Implement strcspn/strpbrk/strspn IFUNC selectors in C
All internal calls within libc.so can use IFUNC on x86-64 since unlike
x86, x86-64 supports PC-relative addressing to access the GOT entry so
that it can call via PLT without using an extra register. For libc.a,
we can't use IFUNC for functions which are called before IFUNC has been
initialized. Use IFUNC internally reduces the icache footprint since
libc.so and other codes in the process use the same implementations.
This patch uses IFUNC for strcspn/strpbrk/strspn functions within libc.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strcspn-sse2, strpbrk-sse2 and strspn-sse2.
* sysdeps/x86_64/strcspn.S (STRPBRK_P): Removed.
Check USE_AS_STRPBRK instead of STRPBRK_P.
* sysdeps/x86_64/strpbrk.S (USE_AS_STRPBRK): New.
* sysdeps/x86_64/multiarch/ifunc-sse4_2.h: New file.
* sysdeps/x86_64/multiarch/strcspn-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strcspn.c: Likewise.
* sysdeps/x86_64/multiarch/strpbrk-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strpbrk.c: Likewise.
* sysdeps/x86_64/multiarch/strspn-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strspn.c: Likewise.
* sysdeps/x86_64/multiarch/strcspn.S: Removed.
* sysdeps/x86_64/multiarch/strpbrk.S: Likewise.
* sysdeps/x86_64/multiarch/strspn.S: Likewise.
* sysdeps/x86_64/multiarch/strpbrk-c.c: Remove "#ifdef SHARED"
and "#endif".
Implement strcat family IFUNC selectors in C.
All internal calls within libc.so can use IFUNC on x86-64 since unlike
x86, x86-64 supports PC-relative addressing to access the GOT entry so
that it can call via PLT without using an extra register. For libc.a,
we can't use IFUNC for functions which are called before IFUNC has been
initialized. Use IFUNC internally reduces the icache footprint since
libc.so and other codes in the process use the same implementations.
This patch uses IFUNC for strcat family functions within libc.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strcat-sse2.
* sysdeps/x86_64/multiarch/strcat-sse2.S: New file.
* sysdeps/x86_64/multiarch/strcat.c: Likewise.
* sysdeps/x86_64/multiarch/strncat.c: Likewise.
* sysdeps/x86_64/multiarch/strcat.S: Removed.
* sysdeps/x86_64/multiarch/strncat.S: Likewise.
Implement memcmp family IFUNC selectors in C.
All internal calls within libc.so can use IFUNC on x86-64 since unlike
x86, x86-64 supports PC-relative addressing to access the GOT entry so
that it can call via PLT without using an extra register. For libc.a,
we can't use IFUNC for functions which are called before IFUNC has been
initialized. Use IFUNC internally reduces the icache footprint since
libc.so and other codes in the process use the same implementations.
This patch uses IFUNC for memcmp family functions within libc.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memcmp-sse2.
* sysdeps/x86_64/multiarch/ifunc-memcmp.h: New file.
* sysdeps/x86_64/multiarch/memcmp-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/memcmp.c: Likewise.
* sysdeps/x86_64/multiarch/wmemcmp.c: Likewise.
* sysdeps/x86_64/multiarch/memcmp.S: Removed.
* sysdeps/x86_64/multiarch/wmemcmp.S: Likewise.
Implement memset family IFUNC selectors in C.
All internal calls within libc.so can use IFUNC on x86-64 since unlike
x86, x86-64 supports PC-relative addressing to access the GOT entry so
that it can call via PLT without using an extra register. For libc.a,
we can't use IFUNC for functions which are called before IFUNC has been
initialized. Use IFUNC internally reduces the icache footprint since
libc.so and other codes in the process use the same implementations.
This patch uses IFUNC for memset functions within libc.
2017-06-07 H.J. Lu <hongjiu.lu@intel.com>
Erich Elsen <eriche@google.com>
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memset-sse2-unaligned-erms, and memset_chk-nonshared.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add test for __memset_chk_erms.
Update comments.
* sysdeps/x86_64/multiarch/ifunc-memset.h: New file.
* sysdeps/x86_64/multiarch/memset-sse2-unaligned-erms.S: Likewise.
* sysdeps/x86_64/multiarch/memset.c: Likewise.
* sysdeps/x86_64/multiarch/memset_chk-nonshared.S: Likewise.
* sysdeps/x86_64/multiarch/memset_chk.c: Likewise.
* sysdeps/x86_64/multiarch/memset.S: Removed.
* sysdeps/x86_64/multiarch/memset_chk.S: Likewise.
* sysdeps/x86_64/multiarch/memset-vec-unaligned-erms.S
(__memset_chk_erms): New function.
All of the major architectures are adopting tunables as a way to add
tuning to the library, from hwcap_mask for aarch64 to HLE for s390 and
ifunc and cache geometry for x86. Given this adoption and the fact
that we don't want additional tuning knobs to be added outside of
tunables, it makes sense to enable tunables by default using this
trivial patch.
Smoke tested on x86 to ensure that tunables code was built without
specifying it as a configure flag. I have kept it as --enabled and
not changed it to --disable since we want to still keep the option of
different kinds of front-ends for tunables.
* configure.ac(--enable-tunables): Enable by default.
* configure: Regenerate.
* NEWS: Mention change.
* manual/install.texi (enable-tunables): Adjust documentation.
* INSTALL: Regenerate.
Current allocate_stack logic for create stacks is to first mmap all
the required memory with the desirable memory and then mprotect the
guard area with PROT_NONE if required. Although it works as expected,
it pessimizes the allocation because it requires the kernel to actually
increase commit charge (it counts against the available physical/swap
memory available for the system).
