Optimised implementations for single and double precision, Advanced
SIMD and SVE, copied from Arm Optimized Routines.
As previously, data tables are used via a barrier to prevent
overly aggressive constant inlining. Special-case handlers are
marked NOINLINE to avoid incurring the penalty of switching call
standards unnecessarily.
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
Optimised implementations for single and double precision, Advanced
SIMD and SVE, copied from Arm Optimized Routines. Log lookup table
added as HIDDEN symbol to allow it to be shared between AdvSIMD and
SVE variants.
As previously, data tables are used via a barrier to prevent
overly aggressive constant inlining. Special-case handlers are
marked NOINLINE to avoid incurring the penalty of switching call
standards unnecessarily.
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
Optimised implementations for single and double precision, Advanced
SIMD and SVE, copied from Arm Optimized Routines.
As previously, data tables are used via a barrier to prevent
overly aggressive constant inlining. Special-case handlers are
marked NOINLINE to avoid incurring the penalty of switching call
standards unnecessarily.
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
Replace the loop-over-scalar placeholder routines with optimised
implementations from Arm Optimized Routines (AOR).
Also add some headers containing utilities for aarch64 libmvec
routines, and update libm-test-ulps.
Data tables for new routines are used via a pointer with a
barrier on it, in order to prevent overly aggressive constant
inlining in GCC. This allows a single adrp, combined with offset
loads, to be used for every constant in the table.
Special-case handlers are marked NOINLINE in order to confine the
save/restore overhead of switching from vector to normal calling
standard. This way we only incur the extra memory access in the
exceptional cases. NOINLINE definitions have been moved to
math_private.h in order to reduce duplication.
AOR exposes a config option, WANT_SIMD_EXCEPT, to enable
selective masking (and later fixing up) of invalid lanes, in
order to trigger fp exceptions correctly (AdvSIMD only). This is
tested and maintained in AOR, however it is configured off at
source level here for performance reasons. We keep the
WANT_SIMD_EXCEPT blocks in routine sources to greatly simplify
the upstreaming process from AOR to glibc.
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
Linux 6.4 adds the riscv_hwprobe syscall on riscv and enables
memfd_secret on s390. Update syscall-names.list and regenerate the
arch-syscall.h headers with build-many-glibcs.py update-syscalls.
Tested with build-many-glibcs.py.
Trying to mount procfs can fail due multiples reasons: proc is locked
due the container configuration, mount syscall is filtered by a
Linux Secuirty Module, or any other security or hardening mechanism
that Linux might eventually add.
The tests does require a new procfs without binding to parent, and
to fully fix it would require to change how the container was created
(which is out of the scope of the test itself). Instead of trying to
foresee any possible scenario, if procfs can not be mount fail with
unsupported.
Checked on aarch64-linux-gnu.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
The tst-ttyname-direct.c checks the ttyname with procfs mounted in
bind mode (MS_BIND|MS_REC), while tst-ttyname-namespace.c checks
with procfs mount with MS_NOSUID|MS_NOEXEC|MS_NODEV in a new
namespace.
Checked on x86_64-linux-gnu and aarch64-linux-gnu.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
These files could be useful to any port that wants to use ld.so.cache.
Signed-off-by: Sergey Bugaev <bugaevc@gmail.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Few tests needed to properly check for asprintf and system calls return
values with _FORTIFY_SOURCE enabled.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
The fread routine return value needs to be checked when fortification
is enabled, hence use xfread helper.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
On i386 and x86_64, for libc.a specifically, __mempcpy_chk calls
mempcpy which leads POSIX routines to call non-POSIX mempcpy indirectly.
This leads the linknamespace test to fail when glibc is built with
__FORTIFY_SOURCE=3.
Since calling mempcpy doesn't bring any benefit for libc.a, directly
call __mempcpy instead.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Replace alloca with a scratch_buffer to avoid potential stack overflows.
Checked on i686-gnu and x86_64-linux-gnu
Message-Id: <20230619144334.2902429-1-josimmon@redhat.com>
There is a potential memory leak for large writes due to writev being a
"shall occur" cancellation point. Add back the cleanup handler removed
in cf30aa43a5.
