Vector variants of the new C23 exp2m1 & exp10m1 routines.
Note: Benchmark inputs for exp2m1 & exp10m1 are identical to exp2 & exp10
respectively, this also includes the floating point variations.
Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
Mode 0777 should be used for directories only because it results
in executable entries (after typical umasks are applied).
Reviewed-by: Arjun Shankar <arjun@redhat.com>
The symbol was unintentionally leaked on ports introduced after
GLIBC_2.34, provide the compat symbol to avoid breaking ABI on them.
Signed-off-by: Xi Ruoyao <xry111@xry111.site>
Reviewed-by: Florian Weimer <fweimer@redhat.com>
Now libpthread is a dummy library and it no longer contains
__pthread_rwlock_unlock at all, thus IS_IN (libpthread) does not make
sense here.
It seems an left over from commit eb29dcde31 ("nptl: Move rwlock
functions with forwarders into libc") and it caused libc.so to export an
unversioned __pthread_rwlock_unlock on Linux ports introduced after the
2.34 release (loongarch and or1k) but the symbol is not ever supposed to
be exported on those new ports. Only since the commit 3b2b88ccee
("elf: early conversion of elf p_flags to mprotect flags") the header
dependency change happened to pull in libc-lockP.h which sets
hidden_proto (__pthread_rwlock_unlock) correctly, the symbol is no
longer exported, breaking the ABI on those ports.
Remove this #if as a clean up and to prevent such a mess from happening
again.
Signed-off-by: Xi Ruoyao <xry111@xry111.site>
Reviewed-by: Florian Weimer <fweimer@redhat.com>
ut_line[] is not a string, it's a fixed-width character field,
and may not be NUL terminated. Thus, the use of strcmp is incorrect.
strncmp is more appropriate as it stops at the field size.
Note that differences beyond the field size do not count here,
as (1) this test doesn't do that, and (2) such differences are
traditionally ignored (i.e. logins that are silently truncated to
8 characters, etc)
While this is "only a test", we should still demonstrate the
correct way of doing things. Also, using strncmp avoids a
"not a string" warning from gcc if you use -O1 or lower,
where it can't deduce that overflow won't happen.
Reviewed-by: Sam James <sam@gentoo.org>
Reviewed-by: Collin Funk <collin.funk1@gmail.com>
Use the flag output constraints feature available in gcc 6+
("=@cc<cond>") instead of explicitly setting a boolean variable
with SETcc instruction. This approach decouples the instruction
that sets the flags from the code that consumes them, allowing
the compiler to create better code when working with flags users.
Instead of e.g.:
lock add %esi,(%rdi)
sets %sil
test %sil,%sil
jne <...>
the compiler now generates:
lock add %esi,(%rdi)
js <...>
No functional changes intended.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: H.J.Lu <hjl.tools@gmail.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Florian Weimer <fweimer@redhat.com>
Add missing "memory" clobber to accessors. The "memory" clobber
tells the compiler that the assembly code performs memory reads
or writes to items other than those listed in the input and output
operands (for example, accessing the memory pointed to by one of
the input parameters).
Use MOVZBL instead of MOVB when reading 1-byte memory location
into a register. MOVB to a register actually inserts into the LSB
of the word-sized register, making the result dependent on
the previous register value. MOVZBL avoids this issue.
Change %P asm operand modifiers to %c. The ‘c’ modifier is a
generic asm operand modifier that requires a constant operand and
prints the constant expression without punctuation.
Replace %b asm operand modifiers with explicit casts. Explicit
casts inform the compiler which part of the register value is used,
allowing it to perform additional optimizations (e.g. narrowing the
preceding operation).
Remove %q asm operand modifiers. Since the value is already cast
to 'long long int', the compiler will emit a 64-bit register name
in the assembly without needing %q.
No functional changes intended.
Tested-by: H.J. Lu <hjl.tools@gmail.com>
Co-Authored-By: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
wc -l pads the output with leading spaces on some systems, e.g. FreeBSD.
