Luna Lamb [Fri, 3 Jan 2025 20:15:17 +0000 (20:15 +0000)]
AArch64: Improve codegen in SVE expm1f and users
Use unpredicated muls, use absolute compare and improve memory access.
Expm1f, sinhf and tanhf show 7%, 5% and 1% improvement in throughput
microbenchmark on Neoverse V1.
Joe Ramsay [Fri, 3 Jan 2025 19:13:36 +0000 (19:13 +0000)]
math: Remove no-mathvec flag
More routines are to follow, some of which hit many failures in the
current testsuite due to wrong sign of zero (mathvec routines are not
required to get this right). Instead of disabling a large number of
tests, change the failure condition such that, for vector routines,
tests pass as long as computed == expected == 0.0, regardless of sign.
Affected tests (vector tests for expm1, log1p, sin, tan and tanh) all
still pass.
Yat Long Poon [Fri, 3 Jan 2025 19:09:05 +0000 (19:09 +0000)]
AArch64: Improve codegen for SVE log1pf users
Reduce memory access by using lanewise MLA and reduce number of MOVPRFXs.
Move log1pf implementation to inline helper function.
Speedup on Neoverse V1 for log1pf (10%), acoshf (-1%), atanhf (2%), asinhf (2%).
Yat Long Poon [Fri, 3 Jan 2025 19:07:30 +0000 (19:07 +0000)]
AArch64: Improve codegen for SVE logs
Reduce memory access by using lanewise MLA and moving constants to struct
and reduce number of MOVPRFXs.
Update maximum ULP error for double log_sve from 1 to 2.
Speedup on Neoverse V1 for log (3%), log2 (5%), and log10 (4%).
H.J. Lu [Fri, 3 Jan 2025 02:21:56 +0000 (10:21 +0800)]
Rename have-mtls-descriptor to have-test-mtls-descriptor
Since have-mtls-descriptor is only used for glibc testing, rename it to
have-test-mtls-descriptor. Also enable tst-gnu2-tls2-amx only if
$(have-test-mtls-descriptor) == gnu2.
Tested with GCC 14 and Clang 19/18/17 on x86-64.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
Linux timberdoodle 6.1.60-gentoo-dist-hardened #1 SMP Fri Dec 1 22:10:49 UTC 2023 ppc64 POWER9 (architected), altivec supported CHRP IBM pSeries (emulated by qemu) GNU/Linux
Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
to fix the namespace bug since __mempcpy and __stpcpy were defined as
macros in <bits/string2.h>. These macros call __builtin_mempcpy and
__builtin_stpcpy which may end up calling the C functions mempcpy
and stpcpy. In libc.so, libc_hidden_builtin_proto ensures that calls
to mempcpy and stpcpy are in turn mapped to call __GI_mempcpy and
__GI_stpcpy. The redirections were applied outside of libc.so, including
libc.a, to map mempcpy and stpcpy to __mempcpy and __stpcpy. Since
There is no longer a need for string2.h, so remove it and all mention of it.
Move the redirect for __stpcpy to include/string.h since it is
still required
until all internal uses have been renamed.
This fixes several linknamespace/localplt failures when building with -Os.
removed the __mempcpy and __stpcpy macros from the public header file,
limit these redirections to libc.a to avoid Clang error:
In file included from tst-iconv-sticky-input-error.c:22:
In file included from ./gconv_int.h:24:
../include/string.h:182:44: error: attribute declaration must precede definition [-Werror,-Wignored-attributes]
182 | extern __typeof (mempcpy) mempcpy __asm__ ("__mempcpy");
| ^
../string/bits/string_fortified.h:42:8: note: previous definition is here
42 | __NTH (mempcpy (void *__restrict __dest, const void *__restrict __src,
| ^
when testing with Clang for fortify build.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
H.J. Lu [Wed, 1 Jan 2025 23:22:36 +0000 (07:22 +0800)]
not-cancel.h: Support testing fortify build with Clang
When Clang is used to test fortify glibc build configured with
--enable-fortify-source=N
clang issues errors like
In file included from tst-rfc3484.c:60:
In file included from ./getaddrinfo.c:81:
../sysdeps/unix/sysv/linux/not-cancel.h:36:10: error: reference to overloaded function could not be resolved; did you mean to call it?
