Now that the thread cancellation type is not accessed concurrently
anymore, it is possible to move it out the cancelhandling.
By removing the cancel state out of the internal thread cancel handling
state there is no need to check if cancelled bit was set in CAS
operation.
It allows simplifing the cancellation wrappers and the
CANCEL_CANCELED_AND_ASYNCHRONOUS is removed.
Checked on x86_64-linux-gnu and aarch64-linux-gnu.
Now that thread cancellation state is not accessed concurrently anymore,
it is possible to move it out the 'cancelhandling'.
The code is also simplified: CANCELLATION_P is replaced with a
internal pthread_testcancel call and the CANCELSTATE_BIT{MASK} is
removed.
With this behavior pthread_setcancelstate does not require to act on
cancellation if cancel type is asynchronous (is already handled either
by pthread_setcanceltype or by the signal handler).
Checked on x86_64-linux-gnu and aarch64-linux-gnu.
To setup either the thread scheduling parameters or affinity,
pthread_create enforce synchronization on created thread to wait until
its parent either release PD ownership or send a cancellation signal if
a failure occurs.
However, cancelling the thread does not deallocate the newly created
stack since cancellation expects that a pthread_join to deallocate any
allocated thread resouces (threads stack or TLS).
This patch changes on how the thread resource is deallocate in case of
failure to be synchronous, where the creating thread will signal the
created thread to exit early so it could be joined. The creating thread
will be reponsible for the resource cleanup before returning to the
caller.
To signal the creating thread that a failure has occured, an unused
'struct pthread' member, parent_cancelhandling_unsed, now indicates
whether the setup has failed so creating thread can proper exit.
This strategy also simplifies by not using thread cancellation and
thus not running libgcc_so load in the signal handler (which is
avoided in thread cancellation since 'pthread_cancel' is the one
responsible to dlopen libgcc_s). Another advantage is since the
early exit is move to first step at thread creation, the signal
mask is not already set and thus it can not act on change ID setxid
handler.
Checked on x86_64-linux-gnu and aarch64-linux-gnu.
Use the __nptl_tls_static_size_for_stack inline function instead,
and the GLRO (dl_tls_static_align) value directly.
The computation of GLRO (dl_tls_static_align) in
_dl_determine_tlsoffset ensures that the alignment is at least
TLS_TCB_ALIGN, which at least STACK_ALIGN (see allocate_stack).
Therefore, the additional rounding-up step is removed.
ALso move the initialization of the default stack size from
__pthread_initialize_minimal_internal to __pthread_early_init.
This introduces an extra system call during single-threaded startup,
but this simplifies the initialization sequence. No locking is
needed around the writes to __default_pthread_attr because the
process is single-threaded at this point.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
This replaces the FREE_P macro with the __nptl_stack_in_use inline
function. stack_list_del is renamed to __nptl_stack_list_del,
stack_list_add to __nptl_stack_list_add, __deallocate_stack to
__nptl_deallocate_stack, free_stacks to __nptl_free_stacks.
It is convenient to move __libpthread_freeres into libc at the
same time. This removes the temporary __default_pthread_attr_freeres
export and restores full freeres coverage for __default_pthread_attr.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
stack_list_del overwrites the in-flight stack variable.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
All the stack lists are now in _rtld_global, so it is possible
to change stack permissions directly from there, instead of
calling into libpthread to do the change.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
Only ia64 needs the page mask, and it is straightforward
to compute the value within the function itself.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
Permissions of the cached stacks may have to be updated if an object
is loaded that requires executable stacks, so the dynamic loader
needs to know about these cached stacks.
The move of in_flight_stack and stack_cache_actsize is a requirement for
merging __reclaim_stacks into the fork implementation in libc.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
It is no longer needed after the SINGLE_THREADED_P consolidation.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
This allows the elimination of the __libc_multiple_threads_ptr
variable in libpthread and its initialization procedure.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
The stack list is available in ld.so since commit
1daccf403b ("nptl: Move stack list
variables into _rtld_global"), so it's possible to walk the stack
list directly in ld.so and perform the initialization there.
This eliminates an unprotected function pointer from _rtld_global
and reduces the libpthread initialization code.