This results in the check `test "$count" = 1` failing. Use -eq for integer
comparison instead.
Signed-off-by: Henrik Lindström <henrik@lxm.se>
Reviewed-by: Arjun Shankar <arjun@redhat.com>
This patch replaces _dl_stack_flags global variable by
_dl_stack_prot_flags.
The advantage is that any convertion from p_flags to final used mprotect
flags occurs at loading of p_flags. It avoids repeated spurious
convertions of _dl_stack_flags, for example in allocate_thread_stack.
This modification was suggested in:
https://sourceware.org/pipermail/libc-alpha/2025-March/165537.html
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Add mremap_chunk support for mmap()ed chunks using hugepages by accounting for
their alignment, to prevent the mremap call failing in most cases where the
size passed is not a hugepage size multiple. It also improves robustness for
reallocating hugepages since mremap is much less likely to fail, so running
out of memory when reallocating a larger size and having to copy the old
contents after mremap fails is also less likely.
To track whether an mmap()ed chunk uses hugepages, have a flag in the lowest
bit of the mchunk_prev_size field which is set after a call to sysmalloc_mmap,
and accessed later in mremap_chunk. Create macros for getting and setting this
bit, and for mapping the bit off when accessing the field for mmap()ed chunks.
Since the alignment cannot be lower than 8 bytes, this flag cannot affect the
alignment data.
Add malloc/tst-tcfree4-malloc-check to the tests-exclude-malloc-check list as
malloc-check prevents the tcache from being used to store chunks. This test
caused failures due to a bug in mem2chunk_check to be fixed in a later patch.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Change the mmap chunk layout to be identical to a normal chunk. This makes it
safe for tcache to hold mmap chunks and simplifies size calculations in
memsize and musable. Add mmap_base() and mmap_size() macros to simplify code.
Reviewed-by: Cupertino Miranda <cupertino.miranda@oracle.com>
* manual/message.texi (Advanced gettext functions): Refer to
libc-alpha@sourceware.org instead of bug-glibc-manual@gnu.org which no
longer exists.
* NEWS: Likewise.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
LIBC_TRY_TEST_CC_OPTION is defined with LIBC_TRY_CC_OPTION:
dnl Test a compiler option or options with an empty input file.
dnl LIBC_TRY_CC_OPTION([options], [action-if-true], [action-if-false])
AC_DEFUN([LIBC_TRY_CC_OPTION],
[AS_IF([AC_TRY_COMMAND([${CC-cc} $1 -xc /dev/null -S -o /dev/null])],
[$2], [$3])])
which passes -S to compiler. Unlike gcc, when -c is also passed to clang
20, we get
configure:7838: clang -c -Werror -fsemantic-interposition -xc /dev/null -S -o /dev/null
clang: error: argument unused during compilation: '-c' [-Werror,-Wunused-command-line-argument]
Don't pass -c to LIBC_TRY_TEST_CC_OPTION since -c isn't needed.
This fixes BZ #33318.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Sam James <sam@gentoo.org>
Since have-mamx-tile is for TEST_CC only as shown in
sysdeps/x86_64/configure.ac:
LIBC_CONFIG_VAR([have-mamx-tile], [$libc_cv_test_x86_have_amx_tile])
remove test-have-mamx-tile from Makefile.in.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Sam James <sam@gentoo.org>
Don't use -Wfree-labels/-Wmissing-parameter-name for "make check" if
they are unsupported by TEST_CC.
This fixes BZ #33310.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Sam James <sam@gentoo.org>
Since clang 20 defines __INT64_C and __UINT64_C as built-in macros, undef
them for glibc build and test. This fixes BZ #33311.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Collin Funk <collin.funk1@gmail.com>
Don't use -Wfree-labels/-Wmissing-parameter-name for "make check" if
they are unsupported by TEST_CC.