The only issue is to actually check this change since side-effects are
really Linux specific and to actually account them it would require a
kernel specific tests to parse the system wide information. On the kernel
I checked /proc/self/statm does not show any meaningful difference for
vmm and/or rss before and after thread creation. I could only see
really meaningful information checking on system wide /proc/meminfo
between thread creation: MemFree, MemAvailable, and Committed_AS shows
large difference without the patch. I think trying to use these
kind of information on a testcase is fragile.
The BZ#18988 reports shows that the commit pages are easily seen with
mlockall (MCL_FUTURE) (with lock all pages that become mapped in the
process) however a more straighfoward testcase shows that pthread_create
could be faster using this patch:
--
static const int inner_count = 256;
static const int outer_count = 128;
static
void *thread1(void *arg)
{
return NULL;
}
static
void *sleeper(void *arg)
{
pthread_t ts[inner_count];
for (int i = 0; i < inner_count; i++)
pthread_create (&ts[i], &a, thread1, NULL);
for (int i = 0; i < inner_count; i++)
pthread_join (ts[i], NULL);
return NULL;
}
int main(void)
{
pthread_attr_init(&a);
pthread_attr_setguardsize(&a, 1<<20);
pthread_attr_setstacksize(&a, 1134592);
pthread_t ts[outer_count];
for (int i = 0; i < outer_count; i++)
pthread_create(&ts[i], &a, sleeper, NULL);
for (int i = 0; i < outer_count; i++)
pthread_join(ts[i], NULL);
assert(r == 0);
}
return 0;
}
--
On x86_64 (4.4.0-45-generic, gcc 5.4.0) running the small benchtests
I see:
$ time ./test
real 0m3.647s
user 0m0.080s
sys 0m11.836s
While with the patch I see:
$ time ./test
real 0m0.696s
user 0m0.040s
sys 0m1.152s
So I added a pthread_create benchtest (thread_create) which check
the thread creation latency. As for the simple benchtests, I saw
improvements in thread creation on all architectures I tested the
change.
Checked on x86_64-linux-gnu, i686-linux-gnu, aarch64-linux-gnu,
arm-linux-gnueabihf, powerpc64le-linux-gnu, sparc64-linux-gnu,
and sparcv9-linux-gnu.
[BZ #18988]
* benchtests/thread_create-inputs: New file.
* benchtests/thread_create-source.c: Likewise.
* support/xpthread_attr_setguardsize.c: Likewise.
* support/Makefile (libsupport-routines): Add
xpthread_attr_setguardsize object.
* support/xthread.h: Add xpthread_attr_setguardsize prototype.
* benchtests/Makefile (bench-pthread): Add thread_create.
* nptl/allocatestack.c (allocate_stack): Call mmap with PROT_NONE and
then mprotect the required area.
Implement memmove family IFUNC selectors in C.
All internal calls within libc.so can use IFUNC on x86-64 since unlike
x86, x86-64 supports PC-relative addressing to access the GOT entry so
that it can call via PLT without using an extra register. For libc.a,
we can't use IFUNC for functions which are called before IFUNC has been
initialized. Use IFUNC internally reduces the icache footprint since
libc.so and other codes in the process use the same implementations.
This patch uses IFUNC for memmove family functions within libc.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memmove-sse2-unaligned-erms, memcpy_chk-nonshared,
mempcpy_chk-nonshared and memmove_chk-nonshared.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __memmove_chk_erms,
__memcpy_chk_erms and __mempcpy_chk_erms. Update comments.
* sysdeps/x86_64/multiarch/ifunc-memmove.h: New file.
* sysdeps/x86_64/multiarch/memcpy.c: Likewise.
* sysdeps/x86_64/multiarch/memcpy_chk-nonshared.S: Likewise.
* sysdeps/x86_64/multiarch/memcpy_chk.c: Likewise.
* sysdeps/x86_64/multiarch/memmove-sse2-unaligned-erms.S: Likewise.
* sysdeps/x86_64/multiarch/memmove.c: Likewise.
* sysdeps/x86_64/multiarch/memmove_chk-nonshared.S: Likewise.
* sysdeps/x86_64/multiarch/memmove_chk.c: Likewise.
* sysdeps/x86_64/multiarch/mempcpy.c: Likewise.
* sysdeps/x86_64/multiarch/mempcpy_chk-nonshared.S: Likewise.
* sysdeps/x86_64/multiarch/mempcpy_chk.c: Likewise.
* sysdeps/x86_64/multiarch/memcpy.S: Removed.
* sysdeps/x86_64/multiarch/memcpy_chk.S: Likewise.
* sysdeps/x86_64/multiarch/memmove.S: Likewise.
* sysdeps/x86_64/multiarch/memmove_chk.S: Likewise.
* sysdeps/x86_64/multiarch/mempcpy.S: Likewise.
* sysdeps/x86_64/multiarch/mempcpy_chk.S: Likewise.
* sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S
(__mempcpy_chk_erms): New function.
(__memmove_chk_erms): Likewise.
(__memcpy_chk_erms): New alias.
Since commit d957c4d3fa (i386: Compile
rtld-*.os with -mno-sse -mno-mmx -mfpmath=387), vector intrinsics can
no longer be used in ld.so, even if the compiled code never makes it
into the final ld.so link. This commit adds the missing IS_IN (libc)
guard to the SSE 4.2 strcspn implementation, so that it can be used from
ld.so in the future.
This is fairly complicated, not because the users of __need_Emath and
__need_error_t have complicated requirements, but because the core
changes had a lot of fallout.