The signal handler is exported as __nptl_setxid_sighandler, so
that the libpthread initialization code can install it. This
is sufficient for now because it is guarantueed to happen before
the first pthread_create call.
I used these shell commands:
../glibc/scripts/update-copyrights $PWD/../gnulib/build-aux/update-copyright
(cd ../glibc && git commit -am"[this commit message]")
and then ignored the output, which consisted lines saying "FOO: warning:
copyright statement not found" for each of 6694 files FOO.
I then removed trailing white space from benchtests/bench-pthread-locks.c
and iconvdata/tst-iconv-big5-hkscs-to-2ucs4.c, to work around this
diagnostic from Savannah:
remote: *** pre-commit check failed ...
remote: *** error: lines with trailing whitespace found
remote: error: hook declined to update refs/heads/master
Now __thread_gscope_wait (the function behind THREAD_GSCOPE_WAIT,
formerly __wait_lookup_done) can be implemented directly in ld.so,
eliminating the unprotected GL (dl_wait_lookup_done) function
pointer.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
There are several compiler implementations that allow large stack
allocations to jump over the guard page at the end of the stack and
corrupt memory beyond that. See CVE-2017-1000364.
Compilers can emit code to probe the stack such that the guard page
cannot be skipped, but on aarch64 the probe interval is 64K by default
instead of the minimum supported page size (4K).
This patch enforces at least 64K guard on aarch64 unless the guard
is disabled by setting its size to 0. For backward compatibility
reasons the increased guard is not reported, so it is only observable
by exhausting the address space or parsing /proc/self/maps on linux.
On other targets the patch has no effect. If the stack probe interval
is larger than a page size on a target then ARCH_MIN_GUARD_SIZE can
be defined to get large enough stack guard on libc allocated stacks.
The patch does not affect threads with user allocated stacks.
Fixes bug 26691.
The per-thread state is refactored two use two strategies:
1. The default one uses a TLS structure, which will be placed in the
static TLS space (using __thread keyword).
2. Linux allocates via struct pthread and access it through THREAD_*
macros.
The default strategy has the disadvantage of increasing libc.so static
TLS consumption and thus decreasing the possible surplus used in
some scenarios (which might be mitigated by BZ#25051 fix).
It is used only on Hurd, where accessing the thread storage in the in
single thread case is not straightforward (afaiu, Hurd developers could
correct me here).
The fallback static allocation used for allocation failure is also
removed: defining its size is problematic without synchronizing with
translated messages (to avoid partial translation) and the resulting
usage is not thread-safe.
Checked on x86-64-linux-gnu, i686-linux-gnu, powerpc64le-linux-gnu,
and s390x-linux-gnu.
Tested-by: Carlos O'Donell <carlos@redhat.com>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
union pthread_attr_transparent has always the correct size, even if
pthread_attr_t has padding that is not present in struct pthread_attr.
This should not result in an observable behavioral change. The
existing code appears to have been correct, but it was brittle because
it was not clear which functions were allowed to write to an entire
pthread_attr_t argument (e.g., by copying it).
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
With all Linux ABIs using the expected Linux kABI to indicate
syscalls errors, the INTERNAL_SYSCALL_DECL is an empty declaration
on all ports.
This patch removes the 'err' argument on INTERNAL_SYSCALL* macro
and remove the INTERNAL_SYSCALL_DECL usage.
Checked with a build against all affected ABIs.
All nptl targets have these signal definitions nowadays. This
changes also replaces the nptl-generic version of pthread_sigmask
with the Linux version.
Tested on x86_64-linux-gnu and i686-linux-gnu. Built with
build-many-glibcs.py.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
This patch removes CLOCK_THREAD_CPUTIME_ID and CLOCK_PROCESS_CPUTIME_ID support
from clock_gettime and clock_settime generic implementation. For Linux, kernel
already provides supports through the syscall and Hurd HTL lacks
__pthread_clock_gettime and __pthread_clock_settime internal implementation.
As described in clock_gettime man-page [1] on 'Historical note for SMP
system', implementing CLOCK_{THREAD,PROCESS}_CPUTIME_ID with timer registers
is error-prone and susceptible to timing and accurary issues that the libc
can not deal without kernel support.
