clarify and invert second argument of _dl_allocate_tls_init
Florian Weimer [Thu, 1 Aug 2024 21:31:23 +0000 (23:31 +0200)]
elf: Clarify and invert second argument of _dl_allocate_tls_init

Also remove an outdated comment: _dl_allocate_tls_init is
called as part of pthread_create.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>
10 days agoFix name space violation in fortify wrappers (bug 32052)
Andreas Schwab [Mon, 5 Aug 2024 08:55:51 +0000 (10:55 +0200)]
Fix name space violation in fortify wrappers (bug 32052)

Rename the identifier sz to __sz everywhere.

Fixes: a643f60c53 ("Make sure that the fortified function conditionals are constant")
10 days agoiconv: Fix matching of multi-character transliterations (bug 31859)
Andreas Schwab [Mon, 10 Jun 2024 10:19:17 +0000 (12:19 +0200)]
iconv: Fix matching of multi-character transliterations (bug 31859)

Only return __GCONV_INCOMPLETE_INPUT for a partial match when the end of
the input buffer is reached.  Otherwise it is a non-match, and other
patterns should be tried.

13 days agox86: Tunables may incorrectly set Prefer_PMINUB_for_stringop (bug 32047)
Florian Weimer [Fri, 2 Aug 2024 13:22:14 +0000 (15:22 +0200)]
x86: Tunables may incorrectly set Prefer_PMINUB_for_stringop (bug 32047)

Fixes commit 5bcf6265f215326d14dfacdce8532792c2c7f8f8 ("x86:
Disable non-temporal memset on Skylake Server").

Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
13 days agox86: Add missing switch/case fall-through markers to init_cpu_features
Florian Weimer [Fri, 2 Aug 2024 13:22:07 +0000 (15:22 +0200)]
x86: Add missing switch/case fall-through markers to init_cpu_features

The commits introducing these fall-throughs intended them to
happen.

Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
13 days agostdlib: Link tst-concurrent-exit with $(shared-thread-library)
Florian Weimer [Fri, 2 Aug 2024 12:45:10 +0000 (14:45 +0200)]
stdlib: Link tst-concurrent-exit with $(shared-thread-library)

This avoids a Hurd build failure.  Fixes commit f6ba993e0cda0ca
("stdlib: Allow concurrent exit (BZ 31997)").

13 days agohurd: Fix missing pthread_ compat symbol in libc
Samuel Thibault [Wed, 17 Jul 2024 00:03:13 +0000 (02:03 +0200)]
hurd: Fix missing pthread_ compat symbol in libc

5476f8cd2e68 ("htl: move pthread_self info libc.") and
9dfa2562162b ("htl: move pthread_equal into libc") to
1dc0bc8f0748 ("htl: move pthread_attr_setdetachstate into libc")
moved some pthread_ symbols from libpthread.so to libc.so, but missed
adding the compat version like 5476f8cd2e68 ("htl: move pthread_self
info libc.") did: libc already had these symbols as forwards,
but versioned GLIBC_2.21, while the symbols in libpthread.so were
versioned GLIBC_2.12.

To fix running executables built before this, we thus have to add the
GLIBC_2.12 version, otherwise execution fails with e.g.

/usr/lib/i386-gnu/libglib-2.0.so: symbol lookup error: /usr/lib/i386-gnu/libglib-2.0.so: undefined symbol: pthread_attr_setinheritsched, version GLIBC_2.12

2 weeks agoresolv: Fix tst-resolv-short-response for older GCC (bug 32042)
Florian Weimer [Thu, 1 Aug 2024 08:46:10 +0000 (10:46 +0200)]
resolv: Fix tst-resolv-short-response for older GCC (bug 32042)

Previous GCC versions do not support the C23 change that
allows labels on declarations.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 weeks agoAdd mremap tests
H.J. Lu [Wed, 24 Jul 2024 21:05:15 +0000 (14:05 -0700)]
Add mremap tests

Add tests for MREMAP_MAYMOVE and MREMAP_FIXED.  On Linux, also test
MREMAP_DONTUNMAP.

Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 weeks agomremap: Update manual entry
H.J. Lu [Wed, 24 Jul 2024 21:05:14 +0000 (14:05 -0700)]
mremap: Update manual entry

Update mremap manual entry:

1. Change mremap to variadic.
2. Document MREMAP_FIXED and MREMAP_DONTUNMAP.

Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 weeks agolinux: Update the mremap C implementation [BZ #31968]
H.J. Lu [Wed, 24 Jul 2024 21:05:13 +0000 (14:05 -0700)]
linux: Update the mremap C implementation [BZ #31968]

Update the mremap C implementation to support the optional argument for
MREMAP_DONTUNMAP added in Linux 5.7 since it may not always be correct
to implement a variadic function as a non-variadic function on all Linux
targets.  Return MAP_FAILED and set errno to EINVAL for unknown flag bits.
This fixes BZ #31968.

Note: A test must be added when a new flag bit is introduced.

Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 weeks agoEnhanced test coverage for strncmp, wcsncmp
Florian Weimer [Thu, 27 Jun 2024 14:26:56 +0000 (16:26 +0200)]
Enhanced test coverage for strncmp, wcsncmp

Add string/test-strncmp-nonarray and
wcsmbs/test-wcsncmp-nonarray.

This is the test that uncovered bug 31934.  Test run time
is more than one minute on a fairly current system, so turn
these into xtests that do not run automatically.

Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
2 weeks agoEnhance test coverage for strnlen, wcsnlen
Florian Weimer [Wed, 26 Jun 2024 09:27:54 +0000 (11:27 +0200)]
Enhance test coverage for strnlen, wcsnlen

This commit adds string/test-strnlen-nonarray and
wcsmbs/test-wcsnlen-nonarray.

Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
2 weeks agostdlib: Mark `abort` as `cold`
Noah Goldstein [Mon, 29 Jul 2024 07:34:33 +0000 (15:34 +0800)]
stdlib: Mark `abort` as `cold`

This helps HotColdSplitting in GCC/LLVM.

Thought about doing `exit` as well since its only called once per
process, but since its easy to imagine a hot path leading into
`exit(0)`, its less clear if its profitable.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 weeks agox86_64 hurd: ensure we have a large enough buffer to receive exception_raise requests.
Flavio Cruz [Tue, 30 Jul 2024 04:51:20 +0000 (00:51 -0400)]
x86_64 hurd: ensure we have a large enough buffer to receive exception_raise requests.

Message-ID: <gtxd6s4s7fi7hdrlb7zayq3akij7x6jqawwq3zfl3v4nqspulo@euucuzeonrl6>

2 weeks agostdlib: Allow concurrent exit (BZ 31997)
Adhemerval Zanella [Thu, 25 Jul 2024 18:41:44 +0000 (15:41 -0300)]
stdlib: Allow concurrent exit (BZ 31997)

Even if C/POSIX standard states that exit is not formally thread-unsafe,
calling it more than once is UB.  The glibc already supports
it for the single-thread, and both elf/nodelete2.c and tst-rseq-disable.c
call exit from a DSO destructor (which is called by _dl_fini, registered
at program startup with __cxa_atexit).

However, there are still race issues when it is called more than once
concurrently by multiple threads.  A recent Rust PR triggered this
issue [1], which resulted in an Austin Group ask for clarification [2].
Besides it, there is a discussion to make concurrent calling not UB [3],
wtih a defined semantic where any remaining callers block until the first
call to exit has finished (reentrant calls, leaving through longjmp, and
exceptions are still undefined).

For glibc, at least reentrant calls are required to be supported to avoid
changing the current behaviour.  This requires locking using a recursive
lock, where any exit called by atexit() handlers resumes at the point of
the current handler (thus avoiding calling the current handle multiple
times).

Checked on x86_64-linux-gnu and aarch64-linux-gnu.