Checked on i686-gnu and x86_64-linux-gnu.
Message-Id: <20230619143842.2901522-1-josimmon@redhat.com>
With fortification enabled, read calls return result needs to be checked,
has it gets the __wur macro enabled.
Note on read call removal from sysdeps/pthread/tst-cancel20.c and
sysdeps/pthread/tst-cancel21.c:
It is assumed that this second read call was there to overcome the race
condition between pipe closure and thread cancellation that could happen
in the original code. Since this race condition got fixed by
d0e3ffb7a5 the second call seems
superfluous. Hence, instead of checking for the return value of read, it
looks reasonable to simply remove it.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Use a scratch_buffer rather than alloca to avoid potential stack
overflows.
Checked on i686-gnu and x86_64-linux-gnu
Message-Id: <20230608155844.976554-1-josimmon@redhat.com>
These functions are about to be added to POSIX, under Austin Group
issue 986.
The fortified strlcat implementation does not raise SIGABRT if the
destination buffer does not contain a null terminator, it just
inherits the non-failing regular strlcat behavior.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
With fortification enabled, fgets calls return result needs to be checked,
has it gets the __wur macro enabled.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Different systems prefer a different divisors.
From benchmarks[1] so far the following divisors have been found:
ICX : 2
SKX : 2
BWD : 8
For Intel, we are generalizing that BWD and older prefers 8 as a
divisor, and SKL and newer prefers 2. This number can be further tuned
as benchmarks are run.
[1]: https://github.com/goldsteinn/memcpy-nt-benchmarks
Reviewed-by: DJ Delorie <dj@redhat.com>
This patch should have no affect on existing functionality.
The current code, which has a single switch for model detection and
setting prefered features, is difficult to follow/extend. The cases
use magic numbers and many microarchitectures are missing. This makes
it difficult to reason about what is implemented so far and/or
how/where to add support for new features.
This patch splits the model detection and preference setting stages so
that CPU preferences can be set based on a complete list of available
microarchitectures, rather than based on model magic numbers.
Reviewed-by: DJ Delorie <dj@redhat.com>
Current `non_temporal_threshold` set to roughly '3/4 * sizeof_L3 /
ncores_per_socket'. This patch updates that value to roughly
'sizeof_L3 / 4`
The original value (specifically dividing the `ncores_per_socket`) was
done to limit the amount of other threads' data a `memcpy`/`memset`
could evict.
Dividing by 'ncores_per_socket', however leads to exceedingly low
non-temporal thresholds and leads to using non-temporal stores in
cases where REP MOVSB is multiple times faster.
Furthermore, non-temporal stores are written directly to main memory
so using it at a size much smaller than L3 can place soon to be
accessed data much further away than it otherwise could be. As well,
modern machines are able to detect streaming patterns (especially if
REP MOVSB is used) and provide LRU hints to the memory subsystem. This
in affect caps the total amount of eviction at 1/cache_associativity,
far below meaningfully thrashing the entire cache.
As best I can tell, the benchmarks that lead this small threshold
where done comparing non-temporal stores versus standard cacheable
stores. A better comparison (linked below) is to be REP MOVSB which,
on the measure systems, is nearly 2x faster than non-temporal stores
at the low-end of the previous threshold, and within 10% for over
100MB copies (well past even the current threshold). In cases with a
low number of threads competing for bandwidth, REP MOVSB is ~2x faster
up to `sizeof_L3`.
The divisor of `4` is a somewhat arbitrary value. From benchmarks it
seems Skylake and Icelake both prefer a divisor of `2`, but older CPUs
such as Broadwell prefer something closer to `8`. This patch is meant
to be followed up by another one to make the divisor cpu-specific, but
in the meantime (and for easier backporting), this patch settles on
`4` as a middle-ground.