This allows removes unused code which, however, still incur in some runtime
overhead in thread creation (the struct pthread cpuclock_offset
initialization).
If hurd eventually wants to support them it should either either implement as
a kernel facility (or something related due its architecture) or in system
specific implementation.
Checked on aarch64-linux-gnu, x86_64-linux-gnu, and i686-linux-gnu. I also
checked on a i686-gnu build.
* nptl/Makefile (libpthread-routines): Remove pthread_clock_gettime and
pthread_clock_settime.
* nptl/pthreadP.h (__find_thread_by_id): Remove prototype.
* elf/dl-support.c [!HP_TIMING_NOAVAIL] (_dl_cpuclock_offset): Remove.
(_dl_non_dynamic_init): Remove _dl_cpuclock_offset setting.
* elf/rtld.c (_dl_start_final): Likewise.
* nptl/allocatestack.c (__find_thread_by_id): Remove function.
* sysdeps/generic/ldsodefs.h [!HP_TIMING_NOAVAIL] (_dl_cpuclock_offset):
Remove.
* sysdeps/mach/hurd/dl-sysdep.c [!HP_TIMING_NOAVAIL]
(_dl_cpuclock_offset): Remove.
* nptl/descr.h (struct pthread): Rename cpuclock_offset to
cpuclock_offset_ununsed.
* nptl/nptl-init.c (__pthread_initialize_minimal_internal): Remove
cpuclock_offset set.
* nptl/pthread_create.c (START_THREAD_DEFN): Likewise.
* sysdeps/nptl/fork.c (__libc_fork): Likewise.
* nptl/pthread_clock_gettime.c: Remove file.
* nptl/pthread_clock_settime.c: Likewise.
* sysdeps/unix/clock_gettime.c (hp_timing_gettime): Remove function.
[HP_TIMING_AVAIL] (realtime_gettime): Remove CLOCK_THREAD_CPUTIME_ID
and CLOCK_PROCESS_CPUTIME_ID support.
* sysdeps/unix/clock_settime.c (hp_timing_gettime): Likewise.
[HP_TIMING_AVAIL] (realtime_gettime): Likewise.
* sysdeps/posix/clock_getres.c (hp_timing_getres): Likewise.
[HP_TIMING_AVAIL] (__clock_getres): Likewise.
* sysdeps/unix/clock_nanosleep.c (CPUCLOCK_P, INVALID_CLOCK_P):
Likewise.
(__clock_nanosleep): Remove CPUCLOCK_P and INVALID_CLOCK_P usage.
[1] http://man7.org/linux/man-pages/man2/clock_gettime.2.html
One group of warnings seen with -Wextra is warnings for static or
inline not at the start of a declaration (-Wold-style-declaration).
This patch fixes various such cases for inline, ensuring it comes at
the start of the declaration (after any static). A common case of the
fix is "static inline <type> __always_inline"; the definition of
__always_inline starts with __inline, so the natural change is to
"static __always_inline <type>". Other cases of the warning may be
harder to fix (one pattern is a function definition that gets
rewritten to be static by an including file, "#define funcname static
wrapped_funcname" or similar), but it seems worth fixing these cases
with inline anyway.
Tested for x86_64.
* elf/dl-load.h (_dl_postprocess_loadcmd): Use __always_inline
before return type, without separate inline.
* elf/dl-tunables.c (maybe_enable_malloc_check): Likewise.
* elf/dl-tunables.h (tunable_is_name): Likewise.
* malloc/malloc.c (do_set_trim_threshold): Likewise.
(do_set_top_pad): Likewise.
(do_set_mmap_threshold): Likewise.
(do_set_mmaps_max): Likewise.
(do_set_mallopt_check): Likewise.
(do_set_perturb_byte): Likewise.
(do_set_arena_test): Likewise.
(do_set_arena_max): Likewise.
(do_set_tcache_max): Likewise.
(do_set_tcache_count): Likewise.
(do_set_tcache_unsorted_limit): Likewise.
* nis/nis_subr.c (count_dots): Likewise.
* nptl/allocatestack.c (advise_stack_range): Likewise.