__need_error_t exists for gnulib compatibility in argz.h and argp.h.
error_t itself is a Hurdism, an enum containing all the E-constants,
so you can do 'p (error_t) errno' in gdb and get a symbolic value.
argz.h and argp.h use it for function return values, and they want to
fall back to 'int' when that's not available. There is no reason why
these nonstandard headers cannot just go ahead and include all of
errno.h; so we do that.
__need_Emath is defined only by .S files; what they _really_ need is
for errno.h to avoid declaring anything other than the E-constants
(e.g. 'extern int __errno_location(void);' is a syntax error in
assembly language). This is replaced with a check for __ASSEMBLER__ in
errno.h, plus a carefully documented requirement for bits/errno.h not
to define anything other than macros. That in turn has the
consequence that bits/errno.h must not define errno - fortunately, all
live ports use the same definition of errno, so I've moved it to
errno.h. The Hurd bits/errno.h must also take care not to define
error_t when __ASSEMBLER__ is defined, which involves repeating all of
the definitions twice, but it's a generated file so that's okay.
* stdlib/errno.h: Remove __need_Emath and __need_error_t logic.
Reorganize file. Declare errno here. When __ASSEMBLER__ is
defined, don't declare anything other than the E-constants.
* include/errno.h: Change conditional for exposing internal
declarations to (not _ISOMAC and not __ASSEMBLER__).
* bits/errno.h: Remove logic for __need_Emath. Document
requirements for a port-specific bits/errno.h.
* sysdeps/unix/sysv/linux/bits/errno.h
* sysdeps/unix/sysv/linux/alpha/bits/errno.h
* sysdeps/unix/sysv/linux/hppa/bits/errno.h
* sysdeps/unix/sysv/linux/mips/bits/errno.h
* sysdeps/unix/sysv/linux/sparc/bits/errno.h:
Add multiple-include guard and check against improper inclusion.
Remove __need_Emath logic. Don't declare errno here. Ensure all
constants are defined as simple integer literals. Consistent
formatting.
* sysdeps/mach/hurd/errnos.awk: Likewise. Only define error_t and
enum __error_t_codes if __ASSEMBLER__ is not defined.
* sysdeps/mach/hurd/bits/errno.h: Regenerate.
* argp/argp.h, string/argz.h: Don't define __need_error_t before
including errno.h.
* sysdeps/i386/i686/fpu/multiarch/s_cosf-sse2.S
* sysdeps/i386/i686/fpu/multiarch/s_sincosf-sse2.S
* sysdeps/i386/i686/fpu/multiarch/s_sinf-sse2.S
* sysdeps/x86_64/fpu/s_cosf.S
* sysdeps/x86_64/fpu/s_sincosf.S
* sysdeps/x86_64/fpu/s_sinf.S:
Just include errno.h; don't define __need_Emath or include
bits/errno.h directly.
__need_FOPEN_MAX wasn't being used anywhere. __need_IOV_MAX was more
complicated; the basic deal is that sys/uio.h wants to define a
constant named UIO_MAXIOV and bits/xopen_lim.h wants to define a
constant named IOV_MAX, with the same meaning. For no apparent reason
this was being handled via bits/stdio_lim.h -- stdio.h is NOT supposed
to define IOV_MAX -- and some mess in Makerules. Also, bits/uio.h on
Linux was being used as a dumping ground for extension functions.
So now we have bits/uio_lim.h, which defines __IOV_MAX.
bits/xopen_lim.h and sys/uio.h use that to define their respective
constants. We also now have bits/uio-ext.h, which is the official
Proper Home for extensions to sys/uio.h. bits/uio.h is removed, and
stdio_lim.h doesn't define IOV_MAX at all.
* bits/uio_lim.h, sysdeps/unix/sysv/linux/bits/uio_lim.h
* bits/uio-ext.h, sysdeps/unix/sysv/linux/bits/uio-ext.h: New file.
* bits/uio.h, sysdeps/unix/sysv/linux/bits/uio.h: Delete file.
* include/bits/xopen_lim.h: Use bits/uio_lim.h to get the value
for IOV_MAX.
* misc/Makefile: Install bits/uio-ext.h and bits/uio_lim.h.
Don't install bits/uio.h.
* misc/sys/uio.h: Don't include bits/uio.h. Do include
bits/types/struct_iovec.h and bits/uio_lim.h. Set UIO_MAXIOV
based on __IOV_MAX. Under __USE_GNU, also include bits/uio-ext.h.
* stdio-common/stdio_lim.h.in: Remove logic for __need_FOPEN_MAX
and __need_IOV_MAX. Don't define IOV_MAX at all.
* Makerules (stdio_lim.h): Remove logic for setting IOV_MAX.
* sysdeps/unix/sysv/linux/bits/fcntl-linux.h:
Include bits/types/struct_iovec.h, not bits/uio.h.
Use __ssize_t, not ssize_t, in function prototypes.
Don't use hard TAB for double space after period in comments.
ELFv2 functions with localentry:0 are those with a single entry point,
ie. global entry == local entry, that have no requirement on r2 or
r12 and guarantee r2 is unchanged on return. Such an external
function can be called via the PLT without saving r2 or restoring it
on return, avoiding a common load-hit-store for small functions.
This patch implements the ld.so changes necessary for this
optimization. ld.so needs to check that an optimized plt call
sequence is in fact calling a function implemented with localentry:0,
end emit a fatal error otherwise.
The elf/testobj6.c change is to stop "error while loading shared
libraries: expected localentry:0 `preload'" when running
elf/preloadtest, which we'd get otherwise.
* elf/elf.h (PPC64_OPT_LOCALENTRY): Define.
* sysdeps/alpha/dl-machine.h (elf_machine_fixup_plt): Add
refsym and sym parameters. Adjust callers.
