platform/upstream/glibc.git
2 years agox86_64: State assembler is being tested on sysdeps/x86/configure
Adhemerval Zanella [Tue, 6 Dec 2022 16:47:47 +0000 (13:47 -0300)]
x86_64: State assembler is being tested on sysdeps/x86/configure

2 years agoconfigure: Remove AS check
Adhemerval Zanella [Fri, 2 Dec 2022 19:00:28 +0000 (16:00 -0300)]
configure: Remove AS check

The assembler is not issued directly, but rather always through CC
wrapper.  The binutils version check if done with LD instead.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 years agoconfigure: Remove check if ld is GNU
Adhemerval Zanella [Fri, 2 Dec 2022 19:00:27 +0000 (16:00 -0300)]
configure: Remove check if ld is GNU

Assume linker has gnu argument input style.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 years agoconfigure: Remove check if as is GNU
Adhemerval Zanella [Fri, 2 Dec 2022 19:00:26 +0000 (16:00 -0300)]
configure: Remove check if as is GNU

It is not used in any place.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 years agoconfigure: Move locale tools early
Adhemerval Zanella [Fri, 2 Dec 2022 19:00:25 +0000 (16:00 -0300)]
configure: Move locale tools early

When using --with-binutils, the configure might not use the specified
linker or assembler while checking for expected support.  Move the
tools check early, before any compiler usage test.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
2 years agohurd: Make getrandom cache the server port
Sergey Bugaev [Fri, 2 Dec 2022 13:55:58 +0000 (16:55 +0300)]
hurd: Make getrandom cache the server port

Previously, getrandom would, each time it's called, traverse the file
system to find /dev/urandom, fetch some random data from it, then throw
away that port. This is quite slow, while calls to getrandom are
genrally expected to be fast.

Additionally, this means that getrandom can not work when /dev/urandom
is unavailable, such as inside a chroot that lacks one. User programs
expect calls to getrandom to work inside a chroot if they first call
getrandom outside of the chroot.

In particular, this is known to break the OpenSSH server, and in that
case the issue is exacerbated by the API of arc4random, which prevents
it from properly reporting errors, forcing glibc to abort on failure.
This causes sshd to just die once it tries to generate a random number.

Caching the random server port, in a manner similar to how socket
server ports are cached, both improves the performance and works around
the chroot issue.

Tested on i686-gnu with the following program:

pthread_barrier_t barrier;

void *worker(void*) {
    pthread_barrier_wait(&barrier);
    uint32_t sum = 0;
    for (int i = 0; i < 10000; i++) {
        sum += arc4random();
    }
    return (void *)(uintptr_t) sum;
}

int main() {
    pthread_t threads[THREAD_COUNT];

    pthread_barrier_init(&barrier, NULL, THREAD_COUNT);

    for (int i = 0; i < THREAD_COUNT; i++) {
        pthread_create(&threads[i], NULL, worker, NULL);
    }
    for (int i = 0; i < THREAD_COUNT; i++) {
        void *retval;
        pthread_join(threads[i], &retval);
        printf("Thread %i: %lu\n", i, (unsigned long)(uintptr_t) retval);
    }

In my totally unscientific benchmark, with this patch, this completes
in about 7 seconds, whereas previously it took about 50 seconds. This
program was also used to test that getrandom () doesn't explode if the
random server dies, but instead reopens the /dev/urandom anew. I have
also verified that with this patch, OpenSSH can once again accept
connections properly.

Signed-off-by: Sergey Bugaev <bugaevc@gmail.com>
Message-Id: <20221202135558.23781-1-bugaevc@gmail.com>

2 years agopowerpc64: Remove old strncmp optimization
Rajalakshmi Srinivasaraghavan [Fri, 2 Dec 2022 20:26:41 +0000 (14:26 -0600)]
powerpc64: Remove old strncmp optimization

This patch cleans up the power4 strncmp optimization for powerpc64 which
is unlikely to be used anywhere.

Tested on ppc64le with and without --disable-multi-arch flag.

Reviewed-by: Paul E. Murphy <murphyp@linux.ibm.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agox86-64 strncpy: Properly handle the length parameter [BZ# 29839]
H.J. Lu [Fri, 2 Dec 2022 00:36:02 +0000 (16:36 -0800)]
x86-64 strncpy: Properly handle the length parameter [BZ# 29839]

On x32, the size_t parameter may be passed in the lower 32 bits of a
64-bit register with the non-zero upper 32 bits.  The string/memory
functions written in assembly can only use the lower 32 bits of a
64-bit register as length or must clear the upper 32 bits before using
the full 64-bit register for length.

This pach fixes strncpy for x32.  Tested on x86-64 and x32.  On x86-64,
libc.so is the same with and without the fix.
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
2 years agox86-64 strncat: Properly handle the length parameter [BZ# 24097]
H.J. Lu [Fri, 2 Dec 2022 00:37:11 +0000 (16:37 -0800)]
x86-64 strncat: Properly handle the length parameter [BZ# 24097]

On x32, the size_t parameter may be passed in the lower 32 bits of a
64-bit register with the non-zero upper 32 bits.  The string/memory
functions written in assembly can only use the lower 32 bits of a
64-bit register as length or must clear the upper 32 bits before using
the full 64-bit register for length.

This pach fixes strncat for x32.  Tested on x86-64 and x32.  On x86-64,
libc.so is the same with and without the fix.
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
2 years agoARC: update definitions in elf/elf.h
Shahab Vahedi [Sun, 27 Nov 2022 17:38:25 +0000 (18:38 +0100)]
ARC: update definitions in elf/elf.h

While porting ARCv2 to elfutils [1], it was brought up that the
necessary changes to the project's libelf/elf.h must come from
glibc, because they sync it from glibc [2].  Therefore, this patch
is to update ARC entries in elf/elf.h.

The majority of the update is about adding new definitions,
specially for the relocations.  However, there is one rename, one
deletion, and one change:

- R_ARC_JUMP_SLOT renamed to R_ARC_JMP_SLOT to match binutils.
- R_ARC_B26 removed because it is unused and deprecated.
- R_ARC_TLS_DTPOFF_S9 changed from 0x4a to the correct value 0x49.

Finally, a specific SHT class for ARC has been added to glibcelf.py.
Else, it would result in a collision:

  _register_elf_h(Sht, ranges=True,
 File "/src/glibc/scripts/glibcelf.py", line x, in _register_elf_h
   raise ValueError('duplicate value {}: {}, {}'.format(
                    ValueError: duplicate value 1879048193:
                    SHT_ARC_ATTRIBUTES, SHT_X86_64_UNWIND

[1]
https://sourceware.org/pipermail/elfutils-devel/2022q4/005530.html

[2]
https://sourceware.org/pipermail/elfutils-devel/2022q4/005548.html

No regression has been observed after applying this patch.  Below
follows the result:

