In commit cba595c350 and commit
f81ddabffd, ABI compatibility with
applications was broken by increasing the size of the on-stack
allocated __pthread_unwind_buf_t beyond the oringal size.
Applications only have the origianl space available for
__pthread_unwind_register, and __pthread_unwind_next to use,
any increase in the size of __pthread_unwind_buf_t causes these
functions to write beyond the original structure into other
on-stack variables leading to segmentation faults in common
applications like vlc. The only workaround is to version those
functions which operate on the old sized objects, but this must
happen in glibc 2.28.
Thank you to Andrew Senkevich, H.J. Lu, and Aurelien Jarno, for
submitting reports and tracking the issue down.
The commit reverts the above mentioned commits and testing on
x86_64 shows that the ABI compatibility is restored. A tst-cleanup1
regression test linked with an older glibc now passes when run
with the newly built glibc. Previously a tst-cleanup1 linked with
an older glibc would segfault when run with an affected glibc build.
Tested on x86_64 with no regressions.
Signed-off-by: Carlos O'Donell <carlos@redhat.com>
I verified that without the guard accounting change in commit
630f4cc3aa (Fix stack guard size
accounting) and RTLD_NOW for libgcc_s introduced by commit
f993b87540 (nptl: Open libgcc.so with
RTLD_NOW during pthread_cancel), the tst-minstack-cancel test fails on
an AVX-512F machine. tst-minstack-exit still passes, and either of
the mentioned commit by itself frees sufficient stack space to make
tst-minstack-cancel pass, too.
Reviewed-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.
GCC PR 83641 results in a miscompilation of libpthread, which
causes pthread_exit not to restore callee-saved registers before
running destructors for objects on the stack. This test detects
this situation:
info: unsigned int, direct pthread_exit call
tst-thread-exit-clobber.cc:80: numeric comparison failure
left: 4148288912 (0xf741dd90); from: value
right: 1600833940 (0x5f6ac994); from: magic_values.v2
info: double, direct pthread_exit call
info: unsigned int, indirect pthread_exit call
info: double, indirect pthread_exit call
error: 1 test failures
`make check' sometimes triggers a rebuild of librt.so using
nptl/Makefile, which ignores librt's dependence on libpthread. This
causes the build to blow up when we attempt to run the test suite on
RISC-V.
2018-01-06 Palmer Dabbelt <palmer@sifive.com>
* nptl/Makefile (/librt.so): Always depend on
"$(shared-thread-library)".
This patch consolidates the pthread_join and gnu extensions to avoid
code duplication. The function pthread_join, pthread_tryjoin_np, and
pthread_timedjoin_np are now based on pthread_timedjoin_ex.
It also fixes some inconsistencies on ESRCH, EINVAL, EDEADLK handling
(where each implementation differs from each other) and also on
clenup handler (which now always use a CAS).
Checked on i686-linux-gnu and x86_64-linux-gnu.
* nptl/pthreadP.h (__pthread_timedjoin_np): Define.
* nptl/pthread_join.c (pthread_join): Use __pthread_timedjoin_np.
* nptl/pthread_tryjoin.c (pthread_tryjoin): Likewise.
* nptl/pthread_timedjoin.c (cleanup): Use CAS on argument setting.
(pthread_timedjoin_np): Define internal symbol and common code from
pthread_join.
* sysdeps/unix/sysv/linux/i386/lowlevellock.h (__lll_timedwait_tid):
Remove superflous checks.
* sysdeps/unix/sysv/linux/x86_64/lowlevellock.h (__lll_timedwait_tid):
Likewise.
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
All binaries use TLS and thus need a properly set up TCB, so we can
simply return its address directly, instead of forwarding to the
libpthread implementation from libc.
For versioned symbols, the dynamic linker checks that the soname matches
the name supplied by the link editor, so a compatibility symbol in
libpthread is needed.
To avoid linking against the libpthread function in all cases, we would
have to bump the symbol version of libpthread in libc.so and supply a
compat symbol. This commit does not do that because the function
implementation is so small, so the overhead by two active copies of the
same function might well be smaller than the increase in symbol table
size.
On x86, padding in struct __jmp_buf_tag is used for shadow stack pointer
to support shadow stack in Intel Control-flow Enforcemen Technology.
Since the cancel_jmp_buf array is passed to setjmp and longjmp by
casting it to pointer to struct __jmp_buf_tag, it should be as large
as struct __jmp_buf_tag. Otherwise when shadow stack is enabled,
setjmp and longjmp will write and read beyond cancel_jmp_buf when saving
and restoring shadow stack pointer.