36 | __typeof (open64) __open64_nocancel;
| ^~~~~~~~
../include/bits/../../io/bits/fcntl2.h:127:1: note: possible target for call
127 | open64 (__fortify_clang_overload_arg (const char *, ,__path), int __oflag,
| ^
../include/bits/../../io/bits/fcntl2.h:118:1: note: possible target for call
118 | open64 (__fortify_clang_overload_arg (const char *, ,__path), int __oflag)
| ^
../include/bits/../../io/bits/fcntl2.h:114:1: note: possible target for call
114 | open64 (const char *__path, int __oflag, mode_t __mode, ...)
| ^
../io/fcntl.h:219:12: note: possible target for call
219 | extern int open64 (const char *__file, int __oflag, ...) __nonnull ((1));
| ^
because clang fortify support for functions with variable arguments relies
on function overload. Update not-cancel.h to avoid __typeof on functions
with variable arguments.
Co-Authored-By: Adhemerval Zanella <adhemerval.zanella@linaro.org> Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
H.J. Lu [Thu, 2 Jan 2025 11:03:47 +0000 (19:03 +0800)]
tst-unique[34].cc: Use explicit instantiation declaration/definition
Use explicit instantiation declaration and definition to silence Clang
error:
tst-unique3.cc:6:18: error: instantiation of variable 'S<char>::i' required here, but no definition is available [-Werror,-Wundefined-var-template]
6 | int t = S<char>::i;
| ^
./tst-unique3.h:5:14: note: forward declaration of template entity is here
5 | static int i;
| ^
tst-unique3.cc:6:18: note: add an explicit instantiation declaration to suppress this warning if 'S<char>::i' is explicitly instantiated in another translation unit
6 | int t = S<char>::i;
| ^
Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
Sam James [Tue, 10 Dec 2024 01:21:46 +0000 (01:21 +0000)]
stdlib: random_r: fix unaligned access in initstate and initstate_r [BZ #30584]
The initstate{,_r} interfaces are documented in BSD as needing an aligned
array of 32-bit values, but neither POSIX nor glibc's own documentation
require it to be aligned. glibc's documentation says it "should" be a power
of 2, but not must.
Use memcpy to read and write to `state` to handle such an unaligned
argument.
assert: Remove the use of %n from __assert_fail_base (BZ #32456)
The require size for mmap can be inferred from __vasprintf return
value. It also fixes tst-assert-2 when building with --enable-fortify,
where even if the format is not translated, __readonly_area fails
because malloc can not be used.
Florian Weimer [Thu, 2 Jan 2025 12:45:27 +0000 (13:45 +0100)]
elf: Use TLS_DTV_OFFSET in __tls_get_addr
This fixes commit 5e249192cac7354af02a7347a0d8c984e0c88ed3 ("elf:
Remove the GET_ADDR_ARGS and related macros from the TLS code"):
GET_ADDR_ARGS was indeed unused, but GET_ADDR_OFFSET was used
on several targets, those that define TLS_DTV_OFFSET. Instead
of reintroducing GET_ADDR_OFFSET, use TLS_DTV_OFFSET directly,
now that it is defined on all targets.
In the new tls_get_addr_adjust helper function, add a cast to
uintptr_t to help the s390 case, where the offset can be positive or
negative, depending on the addresses malloc returns. The cast avoids
pointer wraparound/overflow. The outer uintptr_t cast is needed
to suppress a warning on x86-64 x32 about mismatched integer/pointer
sizes.
Eventually this offset should be folded into the DTV addresses
themselves, to eliminate the subtraction on the TLS fast path.
This will require an adjustment to libthread_db because the
debugger interface currently returns unadjusted pointers.
Samuel Thibault [Wed, 1 Jan 2025 21:07:42 +0000 (22:07 +0100)]
bits/socket.h: Update to recent BSD definition
The old BSD 4.4 definition (not used by Linux) was not 64b-proof: the
cmsg_data field is supposed to CMSG_ALIGN'ed (as can be also seen in the
CMSG_LEN macro).
Paul Eggert [Wed, 1 Jan 2025 20:16:06 +0000 (12:16 -0800)]
Don't update copyright notices in Linux licenses
* scripts/update-copyrights: Do not update copyright notices
in licenses imported from the Linux kernel.