UNSUPPORTED: crypt/cert
UNSUPPORTED: elf/tst-audit22
FAIL: elf/tst-audit25a
FAIL: elf/tst-audit25b
FAIL: elf/tst-bz15311
FAIL: elf/tst-bz28937
FAIL: elf/tst-dlmopen4
UNSUPPORTED: elf/tst-dlopen-self-container
UNSUPPORTED: elf/tst-dlopen-tlsmodid-container
UNSUPPORTED: elf/tst-glibc-hwcaps-prepend-cache
UNSUPPORTED: elf/tst-ldconfig-bad-aux-cache
UNSUPPORTED: elf/tst-ldconfig-ld_so_conf-update
UNSUPPORTED: elf/tst-pldd
UNSUPPORTED: elf/tst-preload-pthread-libc
XPASS: elf/tst-protected1a
XPASS: elf/tst-protected1b
FAIL: elf/tst-tls-allocation-failure-static-patched
FAIL: elf/tst-tls1
FAIL: elf/tst-tls3
FAIL: elf/tst-tlsalign-extern
UNSUPPORTED: elf/tst-valgrind-smoke
UNSUPPORTED: grp/tst-initgroups1
UNSUPPORTED: grp/tst-initgroups2
UNSUPPORTED: io/tst-getcwd-smallbuff
UNSUPPORTED: locale/tst-localedef-path-norm
FAIL: localedata/sort-test
UNSUPPORTED: localedata/tst-localedef-hardlinks
FAIL: malloc/tst-malloc-thread-fail-malloc-check
FAIL: malloc/tst-malloc_info-malloc-check
UNSUPPORTED: math/test-fesetexcept-traps
UNSUPPORTED: math/test-fexcept-traps
UNSUPPORTED: math/test-nearbyint-except
UNSUPPORTED: math/test-nearbyint-except-2
UNSUPPORTED: misc/tst-adjtimex
UNSUPPORTED: misc/tst-clock_adjtime
FAIL: misc/tst-misalign-clone
FAIL: misc/tst-misalign-clone-internal
UNSUPPORTED: misc/tst-ntp_adjtime
UNSUPPORTED: misc/tst-pkey
UNSUPPORTED: misc/tst-rseq
UNSUPPORTED: misc/tst-rseq-disable
UNSUPPORTED: misc/tst-syslog
UNSUPPORTED: misc/tst-ttyname
FAIL: nptl/test-cond-printers
FAIL: nptl/test-condattr-printers
FAIL: nptl/test-mutex-printers
FAIL: nptl/test-mutexattr-printers
FAIL: nptl/test-rwlock-printers
FAIL: nptl/test-rwlockattr-printers
UNSUPPORTED: nptl/tst-pthread-gdb-attach
UNSUPPORTED: nptl/tst-pthread-gdb-attach-static
UNSUPPORTED: nptl/tst-pthread-getattr
UNSUPPORTED: nptl/tst-rseq-nptl
UNSUPPORTED: nss/tst-nss-compat1
UNSUPPORTED: nss/tst-nss-db-endgrent
UNSUPPORTED: nss/tst-nss-db-endpwent
UNSUPPORTED: nss/tst-nss-files-hosts-long
UNSUPPORTED: nss/tst-nss-gai-actions
UNSUPPORTED: nss/tst-nss-test3
UNSUPPORTED: nss/tst-reload1
UNSUPPORTED: nss/tst-reload2
UNSUPPORTED: posix/bug-ga2
UNSUPPORTED: posix/bug-ga2-mem
FAIL: posix/globtest
UNSUPPORTED: posix/tst-vfork3
UNSUPPORTED: posix/tst-vfork3-mem
UNSUPPORTED: resolv/mtrace-tst-leaks2
UNSUPPORTED: resolv/tst-leaks2
UNSUPPORTED: resolv/tst-resolv-ai_idn
UNSUPPORTED: resolv/tst-resolv-ai_idn-latin1
UNSUPPORTED: resolv/tst-resolv-res_init
UNSUPPORTED: resolv/tst-resolv-res_init-thread
UNSUPPORTED: rt/tst-bz28213
UNSUPPORTED: rt/tst-mqueue1
UNSUPPORTED: rt/tst-mqueue10
UNSUPPORTED: rt/tst-mqueue2
UNSUPPORTED: rt/tst-mqueue3
UNSUPPORTED: rt/tst-mqueue4
UNSUPPORTED: rt/tst-mqueue5
UNSUPPORTED: rt/tst-mqueue6
UNSUPPORTED: rt/tst-mqueue8
UNSUPPORTED: rt/tst-mqueue8x
UNSUPPORTED: rt/tst-mqueue9
UNSUPPORTED: stdlib/test-bz22786
UNSUPPORTED: stdlib/tst-system
UNSUPPORTED: string/test-bcopy
UNSUPPORTED: string/test-memmove
UNSUPPORTED: string/tst-memmove-overflow
UNSUPPORTED: string/tst-strerror
UNSUPPORTED: string/tst-strsignal
UNSUPPORTED: time/tst-clock_settime
UNSUPPORTED: time/tst-settimeofday
Summary of test results:
     21 FAIL
   4184 PASS
     69 UNSUPPORTED
     16 XFAIL
      2 XPASS

Signed-off-by: Shahab Vahedi <shahab@synopsys.com>
Signed-off-by: Vineet Gupta <vineet.gupta@linux.dev>
2 years agoscripts: Add "|" operator support to glibcpp's parsing
Shahab Vahedi [Sun, 27 Nov 2022 17:38:24 +0000 (18:38 +0100)]
scripts: Add "|" operator support to glibcpp's parsing

From the tests point of view, this is a necessary step for another
patch [1] and allows parsing macros such as "#define A | B".  Without
it, a few tests [2] choke when the other patch [1] is applied:

/src/glibc/scripts/../elf/elf.h:4167: error: uninterpretable macro
token sequence: ( EF_ARC_MACH_MSK | EF_ARC_OSABI_MSK )
Traceback (most recent call last):
    File "/src/glibc/elf/tst-glibcelf.py", line 23, in <module>
      import glibcelf
    File "/src/glibc/scripts/glibcelf.py", line 226, in <module>
      _elf_h = _parse_elf_h()
               ^^^^^^^^^^^^^^
    File "/src/glibc/scripts/glibcelf.py", line 223, in _parse_elf_h
      raise IOError('parse error in elf.h')
  OSError: parse error in elf.h

[1] ARC: update definitions in elf/elf.h
https://sourceware.org/pipermail/libc-alpha/2022-November/143503.html

[2]
tst-glibcelf, tst-relro-ldso, and tst-relro-libc

Reviewed-by: Florian Weimer <fweimer@redhat.com>
Signed-off-by: Shahab Vahedi <shahab@synopsys.com>
2 years agoApply asm redirections in syslog.h before first use [BZ #27087]
Tulio Magno Quites Machado Filho [Fri, 11 Nov 2022 20:00:15 +0000 (17:00 -0300)]
Apply asm redirections in syslog.h before first use [BZ #27087]

Similar to d0fa09a770, but for syslog.h when _FORTIFY_SOURCE > 0.
Fixes [BZ #27087] by applying long double-related asm redirections
before using functions in bits/syslog.h.

Tested with build-many-glibcs.py.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoLoongArch: Add support for ilogb[f]
Xiaolin Tang [Wed, 23 Nov 2022 03:49:26 +0000 (11:49 +0800)]
LoongArch: Add support for ilogb[f]

Add inline assembler for the ilogb functions. Passes GLIBC regression.

2 years agoLoongArch: Add support for scalb[f]
Xiaolin Tang [Wed, 23 Nov 2022 03:49:25 +0000 (11:49 +0800)]
LoongArch: Add support for scalb[f]

Add inline assembler for the scalb functions. Passes GLIBC regression.

2 years agoLoongArch: Add support for scalbn[f]
Xiaolin Tang [Wed, 23 Nov 2022 03:49:24 +0000 (11:49 +0800)]
LoongArch: Add support for scalbn[f]

Add inline assembler for the scalbn functions. Passes GLIBC regression.

GCC 13, LoongArch support ___builtin_scalbn{,f} with -fno-math-errno,
but only "libm" can use -fno-math-errno in GLIBC, and scalbn is in libc
instead of libm because __printf_fp calls it.

2 years agoLoongArch: Use __builtin_logb{,f} with GCC >= 13
Xiaolin Tang [Wed, 23 Nov 2022 03:45:01 +0000 (11:45 +0800)]
LoongArch: Use __builtin_logb{,f} with GCC >= 13

GCC 13 compiles these built-ins instead of generic
implementation for function logb.

Link: https://gcc.gnu.org/r13-3922
Co-Authored-By: Xi Ruoyao <xry111@xry111.site>
2 years agoUse GCC builtins for logb functions if desired.
Xiaolin Tang [Wed, 23 Nov 2022 03:45:00 +0000 (11:45 +0800)]
Use GCC builtins for logb functions if desired.

This patch is using the corresponding GCC builtin for logbf, logb,
logbl and logbf128 if the USE_FUNCTION_BUILTIN macros are defined to one
in math-use-builtins-function.h.

Co-Authored-By: Xi Ruoyao <xry111@xry111.site>
2 years agoLoongArch: Use __builtin_llrint{,f} with GCC >= 13
Xiaolin Tang [Wed, 23 Nov 2022 03:44:59 +0000 (11:44 +0800)]
LoongArch: Use __builtin_llrint{,f} with GCC >= 13

GCC 13 compiles these built-ins instead of generic
implementation for function llrint.

Link: https://gcc.gnu.org/r13-3920
Co-Authored-By: Xi Ruoyao <xry111@xry111.site>
2 years agoUse GCC builtins for llrint functions if desired.
Xiaolin Tang [Wed, 23 Nov 2022 03:44:58 +0000 (11:44 +0800)]
Use GCC builtins for llrint functions if desired.

This patch is using the corresponding GCC builtin for llrintf, llrint,
llrintl and llrintf128 if the USE_FUNCTION_BUILTIN macros are defined to one
in math-use-builtins-function.h.

Co-Authored-By: Xi Ruoyao <xry111@xry111.site>
2 years agoLoongArch: Use __builtin_lrint{,f} with GCC >= 13
Xiaolin Tang [Wed, 23 Nov 2022 03:44:57 +0000 (11:44 +0800)]
LoongArch: Use __builtin_lrint{,f} with GCC >= 13

GCC 13 compiles these built-ins instead of generic
implementation for function lrint.