* sysdeps/ieee754/dbl-64/s_sin.c (do_cos): Likewise.
(do_sin): Likewise.
(reduce_sincos): Likewise.
(do_sincos): Likewise.
* sysdeps/unix/sysv/linux/x86/elision-conf.c
(do_set_elision_enable): Likewise.
(TUNABLE_CALLBACK_FNDECL): Likewise.
The alignment of TLS variables is wrong if accessed from within a thread
for architectures with tls variant TLS_TCB_AT_TP.
For the main thread the static tls data is properly aligned.
For other threads the alignment depends on the alignment of the thread
pointer as the static tls data is located relative to this pointer.
This patch adds this alignment for TLS_TCB_AT_TP variants in the same way
as it is already done for TLS_DTV_AT_TP. The thread pointer is also already
properly aligned if the user provides its own stack for the new thread.
This patch extends the testcase nptl/tst-tls1.c in order to check the
alignment of the tls variables and it adds a pthread_create invocation
with a user provided stack.
The test itself is migrated from test-skeleton.c to test-driver.c
and the missing support functions xpthread_attr_setstack and xposix_memalign
are added.
ChangeLog:
[BZ #23403]
* nptl/allocatestack.c (allocate_stack): Align pointer pd for
TLS_TCB_AT_TP tls variant.
* nptl/tst-tls1.c: Migrate to support/test-driver.c.
Add alignment checks.
* support/Makefile (libsupport-routines): Add xposix_memalign and
xpthread_setstack.
* support/support.h: Add xposix_memalign.
* support/xthread.h: Add xpthread_attr_setstack.
* support/xposix_memalign.c: New File.
* support/xpthread_attr_setstack.c: Likewise.
The __libc_freeres framework does not extend to non-libc.so objects.
This causes problems in general for valgrind and mtrace detecting
unfreed objects in both libdl.so and libpthread.so. This change is
a pre-requisite to properly moving the malloc hooks out of malloc
since such a move now requires precise accounting of all allocated
data before destructors are run.
This commit adds a proper hook in libc.so.6 for both libdl.so and
for libpthread.so, this ensures that shm-directory.c which uses
freeit () to free memory is called properly. We also remove the
nptl_freeres hook and fall back to using weak-ref-and-check idiom
for a loaded libpthread.so, thus making this process similar for
all DSOs.
Lastly we follow best practice and use explicit free calls for
both libdl.so and libpthread.so instead of the generic hook process
which has undefined order.
Tested on x86_64 with no regressions.
Signed-off-by: DJ Delorie <dj@redhat.com>
Signed-off-by: Carlos O'Donell <carlos@redhat.com>
Previously if user requested S stack and G guard when creating a
thread, the total mapping was S and the actual available stack was
S - G - static_tls, which is not what the user requested.
This patch fixes the guard size accounting by pretending the user
requested S+G stack. This way all later logic works out except
when reporting the user requested stack size (pthread_getattr_np)
or when computing the minimal stack size (__pthread_get_minstack).
Normally this will increase thread stack allocations by one page.
TLS accounting is not affected, that will require a separate fix.
[BZ #22637]
* nptl/descr.h (stackblock, stackblock_size): Update comments.
* nptl/allocatestack.c (allocate_stack): Add guardsize to stacksize.
* nptl/nptl-init.c (__pthread_get_minstack): Remove guardsize from
stacksize.
* nptl/pthread_getattr_np.c (pthread_getattr_np): Likewise.
This patch adds two new internal defines to set the internal
pthread_mutex_t layout required by the supported ABIS:
1. __PTHREAD_MUTEX_NUSERS_AFTER_KIND which control whether to define
__nusers fields before or after __kind. The preferred value for
is 0 for new ports and it sets __nusers before __kind.
2. __PTHREAD_MUTEX_USE_UNION which control whether internal __spins and
__list members will be place inside an union for linuxthreads
compatibility. The preferred value is 0 for ports and it sets
to not use an union to define both fields.
It fixes the wrong offsets value for __kind value on x86_64-linux-gnu-x32.
Checked with a make check run-built-tests=no on all afected ABIs.