This should prevent glitches such as those fixed in my
recent commit.
Paul Eggert [Wed, 1 Jan 2025 18:31:31 +0000 (10:31 -0800)]
Update copyright dates not handled by scripts/update-copyrights
I've updated copyright dates in glibc for 2025. This is the patch for
the changes not generated by scripts/update-copyrights and subsequent
build / regeneration of generated files.
Paul Eggert [Wed, 1 Jan 2025 19:16:35 +0000 (11:16 -0800)]
Pass glibc pre-commit checks
This is needed for the next patch which updates copyright dates.
* assert/test-assert-2.c: Remove trailing white space.
* elf/tst-startup-errno.c: Remove trailing empty lines.
Xi Ruoyao [Thu, 26 Dec 2024 04:51:18 +0000 (12:51 +0800)]
mlock, mlock2, munlock: Tell the compiler we don't dereference the pointer
Since https://gcc.gnu.org/r11-959, the compiler emits
-Wmaybe-uninitialized if a const pointer to an uninitialized buffer is
passed. Tell the compiler we don't dereference the pointer to remove
the false alarm.
Link: https://gcc.gnu.org/PR118194 Signed-off-by: Xi Ruoyao <xry111@xry111.site> Reviewed-by: Sam James <sam@gentoo.org>
The new tunable can be used to control whether executable stacks are
allowed from either the main program or dependencies. The default is
to allow executable stacks.
The executable stacks default permission is checked agains the one
provided by the PT_GNU_STACK from program headers (if present). The
tunable also disables the stack permission change if any dependency
requires an executable stack at loading time.
Checked on x86_64-linux-gnu, i686-linux-gnu, and aarch64-linux-gnu.
elf: Do not change stack permission on dlopen/dlmopen
If some shared library loaded with dlopen/dlmopen requires an executable
stack, either implicitly because of a missing GNU_STACK ELF header
(where the ABI default flags implies in the executable bit) or explicitly
because of the executable bit from GNU_STACK; the loader will try to set
the both the main thread and all thread stacks (from the pthread cache)
as executable.
Besides the issue where any __nptl_change_stack_perm failure does not
undo the previous executable transition (meaning that if the library
fails to load, there can be thread stacks with executable stacks), this
behavior was used on a CVE [1] as a vector for RCE.
This patch changes that if a shared library requires an executable
stack, and the current stack is not executable, dlopen fails. The
change is done only for dynamically loaded modules, if the program
or any dependency requires an executable stack, the loader will still
change the main thread before program execution and any thread created
with default stack configuration.
Florian Weimer [Mon, 30 Dec 2024 11:41:51 +0000 (12:41 +0100)]
x86-64: Reorder dynamic linker list in ldd script (bug 32508)
Move the x86-64 loader first, before the i386 and x32 loaders. In
most cases, it's the loader the script needs. This avoids an error
message if the i386 loader does not work.
The effect of this change to the generated ldd script looks like this:
Michael Jeanson [Fri, 27 Dec 2024 16:41:02 +0000 (17:41 +0100)]
nptl: Add <thread_pointer.h> for hppa
This will be required by the rseq extensible ABI implementation on all
Linux architectures exposing the '__rseq_size' and '__rseq_offset'
symbols to set the initial value of the 'cpu_id' field which can be used
by applications to test if rseq is available and registered. As long as
the symbols are exposed it is valid for an application to perform this
test even if rseq is not yet implemented in libc for this architecture.
Compile tested with build-many-glibcs.py but I don't have access to any
hardware to run the tests.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
I don't think we need a compatibility symbol for this. As the
GStreamer example shows, this change is much more likely to fix bugs
than cause compatibility issues.
Suggested-by: Dmitry V. Levin <ldv@altlinux.org> Suggested-by: Archie Cobbs <archie.cobbs@gmail.com> Suggested-by: Solar Designer <solar@openwall.com> Reviewed-by: Sam James <sam@gentoo.org>
Florian Weimer [Tue, 17 Dec 2024 08:20:20 +0000 (09:20 +0100)]
nptl: More useful padding in struct pthread
The previous use of padding within a union made it impossible to
re-use the padding for GLIBC_PRIVATE ABI preservation because
tcbhead_t could use up all of the padding (as was historically the
case on x86-64). Allocating padding unconditionally addresses this
issue.