Link: https://gcc.gnu.org/r13-3920
Co-Authored-By: Xi Ruoyao <xry111@xry111.site>
2 years agoUse GCC builtins for lrint functions if desired.
Xiaolin Tang [Wed, 23 Nov 2022 03:44:56 +0000 (11:44 +0800)]
Use GCC builtins for lrint functions if desired.

This patch is using the corresponding GCC builtin for lrintf, lrint,
lrintl and lrintf128 if the USE_FUNCTION_BUILTIN macros are defined to one
in math-use-builtins-function.h.

Co-Authored-By: Xi Ruoyao <xry111@xry111.site>
2 years agoLoongArch: Use __builtin_rint{,f} with GCC >= 13
Xi Ruoyao [Wed, 23 Nov 2022 03:44:55 +0000 (11:44 +0800)]
LoongArch: Use __builtin_rint{,f} with GCC >= 13

GCC 13 compiles these built-ins to frint.{d,s} instruction.

Link: https://gcc.gnu.org/r13-3919
2 years agox86/fpu: Factor out shared avx2/avx512 code in svml_{s|d}_wrapper_impl.h
Noah Goldstein [Sat, 19 Nov 2022 00:13:32 +0000 (16:13 -0800)]
x86/fpu: Factor out shared avx2/avx512 code in svml_{s|d}_wrapper_impl.h

Code is exactly the same for the two so better to only maintain one
version.

All math and mathvec tests pass on x86.

2 years agox86/fpu: Cleanup code in svml_{s|d}_wrapper_impl.h
Noah Goldstein [Sat, 19 Nov 2022 00:13:31 +0000 (16:13 -0800)]
x86/fpu: Cleanup code in svml_{s|d}_wrapper_impl.h

1. Remove unnecessary spills.
2. Fix some small nit missed optimizations.

All math and mathvec tests pass on x86.

2 years agox86/fpu: Reformat svml_{s|d}_wrapper_impl.h
Noah Goldstein [Sat, 19 Nov 2022 00:13:30 +0000 (16:13 -0800)]
x86/fpu: Reformat svml_{s|d}_wrapper_impl.h

Just reformat with the style convention used in other x86 assembler
files.  This doesn't change libm.so or libmvec.so.

2 years agox86/fpu: Fix misspelled evex512 section in variety of svml files
Noah Goldstein [Fri, 18 Nov 2022 20:00:36 +0000 (12:00 -0800)]
x86/fpu: Fix misspelled evex512 section in variety of svml files

```
.section .text.evex512, "ax", @progbits
```

With misspelled as:

```
.section .text.exex512, "ax", @progbits
```

2 years agox86/fpu: Add missing ISA sections to variety of svml files
Noah Goldstein [Fri, 18 Nov 2022 21:06:11 +0000 (13:06 -0800)]
x86/fpu: Add missing ISA sections to variety of svml files

Many sse4/avx2/avx512 files where just in .text.

2 years agostdio-common: Add missing dependencies (bug 29780)
Andreas Schwab [Mon, 21 Nov 2022 12:08:17 +0000 (13:08 +0100)]
stdio-common: Add missing dependencies (bug 29780)

Handle all object suffixes for dependencies of errlist-data and siglist
objects.

2 years agoi386: Avoid rely on linker optimization to avoid relocation
Adhemerval Zanella Netto [Thu, 17 Nov 2022 18:13:08 +0000 (15:13 -0300)]
i386: Avoid rely on linker optimization to avoid relocation

lld does not implement all the linker optimization to avoid the GOT
relocation as done by binutils (bfd/elf32-i386.c:elf_i386_convert_load_reloc).
The current 'movl main@GOT(%ebx), %eax' will then create a GOT
relocation when building with lld, which make static-pie status to
not being able to start the provided main function.

The change uses a __wrap_main local symbol, which in turn calls main
(similar as used by aarch64 and s390x).

Checked on i686-linux-gnu with binutils and lld.
Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoelf: Fix rtld-audit trampoline for aarch64
Vladislav Khmelevsky [Thu, 17 Nov 2022 08:47:29 +0000 (12:47 +0400)]
elf: Fix rtld-audit trampoline for aarch64

This patch fixes two problems with audit:

  1. The DL_OFFSET_RV_VPCS offset was mixed up with DL_OFFSET_RG_VPCS,
     resulting in x2 register value nulling in RG structure.

  2. We need to preserve the x8 register before function call, but
     don't have to save it's new value and restore it before return.

Anyway the final restore was using OFFSET_RV instead of OFFSET_RG value
which is wrong (althoug doesn't affect anything).

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoDefine in_int32_t_range to check if the 64 bit time_t syscall should be used
YunQiang Su [Tue, 8 Nov 2022 04:49:46 +0000 (12:49 +0800)]
Define in_int32_t_range to check if the 64 bit time_t syscall should be used

Currently glibc uses in_time_t_range to detects time_t overflow,
and if it occurs fallbacks to 64 bit syscall version.

The function name is confusing because internally time_t might be
either 32 bits or 64 bits (depending on __TIMESIZE).

This patch refactors the in_time_t_range by replacing it with
in_int32_t_range for the case to check if the 64 bit time_t syscall
should be used.

The in_time_t range is used to detect overflow of the
syscall return value.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoelf/tst-tlsopt-powerpc fails when compiled with -mcpu=power10 (BZ# 29776)
Alan Modra [Sat, 12 Nov 2022 00:50:31 +0000 (11:20 +1030)]
elf/tst-tlsopt-powerpc fails when compiled with -mcpu=power10 (BZ# 29776)

Supports pcrel addressing of TLS GOT entry.  Also tweak the non-pcrel
asm constraint to better reflect how the reg is used.

2 years agoLoongArch: Hard Float Support for fmaximum_mag_num{f/ }, fminimum_mag_num{f/ }.
Xiaolin Tang [Mon, 7 Nov 2022 14:08:21 +0000 (22:08 +0800)]
LoongArch: Hard Float Support for fmaximum_mag_num{f/ }, fminimum_mag_num{f/ }.

Use hardware Floating-point instruction f{maxa/mina}.{s/d}, fclass.{s/d}
to implement fmaximum_mag_num{f/ }, fminimum_mag_num{f/ }.

       *  sysdeps/loongarch/fpu/s_fmaximum_mag_num.c: New file.
       *  sysdeps/loongarch/fpu/s_fmaximum_mag_numf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum_mag_num.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum_mag_numf.c: Likewise.

2 years agoLoongArch: Hard Float Support for fmaximum_mag{f/ }, fminimum_mag{f/ }.
Xiaolin Tang [Mon, 7 Nov 2022 14:08:20 +0000 (22:08 +0800)]
LoongArch: Hard Float Support for fmaximum_mag{f/ }, fminimum_mag{f/ }.

Use hardware Floating-point instruction f{maxa/mina}.{s/d}, fclass.{s/d}
to implement fmaximum_mag{f/ }, fminimum_mag{f/ }.

       *  sysdeps/loongarch/fpu/s_fmaximum_mag.c: New file.
       *  sysdeps/loongarch/fpu/s_fmaximum_magf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum_mag.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum_magf.c: Likewise.

2 years agoLoongArch: Hard Float Support for fmaxmag{f/ }, fminmag{f/ }.
Xiaolin Tang [Mon, 7 Nov 2022 14:08:19 +0000 (22:08 +0800)]
LoongArch: Hard Float Support for fmaxmag{f/ }, fminmag{f/ }.

Use hardware Floating-point instruction f{maxa/mina}.{s/d},
to implement fmaxmag{f/ }, fminmag{f/ }.

       *  sysdeps/loongarch/fpu/s_fmaxmag.c: New file.
       *  sysdeps/loongarch/fpu/s_fmaxmagf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminmag.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminmagf.c: Likewise.

2 years agoLoongArch: Hard Float Support for fmaximum_num{f/ }, fminimum_num{f/ }.
Xiaolin Tang [Mon, 7 Nov 2022 14:07:48 +0000 (22:07 +0800)]
LoongArch: Hard Float Support for fmaximum_num{f/ }, fminimum_num{f/ }.

Use hardware Floating-point instruction f{max/min}.{s/d}, fclass.{s/d}
to implement fmaximum_num{f/ }, fminimum_num{f/ }.

       *  sysdeps/loongarch/fpu/s_fmaximum_num.c: New file.
       *  sysdeps/loongarch/fpu/s_fmaximum_numf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum_num.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum_numf.c: Likewise.

2 years agoLoongArch: Hard Float Support for fmaximum{f/ }, fminimum{f/ }.
Xiaolin Tang [Mon, 7 Nov 2022 14:05:19 +0000 (22:05 +0800)]
LoongArch: Hard Float Support for fmaximum{f/ }, fminimum{f/ }.

Use hardware Floating-point instruction f{max/min}.{s/d}, fclass.{s/d}
to implement fmaximum{f/ }, fminimum{f/ }.