This patch adds bits/types/__cancel_jmp_buf_tag.h to define struct
__cancel_jmp_buf_tag so that Linux/x86 can add saved_mask to
cancel_jmp_buf.
Tested natively on i386, x86_64 and x32. Tested hppa-linux-gnu with
build-many-glibcs.py.
[BZ #22563]
* bits/types/__cancel_jmp_buf_tag.h: New file.
* sysdeps/unix/sysv/linux/x86/bits/types/__cancel_jmp_buf_tag.h
* sysdeps/unix/sysv/linux/x86/pthreaddef.h: Likewise.
* sysdeps/unix/sysv/linux/x86/nptl/pthreadP.h: Likewise.
* nptl/Makefile (headers): Add
bits/types/__cancel_jmp_buf_tag.h.
* nptl/descr.h [NEED_SAVED_MASK_IN_CANCEL_JMP_BUF]
(pthread_unwind_buf): Add saved_mask to cancel_jmp_buf.
* sysdeps/nptl/pthread.h: Include
<bits/types/__cancel_jmp_buf_tag.h>.
(__pthread_unwind_buf_t): Use struct __cancel_jmp_buf_tag with
__cancel_jmp_buf.
* sysdeps/unix/sysv/linux/hppa/pthread.h: Likewise.
nptl/tst-attr3 fails to build with GCC mainline because of
(deliberate) aliasing between the second (attributes) and fourth
(argument to thread start routine) arguments to pthread_create.
Although both those arguments are restrict-qualified in POSIX,
pthread_create does not actually dereference its fourth argument; it's
an opaque pointer passed to the thread start routine. Thus, the
aliasing is actually valid in this case, and it's deliberate in the
test. So this patch makes the test disable -Wrestrict for the two
pthread_create calls in question. (-Wrestrict was added in GCC 7,
hence the __GNUC_PREREQ conditions, but the particular warning in
question is new in GCC 8.)
Tested compilation with build-many-glibcs.py for aarch64-linux-gnu.
* nptl/tst-attr3.c: Include <libc-diag.h>.
(do_test) [__GNUC_PREREQ (7, 0)]: Ignore -Wrestrict for two tests.
This patch adds several new tunables to control the behavior of
elision on supported platforms[1]. Since elision now depends
on tunables, we should always *compile* with elision enabled,
and leave the code disabled, but available for runtime
selection. This gives us *much* better compile-time testing of
the existing code to avoid bit-rot[2].
Tested on ppc, ppc64, ppc64le, s390x and x86_64.
[1] This part of the patch was initially proposed by
Paul Murphy but was "staled" because the framework have changed
since the patch was originally proposed:
https://patchwork.sourceware.org/patch/10342/
[2] This part of the patch was inititally proposed as a RFC by
Carlos O'Donnell. Make sense to me integrate this on the patch:
https://sourceware.org/ml/libc-alpha/2017-05/msg00335.html
* elf/dl-tunables.list: Add elision parameters.
* manual/tunables.texi: Add entries about elision tunable.
* sysdeps/unix/sysv/linux/powerpc/elision-conf.c:
Add callback functions to dynamically enable/disable elision.
Add multiple callbacks functions to set elision parameters.
Deleted __libc_enable_secure check.
* sysdeps/unix/sysv/linux/s390/elision-conf.c: Likewise.
* sysdeps/unix/sysv/linux/x86/elision-conf.c: Likewise.
* configure: Regenerated.
* configure.ac: Option enable_lock_elision was deleted.
* config.h.in: ENABLE_LOCK_ELISION flag was deleted.
* config.make.in: Remove references to enable_lock_elision.
* manual/install.texi: Elision configure option was removed.
* INSTALL: Regenerated to remove enable_lock_elision.
* nptl/Makefile:
Disable elision so it can verify error case for destroying a mutex.
* sysdeps/powerpc/nptl/elide.h:
Cleanup ENABLE_LOCK_ELISION check.
Deleted macros for the case when ENABLE_LOCK_ELISION was not defined.
* sysdeps/s390/configure: Regenerated.
* sysdeps/s390/configure.ac: Remove references to enable_lock_elision..
* nptl/tst-mutex8.c:
Deleted all #ifndef ENABLE_LOCK_ELISION from the test.