* sysdeps/aarch64/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/arm/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/generic/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/hppa/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/i386/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/ia64/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/m68k/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/microblaze/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/mips/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/nios2/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/powerpc/powerpc32/dl-machine.h (elf_machine_fixup_plt):
Likewise.
* sysdeps/s390/s390-32/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/s390/s390-64/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/sh/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/sparc/sparc32/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/sparc/sparc64/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/tile/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/x86_64/dl-machine.h (elf_machine_fixup_plt): Likewise.
* sysdeps/powerpc/powerpc64/dl-machine.c (_dl_error_localentry): New.
(_dl_reloc_overflow): Increase buffser size. Formatting.
* sysdeps/powerpc/powerpc64/dl-machine.h (ppc64_local_entry_offset):
Delete reloc param, add refsym and sym. Check optimized plt
call stubs for localentry:0 functions. Adjust callers.
(elf_machine_fixup_plt, elf_machine_plt_conflict): Add refsym
and sym parameters. Adjust callers.
(_dl_reloc_overflow): Move attribute.
(_dl_error_localentry): Declare.
* elf/dl-runtime.c (_dl_fixup): Save original sym. Pass
refsym and sym to elf_machine_fixup_plt.
* elf/testobj6.c (preload): Call printf.
Makes __stpncpy_power8 call __memset_power8 directly rather than via an
IFUNC. Fixes a missing _mcount, and removes some redundant NOPS. The
*_is_local defines are also used in a followup patch.
* sysdeps/powerpc/powerpc64/multiarch/strncpy-power7.S: Define
MEMSET_is_local.
* sysdeps/powerpc/powerpc64/multiarch/strncpy-power8.S: Likewise.
* sysdeps/powerpc/powerpc64/multiarch/stpncpy-power7.S: Likewise.
* sysdeps/powerpc/powerpc64/multiarch/stpncpy-power8.S: Likewise.
Define MEMSET.
* sysdeps/powerpc/powerpc64/multiarch/strstr-power7.S: Define
STRLEN_is_local, STRNLEN_is_local, and STRCHR_is_local.
* sysdeps/powerpc/powerpc64/power7/strstr.S: Likewise. Don't add
nop after local calls.
* sysdeps/powerpc/powerpc64/power7/strncpy.S: Define MEMSET_is_local.
Don't add nop after local call.
* sysdeps/powerpc/powerpc64/power8/strncpy.S: Likewise. Add missing
CALL_MCOUNT.
.align on some targets takes a byte alignment, on others like powerpc,
log2 of the byte alignment. It's a good idea to avoid .align,
particularly since x86 and powerpc are different. This patch fixes
the occurrences of .align in powerpc64/sysdep.h, renames DOT_LABEL
since the macro doesn't have anything to do with adding dots, removes
extraneous semicolons, and fixes some formatting.
* sysdeps/powerpc/powerpc64/sysdep.h: Formatting.
(FUNC_LABEL): Rename from DOT_LABEL.
(ENTRY_1): Use FUNC_LABEL and remove leading space from label.
Use .p2align rather than .align.
(TRACEBACK, TRACEBACK_MASK): Use .p2align rather than .align.
(ABORT_TRANSACTION): Likewise.
(ENTRY_1, ENTRY_2, END_2, LOCALENTRY): Remove unnecessary semicolons,
particularly at end. Add semicolon at invocation as necessary.
(TRACEBACK, TRACEBACK_MASK, PSEUDO, PSEUDO_NOERRNO): Likewise.
(PSEUDO_ERRVAL, PPC64_LOAD_FUNCPTR, OPD_ENT): Likewise.
* sysdeps/powerpc/powerpc64/multiarch/strrchr-power8.S (ENTRY,
END): Adjust to suit.
I think FRAME_PARM[1-9]_SAVE confuse the code, particularly
FRAME_PARM9_SAVE. There are only 8 parameter save slots!
* sysdeps/powerpc/powerpc64/sysdep.h: (FRAME_BACKCHAIN,
FRAME_CR_SAVE, FRAME_LR_SAVE): Move out of conditional.
(FRAME_PARM1_SAVE, FRAME_PARM2_SAVE, FRAME_PARM3_SAVE,
FRAME_PARM4_SAVE, FRAME_PARM5_SAVE, FRAME_PARM6_SAVE,
FRAME_PARM7_SAVE, FRAME_PARM8_SAVE, FRAME_PARM9_SAVE): Delete.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/makecontext.S: Replace
uses of FRAME_PARM[1-9]_SAVE with FRAME_PARM_SAVE plus offset.
The macros used in assembly were broken on powerpc64 ELFv1.
* sysdeps/powerpc/powerpc64/sysdep.h: (call_mcount_parm_offset): Delete.
(SAVE_ARG, REST_ARG, CFI_SAVE_ARG): Correct.
General purpose registers in mcontext_t structure
are 8 bytes long for both MIPS32/MIPS64.
get/set/make/swap context implementations for MIPS O32
incorrectly assume that general purpose registers
in this structure are 4 bytes long.
This patch is fixing that.
Tested for MIPS O32 LE and BE.
Compared objdump of modified functions for mips n32 and mips n64.
[BZ #21548]
* sysdeps/unix/sysv/linux/mips/getcontext.S: Define MCONTEXT_SZGREG as
8 and use it when copying general purpose registers.
* sysdeps/unix/sysv/linux/mips/makecontext.S: Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/Makefile: Include new test for
mips o32.
* sysdeps/unix/sysv/linux/mips/mips32/bug-getcontext-mips-gp.c: Added
new test for mips o32.
* sysdeps/unix/sysv/linux/mips/setcontext.S: Define MCONTEXT_SZGREG as
8 and use it when copying general purpose registers.