       *  sysdeps/loongarch/fpu/s_fmaximum.c: New file.
       *  sysdeps/loongarch/fpu/s_fmaximumf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimum.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fminimumf.c: Likewise.

2 years agoLoongArch: Hard Float Support for float-point classification functions.
Xiaolin Tang [Mon, 7 Nov 2022 14:05:16 +0000 (22:05 +0800)]
LoongArch: Hard Float Support for float-point classification functions.

Use hardware Floating-point instruction fclass.{s/d} to implement
classification functions, i.e finite{f/ }, fpclassify{f/ }, isnan{f/ },
isinf{f/ }, issignaling{f/ }.

       *  sysdeps/loongarch/fpu/s_finite.c: New file.
       *  sysdeps/loongarch/fpu/s_finitef.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fpclassify.c: Likewise.
       *  sysdeps/loongarch/fpu/s_fpclassifyf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_isinf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_isinff.c: Likewise.
       *  sysdeps/loongarch/fpu/s_isnan.c: Likewise.
       *  sysdeps/loongarch/fpu/s_isnanf.c: Likewise.
       *  sysdeps/loongarch/fpu/s_issignaling.c: Likewise.
       *  sysdeps/loongarch/fpu/s_issignalingf.c: Likewise.
       *  sysdeps/loongarch/fpu_control.h: Add _FCLASS_* macro.

2 years agoLoongArch: Use __builtin_{fma, fmaf} to implement function {fma, fmaf}.
Xiaolin Tang [Mon, 7 Nov 2022 14:05:15 +0000 (22:05 +0800)]
LoongArch: Use __builtin_{fma, fmaf} to implement function {fma, fmaf}.

Use __builtin_{fma, fmaf} to implement function {fma, fmaf} instead of
the generic implementation.

       *  sysdeps/loongarch/fpu/math-use-builtins-fma.h: New file.

2 years agoLinux: Support __IPC_64 in sysvctl *ctl command arguments (bug 29771)
Florian Weimer [Tue, 8 Nov 2022 13:15:02 +0000 (14:15 +0100)]
Linux: Support __IPC_64 in sysvctl *ctl command arguments (bug 29771)

Old applications pass __IPC_64 as part of the command argument because
old glibc did not check for unknown commands, and passed through the
arguments directly to the kernel, without adding __IPC_64.
Applications need to continue doing that for old glibc compatibility,
so this commit enables this approach in current glibc.

For msgctl and shmctl, if no translation is required, make
direct system calls, as we did before the time64 changes.  If
translation is required, mask __IPC_64 from the command argument.

For semctl, the union-in-vararg argument handling means that
translation is needed on all architectures.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoriscv: Get level 3 cache's information
Zong Li [Wed, 9 Nov 2022 14:40:59 +0000 (11:40 -0300)]
riscv: Get level 3 cache's information

RISC-V architecture extends the cache information for level 3 cache
in AUX vector in Linux v.6.1-rc1. This patch supports sysconf to get
the level 3 cache information.

Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
2 years agodebug: Fix typo in tests-unsupported rule
Adhemerval Zanella [Wed, 9 Nov 2022 14:24:45 +0000 (11:24 -0300)]
debug: Fix typo in tests-unsupported rule

From commit 2e274cd8c1ebd0bd0c43a7f2e5433685740938ca.

2 years agoiconvdata/tst-table-charmap.sh: remove handling of old, borrowed format
наб [Wed, 9 Nov 2022 12:48:46 +0000 (13:48 +0100)]
iconvdata/tst-table-charmap.sh: remove handling of old, borrowed format

This "Old POSIX/DKUUG borrowed format" handling is original to the file
and doesn't seem to have ever been used, i.e. id/t-t-c doesn't seem to
have ever been called with argv[1] == POSIX.

Upcoming is a POSIX charmap, which would inadvertently trigger this.

Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoMakerules: Generate shlib.lds with -fuse-ld=bfd
Fangrui Song [Wed, 9 Nov 2022 04:56:59 +0000 (20:56 -0800)]
Makerules: Generate shlib.lds with -fuse-ld=bfd

lld does not dump a linker script with --verbose (it does not use a
linker script driven design and lots of linker processing is not
serializable as a linker script anyway).  With the default
--with-default-link=no build, $@T is empty and makes `test -s $@T` fail.

Just dump the linker script with -fuse-ld=bfd.  lld since 15
(https://reviews.llvm.org/D124656) supports custom RELRO sections in the
GNU ld dumped linker script.

Reviewed-by: Sam James <sam@gentoo.org>
2 years agox86: Add avx2 optimized functions for the wchar_t strcpy family
Noah Goldstein [Wed, 9 Nov 2022 01:38:41 +0000 (17:38 -0800)]
x86: Add avx2 optimized functions for the wchar_t strcpy family

Implemented:
    wcscat-avx2  (+ 744 bytes
    wcscpy-avx2  (+ 539 bytes)
    wcpcpy-avx2  (+ 577 bytes)
    wcsncpy-avx2 (+1108 bytes)
    wcpncpy-avx2 (+1214 bytes)
    wcsncat-avx2 (+1085 bytes)

Performance Changes:
    Times are from N = 10 runs of the benchmark suite and are reported
    as geometric mean of all ratios of New Implementation / Best Old
    Implementation. Best Old Implementation was determined with the
    highest ISA implementation.

    wcscat-avx2     -> 0.975
    wcscpy-avx2     -> 0.591
    wcpcpy-avx2     -> 0.698
    wcsncpy-avx2    -> 0.730
    wcpncpy-avx2    -> 0.711
    wcsncat-avx2    -> 0.954

Code Size Changes:
    This change  increase the size of libc.so by ~5.5kb bytes. For
    reference the patch optimizing the normal strcpy family functions
    decreases libc.so by ~5.2kb.

Full check passes on x86-64 and build succeeds for all ISA levels w/
and w/o multiarch.

2 years agox86: Add evex optimized functions for the wchar_t strcpy family
Noah Goldstein [Wed, 9 Nov 2022 01:38:40 +0000 (17:38 -0800)]
x86: Add evex optimized functions for the wchar_t strcpy family

Implemented:
    wcscat-evex  (+ 905 bytes)
    wcscpy-evex  (+ 674 bytes)
    wcpcpy-evex  (+ 709 bytes)
    wcsncpy-evex (+1358 bytes)
    wcpncpy-evex (+1467 bytes)
    wcsncat-evex (+1213 bytes)

Performance Changes:
    Times are from N = 10 runs of the benchmark suite and are reported
    as geometric mean of all ratios of New Implementation / Best Old
    Implementation. Best Old Implementation was determined with the
    highest ISA implementation.

    wcscat-evex     -> 0.991
    wcscpy-evex     -> 0.587
    wcpcpy-evex     -> 0.695
    wcsncpy-evex    -> 0.719
    wcpncpy-evex    -> 0.694
    wcsncat-evex    -> 0.979

Code Size Changes:
    This change  increase the size of libc.so by ~6.3kb bytes. For
    reference the patch optimizing the normal strcpy family functions
    decreases libc.so by ~5.7kb.

Full check passes on x86-64 and build succeeds for all ISA levels w/
and w/o multiarch.

2 years agox86: Optimize and shrink st{r|p}{n}{cat|cpy}-avx2 functions
Noah Goldstein [Wed, 9 Nov 2022 01:38:39 +0000 (17:38 -0800)]
x86: Optimize and shrink st{r|p}{n}{cat|cpy}-avx2 functions

Optimizations are:
    1. Use more overlapping stores to avoid branches.
    2. Reduce how unrolled the aligning copies are (this is more of a
       code-size save, its a negative for some sizes in terms of
       perf).
    3. For st{r|p}n{cat|cpy} re-order the branches to minimize the
       number that are taken.

Performance Changes:

    Times are from N = 10 runs of the benchmark suite and are
    reported as geometric mean of all ratios of
    New Implementation / Old Implementation.

    strcat-avx2      -> 0.998
    strcpy-avx2      -> 0.937
    stpcpy-avx2      -> 0.971

    strncpy-avx2     -> 0.793
    stpncpy-avx2     -> 0.775

    strncat-avx2     -> 0.962

Code Size Changes:
    function         -> Bytes New / Bytes Old -> Ratio

    strcat-avx2      ->  685 / 1639 -> 0.418
    strcpy-avx2      ->  560 /  903 -> 0.620
    stpcpy-avx2      ->  592 /  939 -> 0.630

    strncpy-avx2     -> 1176 / 2390 -> 0.492
    stpncpy-avx2     -> 1268 / 2438 -> 0.520

    strncat-avx2     -> 1042 / 2563 -> 0.407

Notes:
    1. Because of the significant difference between the
       implementations they are split into three files.

           strcpy-avx2.S    -> strcpy, stpcpy, strcat
           strncpy-avx2.S   -> strncpy
           strncat-avx2.S    > strncat

       I couldn't find a way to merge them without making the
       ifdefs incredibly difficult to follow.