Benchmarks comparing non-temporal stores, REP MOVSB, and cacheable
stores where done using:
https://github.com/goldsteinn/memcpy-nt-benchmarks
Sheets results (also available in pdf on the github):
https://docs.google.com/spreadsheets/d/e/2PACX-1vS183r0rW_jRX6tG_E90m9qVuFiMbRIJvi5VAE8yYOvEOIEEc3aSNuEsrFbuXw5c3nGboxMmrupZD7K/pubhtml
Reviewed-by: DJ Delorie <dj@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
Container management default seccomp filter [1] only accepts
personality(2) with PER_LINUX, (0x0), UNAME26 (0x20000),
PER_LINUX32 (0x8), UNAME26 | PER_LINUX32, and 0xffffffff (to query
current personality)
Although the documentation only state it is blocked to prevent
'enabling BSD emulation' (PER_BSD, not implemented by Linux), checking
on repository log the real reason is to block ASLR disable flag
(ADDR_NO_RANDOMIZE) and other poorly support emulations.
So handle EPERM and fail as UNSUPPORTED if we can really check for
BZ#19408.
Checked on aarch64-linux-gnu.
[1] https://github.com/moby/moby/blob/master/profiles/seccomp/default.json
Reviewed-by: Florian Weimer <fweimer@redhat.com>
Since the area of the user's stack we use for the registers dump (and
otherwise as __sigreturn2's stack) can and does overlap the sigcontext,
we have to be very careful about the order of loads and stores that we
do. In particular we have to load sc_reply_port before we start
clobbering the sigcontext.
Signed-off-by: Sergey Bugaev <bugaevc@gmail.com>
epoll_pwait2(2)'s second argument should be nonnull. We're going to add
__nonnull to the prototype, so let's fix the test accordingly. We can
use a dummy variable to avoid passing NULL.
Reported-by: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>
With fortification enabled, few function calls return result need to be
checked, has they get the __wur macro enabled.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Since the assembly source file with -evex suffix should use YMM registers,
not ZMM registers, include x86-evex256-vecs.h by default to use YMM
registers in memcmpeq-evex.S
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
Different than other 64 bit architectures, powerpc64 defines the
LFS POSIX lock constants with values similar to 32 ABI, which
are meant to be used with fcntl64 syscall. Since powerpc64 kABI
does not have fcntl, the constants are adjusted with the
FCNTL_ADJUST_CMD macro.
The 4d0fe291ae changed the logic of generic constants
LFS value are equal to the default values; which is now wrong
for powerpc64.
Fix the value by explicit define the previous glibc constants
(powerpc64 does not need to use the 32 kABI value, but it simplifies
the FCNTL_ADJUST_CMD which should be kept as compatibility).
Checked on powerpc64-linux-gnu and powerpc-linux-gnu.
For architecture with default 64 bit time_t support, the kernel
does not provide LFS and non-LFS values for F_GETLK, F_GETLK, and
F_GETLK (the default value used for 64 bit architecture are used).
This is might be considered an ABI break, but the currenct exported
values is bogus anyway.
The POSIX lockf is not affected since it is aliased to lockf64,
which already uses the LFS values.
Checked on i686-linux-gnu and the new tests on a riscv32.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
The LoongArch glibc was using the value of the SHMLBA macro from common code,
which is __getpagesize() (16k), but this was inconsistent with the value of
the SHMLBA macro in the kernel, which is SZ_64K (64k). This caused several
shmat-related tests in LTP (Linux Test Project) to fail. This commit fixes
the issue by ensuring that the glibc's SHMLBA macro value matches the value
used in the kernel like other architectures.
Use a scratch_buffer rather than either alloca or malloc to reduce the
possibility of a stack overflow.
Suggested-by: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
If `non_temporal_threshold` is below `minimum_non_temporal_threshold`,
it almost certainly means we failed to read the systems cache info.
In this case, rather than defaulting the minimum correct value, we
should default to a value that gets at least reasonable
performance. 64MB is chosen conservatively to be at the very high
end. This should never cause non-temporal stores when, if we had read
cache info, we wouldn't have otherwise.
Reviewed-by: Florian Weimer <fweimer@redhat.com>