* sysdeps/unix/sysv/linux/mips/swapcontext.S: Likewise.
<bits/std_abs.h> from GCC 7 will include /usr/include/stdlib.h from
"#include_next" (instead of stdlib/stdlib.h in the glibc source
directory), and this turns up as a make dependency. Also make a copy
of <bits/std_abs.h> to prevent it from including /usr/include/stdlib.h.
[BZ #21573]
* Makerules [$(c++-bits-std_abs-h) != ""] (before-compile): Add
$(common-objpfx)bits/std_abs.h.
[$(c++-bits-std_abs-h) != ""] ($(common-objpfx)bits/std_abs.h):
New target.
* config.make.in (c++-bits-std_abs-h): New.
* configure.ac (find_cxx_header): Use "\,$1," with sed.
(CXX_BITS_STD_ABS_H): New.
(AC_SUBST(CXX_BITS_STD_ABS_H)): Likewise.
* configure: Regenerated.
bits/sched.h has logic to expose only an impl-namespace variant of
struct sched_param (i.e. struct __sched_param), but nothing uses it,
and the only header that includes bits/sched.h is sched.h. The
__need_schedparam logic can therefore be removed.
bits/sched.h also has a great deal of code relating to cpu_set_t
objects that was *almost* the same between the two versions of
bits/sched.h in the tree; a little spelunking indicated that this is
because some bug fixes got applied to the Linux-specific bits/sched.h
but not the generic one. Introduce a new header, bits/cpu-set.h,
containing the version of that code with the bugfixes, have sched.h
include it directly, and delete all of the code from both versions of
bits/sched.h.
Also remove the unnecessary name mangling in the definition of struct
sched_param -- POSIX specifies a field 'sched_priority', so there is
no reason to define it as '__sched_priority' and then paper over that
with a macro. (Just in case someone was using the internal name,
'sched_priority' remains a macro defined to expand to itself, and
'__sched_priority' now expands to 'sched_priority'.)
Finally, as long as I'm touching these files anyway, merge new
constants from linux/sched.h into the Linux bits/sched.h.
* bits/sched.h: Remove __need_schedparam logic and replace with a
normal multiple-include guard. Change field name in struct
sched_param from __sched_priority to sched_priority. Delete
everything under #ifndef __cpu_set_t_defined.
* sysdeps/unix/sysv/linux/bits/sched.h: Likewise. Also sync with
kernel sched.h, adding SCHED_ISO and SCHED_DEADLINE constants.
* posix/sched.h: Include bits/cpu-set.h as well as bits/sched.h.
For compatibility, #define sched_priority to itself, and #define
__sched_priority as sched_priority.
* posix/bits/cpu-set.h: New file containing, verbatim, the code
that was under #ifndef __cpu_set_t_defined in
sysdeps/unix/sysv/linux/bits/sched.h.
* include/bits/cpu-set.h: New wrapper.
* posix/Makefile: Install bits/cpu-set.h.
Add an undef of __stpcpy in string-inlines.c to avoid a redefinition
error on x86.
* string/string-inlines.c: Add undef of __stpcpy to fix build issue.
This adds test support for float128, and lays some groundwork for future
_FloatN types.
* math/gen-libm-test.pl (@all_floats): Add ifloat128 and float128.
(%all_floats_pfx): Add macro prefix for float128 (FLT128).
* math/libm-test-exp10.inc (pow10_test): Do not test for _FloatN,
* math/libm-test-isfinite.inc (finite_test): Likewise.
* math/libm-test-lgamma.inc (gamma_test): Likewise.
* math/libm-test-nexttoward.inc (nexttoward_test): Likewise.
(nexttoward_test_data}: Likewise.
* math/libm-test-remainder.inc (drem_test): Likewise.
* math/libm-test-scalb.inc (scalb_test): Likewise.
(scalb_test_data): Likewise.
* math/libm-test-significand.inc (significand_test): Likewise.
(significand_test_data): Likewise.
* math/libm-test-support.c (check_complex): Replace __complex__
FLOAT with CFLOAT to get the support for old compiler.
* math/libm-test-support.h (check_complex): Likewise.
* math/test-double.h (CFLOAT, TEST_FLOATN): New macros.
* math/test-float.h (CFLOAT, TEST_FLOATN): Likewise.
* math/test-ldouble.h (CFLOAT, TEST_FLOATN): Likewise.
* math/test-float128.h: New file.
* math/test-math-floatn.h: New file.
TS 18661-3 specifies that macros prefixed with FLTN_ can be defined in
float.h, only if __STDC_WANT_IEC_60559_TYPES_EXT__ is defined as a macro
before the inclusion of float.h. Since GCC 7.0, these macros are provided
under this condition, however, for older versions of GCC, these macros are
not provided at all. This patch allows the definitions of such macros in
include/float.h for older compilers, if the condition above is met, and
even if _ISOMAC is defined.
* include/float.h: Allow the definition of macros prefixed with
FLT128 even if _ISOMAC is defined, but provided that
__STDC_WANT_IEC_60559_TYPES_EXT__ is defined as a macro.
* manual/math.texi (Mathematics): Add description for _FloatN and
_FloatNx types and mentions that they are not support in glibc for
any architecture, so far.
* manual/arith.texi (wcstof, wcstold): Replace the mention to
stdlib.h with wchar.h.
(Parsing of Floats): Add descriptions for strtofN and wcstofN.
(Printing of Floats): Add description for strfromfN.
The implementations are contained with sysdeps/ieee754/float128 as
they are only built when _Float128 is enabled within libc/m.
* include/gmp.h (__mpn_construct_float128): New declaration.