Full check passes on x86-64 and build succeeds for all ISA levels w/
and w/o multiarch.

2 years agox86: Optimize and shrink st{r|p}{n}{cat|cpy}-evex functions
Noah Goldstein [Wed, 9 Nov 2022 01:38:38 +0000 (17:38 -0800)]
x86: Optimize and shrink st{r|p}{n}{cat|cpy}-evex functions

Optimizations are:
    1. Use more overlapping stores to avoid branches.
    2. Reduce how unrolled the aligning copies are (this is more of a
       code-size save, its a negative for some sizes in terms of
       perf).
    3. Improve the loop a bit (similiar to what we do in strlen with
       2x vpminu + kortest instead of 3x vpminu + kmov + test).
    4. For st{r|p}n{cat|cpy} re-order the branches to minimize the
       number that are taken.

Performance Changes:

    Times are from N = 10 runs of the benchmark suite and are
    reported as geometric mean of all ratios of
    New Implementation / Old Implementation.

    stpcpy-evex      -> 0.922
    strcat-evex      -> 0.985
    strcpy-evex      -> 0.880

    strncpy-evex     -> 0.831
    stpncpy-evex     -> 0.780

    strncat-evex     -> 0.958

Code Size Changes:
    function         -> Bytes New / Bytes Old -> Ratio

    strcat-evex      ->  819 / 1874 -> 0.437
    strcpy-evex      ->  700 / 1074 -> 0.652
    stpcpy-evex      ->  735 / 1094 -> 0.672

    strncpy-evex     -> 1397 / 2611 -> 0.535
    stpncpy-evex     -> 1489 / 2691 -> 0.553

    strncat-evex     -> 1184 / 2832 -> 0.418

Notes:
    1. Because of the significant difference between the
       implementations they are split into three files.

           strcpy-evex.S    -> strcpy, stpcpy, strcat
           strncpy-evex.S   -> strncpy
           strncat-evex.S    > strncat

       I couldn't find a way to merge them without making the
       ifdefs incredibly difficult to follow.

    2. All implementations can be made evex512 by including
       "x86-evex512-vecs.h" at the top.

    3. All implementations have an optional define:
        `USE_EVEX_MASKED_STORE`
       Setting to one uses evex-masked stores for handling short
       strings.  This saves code size and branches.  It's disabled
       for all implementations are the moment as there are some
       serious drawbacks to masked stores in certain cases, but
       that may be fixed on future architectures.

Full check passes on x86-64 and build succeeds for all ISA levels w/
and w/o multiarch.

2 years agobenchtests: Make str{n}{cat|cpy} benchmarks output json
Noah Goldstein [Fri, 4 Nov 2022 08:20:21 +0000 (01:20 -0700)]
benchtests: Make str{n}{cat|cpy} benchmarks output json

Json output is easier to parse and most other benchmarks already do
the same.

2 years agox86: Use VMM API in memcmpeq-evex.S and minor changes
Noah Goldstein [Sat, 29 Oct 2022 20:19:59 +0000 (15:19 -0500)]
x86: Use VMM API in memcmpeq-evex.S and minor changes

Changes to generated code are:
    1. In a few places use `vpcmpeqb` instead of `vpcmpneq` to save a
       byte of code size.
    2. Add a branch for length <= (VEC_SIZE * 6) as opposed to doing
       the entire block of [VEC_SIZE * 4 + 1, VEC_SIZE * 8] in a
       single basic-block (the space to add the extra branch without
       changing code size is bought with the above change).

Change (2) has roughly a 20-25% speedup for sizes in [VEC_SIZE * 4 +
1, VEC_SIZE * 6] and negligible to no-cost for [VEC_SIZE * 6 + 1,
VEC_SIZE * 8]

From N=10 runs on Tigerlake:

align1,align2 ,length ,result               ,New Time ,Cur Time ,New Time / Old Time
0     ,0      ,129    ,0                    ,5.404    ,6.887    ,0.785
0     ,0      ,129    ,1                    ,5.308    ,6.826    ,0.778
0     ,0      ,129    ,18446744073709551615 ,5.359    ,6.823    ,0.785
0     ,0      ,161    ,0                    ,5.284    ,6.827    ,0.774
0     ,0      ,161    ,1                    ,5.317    ,6.745    ,0.788
0     ,0      ,161    ,18446744073709551615 ,5.406    ,6.778    ,0.798

0     ,0      ,193    ,0                    ,6.804    ,6.802    ,1.000
0     ,0      ,193    ,1                    ,6.950    ,6.754    ,1.029
0     ,0      ,193    ,18446744073709551615 ,6.792    ,6.719    ,1.011
0     ,0      ,225    ,0                    ,6.625    ,6.699    ,0.989
0     ,0      ,225    ,1                    ,6.776    ,6.735    ,1.003
0     ,0      ,225    ,18446744073709551615 ,6.758    ,6.738    ,0.992
0     ,0      ,256    ,0                    ,5.402    ,5.462    ,0.989
0     ,0      ,256    ,1                    ,5.364    ,5.483    ,0.978
0     ,0      ,256    ,18446744073709551615 ,5.341    ,5.539    ,0.964

Rewriting with VMM API allows for memcmpeq-evex to be used with
evex512 by including "x86-evex512-vecs.h" at the top.

Complete check passes on x86-64.

2 years agox86: Use VMM API in memcmp-evex-movbe.S and minor changes
Noah Goldstein [Sat, 29 Oct 2022 20:19:58 +0000 (15:19 -0500)]
x86: Use VMM API in memcmp-evex-movbe.S and minor changes

The only change to the existing generated code is `tzcnt` -> `bsf` to
save a byte of code size here and there.

Rewriting with VMM API allows for memcmp-evex-movbe to be used with
evex512 by including "x86-evex512-vecs.h" at the top.

Complete check passes on x86-64.

2 years agostring: Add len=0 to {w}memcmp{eq} tests and benchtests
Noah Goldstein [Sat, 29 Oct 2022 20:19:57 +0000 (15:19 -0500)]
string: Add len=0 to {w}memcmp{eq} tests and benchtests

len=0 is valid and fairly common so should be tested.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
2 years agoLinux: Add ppoll fortify symbol for 64 bit time_t (BZ# 29746)
Adhemerval Zanella [Fri, 4 Nov 2022 19:02:52 +0000 (16:02 -0300)]
Linux: Add ppoll fortify symbol for 64 bit time_t (BZ# 29746)

Similar to ppoll, the poll.h header needs to redirect the poll call
to a proper fortified ppoll with 64 bit time_t support.

The implementation is straightforward, just need to add a similar
check as __poll_chk and call the 64 bit time_t ppoll version.  The
debug fortify tests are also extended to cover 64 bit time_t for
affected ABIs.

Unfortunately it requires an aditional symbol, which makes backport
tricky.  One possibility is to add a static inline version if compiler
supports is and call abort instead of __chk_fail, so fortified version
will call __poll64 in the end.

Another possibility is to just remove the fortify support for
_TIME_BITS=64.

Checked on i686-linux-gnu.

2 years agohurd: Add sigtimedwait and sigwaitinfo support
Samuel Thibault [Mon, 7 Nov 2022 20:14:39 +0000 (21:14 +0100)]
hurd: Add sigtimedwait and sigwaitinfo support

This simply needed to add the timeout parameter to mach_msg, and copy
information from struct hurd_signal_detail.

2 years agoApply asm redirection in gmp.h before first use
Adhemerval Zanella [Thu, 10 Mar 2022 22:04:47 +0000 (19:04 -0300)]
Apply asm redirection in gmp.h before first use

For clang the redeclaration after the first use, the visibility attribute
is silently ignored (symbol is STV_DEFAULT) while the asm label attribute
causes an error.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoRewrite find_cxx_header config configure.ac
Adhemerval Zanella [Wed, 2 Nov 2022 14:55:57 +0000 (11:55 -0300)]
Rewrite find_cxx_header config configure.ac

With clang if the main file is <stdin>, the first non-main-file
dependency is not listed in the -MP output.  Although it was fixed
on clang-16 [1], this change adds portability for older version.

[1] https://github.com/llvm/llvm-project/commit/ff9576f74514b836e1ba0268409a2ecb919d7118

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoelf/tlsdeschtab.h: Add the Malloc return value check in _dl_make_tlsdesc_dynamic()
Xiaoming Ni [Fri, 4 Nov 2022 09:30:00 +0000 (17:30 +0800)]
elf/tlsdeschtab.h: Add the Malloc return value check in _dl_make_tlsdesc_dynamic()

Check the return value of malloc based on the function header comment of
 _dl_make_tlsdesc_dynamic(). If the return value fails, NULL is returned.