[1] https://github.com/rust-lang/rust/issues/126600
[2] https://austingroupbugs.net/view.php?id=1845
[3] https://www.openwall.com/lists/libc-coord/2024/07/24/4
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 weeks agoAdd F_DUPFD_QUERY from Linux 6.10 to bits/fcntl-linux.h
Adhemerval Zanella [Tue, 16 Jul 2024 17:08:49 +0000 (17:08 +0000)]
Add F_DUPFD_QUERY from Linux 6.10 to bits/fcntl-linux.h

It was added by commit c62b758bae6af16 as a way for userspace to
check if two file descriptors refer to the same struct file.

Checked on aarch64-linux-gnu.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agoAdd STATX_SUBVOL from Linux 6.10 to bits/statx-generic.h
Adhemerval Zanella [Tue, 16 Jul 2024 17:08:48 +0000 (17:08 +0000)]
Add STATX_SUBVOL from Linux 6.10 to bits/statx-generic.h

Tested for aarch64-linux-gnu.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agoUpdate kernel version to 6.10 in header constant tests
Adhemerval Zanella [Tue, 16 Jul 2024 17:08:47 +0000 (17:08 +0000)]
Update kernel version to 6.10 in header constant tests

This patch updates the kernel version in the tests tst-mman-consts.py,
tst-mount-consts.py, and tst-pidfd-consts.py to 6.9.

There are no new constants covered by these tests in 6.10.

Tested with build-many-glibcs.py.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agoUpdate syscall lists for Linux 6.10
Adhemerval Zanella [Tue, 16 Jul 2024 17:08:46 +0000 (17:08 +0000)]
Update syscall lists for Linux 6.10

Linux 6.10 changes for syscall are:

  * mseal for all architectures.
  * map_shadow_stack for x32.
  * Replace sync_file_range with sync_file_range2 for csky (which
    fixes a broken sync_file_range usage).

Update syscall-names.list and regenerate the arch-syscall.h headers
with build-many-glibcs.py update-syscalls.

Tested with build-many-glibcs.py.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agoUse Linux 6.10 in build-many-glibcs.py
Adhemerval Zanella [Tue, 16 Jul 2024 17:08:45 +0000 (17:08 +0000)]
Use Linux 6.10 in build-many-glibcs.py

Tested with build-many-glibcs.py (host-libraries, compilers and glibcs
builds).
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agoMitigation for "clone on sparc might fail with -EFAULT for no valid reason" (bz 31394)
Michael Karcher [Sun, 28 Jul 2024 13:30:57 +0000 (15:30 +0200)]
Mitigation for "clone on sparc might fail with -EFAULT for no valid reason" (bz 31394)

It seems the kernel can not deal with uncommitted stack space in the area intended
for the register window when executing the clone() system call. So create a nested
frame (proxy for the kernel frame) and flush it from the processor to memory to
force committing pages to the stack before invoking the system call.

Bug: https://www.mail-archive.com/debian-glibc@lists.debian.org/msg62592.html
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=31394
See-also: https://lore.kernel.org/sparclinux/62f9be9d-a086-4134-9a9f-5df8822708af@mkarcher.dialup.fu-berlin.de/
Signed-off-by: Michael Karcher <sourceware-bugzilla@mkarcher.dialup.fu-berlin.de>
Reviewed-by: DJ Delorie <dj@redhat.com>
2 weeks agomanual: make setrlimit() description less ambiguous
Lukas Bulwahn [Mon, 29 Jul 2024 09:08:17 +0000 (11:08 +0200)]
manual: make setrlimit() description less ambiguous

The existing description for setrlimit() has some ambiguity. It could be
understood to have the semantics of getrlimit(), i.e., the limits from the
process are stored in the provided rlp pointer.

Make the description more explicit that rlp are the input values, and that
the limits of the process is changed with this function.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agomanual/stdio: Clarify putc and putwc
Arjun Shankar [Mon, 29 Jul 2024 12:30:59 +0000 (14:30 +0200)]
manual/stdio: Clarify putc and putwc

The manual entry for `putc' described what "most systems" do instead of
describing the glibc implementation and its guarantees.  This commit
fixes that by warning that putc may be implemented as a macro that
double-evaluates `stream', and removing the performance claim.