* include/stdlib.h: Include bits/floatn.h for _Float128 tests.
(__strtof128_l): New declaration.
(__strtof128_nan): Likewise.
(__wcstof128_nan): Likewise.
(__strtof128_internal): Likewise.
(____strtof128_l_internal): Likewise.
* include/wchar.h: Include bits/floatn.h for _Float128 tests.
(__wcstof128_l): New declaration.
(__wcstof128_internal): Likewise.
* stdlib/Makefile (bug-strtod2): Link libm too.
* stdlib/stdlib.h (strtof128): New declaration.
(strtof128_l): Likewise.
* stdlib/tst-strtod-nan-locale-main.c: Updated to use
tst-strtod.h macros to ensure float128 gets tested too.
* stdlib/tst-strtod-round-skeleton.c (CHOOSE_f128): New macro.
* stdlib/tst-strtod.h: Include bits/floatn.h for _Float128
tests.
(IF_FLOAT128): New macro.
(GEN_TEST_STRTOD): Update to optionally include _Float128 in
the tests.
(STRTOD_TEST_FOREACH): Likewise.
* sysdeps/ieee754/float128/Makefile: Insert new strtof128 and
wcstof128 functions into libc.
* sysdeps/ieee754/float128/Versions: Add exports for the above
new functions.
* sysdeps/ieee754/float128/mpn2float128.c: New file.
* sysdeps/ieee754/float128/strtod_nan_float128.h: New file.
* sysdeps/ieee754/float128/strtof128.c: New file.
* sysdeps/ieee754/float128/strtof128_l.c: New file.
* sysdeps/ieee754/float128/strtof128_nan.c: New file.
* sysdeps/ieee754/float128/wcstof128.c: New file.
* sysdeps/ieee754/float128/wcstof128_l.c: New file.
* sysdeps/ieee754/float128/wcstof128_nan.c: New fike.
* wcsmbs/Makefile: (CFLAGS-wcstof128.c): Append strtox-CFLAGS.
(CFLAGS-wcstof128_l): Likewise.
* wcsmbs/wchar.h: Include bits/floatn.h for _Float128 tests.
(wcstof128): New declaration.
(wcstof128_l): Likewise.
bsd_signal was added in XPG4.2, so should be declared for
__USE_XOPEN_EXTENDED, instead of the present __USE_XOPEN condition.
This patch fixes the condition accordingly.
Tested for x86_64.
[BZ #21552]
* signal/signal.h (bsd_signal): Declare if [__USE_XOPEN_EXTENDED],
not [__USE_XOPEN].
* conform/Makefile (test-xfail-XPG4/signal.h/conform): Remove.
sigwait was added to POSIX in 1995; it is not in older POSIX versions
or XPG4 / XPG4.2. Thus it should be declared only if
__USE_POSIX199506; signal.h wrongly declares it for __USE_POSIX.
(sigwaitinfo is correctly conditioned on __USE_POSIX199309.) This
patch fixes the condition on the declaration of sigwait.
Tested for x86_64. Because of other namespace problems this does not
allow any conform/ XFAILs to be removed.
[BZ #21550]
* signal/signal.h (sigwait): Only declare if [__USE_POSIX199506].
Implement strcpy family IFUNC selectors in C.
All internal calls within libc.so can use IFUNC on x86-64 since unlike
x86, x86-64 supports PC-relative addressing to access the GOT entry so
that it can call via PLT without using an extra register. For libc.a,
we can't use IFUNC for functions which are called before IFUNC has been
initialized. Use IFUNC internally reduces the icache footprint since
libc.so and other codes in the process use the same implementations.
This patch uses IFUNC for strcpy family functions within libc.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strcpy-sse2 and stpcpy-sse2.
* sysdeps/x86_64/multiarch/ifunc-unaligned-ssse3.h: New file.
* sysdeps/x86_64/multiarch/stpcpy-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/stpcpy.c: Likewise.
* sysdeps/x86_64/multiarch/stpncpy.c: Likewise.
* sysdeps/x86_64/multiarch/strcpy-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strcpy.c: Likewise.
* sysdeps/x86_64/multiarch/strncpy.c: Likewise.
* sysdeps/x86_64/multiarch/stpcpy.S: Removed.
* sysdeps/x86_64/multiarch/stpncpy.S: Likewise.
* sysdeps/x86_64/multiarch/strcpy.S: Likewise.
* sysdeps/x86_64/multiarch/strncpy.S: Likewise.
* sysdeps/x86_64/multiarch/stpncpy-c.c (weak_alias): New.
(libc_hidden_def): Always defined as empty.
* sysdeps/x86_64/multiarch/strncpy-c.c (libc_hidden_builtin_def):
Always Defined as empty.
There is no longer a need for string2.h, so remove it and all mention of it.
Move the redirect for __stpcpy to include/string.h since it is still required
until all internal uses have been renamed.
This fixes several linknamespace/localplt failures when building with -Os.
[BZ #15105]
[BZ #19463]
* include/string.h: Add internal redirect for __stpcpy.
* string/Makefile: Remove bits/string2.h.
* string/string.h: Update comment.
* string/string-inlines.c: Remove bits/string2.h include and comment.
* string/bits/string2.h: Remove file.
There is currently no "cross-platform" (x86 and POWER) support for
determining the cacheline size.
This patch adds support to sysconf() to correctly report cacheline sizes
based on the information in the auxilliary vector.
Thus, using sysconf() is a cross-platform (x86 and POWER) solution for
determining cacheline sizes.