This fixes BZ #33310.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Collin Funk <collin.funk1@gmail.com>
If the building compiler enables no direct external data access by
default, access to protected data in shared libraries from executables
must be compiled with no direct external data access. If the testing
compiler doesn't support it, set have-protected-data to no to disable
the tests which requires no direct external data access.
Add LIBC_TRY_CC_COMMAND to test a building compiler option or options
with an input file.
This fixes BZ #33286.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Sam James <sam@gentoo.org>
Convert 'compare_real', 'read_real', and 'verify_input' macros to
functions so as to improve readability and avoid pitfalls.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Move the inclusion of the data class header from the individual tests to
the data-type-specific skeleton, providing for the use of the data type
under test in the data class header and reducing duplication.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Follow 'scanf' itself and use the system header inclusion variant for
the data-type-specific skeleton consistently across the remaining scanf
family functions so that any sysdeps/ variant takes precedence even in
the presence of a corresponding skeleton in stdio-common/ (though we
have no such arrangement at the moment).
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Update NaN input data with 'n-char-sequence' in reference data matching
data under test, removing test failures with the M68K host.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Fix a null pointer dereference causing a crash in 'read_real' when the
terminating null character is written for use with the subsequent call
to 'nan' for invalid NaN reference input, such as:
%a:nan:1:3:nanny:
by moving all the 'n-char-sequence' handling under the check for the
opening parenthesis.
No test case added as it's a test case issue in the first place.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Fix a null pointer dereference causing a crash in 'read_real' when the
terminating null character is written for use with the subsequent call
to 'nan' for NaN reference input using null 'n-char-sequence', such as:
%a:nan():1:5:nan():
by moving the memory allocation call ahead of the check for the closing
parenthesis.
No test case added as it's a test case issue in the first place.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Remove buffer contents reporting from the real variant of 'verify_input'
where there has been an input data format error making the contents of
data buffers irrelevant.
For example given invalid float input data:
%a:nan:1:3:nan(:
these messages are produced:
error: ./tst-scanf-format-skeleton.c:240: input buffer: `0000c07f'
error: ./tst-scanf-format-skeleton.c:240: value buffer: `0000c07f'
error: ./tst-scanf-format-skeleton.c:242: input line 1: input data format error
with the two former lines irrelevant. Remove them from output then,
only leaving:
error: ./tst-scanf-format-skeleton.c:242: input line 1: input data format error
No test case added as it's a test case issue in the first place.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Reject invalid formatted scanf character data with the 'c' conversion
where there is not enough input available to satisfy the field width
requested. It is required by ISO C that this conversion matches a
sequence of characters of exactly the number specified by the field
width and it is also already documented as such in our own manual:
"It reads precisely the next N characters, and fails if it cannot get
that many."
Currently a matching success is instead incorrectly produced where the
EOF condition is encountered before the required number of characters
has been retrieved, and the characters actually obtained are stored in
the buffer provided.
Add test cases accordingly and remove placeholders from 'c' conversion
input data for the existing scanf tests.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Clang's -Wimplicit-fallthrough warning, which flags unannotated
fall-through in switch statements, does not recognize specific comments
like /* FALLTHROUGH */ for suppressing the warning, unlike GCC. Since
fall through comments are used extensively in glibc, disable
-Wimplicit-fallthrough when clang is in use.
This fixes BZ #33312.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Collin Funk <collin.funk1@gmail.com>
commit 3020f72618
Author: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
Date: Tue Dec 27 18:11:43 2022 -0300
libio: Remove the usage of __libc_IO_vtables
added
#define libio_static_fn_required(name) __asm (".globl " #name);
to link in libio functions in static binaries. But there is no relocation
in
.globl _IO_file_open
and "strip --strip-unneeded" will remove such unreferenced symbols which
breaks static binaries. Redefine libio_static_fn_required to create a
reference to the required function with
static __typeof (name) *const name##_p __attribute__((used)) = name;
This fixes BZ #33300.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Tested-by: Xi Ruoyao <xry111@xry111.site>
Use __seg_fs named address space qualifiers to cast NPTL accessors
to %fs: prefixed addresses. Use volatile access only where
strictly necessary.