Even though the current `putc' implementation does not double-evaluate
`stream', offering this obscure guarantee as an extension to what
POSIX allows does not seem very useful.

The entry for `putwc' is also edited to bring it in line with `putc'.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 weeks agoMIPS: Regenerate ulps
Julian Zhu [Sat, 27 Jul 2024 10:27:52 +0000 (18:27 +0800)]
MIPS: Regenerate ulps

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

Signed-off-by: Julian Zhu <jz531210@gmail.com>
2 weeks agomalloc: Link threading tests with $(shared-thread-library)
Florian Weimer [Sat, 27 Jul 2024 14:23:55 +0000 (16:23 +0200)]
malloc: Link threading tests with $(shared-thread-library)

Fixes build failures on Hurd.

2 weeks agoassert: Mark `__assert_fail` as `cold`
Noah Goldstein [Fri, 26 Jul 2024 08:35:13 +0000 (16:35 +0800)]
assert: Mark `__assert_fail` as `cold`

This helps compilers split the codegen for setting up the arguments
(`__expression`, `__filename`, etc...) from the potentially hot cold
where the `assert` is to a presumably cold region on the assertion
failure path.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
Reviewed-by: Sam James <sam@gentoo.org>
2 weeks agostdio-common: Add test for vfscanf with matches longer than INT_MAX [BZ #27650]
Maciej W. Rozycki [Fri, 26 Jul 2024 12:21:34 +0000 (13:21 +0100)]
stdio-common: Add test for vfscanf with matches longer than INT_MAX [BZ #27650]

Complement commit b03e4d7bd25b ("stdio: fix vfscanf with matches longer
than INT_MAX (bug 27650)") and add a test case for the issue, inspired
by the reproducer provided with the bug report.

This has been verified to succeed as from the commit referred and fail
beforehand.

As the test requires 2GiB of data to be passed around its performance
has been evaluated using a choice of systems and the execution time
determined to be respectively in the range of 9s for POWER9@2.166GHz,
24s for FU740@1.2GHz, and 40s for 74Kf@950MHz.  As this is on the verge
of and beyond the default timeout it has been increased by the factor of
8.  Regardless, following recent practice the test has been added to the
standard rather than extended set.

Reviewed-by: DJ Delorie <dj@redhat.com>
2 weeks agosupport: Add FAIL test failure helper
Maciej W. Rozycki [Fri, 26 Jul 2024 12:21:34 +0000 (13:21 +0100)]
support: Add FAIL test failure helper

Add a FAIL test failure helper analogous to FAIL_RET, that does not
cause the current function to return, providing a standardized way to
report a test failure with a message supplied while permitting the
caller to continue executing, for further reporting, cleaning up, etc.

Update existing test cases that provide a conflicting definition of FAIL
by removing the local FAIL definition and then as follows:

- tst-fortify-syslog: provide a meaningful message in addition to the
  file name already added by <support/check.h>; 'support_record_failure'
  is already called by 'support_print_failure_impl' invoked by the new
  FAIL test failure helper.

- tst-ctype: no update to FAIL calls required, with the name of the file
  and the line number within of the failure site additionally included
  by the new FAIL test failure helper, and error counting plus count
  reporting upon test program termination also already provided by
  'support_record_failure' and 'support_report_failure' respectively,
  called by 'support_print_failure_impl' and 'adjust_exit_status' also
  respectively.  However in a number of places 'printf' is called and
  the error count adjusted by hand, so update these places to make use
  of FAIL instead.  And last but not least adjust the final summary just
  to report completion, with any error count following as reported by
  the test driver.

- test-tgmath2: no update to FAIL calls required, with the name of the
  file of the failure site additionally included by the new FAIL test
  failure helper.  Also there is no need to track the return status by
  hand as any call to FAIL will eventually cause the test case to return
  an unsuccesful exit status regardless of the return status from the
  test function, via a call to 'adjust_exit_status' made by the test
  driver.