Signed-off-by: Xiaoming Ni <nixiaoming@huawei.com>
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2 years agoelf: Disable some subtests of ifuncmain1, ifuncmain5 for !PIE
Florian Weimer [Fri, 4 Nov 2022 17:37:16 +0000 (18:37 +0100)]
elf: Disable some subtests of ifuncmain1, ifuncmain5 for !PIE

2 years agoposix: Make posix_spawn extensions available by default
Florian Weimer [Fri, 4 Nov 2022 06:43:59 +0000 (07:43 +0100)]
posix: Make posix_spawn extensions available by default

Some sources merely include <spawn.h> without -D_GNU_SOURCE and expect
declarations for posix_spawn_file_actions_addchdir_np to be available.
For consistency, declare posix_spawn_file_actions_addfchdir_np,
posix_spawn_file_actions_addclosefrom_np,
posix_spawn_file_actions_addtcsetpgrp_np as well.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agox86_64: Implement evex512 version of strrchr and wcsrchr
Sunil K Pandey [Tue, 9 Aug 2022 14:57:29 +0000 (07:57 -0700)]
x86_64: Implement evex512 version of strrchr and wcsrchr

Changes from v1:
  Use vec api for register.
  Replace VPCMP with VPCMPEQ
  Restructure and remove 1 unconditional jump.
  Change page cross logic to use sall.

This patch implements following evex512 version of string functions.
evex512 version takes up to 30% less cycle as compared to evex,
depending on length and alignment.

- strrchr function using 512 bit vectors.
- wcsrchr function using 512 bit vectors.

Code size data:

strrchr-evex.o 879 byte
strrchr-evex512.o 601 byte (-32%)

wcsrchr-evex.o 882 byte
wcsrchr-evex512.o 572 byte (-35%)

Placeholder function, not used by any processor at the moment.

Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
2 years agoelf: Introduce <dl-call_tls_init_tp.h> and call_tls_init_tp (bug 29249)
Florian Weimer [Thu, 3 Nov 2022 16:28:03 +0000 (17:28 +0100)]
elf: Introduce <dl-call_tls_init_tp.h> and call_tls_init_tp (bug 29249)

This makes it more likely that the compiler can compute the strlen
argument in _startup_fatal at compile time, which is required to
avoid a dependency on strlen this early during process startup.

Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2 years agoLoongArch: Fix ABI related macros in elf.h to keep consistent with binutils[1].
caiyinyu [Wed, 12 Oct 2022 12:28:42 +0000 (20:28 +0800)]
LoongArch: Fix ABI related macros in elf.h to keep consistent with binutils[1].

[1]:
https://sourceware.org/git/?p=binutils-gdb.git;a=commitdiff;h=c4a7e6b56218e1d5a858682186b542e2eae01a4a;hp=0d94a8735055432029237612a6eb9165db1ec9dd
[2]:
Reference: https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html#_e_flags_identifies_abi_type_and_version

2 years agoscripts/glibcelf.py: Properly report <elf.h> parsing failures
Florian Weimer [Thu, 3 Nov 2022 11:24:17 +0000 (12:24 +0100)]
scripts/glibcelf.py: Properly report <elf.h> parsing failures

Without this change, parse failures result in an exception:

Traceback (most recent call last):
  File "tst-glibcelf.py", line 23, in <module>
    import glibcelf
  File "/path/to/git/scripts/glibcelf.py", line 226, in <module>
    _elf_h = _parse_elf_h()
  File "/path/to/git/scripts/glibcelf.py", line 221, in _parse_elf_h
    result = glibcpp.macro_eval(glibcpp.macro_definitions(tokens), reporter)
  File "/path/to/git/scripts/glibcpp.py", line 379, in macro_eval
    reporter.error(md.line, 'macro {} redefined'.format(md.name))
  File "/path/to/git/scripts/glibcelf.py", line 214, in error
    errors += 1
UnboundLocalError: local variable 'errors' referenced before assignment

2 years agoelf: Rework exception handling in the dynamic loader [BZ #25486]
Florian Weimer [Thu, 3 Nov 2022 08:39:31 +0000 (09:39 +0100)]
elf: Rework exception handling in the dynamic loader [BZ #25486]

The old exception handling implementation used function interposition
to replace the dynamic loader implementation (no TLS support) with the
libc implementation (TLS support).  This results in problems if the
link order between the dynamic loader and libc is reversed (bug 25486).

The new implementation moves the entire implementation of the
exception handling functions back into the dynamic loader, using
THREAD_GETMEM and THREAD_SETMEM for thread-local data support.
These depends on Hurd support for these macros, added in commit
b65a82e4e757c1e6cb7073916 ("hurd: Add THREAD_GET/SETMEM/_NC").

One small obstacle is that the exception handling facilities are used
before the TCB has been set up, so a check is needed if the TCB is
available.  If not, a regular global variable is used to store the
exception handling information.

Also rename dl-error.c to dl-catch.c, to avoid confusion with the
dlerror function.

Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
2 years agolinux: Drop useless include from fstatat.c
Aurelien Jarno [Tue, 1 Nov 2022 19:49:32 +0000 (20:49 +0100)]
linux: Drop useless include from fstatat.c

It is a left-over from previous refactorings.

Reviewed by: Adhemerval Zanella  <adhemerval.zanella@linaro.org>

2 years agoFix OOB read in stdlib thousand grouping parsing [BZ #29727]
Szabolcs Nagy [Tue, 11 Oct 2022 14:24:41 +0000 (15:24 +0100)]
Fix OOB read in stdlib thousand grouping parsing [BZ #29727]

__correctly_grouped_prefixmb only worked with thousands_len == 1,
otherwise it read past the end of cp or thousands.

This affects scanf formats like %'d, %'f and the internal but
exposed __strto{l,ul,f,d,..}_internal with grouping flag set
and an LC_NUMERIC locale where thousands_len > 1.

Avoid OOB access by considering thousands_len when initializing cp.
This fixes bug 29727.

Found by the morello port with strict bounds checking where

FAIL: stdlib/tst-strtod4
FAIL: stdlib/tst-strtod5i

crashed using a locale with thousands_len==3.

2 years agolinux: Fix fstatat on MIPSn64 (BZ #29730)
Aurelien Jarno [Tue, 1 Nov 2022 19:43:55 +0000 (20:43 +0100)]
linux: Fix fstatat on MIPSn64 (BZ #29730)

Commit 6e8a0aac2f883 ("time: Fix overflow itimer tests on 32-bit
systems") changed in_time_t_range to assume a 32-bit time_t. This broke
fstatat on MIPSn64 that was using it with a 64-bit time_t due to
difference between stat and stat64. This commit fix that by adding a
MIPSn64 specific version, which bypasses the EOVERFLOW tests.

Resolves: BZ #29730

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoelf: Remove allocate use on _dl_debug_printf
Adhemerval Zanella [Mon, 31 Oct 2022 18:46:38 +0000 (15:46 -0300)]
elf: Remove allocate use on _dl_debug_printf

The maximum number of directives is already limited by the maximum
value of iovec, and current padding usage on _dl_map_object_from_fd
specifies a value of 16 (2 times sizeof (void *)) in hexa, which is
less than the INT_STRLEN_BOUND(void *) (20 for LP64).

This works if pointers are larger than 8 bytes, for instance 16.
In this case the maximum padding would be 32 and the IFMTSIZE would
be 40.

The resulting code does use a slightly larger static stack, the
output of -fstack-usage (for x86_64):

 * master:
   dl-printf.c:35:1:_dl_debug_vdprintf     1344    dynamic

 * patch:
   dl-printf.c:36:1:_dl_debug_vdprintf     2416    static

However, there is an improvement in code generation:

 * master
   text    data     bss     dec     hex filename
   3309       0       0    3309     ced elf/dl-printf.os

 * patch
   text    data     bss     dec     hex filename
   3151       0       0    3151     c4f elf/dl-printf.os

Checked on x86_64-linux-gnu.

Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2 years agonptl: Fix pthread_create.c build with clang
Adhemerval Zanella [Tue, 25 Oct 2022 14:07:59 +0000 (11:07 -0300)]
nptl: Fix pthread_create.c build with clang

clang complains that libc_hidden_data_def (__nptl_threads_events)
creates an invalid alias:

  pthread_create.c:50:1: error: alias must point to a defined variable or function
  libc_hidden_data_def (__nptl_threads_events)
  ^
  ../include/libc-symbols.h:621:37: note: expanded from macro
  'libc_hidden_data_def'

It seems that clang requires that a proper prototype is defined prior
the hidden alias creation.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoallocate_once: Apply asm redirection before first use
Adhemerval Zanella [Tue, 25 Oct 2022 13:42:16 +0000 (10:42 -0300)]
allocate_once: Apply asm redirection before first use

Compilers may not be able to apply asm redirections to functions
after these functions are used for the first time, e.g. clang 15.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoalloc_buffer: Apply asm redirection before first use
Adhemerval Zanella [Fri, 11 Mar 2022 18:37:57 +0000 (15:37 -0300)]
alloc_buffer: Apply asm redirection before first use

Compilers may not be able to apply asm redirections to functions after
these functions are used for the first time, e.g. clang 15.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoconfigure: Use -Wno-ignored-attributes if compiler warns about multiple aliases
Adhemerval Zanella [Fri, 11 Mar 2022 13:40:44 +0000 (10:40 -0300)]
configure: Use -Wno-ignored-attributes if compiler warns about multiple aliases

clang emits an warning when a double alias redirection is used, to warn
the the original symbol will be used even when weak definition is
overridden.  However, this is a common pattern for weak_alias, where
multiple alias are set to same symbol.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoDisable use of -fsignaling-nans if compiler does not support it
Adhemerval Zanella [Mon, 7 Mar 2022 18:09:53 +0000 (15:09 -0300)]
Disable use of -fsignaling-nans if compiler does not support it

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agointl: Fix clang -Wunused-but-set-variable on plural.c
Adhemerval Zanella [Wed, 24 Aug 2022 14:13:23 +0000 (11:13 -0300)]
intl: Fix clang -Wunused-but-set-variable on plural.c

Clang warns that '__gettextnerrs' set but not used:

  intl/plural.c:1034:9: error: variable '__gettextnerrs' set but not used
  [-Werror,-Wunused-but-set-variable]
      int yynerrs = 0;
        ^

Clang 15 (https://reviews.llvm.org/D122271) -Wunused-but-set-variable
gives a warning while GCC doesn't.  The -Wunused-but-set-variable is
available in GCC 4.6, lower than the minimum required version 6.2.

Since the file is auto-generated, suppress the warning with a compiler
flag.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agoApply asm redirection in not-cancel before first use
Adhemerval Zanella [Mon, 24 Oct 2022 19:12:15 +0000 (16:12 -0300)]
Apply asm redirection in not-cancel before first use

It is avoid a build failure on clang where it can not redeclare function
attribute after its first use.

Reviewed-by: Fangrui Song <maskray@google.com>
2 years agomalloc: Use uintptr_t for pointer alignment
Carlos Eduardo Seo [Wed, 26 Jan 2022 19:00:13 +0000 (16:00 -0300)]
malloc: Use uintptr_t for pointer alignment

Avoid integer casts that assume unsigned long can represent pointers.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoUse uintptr_t in fts for pointer alignment
Szabolcs Nagy [Tue, 12 Jul 2022 09:59:16 +0000 (10:59 +0100)]
Use uintptr_t in fts for pointer alignment

The code assumed unsigned long can represent pointers.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoFix build with GCC 13 _FloatN, _FloatNx built-in functions
Joseph Myers [Mon, 31 Oct 2022 23:20:08 +0000 (23:20 +0000)]
Fix build with GCC 13 _FloatN, _FloatNx built-in functions

GCC 13 has added more _FloatN and _FloatNx versions of existing
<math.h> and <complex.h> built-in functions, for use in libstdc++-v3.

This breaks the glibc build because of how those functions are defined
as aliases to functions with the same ABI but different types.  Add
appropriate -fno-builtin-* options for compiling relevant files, as
already done for the case of long double functions aliasing double
ones and based on the list of files used there.

I fixed some mistakes in that list of double files that I noticed
while implementing this fix, but there may well be more such
(harmless) cases, in this list or the new one (files that don't
actually exist or don't define the named functions as aliases so don't
need the options).  I did try to exclude cases where glibc doesn't
define certain functions for _FloatN or _FloatNx types at all from the
new uses of -fno-builtin-* options.  As with the options for double
files (see the commit message for commit
49348beafe9ba150c9bd48595b3f372299bddbb0, "Fix build with GCC 10 when
long double = double."), it's deliberate that the options are used
even if GCC currently doesn't have a built-in version of a given
functions, so providing some level of future-proofing against more
such built-in functions being added in future.

Tested with build-many-glibcs.py for aarch64-linux-gnu
powerpc-linux-gnu powerpc64le-linux-gnu x86_64-linux-gnu (compilers
and glibcs builds) with GCC mainline.

2 years agoelf: Build tst-relr-mod[34]a.so with $(LDFLAGS-rpath-ORIGIN)
H.J. Lu [Fri, 28 Oct 2022 22:16:32 +0000 (15:16 -0700)]
elf: Build tst-relr-mod[34]a.so with $(LDFLAGS-rpath-ORIGIN)

When --enable-hardcoded-path-in-tests is used only with DT_RUNPATH,
elf/tst-relr3 and elf/tst-relr4 failed to run.  Their dependency
libraries, tst-relr-mod3a.so and tst-relr-mod4a.so, are failed to
load since DT_RUNPATH on executable doesn't apply to them.  Build
tst-relr-mod3a.so and tst-relr-mod4a.so with $(LDFLAGS-rpath-ORIGIN)
to add DT_RUNPATH for their dependency libraries.
Reviewed-by: Fangrui Song <maskray@google.com>
2 years agox86-64: Improve evex512 version of strlen functions
Sunil K Pandey [Mon, 3 Oct 2022 19:00:53 +0000 (12:00 -0700)]
x86-64: Improve evex512 version of strlen functions

This patch improves following functionality
- Replace VPCMP with VPCMPEQ.
- Replace page cross check logic with sall.
- Remove extra lea from align_more.
- Remove uncondition loop jump.
- Use bsf to check max length in first vector.

Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
2 years agoCorrectly determine libc.so 'OUTPUT_FORMAT' when cross-compiling.
Ludovic Courtès [Thu, 1 Jul 2021 21:00:19 +0000 (23:00 +0200)]
Correctly determine libc.so 'OUTPUT_FORMAT' when cross-compiling.

Commit 87d583c6e8cd0e49f64da76636ebeec033298b4d replaces the sed script
with an "objdump -f" invocation to determine the 'OUTPUT_FORMAT' bit of
the libc.so linker script.

However, when cross-compiling, for example from x86_64-linux-gnu to
aarch64-linux-gnu, "objdump -f" would report the wrong
format ("elf64-little").  Conversely, "aarch64-linux-gnu-objdump -f"
reports "elf64-littleaarch64" as expected.

This patch changes 'configure.ac' to use AC_CHECK_TOOL rather than
'$CC -print-prog-name=objdump' to determine the value of the OBJDUMP
variable.  That way, OBJDUMP is set to TRIPLET-objdump when
cross-compiling for TRIPLET.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
Tested-by: Carlos O'Donell <carlos@redhat.com>
2 years agoRemove unused scratch_buffer_dupfree
Szabolcs Nagy [Tue, 11 Oct 2022 12:23:25 +0000 (13:23 +0100)]
Remove unused scratch_buffer_dupfree

Turns out scratch_buffer_dupfree internal API was unused since

commit ef0700004bf0dccf493a5e8e21f71d9e7972ea9f
stdlib: Simplify buffer management in canonicalize

And the related test in malloc/tst-scratch_buffer had issues
so it's better to remove it completely.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoFix elf/tst-dlmopen-twice not to exhaust static TLS
Szabolcs Nagy [Tue, 18 Oct 2022 15:33:52 +0000 (16:33 +0100)]
Fix elf/tst-dlmopen-twice not to exhaust static TLS

By default glibc only allocates enough static TLS for 4 link namespaces
including the initial one. So only use 3 dlmopens in the test.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoUse uintptr_t in string/tester for pointer alignment
Szabolcs Nagy [Mon, 21 Mar 2022 12:36:16 +0000 (12:36 +0000)]
Use uintptr_t in string/tester for pointer alignment

The code assumed unsigned long can represent pointers.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agostdlib/strfrom: Add copysign to fix NAN issue on riscv (BZ #29501)
Letu Ren [Fri, 21 Oct 2022 14:54:50 +0000 (22:54 +0800)]
stdlib/strfrom: Add copysign to fix NAN issue on riscv (BZ #29501)

According to the specification of ISO/IEC TS 18661-1:2014,

The strfromd, strfromf, and strfroml functions are equivalent to
snprintf(s, n, format, fp) (7.21.6.5), except the format string contains only
the character %, an optional precision that does not contain an asterisk *, and
one of the conversion specifiers a, A, e, E, f, F, g, or G, which applies to
the type (double, float, or long double) indicated by the function suffix
(rather than  by a length modifier). Use of these functions with any other 20
format string results in undefined behavior.

strfromf will convert the arguement with type float to double first.