[BZ #22298]
* nptl/allocatestack.c (allocate_stack): Check if
__PTHREAD_MUTEX_HAVE_PREV is non-zero, instead if
__PTHREAD_MUTEX_HAVE_PREV is defined.
* nptl/descr.h (pthread): Likewise.
* nptl/nptl-init.c (__pthread_initialize_minimal_internal):
Likewise.
* nptl/pthread_create.c (START_THREAD_DEFN): Likewise.
* sysdeps/nptl/fork.c (__libc_fork): Likewise.
* sysdeps/nptl/pthread.h (PTHREAD_MUTEX_INITIALIZER): Likewise.
* sysdeps/nptl/bits/thread-shared-types.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION): New
defines.
(__pthread_internal_list): Check __PTHREAD_MUTEX_USE_UNION instead
of __WORDSIZE for internal layout.
(__pthread_mutex_s): Check __PTHREAD_MUTEX_NUSERS_AFTER_KIND instead
of __WORDSIZE for internal __nusers layout and __PTHREAD_MUTEX_USE_UNION
instead of __WORDSIZE whether to use an union for __spins and __list
fields.
(__PTHREAD_MUTEX_HAVE_PREV): Define also for __PTHREAD_MUTEX_USE_UNION
case.
* sysdeps/aarch64/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION): New
defines.
* sysdeps/alpha/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/arm/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/hppa/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/ia64/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/m68k/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/microblaze/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/mips/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/nios2/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/powerpc/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/s390/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/sh/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/sparc/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/tile/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
* sysdeps/x86/nptl/bits/pthreadtypes-arch.h
(__PTHREAD_MUTEX_NUSERS_AFTER_KIND, __PTHREAD_MUTEX_USE_UNION):
Likewise.
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
This patch fixes ia64 failures on thread exit by madvise the required
area taking in consideration its disjoing stacks
(NEED_SEPARATE_REGISTER_STACK). Also the snippet that setup the
madvise call to advertise kernel the area won't be used anymore in
near future is reallocated in allocatestack.c (for consistency to
put all stack management function in one place).
Checked on x86_64-linux-gnu and i686-linux-gnu for sanity (since
it is not expected code changes for architecture that do not
define NEED_SEPARATE_REGISTER_STACK) and also got a report that
it fixes ia64-linux-gnu failures from Sergei Trofimovich
<slyfox@gentoo.org>.
[BZ #21672]
* nptl/allocatestack.c [_STACK_GROWS_DOWN] (setup_stack_prot):
Set to use !NEED_SEPARATE_REGISTER_STACK as well.
(advise_stack_range): New function.
* nptl/pthread_create.c (START_THREAD_DEFN): Move logic to mark
stack non required to advise_stack_range at allocatestack.c
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.
This patch removes the COLORING_INCREMENT define and usage on allocatestack.c.
It has not been used since 564cd8b67e (glibc-2.3.3) by any architecture.
The idea is to simplify the code by removing obsolete code.
* nptl/allocatestack.c [COLORING_INCREMENT] (nptl_ncreated): Remove.
(allocate_stack): Remove COLORING_INCREMENT usage.
* nptl/stack-aliasing.h (COLORING_INCREMENT). Likewise.
* sysdeps/i386/i686/stack-aliasing.h (COLORING_INCREMENT): Likewise.
In _dl_nothread_init_static_tls() and init_one_static_tls() we must not
touch the DTV of other threads since we do not have ownership of them.
The DTV need not be initialized at this point anyway since only LD/GD
accesses will use them. If LD/GD accesses occur they will take care to
initialize their own thread's DTV.
Concurrency comments were removed from the patch since they need to be
reworked along with a full description of DTV ownership and when it is
or is not safe to modify these structures.
Alexandre Oliva's original patch and discussion:
https://sourceware.org/ml/libc-alpha/2016-09/msg00512.html
This patch remove the PID cache and usage in current GLIBC code. Current
usage is mainly used a performance optimization to avoid the syscall,
however it adds some issues:
- The exposed clone syscall will try to set pid/tid to make the new
thread somewhat compatible with current GLIBC assumptions. This cause
a set of issue with new workloads and usecases (such as BZ#17214 and
[1]) as well for new internal usage of clone to optimize other algorithms
(such as clone plus CLONE_VM for posix_spawn, BZ#19957).