Florian Weimer [Mon, 23 Dec 2024 12:55:49 +0000 (13:55 +0100)]
elf: Remove the GET_ADDR_ARGS and related macros from the TLS code
This was used to manage an IA-64 ABI divergence is no longere needed
after the IA-64 removal.
(It should be possible to encode all the required information in
one machine word, so the pointer indirection is really unnecessary.
Technically, none of this is part of the ABI, so perhaps it's
possible to do this retroactively. See bug 27404.)
Miao Wang [Mon, 23 Dec 2024 09:30:43 +0000 (10:30 +0100)]
io: statx, fstatat: Drop nonnull attribute on the path argument
Since Linux 6.11 the kernel allows path to be NULL if flags &
AT_EMPTY_PATH. Let's allow users to take the advantage if they don't
care running on old kernels.
Signed-off-by: Miao Wang <shankerwangmiao@gmail.com> Signed-off-by: Xi Ruoyao <xry111@xry111.site> Reviewed-by: Florian Weimer <fweimer@redhat.com>
Add valid_decimal_value to check valid decimal value in a string to
avoid uninitialized endp in add_prefixlist and gaiconf_init as reported
by Clang 19:
./getaddrinfo.c:1884:11: error: variable 'endp' is used uninitialized whenever '||' condition is true [-Werror,-Wsometimes-uninitialized]
1884 | && (cp == NULL
| ^~~~~~~~~~
./getaddrinfo.c:1887:11: note: uninitialized use occurs here
1887 | && *endp == '\0'
| ^~~~
./getaddrinfo.c:1884:11: note: remove the '||' if its condition is always false
1884 | && (cp == NULL
| ^~~~~~~~~~
1885 | || (bits = strtoul (cp, &endp, 10)) != ULONG_MAX
| ~~
./getaddrinfo.c:1875:13: note: initialize the variable 'endp' to silence this warning
1875 | char *endp;
| ^
| = NULL
This fixes BZ #32465.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
H.J. Lu [Sun, 22 Dec 2024 21:12:03 +0000 (05:12 +0800)]
stdio-common: Suppress Clang warnings on scanf13.c
Suppress Clang warnings on stdio-common/scanf13.c:
1. Before this commit:
scanf13.c:43:17: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
43 | "A%mS%10mls%4ml[bcd]%4mCB", &lsp1, &lsp2, &lsp3, &lsp4) != 4)
| ~~~~^
scanf13.c:43:22: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
43 | "A%mS%10mls%4ml[bcd]%4mCB", &lsp1, &lsp2, &lsp3, &lsp4) != 4)
| ~~~~^
scanf13.c:43:50: error: data argument not used by format string [-Werror,-Wformat-extra-args]
43 | "A%mS%10mls%4ml[bcd]%4mCB", &lsp1, &lsp2, &lsp3, &lsp4) != 4)
| ~~~~~~~~~~~~~~~~~~~~~~~~~~ ^
scanf13.c:145:27: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
145 | if (sscanf (buf, "%2048mls%mlc", &lsp3, &lsp4) != 2)
| ~~~~~~^
scanf13.c:145:31: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
145 | if (sscanf (buf, "%2048mls%mlc", &lsp3, &lsp4) != 2)
| ~~~^
scanf13.c:145:43: error: data argument not used by format string [-Werror,-Wformat-extra-args]
145 | if (sscanf (buf, "%2048mls%mlc", &lsp3, &lsp4) != 2)
| ~~~~~~~~~~~~~~ ^
scanf13.c:161:31: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
161 | if (sscanf (buf, "%4mC%1500ml[dr/]%548ml[abc/d]%3mlc",
| ~~~~~~^
scanf13.c:161:42: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
161 | if (sscanf (buf, "%4mC%1500ml[dr/]%548ml[abc/d]%3mlc",
| ~~~~~~~~~~^
scanf13.c:161:53: error: invalid conversion specifier 'l' [-Werror,-Wformat-invalid-specifier]
161 | if (sscanf (buf, "%4mC%1500ml[dr/]%548ml[abc/d]%3mlc",
| ~~~~~~~~~~^
scanf13.c:162:15: error: data argument not used by format string [-Werror,-Wformat-extra-args]
161 | if (sscanf (buf, "%4mC%1500ml[dr/]%548ml[abc/d]%3mlc",
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
162 | &lsp1, &lsp2, &lsp3, &lsp4) != 4)
| ^
10 errors generated.