According to the latest version of IEEE754 which is published in 2019,

Conversion of a quiet NaN from a narrower format to a wider format in the same
radix, and then back to the same narrower format, should not change the quiet
NaN payload in any way except to make it canonical.

When either an input or result is a NaN, this standard does not interpret the
sign of a NaN. However, operations on bit strings—copy, negate, abs,
copySign—specify the sign bit of a NaN result, sometimes based upon the sign
bit of a NaN operand. The logical predicates totalOrder and isSignMinus are
also affected by the sign bit of a NaN operand. For all other operations, this
standard does not specify the sign bit of a NaN result, even when there is only
one input NaN, or when the NaN is produced from an invalid operation.

converting NAN or -NAN with type float to double doesn't need to keep
the signbit. As a result, this test case isn't mandatory.

The problem is that according to RISC-V ISA manual in chapter 11.3 of
riscv-isa-20191213,

Except when otherwise stated, if the result of a floating-point operation is
NaN, it is the canonical NaN. The canonical NaN has a positive sign and all
significand bits clear except the MSB, a.k.a. the quiet bit. For
single-precision floating-point, this corresponds to the pattern 0x7fc00000.

which means that conversion -NAN from float to double won't keep the signbit.

Since glibc ought to be consistent here between types and architectures, this
patch adds copysign to fix this problem if the string is NAN. This patch
adds two different functions under sysdeps directory to work around the
issue.

This patch has been tested on x86_64 and riscv64.

Resolves: BZ #29501

v2: Change from macros to different inline functions.
v3: Add unlikely check to isnan.
v4: Fix wrong commit message header.
v5: Fix style: add space before parentheses.
v6: Add copyright.
Signed-off-by: Letu Ren <fantasquex@gmail.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoFix resource/bug-ulimit1 test
Szabolcs Nagy [Fri, 6 May 2022 13:19:54 +0000 (14:19 +0100)]
Fix resource/bug-ulimit1 test

ulimit is a variadic function and the second argument must have type
long (or unsigned long).

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoFix missing NUL terminator in stdio-common/scanf13 test
Szabolcs Nagy [Tue, 11 Oct 2022 13:57:16 +0000 (14:57 +0100)]
Fix missing NUL terminator in stdio-common/scanf13 test

sscanf is only defined on nul terminated string input, but '\0' was
missing in this test which caused _IO_str_init_static_internal to
read OOB on the stack when computing the bounds of the string.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoFix off-by-one OOB read in elf/tst-tls20
Szabolcs Nagy [Tue, 11 Oct 2022 12:07:59 +0000 (13:07 +0100)]
Fix off-by-one OOB read in elf/tst-tls20

The int mods[nmods] array on the stack was overread by one.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoelf: Fix alloca size in _dl_debug_vdprintf
Szabolcs Nagy [Tue, 11 Oct 2022 13:22:35 +0000 (14:22 +0100)]
elf: Fix alloca size in _dl_debug_vdprintf

The alloca size did not consider the optional width parameter for
padding which could cause buffer underflow. The width is currently used
e.g. by _dl_map_object_from_fd which passes 2 * sizeof(void *) which
can be larger than the alloca buffer size on targets where
sizeof(void *) >= 2 * sizeof(unsigned long).

Even if large width is not used on existing targets it is better to fix
the formatting code to avoid surprises.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agomalloc: Use uintptr_t in alloc_buffer
Szabolcs Nagy [Wed, 16 Mar 2022 12:09:15 +0000 (12:09 +0000)]
malloc: Use uintptr_t in alloc_buffer

The values represnt pointers and not sizes. The members of struct
alloc_buffer are already uintptr_t.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoFix invalid pointer dereference in wcpcpy_chk
Szabolcs Nagy [Tue, 21 Jun 2022 14:57:48 +0000 (15:57 +0100)]
Fix invalid pointer dereference in wcpcpy_chk

The src pointer is const and points to a different object, so accessing
dest via src is invalid.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoFix invalid pointer dereference in wcscpy_chk
Szabolcs Nagy [Tue, 21 Jun 2022 13:43:30 +0000 (14:43 +0100)]
Fix invalid pointer dereference in wcscpy_chk

The src pointer is const and points to a different object, so accessing
dest via src is invalid.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agoaarch64: Fix the extension header write in getcontext and swapcontext
Szabolcs Nagy [Fri, 29 Jul 2022 07:00:06 +0000 (08:00 +0100)]
aarch64: Fix the extension header write in getcontext and swapcontext

The extension header is two 32bit words and in the last header both
should be 0. There is plenty space in the __reserved area, but it's
better not to write more than we mean to.

2 years agoaarch64: Don't build wordcopy
Szabolcs Nagy [Thu, 28 Jul 2022 13:29:19 +0000 (14:29 +0100)]
aarch64: Don't build wordcopy

Use an empty wordcopy.c to avoid building the generic one.
It does not seem to be used anywhere.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agoscripts: Use bool in tunables initializer
Szabolcs Nagy [Tue, 8 Jun 2021 13:03:38 +0000 (14:03 +0100)]
scripts: Use bool in tunables initializer

The initializer for a tunable_t set the bool initialized flag to NULL.
This causes a build failure when pointer to bool conversion warns.

Reviewed-by: Florian Weimer <fweimer@redhat.com>
2 years agolonglong.h: update from GCC for LoongArch clz/ctz support
Xi Ruoyao [Sat, 15 Oct 2022 06:12:13 +0000 (14:12 +0800)]
longlong.h: update from GCC for LoongArch clz/ctz support

Update longlong.h to GCC r13-3269.  Keep our local change (prefer https
for gnu.org URL).

2 years agoelf: Introduce to _dl_call_fini
Florian Weimer [Thu, 27 Oct 2022 09:36:44 +0000 (11:36 +0200)]
elf: Introduce to _dl_call_fini

This consolidates the destructor invocations from _dl_fini and
dlclose.  Remove the micro-optimization that avoids
calling _dl_call_fini if they are no destructors (as dlclose is quite
expensive anyway).  The debug log message is now printed
unconditionally.

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
2 years agold.so: Export tls_init_tp_called as __rtld_tls_init_tp_called
Florian Weimer [Thu, 27 Oct 2022 09:36:44 +0000 (11:36 +0200)]
ld.so: Export tls_init_tp_called as __rtld_tls_init_tp_called

This allows the rest of dynamic loader to check whether the TCB
has been set up (and THREAD_GETMEM and THREAD_SETMEM will work).

Reviewed-by: Siddhesh Poyarekar <siddhesh@gotplt.org>
2 years agoscripts/localplt.awk: Handle DT_JMPREL with empty PLT (for C-SKY)
Florian Weimer [Thu, 27 Oct 2022 09:36:44 +0000 (11:36 +0200)]
scripts/localplt.awk: Handle DT_JMPREL with empty PLT (for C-SKY)

On csky-linux-gnuabiv2, binutils 2.33 produces a DT_JMPREL entry
for the dynamic loader if it does not contain any PLT relocations:

Dynamic section at offset 0x1df48 contains 19 entries:
  Tag        Type                         Name/Value
 0x0000000e (SONAME)                     Library soname: [ld-linux-cskyv2-hf.so.1]
 0x00000004 (HASH)                       0xd4
 0x6ffffef5 (GNU_HASH)                   0x1a8
 0x00000005 (STRTAB)                     0x4ac
 0x00000006 (SYMTAB)                     0x28c
 0x0000000a (STRSZ)                      527 (bytes)
 0x0000000b (SYMENT)                     16 (bytes)
 0x00000003 (PLTGOT)                     0x1f000
 0x00000002 (PLTRELSZ)                   0 (bytes)
 0x00000014 (PLTREL)                     RELA
 0x00000017 (JMPREL)                     0xaa4
 0x00000007 (RELA)                       0x75c
 0x00000008 (RELASZ)                     840 (bytes)
 0x00000009 (RELAENT)                    12 (bytes)
 0x6ffffffc (VERDEF)                     0x700
 0x6ffffffd (VERDEFNUM)                  3
 0x6ffffff0 (VERSYM)                     0x6bc
 0x6ffffff9 (RELACOUNT)                  68
 0x00000000 (NULL)                       0x0

This confuses the script:

Unexpected output from check-localplt: …/elf/ld.so.jmprel:
*** DT_JMPREL does not match any section's address

This commit changes the script to record the DT_PLTRELSZ value and
reject DT_JMPREL values not a section boundary only if DT_PLTRELSZ
is present with a non-zero value.

Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
2 years agoRemove lingering libSegfault Makefile entries
Adhemerval Zanella [Wed, 26 Oct 2022 18:13:17 +0000 (15:13 -0300)]
Remove lingering libSegfault Makefile entries

The library was removed by 65ccd641ba.