* sysdeps/powerpc/powerpc32/sysdep.h:
Deleted all ENABLE_LOCK_ELISION checks.
* sysdeps/powerpc/powerpc64/sysdep.h: Likewise.
* sysdeps/powerpc/sysdep.h: Likewise.
* sysdeps/s390/nptl/bits/pthreadtypes-arch.h: Likewise.
* sysdeps/unix/sysv/linux/powerpc/force-elision.h: Likewise.
* sysdeps/unix/sysv/linux/s390/elision-conf.h: Likewise.
* sysdeps/unix/sysv/linux/s390/force-elision.h: Likewise.
* sysdeps/unix/sysv/linux/s390/lowlevellock.h: Likewise.
* sysdeps/unix/sysv/linux/s390/Makefile: Remove references to
enable-lock-elision.
Reviewed-by: Tulio Magno Quites Machado Filho <tuliom@linux.vnet.ibm.com>
GCC 4.9 (the minimum current supported) emits an warning for universal
zero initializer ({0}) on ASSERT_PTHREAD_INTERNAL_SIZE:
pthread_attr_init.c: In function ‘__pthread_attr_init_2_1’:
pthread_attr_init.c:37:3: error: missing braces around initializer [-Werror=missing-braces]
ASSERT_PTHREAD_INTERNAL_SIZE (pthread_attr_t, struct pthread_attr);
^
pthread_attr_init.c:37:3: error: (near initialization for ‘(anonymous).__size’) [-Werror=missing-braces]
It is fact GCC BZ#53119 [1] fixed in later version (GCC5+). Since
current branch is closed and there is no indication it will be backports
(comment #20 in same bug report) this patch fixes by using a double
bracket to zero initialize the struct.
Checked on x86_64-linux-gnu with GCC 7 and GCC 4.9.
* nptl/pthreadP.h (ASSERT_PTHREAD_INTERNAL_SIZE): Add workarond for
-Wmissing-braces on GCC 4.9.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53119
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Update all sourceware links to https. The website redirects
everything to https anyway so let the web server do a bit less work.
The only reference that remains unchanged is the one in the old
ChangeLog, since it didn't seem worth changing it.
* NEWS: Update sourceware link to https.
* configure.ac: Likewise.
* crypt/md5test-giant.c: Likewise.
* dlfcn/bug-atexit1.c: Likewise.
* dlfcn/bug-atexit2.c: Likewise.
* localedata/README: Likewise.
* malloc/tst-mallocfork.c: Likewise.
* manual/install.texi: Likewise.
* nptl/tst-pthread-getattr.c: Likewise.
* stdio-common/tst-fgets.c: Likewise.
* stdio-common/tst-fwrite.c: Likewise.
* sunrpc/Makefile: Likewise.
* sysdeps/arm/armv7/multiarch/memcpy_impl.S: Likewise.
* wcsmbs/tst-mbrtowc2.c: Likewise.
* configure: Regenerate.
* INSTALL: Regenerate.
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 adds a new build test to check for internal fields
offsets for user visible internal field. Although currently
the only field which is statically initialized to a non zero value
is pthread_mutex_t.__data.__kind value, the tests also check the
offset of __kind, __spins, __elision (if supported), and __list
internal member. A internal header (pthread-offset.h) is added
to each major ABI with the reference value.
Checked on x86_64-linux-gnu and with a build check for all affected
ABIs (aarch64-linux-gnu, alpha-linux-gnu, arm-linux-gnueabihf,
hppa-linux-gnu, i686-linux-gnu, ia64-linux-gnu, m68k-linux-gnu,
microblaze-linux-gnu, mips64-linux-gnu, mips64-n32-linux-gnu,
mips-linux-gnu, powerpc64le-linux-gnu, powerpc-linux-gnu,
s390-linux-gnu, s390x-linux-gnu, sh4-linux-gnu, sparc64-linux-gnu,
sparcv9-linux-gnu, tilegx-linux-gnu, tilegx-linux-gnu-x32,
tilepro-linux-gnu, x86_64-linux-gnu, and x86_64-linux-x32).
* nptl/pthreadP.h (ASSERT_PTHREAD_STRING,
ASSERT_PTHREAD_INTERNAL_OFFSET): New macro.
* nptl/pthread_mutex_init.c (__pthread_mutex_init): Add build time
checks for internal pthread_mutex_t offsets.