Support is added (on powerpc) for:
_SC_LEVEL1_ICACHE_SIZE _SC_LEVEL1_ICACHE_ASSOC _SC_LEVEL1_ICACHE_LINESIZE
_SC_LEVEL1_DCACHE_SIZE _SC_LEVEL1_DCACHE_ASSOC _SC_LEVEL1_DCACHE_LINESIZE
_SC_LEVEL2_CACHE_SIZE _SC_LEVEL2_CACHE_ASSOC _SC_LEVEL2_CACHE_LINESIZE
_SC_LEVEL3_CACHE_SIZE _SC_LEVEL3_CACHE_ASSOC _SC_LEVEL3_CACHE_LINESIZE
* sysdeps/unix/sysv/linux/powerpc/sysconf.c: New file.
Add powerpc-specific overrides for L1, L2, L3 CACHE_SIZEs,
CACHE_ASSOCs, and CACHE_LINESIZEs, retrieving from auxv.
* sysdeps/unix/sysv/linux/powerpc/test-powerpc-linux-sysconf.c:
New file. Invoke newly supported sysconf values for powerpc,
and report results. If none are supported, report so.
* sysdeps/unix/sysv/linux/powerpc/Makefile (tests): Add new test,
tst-sysconf.
In sys/wait.h, waitid and associated constants and types are UX-shaded
in XPG4.2 (so not in XPG4), and XSI-shaded in POSIX before 2008, so
should be appropriately conditional in the headers. This patch fixes
the conditionals accordingly. (WCONTINUED is actually still
XSI-shaded in POSIX.1:2008, but W* is also reserved there without
XSI-shading, so nothing special needs to be done about the
conditionals on WCONTINUED to conform to POSIX.1:2008 namespace
rules.)
Tested for x86_64.
[BZ #21561]
* posix/sys/wait.h (idtype_t): Change [__USE_XOPEN] condition to
[__USE_XOPEN_EXTENDED].
(id_t): Likewise.
(include of <bits/types/siginfo_t.h): Likewise.
(waitid): Likewise.
* sysdeps/unix/sysv/linux/bits/waitflags.h (WSTOPPED): Condition
on [__USE_XOPEN_EXTENDED || __USE_XOPEN2K8].
(WEXITED): Likewise.
(WCONTINUED): Likewise.
(WNOWAIT): Likewise.
* conform/Makefile (test-xfail-XPG4/stdlib.h/conform): Remove.
(test-xfail-XPG4/sys/wait.h/conform): Likewise.
(test-xfail-POSIX/sys/wait.h/conform): Likewise.
A recent GCC change to expand floating-point classification built-in
functions inline using integer rather than floating-point arithmetic
in some cases resulted in localplt test failures for nios2 and sparc32
<https://sourceware.org/ml/libc-testresults/2017-q2/msg00320.html>.
This patch updates the localplt.data files in question to mark the
relevant symbols as optional / add a new optional symbol. (The GCC
patch has been reverted because of other problems it caused, but one
can assume it will be applied again, without changes that would affect
the PLT entries generated, once those issues have been resolved.)
Tested with build-many-glibcs.py.
* sysdeps/unix/sysv/linux/nios2/localplt.data (__gtdf2): Mark
libc.so PLT entry optional.
(__gtsf2): Likewise.
(__unorddf2): Likewise.
(__unordsf2): Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc32/localplt.data (_Q_fgt):
New optional libc.so PLT entry.
The sigpause function is declared for __USE_XOPEN. As it's new in
XPG4.2 it should be declared only for __USE_XOPEN_EXTENDED. This
patch fixes the declaration accordingly.
Tested for x86_64.
[BZ #21554]
* signal/signal.h (sigpause):: Declare if [__USE_XOPEN_EXTENDED],
not [__USE_XOPEN].
sys/wait.h includes signal.h unconditionally. But the permission to
do so is UX-shaded in XPG4.2, and XSI-shaded in POSIX before 2008, so
this should not be unconditional. This patch fixes this
conservatively: the include is kept, but conditioned on the standards
that permit it (meaning it is still present by default, because
non-XSI POSIX.1:2008 is enabled by default). <bits/types.h> is now
included unconditionally to provide the required definition of
__pid_t; it was previously included via <signal.h>. Some standards
require pid_t to be defined here, and all allow it to be defined here;
previously defined via <signal.h>, it's now defined directly in this
header.
Tested for x86_64. This does not fix any of the sys/wait.h
conformtest failures, but substantially reduces the number of
namespace failures for sys/wait.h for XPG4 and POSIX.
[BZ #21560]
* posix/sys/wait.h: Condition include of <signal.h> on
[__USE_XOPEN_EXTENDED || __USE_XOPEN2K8]. Include <bits/types.h>
unconditionally.
[!__pid_t_defined] (pid_t): Define typedef.
This is just a style fix; we always prefer the shorthand macros over
bare uses of __attribute__, even in private headers.
* include/errno.h (__errno_location): Use __attribute_const__
instead of bare __attribute__.
Like basically all before-compile headers, dl-tunable-list.h should be
generated using a stamp file to minimize unnecessary rebuilding; it
wasn't being added to common-generated, so it wouldn't get cleaned up;
and it was in between the rules for libc-modules.h/.stmp and their own
addition to common-generated.
* Makeconfig (dl-tunable-list.h): Generate using a stamp file.
(common-generated): Add libc-modules.h and libc-modules.stmp in a
more appropriate location. Also add dl-tunable-list.h and
dl-tunable-list.stmp.
Optimize strrchr/wcsrchr with AVX2 to check 32 bytes with vector
instructions. It is as fast as SSE2 version for small data sizes
and up to 1X faster for large data sizes on Haswell. Select AVX2
version on AVX2 machines where vzeroupper is preferred and AVX
unaligned load is fast.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strrchr-sse2, strrchr-avx2, wcsrchr-sse2 and wcsrchr-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __strrchr_avx2,
__strrchr_sse2, __wcsrchr_avx2 and __wcsrchr_sse2.
* sysdeps/x86_64/multiarch/strrchr-avx2.S: New file.