Reviewed-by: DJ Delorie <dj@redhat.com>
2 weeks agoposix: Use <support/check.h> facilities in tst-truncate and tst-truncate64
Maciej W. Rozycki [Fri, 26 Jul 2024 12:21:34 +0000 (13:21 +0100)]
posix: Use <support/check.h> facilities in tst-truncate and tst-truncate64

Remove local FAIL macro in favor to FAIL_RET from <support/check.h>,
which provides equivalent reporting, with the name of the file of the
failure site additionally included, for the tst-truncate-common core
shared between the tst-truncate and tst-truncate64 tests.

Reviewed-by: DJ Delorie <dj@redhat.com>
2 weeks agonptl: Use <support/check.h> facilities in tst-setuid3
Maciej W. Rozycki [Fri, 26 Jul 2024 12:21:34 +0000 (13:21 +0100)]
nptl: Use <support/check.h> facilities in tst-setuid3

Remove local FAIL macro in favor to FAIL_EXIT1 from <support/check.h>,
which provides equivalent reporting, with the name of the file and the
line number within of the failure site additionally included.  Remove
FAIL_ERR altogether and include ": %m" explicitly with the format string
supplied to FAIL_EXIT1 as there seems little value to have a separate
macro just for this.

Reviewed-by: DJ Delorie <dj@redhat.com>
3 weeks agosparc: Regenerate ULPs
Adhemerval Zanella [Thu, 25 Jul 2024 14:06:53 +0000 (11:06 -0300)]
sparc: Regenerate ULPs

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

3 weeks agoi386: Regenerate ULPs
Adhemerval Zanella [Thu, 25 Jul 2024 13:49:06 +0000 (10:49 -0300)]
i386: Regenerate ULPs

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

3 weeks agoarm: Regenerate ULPs
Adhemerval Zanella [Thu, 25 Jul 2024 13:41:00 +0000 (10:41 -0300)]
arm: Regenerate ULPs

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

3 weeks agoaarch64: Regenerate ULPs
Adhemerval Zanella [Thu, 25 Jul 2024 13:40:32 +0000 (10:40 -0300)]
aarch64: Regenerate ULPs

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

3 weeks agopowerpc: Regenerate ULPs for soft-fp
Adhemerval Zanella [Thu, 25 Jul 2024 13:33:40 +0000 (10:33 -0300)]
powerpc: Regenerate ULPs for soft-fp

From new tests added by 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61.

3 weeks agopowerpc: Update ulps for fpu
jeevitha [Wed, 24 Jul 2024 05:58:54 +0000 (11:28 +0530)]
powerpc: Update ulps for fpu

Adjust the ULPs for the log2p1 implementation.

3 weeks agoriscv: Update ulps
Khem Raj [Wed, 24 Jul 2024 04:43:56 +0000 (21:43 -0700)]
riscv: Update ulps

Generated with make regen-ulps using gcc14 on a visionfive2 SBC.

Signed-off-by: Khem Raj <raj.khem@gmail.com>
3 weeks agos390x: Regenerate ULPs.
Stefan Liebler [Thu, 25 Jul 2024 12:14:22 +0000 (14:14 +0200)]
s390x: Regenerate ULPs.

Needed due to:
"This patch adds larger ulp errors for the log2p1 function."
commit 4dc22baa84bdb4111c0ac0db7139bf9ab953bf61

3 weeks agox32/cet: Support shadow stack during startup for Linux 6.10
H.J. Lu [Tue, 23 Jul 2024 00:47:22 +0000 (17:47 -0700)]
x32/cet: Support shadow stack during startup for Linux 6.10

Use RXX_LP in RTLD_START_ENABLE_X86_FEATURES.  Support shadow stack during
startup for Linux 6.10:

commit 2883f01ec37dd8668e7222dfdb5980c86fdfe277
Author: H.J. Lu <hjl.tools@gmail.com>
Date:   Fri Mar 15 07:04:33 2024 -0700

    x86/shstk: Enable shadow stacks for x32

    1. Add shadow stack support to x32 signal.
    2. Use the 64-bit map_shadow_stack syscall for x32.
    3. Set up shadow stack for x32.