- The caching complexity also added some bugs in the past [2] [3] and
requires more effort of each port to handle such requirements (for
both clone and vfork implementation).
- Caching performance gain in mainly on getpid and some specific
code paths. The getpid performance leverage is questionable [4],
either by the idea of getpid being a hotspot as for the getpid
implementation itself (if it is indeed a justifiable hotspot a
vDSO symbol could let to a much more simpler solution).
Other usage is mainly for non usual code paths, such as pthread
cancellation signal and handling.
For thread creation (on stack allocation) the code simplification in fact
adds some performance gain due the no need of transverse the stack cache
and invalidate each element pid.
Other thread usages will require a direct getpid syscall, such as
cancellation/setxid signal, thread cancellation, thread fail path (at
create_thread), and thread signal (pthread_kill and pthread_sigqueue).
However these are hardly usual hotspots and I think adding a syscall is
justifiable.
It also simplifies both the clone and vfork arch-specific implementation.
And by review each fork implementation there are some discrepancies that
this patch also solves:
- microblaze clone/vfork does not set/reset the pid/tid field
- hppa uses the default vfork implementation that fallback to fork.
Since vfork is deprecated I do not think we should bother with it.
The patch also removes the TID caching in clone. My understanding for
such semantic is try provide some pthread usage after a user program
issue clone directly (as done by thread creation with CLONE_PARENT_SETTID
and pthread tid member). However, as stated before in multiple discussions
threads, GLIBC provides clone syscalls without further supporting all this
semantics.
I ran a full make check on x86_64, x32, i686, armhf, aarch64, and powerpc64le.
For sparc32, sparc64, and mips I ran the basic fork and vfork tests from
posix/ folder (on a qemu system). So it would require further testing
on alpha, hppa, ia64, m68k, nios2, s390, sh, and tile (I excluded microblaze
because it is already implementing the patch semantic regarding clone/vfork).
[1] https://codereview.chromium.org/800183004/
[2] https://sourceware.org/ml/libc-alpha/2006-07/msg00123.html
[3] https://sourceware.org/bugzilla/show_bug.cgi?id=15368
[4] http://yarchive.net/comp/linux/getpid_caching.html
* sysdeps/nptl/fork.c (__libc_fork): Remove pid cache setting.
* nptl/allocatestack.c (allocate_stack): Likewise.
(__reclaim_stacks): Likewise.
(setxid_signal_thread): Obtain pid through syscall.
* nptl/nptl-init.c (sigcancel_handler): Likewise.
(sighandle_setxid): Likewise.
* nptl/pthread_cancel.c (pthread_cancel): Likewise.
* sysdeps/unix/sysv/linux/pthread_kill.c (__pthread_kill): Likewise.
* sysdeps/unix/sysv/linux/pthread_sigqueue.c (pthread_sigqueue):
Likewise.
* sysdeps/unix/sysv/linux/createthread.c (create_thread): Likewise.
* sysdeps/unix/sysv/linux/getpid.c: Remove file.
* nptl/descr.h (struct pthread): Change comment about pid value.
* nptl/pthread_getattr_np.c (pthread_getattr_np): Remove thread
pid assert.
* sysdeps/unix/sysv/linux/pthread-pids.h (__pthread_initialize_pids):
Do not set pid value.
* nptl_db/td_ta_thr_iter.c (iterate_thread_list): Remove thread
pid cache check.
* nptl_db/td_thr_validate.c (td_thr_validate): Likewise.
* sysdeps/aarch64/nptl/tcb-offsets.sym: Remove pid offset.
* sysdeps/alpha/nptl/tcb-offsets.sym: Likewise.
* sysdeps/arm/nptl/tcb-offsets.sym: Likewise.
* sysdeps/hppa/nptl/tcb-offsets.sym: Likewise.
* sysdeps/i386/nptl/tcb-offsets.sym: Likewise.
* sysdeps/ia64/nptl/tcb-offsets.sym: Likewise.
* sysdeps/m68k/nptl/tcb-offsets.sym: Likewise.
* sysdeps/microblaze/nptl/tcb-offsets.sym: Likewise.