2. With DIAG_IGNORE_NEEDS_COMMENT_CLANG changes in stdio-common/scanf13.c:
scanf13.c:28:40: error: 'sscanf' may overflow; destination buffer in argument 4 has size 8, but the corresponding specifier may require size 11 [-Werror,-Wfortify-source]
28 | "A%ms%10ms%4m[bcd]%4mcB", &sp1, &sp2, &sp3, &sp4) != 4)
| ^
scanf13.c:94:34: error: 'sscanf' may overflow; destination buffer in argument 3 has size 8, but the corresponding specifier may require size 2049 [-Werror,-Wfortify-source]
94 | if (sscanf (buf, "%2048ms%mc", &sp3, &sp4) != 2)
| ^
scanf13.c:110:61: error: 'sscanf' may overflow; destination buffer in argument 4 has size 8, but the corresponding specifier may require size 1501 [-Werror,-Wfortify-source]
110 | if (sscanf (buf, "%4mc%1500m[dr/]%548m[abc/d]%3mc", &sp1, &sp2, &sp3, &sp4)
| ^
scanf13.c:110:67: error: 'sscanf' may overflow; destination buffer in argument 5 has size 8, but the corresponding specifier may require size 549 [-Werror,-Wfortify-source]
110 | if (sscanf (buf, "%4mc%1500m[dr/]%548m[abc/d]%3mc", &sp1, &sp2, &sp3, &sp4)
| ^
4 errors generated.
Co-Authored-By: Adhemerval Zanella <adhemerval.zanella@linaro.org> Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
H.J. Lu [Sun, 22 Dec 2024 19:58:54 +0000 (03:58 +0800)]
Fix build without stack protector [BZ #32494]
Without stack protector, inhibit_stack_protector is undefined during build:
In file included from <command-line>:
./../include/libc-symbols.h:665:3: error: expected ';' before '__typeof'
665 | __typeof (type_name) *name##_ifunc (__VA_ARGS__)
\
| ^~~~~~~~
./../include/libc-symbols.h:676:3: note: in expansion of macro
'__ifunc_resolver'
676 | __ifunc_resolver (type_name, name, expr, init, static, __VA_ARGS__)
| ^~~~~~~~~~~~~~~~
./../include/libc-symbols.h:703:3: note: in expansion of macro '__ifunc_args'
703 | __ifunc_args (type_name, name, expr, init, arg)
| ^~~~~~~~~~~~
./../include/libc-symbols.h:790:3: note: in expansion of macro '__ifunc'
790 | __ifunc (redirected_name, name, expr, void, INIT_ARCH)
| ^~~~~~~
../sysdeps/x86_64/multiarch/memchr.c:29:1: note: in expansion of macro
'libc_ifunc_redirected'
29 | libc_ifunc_redirected (__redirect_memchr, memchr, IFUNC_SELECTOR ());
| ^~~~~~~~~~~~~~~~~~~~~
1. Fix a typo in include/libc-symbols.h to define inhibit_stack_protector
for build.
2. Don't include <config.h> in include/libc-symbols.h since it has been
included in include/libc-misc.h.
3. Change #include "libc-misc.h" to #include <libc-misc.h> in
string/test-string.h.
This fixes BZ #32494.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
H.J. Lu [Thu, 19 Dec 2024 21:52:56 +0000 (05:52 +0800)]
Compile tst-deadline.c with -Wno-ignored-attributes for Clang
Since tst-deadline.c is an internal test, compile tst-deadline.c with
-Wno-ignored-attributes for Clang to silence -Werror,-Wunknown-attributes
errors. Also suppress -Wmaybe-uninitialized only for GCC in net-internal.h.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com> Reviewed-by: Sam James <sam@gentoo.org>
projects / thirdparty / glibc.git / log