Add the map_shadow_stack system call to <fixup-asm-unistd.h> and regenerate
arch-syscall.h.  Tested on Intel Tiger Lake with CET enabled x32.  There
are no regressions with CET enabled x86-64.  There are no changes in CET
enabled x86-64 _dl_start_user.

Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
3 weeks agox86-64: Remove sysdeps/x86_64/x32/dl-machine.h
H.J. Lu [Tue, 23 Jul 2024 00:47:21 +0000 (17:47 -0700)]
x86-64: Remove sysdeps/x86_64/x32/dl-machine.h

Remove sysdeps/x86_64/x32/dl-machine.h by folding x32 ARCH_LA_PLTENTER,
ARCH_LA_PLTEXIT and RTLD_START into sysdeps/x86_64/dl-machine.h.  There
are no regressions on x86-64 nor x32.  There are no changes in x86-64
_dl_start_user.  On x32, _dl_start_user changes are

 <_dl_start_user>:
  mov    %eax,%r12d
+ mov    %esp,%r13d
  mov    (%rsp),%edx
  mov    %edx,%esi
- mov    %esp,%r13d
  and    $0xfffffff0,%esp
  mov    0x0(%rip),%edi        # <_dl_start_user+0x14>
  lea    0x8(%r13,%rdx,4),%ecx

Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
3 weeks agohppa: Update libm-test-ulps
John David Anglin [Wed, 24 Jul 2024 20:43:01 +0000 (16:43 -0400)]
hppa: Update libm-test-ulps

3 weeks agomanual: Do not mention STATIC_TLS in dynamic linker hardening recommendations
Florian Weimer [Wed, 24 Jul 2024 10:50:17 +0000 (12:50 +0200)]
manual: Do not mention STATIC_TLS in dynamic linker hardening recommendations

The current toolchain does not consistently generate it, and
glibc does not use it.

Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
3 weeks agoresolv: Implement strict-error stub resolver option (bug 27929)
Florian Weimer [Wed, 24 Jul 2024 10:06:47 +0000 (12:06 +0200)]
resolv: Implement strict-error stub resolver option (bug 27929)

For now, do not enable this mode by default due to the potential
impact on compatibility with existing deployments.

Reviewed-by: DJ Delorie <dj@redhat.com>
3 weeks agoresolv: Support clearing option flags with a “-” prefix (bug 14799)
Florian Weimer [Wed, 24 Jul 2024 10:06:47 +0000 (12:06 +0200)]
resolv: Support clearing option flags with a “-” prefix (bug 14799)

I think using a “-” prefix is less confusing than introducing
double-negation construct (“no-no-tld-query”).

Reviewed-by: DJ Delorie <dj@redhat.com>
3 weeks agoresolv: Do not wait for non-existing second DNS response after error (bug 30081)
Florian Weimer [Wed, 24 Jul 2024 10:06:47 +0000 (12:06 +0200)]
resolv: Do not wait for non-existing second DNS response after error (bug 30081)

In single-request mode, there is no second response after an error
because the second query has not been sent yet.  Waiting for it
introduces an unnecessary timeout.

Reviewed-by: DJ Delorie <dj@redhat.com>
3 weeks agoresolv: Allow short error responses to match any query (bug 31890)
Florian Weimer [Wed, 24 Jul 2024 10:06:47 +0000 (12:06 +0200)]
resolv: Allow short error responses to match any query (bug 31890)

Reviewed-by: DJ Delorie <dj@redhat.com>
3 weeks agomalloc: add multi-threaded tests for aligned_alloc/calloc/malloc
Miguel Martín [Tue, 16 Jul 2024 15:14:57 +0000 (17:14 +0200)]
malloc: add multi-threaded tests for aligned_alloc/calloc/malloc

Improve aligned_alloc/calloc/malloc test coverage by adding
multi-threaded tests with random memory allocations and with/without
cross-thread memory deallocations.