* sysdeps/x86_64/multiarch/strrchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strrchr.c: Likewise.
* sysdeps/x86_64/multiarch/wcsrchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcsrchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wcsrchr.c: Likewise.
Optimize memrchr with AVX2 to search 32 bytes with a single vector
compare instruction. It is as fast as SSE2 memrchr for small data
sizes and up to 1X faster for large data sizes on Haswell. Select
AVX2 memrchr on AVX2 machines where vzeroupper is preferred and AVX
unaligned load is fast.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memrchr-sse2 and memrchr-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __memrchr_avx2 and
__memrchr_sse2.
* sysdeps/x86_64/multiarch/memrchr-avx2.S: New file.
* sysdeps/x86_64/multiarch/memrchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/memrchr.c: Likewise.
Optimize strchr/strchrnul/wcschr with AVX2 to search 32 bytes with vector
instructions. It is as fast as SSE2 versions for size <= 16 bytes and up
to 1X faster for or size > 16 bytes on Haswell. Select AVX2 version on
AVX2 machines where vzeroupper is preferred and AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strchr-sse2, strchrnul-sse2, strchr-avx2, strchrnul-avx2,
wcschr-sse2 and wcschr-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __strchr_avx2,
__strchrnul_avx2, __strchrnul_sse2, __wcschr_avx2 and
__wcschr_sse2.
* sysdeps/x86_64/multiarch/strchr-avx2.S: New file.
* sysdeps/x86_64/multiarch/strchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strchr.c: Likewise.
* sysdeps/x86_64/multiarch/strchrnul-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/strchrnul-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strchrnul.c: Likewise.
* sysdeps/x86_64/multiarch/wcschr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcschr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wcschr.c: Likewise.
* sysdeps/x86_64/multiarch/strchr.S: Removed.
Optimize strlen/strnlen/wcslen/wcsnlen with AVX2 to check 32 bytes with
a single vector compare instruction. It is as fast as SSE2 versions for
size <= 16 bytes and up to 1X faster for or size > 16 bytes on Haswell.
Select AVX2 version on AVX2 machines where vzeroupper is preferred and
AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
strlen-sse2, strnlen-sse2, strlen-avx2, strnlen-avx2,
wcslen-sse2, wcslen-avx2 and wcsnlen-avx2.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Add tests for __strlen_avx2,
__strlen_sse2, __strnlen_avx2, __strnlen_sse2, __wcslen_avx2,
__wcslen_sse2 and __wcsnlen_avx2.
* sysdeps/x86_64/multiarch/strlen-avx2.S: New file.
* sysdeps/x86_64/multiarch/strlen-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strlen.c: Likewise.
* sysdeps/x86_64/multiarch/strnlen-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/strnlen-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/strnlen.c: Likewise.
* sysdeps/x86_64/multiarch/wcslen-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcslen-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wcslen.c: Likewise.
* sysdeps/x86_64/multiarch/wcsnlen-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wcsnlen.c (OPTIMIZE (avx2)): New.
(IFUNC_SELECTOR): Return OPTIMIZE (avx2) on AVX2 machines where
vzeroupper is preferred and AVX unaligned load is fast.
SSE2 memchr is extended to support wmemchr. AVX2 memchr/rawmemchr/wmemchr
are added to search 32 bytes with a single vector compare instruction.
AVX2 memchr/rawmemchr/wmemchr are as fast as SSE2 memchr/rawmemchr/wmemchr
for small sizes and up to 1.5X faster for larger sizes on Haswell and
Skylake. Select AVX2 memchr/rawmemchr/wmemchr on AVX2 machines where
vzeroupper is preferred and AVX unaligned load is fast.
NB: It uses TZCNT instead of BSF since TZCNT produces the same result
as BSF for non-zero input. TZCNT is faster than BSF and is executed
as BSF if machine doesn't support TZCNT.
* sysdeps/x86_64/memchr.S (MEMCHR): New. Depending on if
USE_AS_WMEMCHR is defined.
(PCMPEQ): Likewise.
(memchr): Renamed to ...
(MEMCHR): This. Support wmemchr if USE_AS_WMEMCHR is defined.
Replace pcmpeqb with PCMPEQ.
* sysdeps/x86_64/multiarch/Makefile (sysdep_routines): Add
memchr-sse2, rawmemchr-sse2, memchr-avx2, rawmemchr-avx2,
wmemchr-sse4_1, wmemchr-avx2 and wmemchr-c.
* sysdeps/x86_64/multiarch/ifunc-avx2.h: New file.
* sysdeps/x86_64/multiarch/memchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/memchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/memchr.c: Likewise.
* sysdeps/x86_64/multiarch/rawmemchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/rawmemchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/rawmemchr.c: Likewise.
* sysdeps/x86_64/multiarch/wmemchr-avx2.S: Likewise.
* sysdeps/x86_64/multiarch/wmemchr-sse2.S: Likewise.
* sysdeps/x86_64/multiarch/wmemchr.c: Likewise.
* sysdeps/x86_64/multiarch/ifunc-impl-list.c
(__libc_ifunc_impl_list): Test __memchr_avx2, __memchr_sse2,
__rawmemchr_avx2, __rawmemchr_sse2, __wmemchr_avx2 and
__wmemchr_sse2.
Previously, the implementation would conditionally exit based on the
status argument, which GCC did not know about. This leads to
false uninitialized variable warnings when data is accessed after a
TEST_VERIFY_EXIT failure (from code which would never execute).