Use existing assembly RSEQ_* accessors for x32 to
work around the GCC bug:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121613
because negative value in __rseq_offset is used
as an offset from %fs.
Co-Authored-By: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Carlos O'Donell <carlos@redhat.com>
After
f6dd43d5f7 i386: Remove stalled __GNUC_PREREQ (6, 0) test in THREAD_SELF()
b0f0c41a5f x86_64: Remove stalled __GNUC_PREREQ (6, 0) test in THREAD_SELF()
removed the unnecessary __GNUC_PREREQ (6, 0) test, remove the extra space
before THREAD_SELF macro name.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Currenty GCC-12 is required as the minimum supported compiler
version. Remove stalled __GNUC_PREREQ (6, 0) test for
GCC compiler version in THREAD_SELF() macro definition.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: H.J.Lu <hjl.tools@gmail.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Use __seg_gs named address space qualifiers to cast NPTL accessors
to %gs: prefixed addresses. Use volatile access only where
strictly necessary.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: H.J.Lu <hjl.tools@gmail.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Currenty GCC-12 is required as the minimum supported compiler
version. Remove stalled __GNUC_PREREQ (6, 0) test for
GCC compiler version in THREAD_SELF() macro definition.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: H.J.Lu <hjl.tools@gmail.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
There is no need to use TESTL when checking the least-significant bit
with a TEST instruction. Use TESTB, which is three bytes shorter:
f6 44 24 04 01 testb $0x1,0x4(%esp)
vs:
f7 44 24 04 01 00 00 testl $0x1,0x4(%esp)
00
for the same effect.
No functional changes intended.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: H.J.Lu <hjl.tools@gmail.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
There is no need to use TESTL when checking the least-significant bit
with a TEST instruction. Use TESTB, which is three bytes shorter:
f6 44 24 f0 01 testb $0x1,-0x10(%rsp)
vs:
f7 44 24 f0 01 00 00 testl $0x1,-0x10(%rsp)
00
for the same effect.
No functional changes intended.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Cc: H.J.Lu <hjl.tools@gmail.com>
Cc: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Fix a bug in predicate logic introduced in last change.
A slight performance improvement from relying on all true
predicates during conversion from single to double.
This fixes BZ #33299.
Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
When testing with some kernel versions, support FUSE infrastructure
encounters a FUSE_GETXATTR request, leading to FUSE tests hanging until
timed out. Therefore, pass FUSE_GETXATTR requests from
support_fuse_handle_mountpoint to support_fuse_handle_directory, and
adjust support_fuse_handle_directory to return ENOSYS so that tests can
proceed.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
When transparent hugepages (THP) are configured to 32MB on x86/loongarch
systems, the current big_size value may not be sufficiently large to
guarantee that free(ptr) [1] will call munmap(ptr_aligned, big_size).
Tested on x86_64 and loongarch64.
PS: Without this patch and using 32M THP, there is a about 50% chance
that malloc/tst-free-errno-malloc-hugetlb1 will fail on both x86_64 and
loongarch64.
[1] malloc/tst-free-errno.c:
...
errno = 1789;
/* This call to free() is supposed to call
munmap (ptr_aligned, big_size);
which increases the number of VMAs by 1, which is supposed
to fail. */
-> free (ptr);
TEST_VERIFY (get_errno () == 1789);
}
...
Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
Hasaan Khan
Florian Weimer
Jonathan Wakely
Xi Ruoyao
DJ Delorie
Uros Bizjak
Henrik Lindström
Cupertino Miranda
Wilco Dijkstra
Paul Zimmermann
Collin Funk
H.J. Lu
Maciej W. Rozycki
Pierre Blanchard
Arjun Shankar
caiyinyu
gfleury