Perform a number of memory allocation calls with random sizes limited
to 0xffff.

Use the existing DSO ('malloc/tst-aligned_alloc-lib.c') to randomize
allocator selection.

The multi-threaded allocation/deallocation is staged as described below:

- Stage 1: Half of the threads will be allocating memory and the
  other half will be waiting for them to finish the allocation.
- Stage 2: Half of the threads will be allocating memory and the
  other half will be deallocating memory.
- Stage 3: Half of the threads will be deallocating memory and the
  second half waiting on them to finish.

Add 'malloc/tst-aligned-alloc-random-thread.c' where each thread will
deallocate only the memory that was previously allocated by itself.

Add 'malloc/tst-aligned-alloc-random-thread-cross.c' where each thread
will deallocate memory that was previously allocated by another thread.

The intention is to be able to utilize existing malloc testing to ensure
that similar allocation APIs are also exposed to the same rigors.
Reviewed-by: Arjun Shankar <arjun@redhat.com>
3 weeks agomalloc: avoid global locks in tst-aligned_alloc-lib.c
Miguel Martín [Tue, 16 Jul 2024 15:14:56 +0000 (17:14 +0200)]
malloc: avoid global locks in tst-aligned_alloc-lib.c

Make sure the DSO used by aligned_alloc/calloc/malloc tests does not get
a global lock on multithreaded tests.
Reviewed-by: Arjun Shankar <arjun@redhat.com>
3 weeks agoThis patch adds larger ulp errors for the log2p1 function.
Paul Zimmermann [Mon, 22 Jul 2024 06:54:23 +0000 (08:54 +0200)]
This patch adds larger ulp errors for the log2p1 function.

Changes in v2:
- added larger error for long double on AMD reported by Adhemerval
  (https://sourceware.org/pipermail/libc-alpha/2024-June/157755.html)

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
3 weeks agoNEWS: Add new section
Andreas K. Hüttel [Sun, 21 Jul 2024 16:50:57 +0000 (18:50 +0200)]
NEWS: Add new section

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agoIncrease version number to 2.40.9000 glibc-2.40.9000
Andreas K. Hüttel [Sun, 21 Jul 2024 16:49:35 +0000 (18:49 +0200)]
Increase version number to 2.40.9000

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agoAdd ChangeLog file glibc-2.40
Andreas K. Hüttel [Sun, 21 Jul 2024 16:33:37 +0000 (18:33 +0200)]
Add ChangeLog file

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agoIncrease version number to 2.40
Andreas K. Hüttel [Sun, 21 Jul 2024 16:23:25 +0000 (18:23 +0200)]
Increase version number to 2.40

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agopo/*: regenerate (only line number changes)
Andreas K. Hüttel [Sun, 21 Jul 2024 15:50:35 +0000 (17:50 +0200)]
po/*: regenerate (only line number changes)

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agocontrib.texi: Fix format of MIPS and RISC-V
Andreas K. Hüttel [Sun, 21 Jul 2024 12:12:30 +0000 (14:12 +0200)]
contrib.texi: Fix format of MIPS and RISC-V

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agolibc.pot: regenerate (only line number changes)
Andreas K. Hüttel [Sat, 20 Jul 2024 22:33:43 +0000 (00:33 +0200)]
libc.pot: regenerate (only line number changes)

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agoinstall.texi: bump "latest verified" versions
Andreas K. Hüttel [Sat, 20 Jul 2024 22:27:35 +0000 (00:27 +0200)]
install.texi: bump "latest verified" versions

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agoNEWS: drop 2.40 section "Changes to build and runtime requirements"
Andreas K. Hüttel [Sat, 20 Jul 2024 21:42:05 +0000 (23:42 +0200)]
NEWS: drop 2.40 section "Changes to build and runtime requirements"

Can't find anything that should go here.

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
3 weeks agocontrib.texi: update
Andreas K. Hüttel [Sat, 20 Jul 2024 17:47:52 +0000 (19:47 +0200)]
contrib.texi: update

Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
This page took 0.088979 seconds and 5 git commands to generate.