* sysdeps/mips/nptl/tcb-offsets.sym: Likewise.
* sysdeps/nios2/nptl/tcb-offsets.sym: Likewise.
* sysdeps/powerpc/nptl/tcb-offsets.sym: Likewise.
* sysdeps/s390/nptl/tcb-offsets.sym: Likewise.
* sysdeps/sh/nptl/tcb-offsets.sym: Likewise.
* sysdeps/sparc/nptl/tcb-offsets.sym: Likewise.
* sysdeps/tile/nptl/tcb-offsets.sym: Likewise.
* sysdeps/x86_64/nptl/tcb-offsets.sym: Likewise.
* sysdeps/unix/sysv/linux/aarch64/clone.S: Remove pid and tid caching.
* sysdeps/unix/sysv/linux/alpha/clone.S: Likewise.
* sysdeps/unix/sysv/linux/arm/clone.S: Likewise.
* sysdeps/unix/sysv/linux/hppa/clone.S: Likewise.
* sysdeps/unix/sysv/linux/i386/clone.S: Likewise.
* sysdeps/unix/sysv/linux/ia64/clone2.S: Likewise.
* sysdeps/unix/sysv/linux/mips/clone.S: Likewise.
* sysdeps/unix/sysv/linux/nios2/clone.S: Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc32/clone.S: Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/clone.S: Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/clone.S: Likewise.
* sysdeps/unix/sysv/linux/s390/s390-64/clone.S: Likewise.
* sysdeps/unix/sysv/linux/sh/clone.S: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc32/clone.S: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc64/clone.S: Likewise.
* sysdeps/unix/sysv/linux/tile/clone.S: Likewise.
* sysdeps/unix/sysv/linux/x86_64/clone.S: Likewise.
* sysdeps/unix/sysv/linux/aarch64/vfork.S: Remove pid set and reset.
* sysdeps/unix/sysv/linux/alpha/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/arm/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/i386/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/ia64/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/m68k/clone.S: Likewise.
* sysdeps/unix/sysv/linux/m68k/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/mips/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/nios2/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc32/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/s390/s390-64/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/sh/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc32/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc64/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/tile/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/x86_64/vfork.S: Likewise.
* sysdeps/unix/sysv/linux/tst-clone2.c (f): Remove direct pthread
struct access.
(clone_test): Remove function.
(do_test): Rewrite to take in consideration pid is not cached anymore.
An earlier fix for TLS dropped early initialization of DTV entries for
modules using static TLS, leaving it for __tls_get_addr to set them
up. That worked on platforms that require the GD access model to be
relaxed to LE in the main executable, but it caused a regression on
platforms that allow GD in the main executable, particularly in
statically-linked programs: they use a custom __tls_get_addr that does
not update the DTV, which fails when the DTV early initialization is
not performed.
In static programs, __libc_setup_tls performs the DTV initialization
for the main thread, but the DTV of other threads is set up in
_dl_allocate_tls_init, so that's the fix that matters.
Restoring the initialization in the remaining functions modified by
this patch was just for uniformity. It's not clear that it is ever
needed: even on platforms that allow GD in the main executable, the
dynamically-linked version of __tls_get_addr would set up the DTV
entries, even for static TLS modules, while updating the DTV counter.
for ChangeLog
[BZ #19826]
* elf/dl-tls.c (_dl_allocate_tls_init): Restore DTV early
initialization of static TLS entries.
* elf/dl-reloc.c (_dl_nothread_init_static_tls): Likewise.
* nptl/allocatestack.c (init_one_static_tls): Likewise.
Instead of a flag which indicates the pointer can be freed, dtv_t
now includes the pointer which should be freed. Due to padding,
the size of dtv_t does not increase.
To avoid using memalign, the new allocate_dtv_entry function
allocates a sufficiently large buffer so that a sub-buffer
can be found in it which starts with an aligned pointer. Both
the aligned and original pointers are kept, the latter for calling
free later.
On arches that set _STACK_GROWS_UP, the stacktop variable is declared
and set, but never actually used. Refactor the code a bit so that the
variable is only declared/set under _STACK_GROWS_DOWN settings.