* sysdeps/aarch64/nptl/pthread-offsets.h
(__PTHREAD_MUTEX_NUSERS_OFFSET, __PTHREAD_MUTEX_KIND_OFFSET,
__PTHREAD_MUTEX_SPINS_OFFSET, __PTHREAD_MUTEX_ELISION_OFFSET,
__PTHREAD_MUTEX_LIST_OFFSET): New macro.
* sysdeps/alpha/nptl/pthread-offsets.h: Likewise.
* sysdeps/arm/nptl/pthread-offsets.h: Likewise.
* sysdeps/hppa/nptl/pthread-offsets.h: Likewise.
* sysdeps/i386/nptl/pthread-offsets.h: Likewise.
* sysdeps/ia64/nptl/pthread-offsets.h: Likewise.
* sysdeps/m68k/nptl/pthread-offsets.h: Likewise.
* sysdeps/microblaze/nptl/pthread-offsets.h: Likewise.
* sysdeps/mips/nptl/pthread-offsets.h: Likewise.
* sysdeps/nios2/nptl/pthread-offsets.h: Likewise.
* sysdeps/powerpc/nptl/pthread-offsets.h: Likewise.
* sysdeps/s390/nptl/pthread-offsets.h: Likewise.
* sysdeps/sh/nptl/pthread-offsets.h: Likewise.
* sysdeps/sparc/nptl/pthread-offsets.h: Likewise.
* sysdeps/tile/nptl/pthread-offsets.h: Likewise.
* sysdeps/x86_64/nptl/pthread-offsets.h: Likewise.
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Various subdirectories of glibc include Banner files to put some text
in the output of executing libc.so.6, under "Available extensions".
Some of those subdirectories (e.g. crypt) may originally have been
add-ons (and so optional, so a particular glibc build might or might
not have included them), but except for libidn they aren't now (or if
only included in some builds, in the case of soft-fp, the inclusion
depends on the architecture for which glibc is configured rather than
having any glibc configuration for which it's an optional feature),
and it doesn't seem useful for the libc.so.6 output to call out a few
features like that.
This patch removes the non-add-on Banner files, updating contrib.texi
where they noted contributions not otherwise mentioned there.
Tested for x86_64.
* crypt/Banner: Remove file.
* nptl/Banner: Likewise.
* resolv/Banner: Likewise.
* soft-fp/Banner: Likewise.
* nptl/Makefile ($(objpfx)banner.h): Remove rule.
($(objpfx)version.d): Remove dependency on banner.h.
($(objpfx)version.os): Likewise.
* nptl/version.c (banner): Do not include banner.h.
* manual/contrib.texi: Update entries for Richard Henderson, Jakub
Jelinek and BIND code.
On very large multi-processor systems, creating hundreds of threads
runs into a test time out. The tests do not seem to benefit from
massive over-scheduling.
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
This patch consolidates all the non cancellable nanosleep calls to use
the __nanosleep_nocancel identifier. For non cancellable targets it will
be just a macro to call the default respective symbol while on Linux
will be a internal one.
Checked on x86_64-linux-gnu, x86_64-linux-gnu-x32, and i686-linux-gnu.
* nptl/pthread_mutex_timedlock.c (__pthread_mutex_timedlock): Replace
nanosleep_not_cancel with __nanosleep_nocancel.
* sysdeps/generic/not-cancel.h (nanosleep_not_cancel): Remove macro.
(__nanosleep_nocancel): New macro.
* sysdeps/unix/sysv/linux/nanosleep.c (__nanosleep_nocancel): New
function.
* sysdeps/unix/sysv/linux/not-cancel.h (nanosleep_not_cancel): Remove
macro.
(__nanosleep_nocancel): New prototype.
This patch consolidates all the non cancellable pause calls to use
the __pause_nocancel identifier. For non cancellable targets it will
be just a macro to call the default respective symbol while on Linux
will be a internal one.
Checked on x86_64-linux-gnu, x86_64-linux-gnu-x32, and i686-linux-gnu.
* nptl/pthread_mutex_lock.c (__pthread_mutex_lock_full): Replace
pause_not_cancel with __pause_nocancel.
* sysdeps/generic/not-cancel.h (pause_not_cancel): Remove macro.
(__pause_nocancel): New macro.
* sysdeps/unix/sysv/linux/not-cancel.h (pause_not_cancel): Remove
macro.
(__pause_nocancel): New prototype.
* sysdeps/unix/sysv/linux/pause.c (__pause_nocancel): New function.
Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
there won't be a problem when the compiler does not turn this into a tail
call.
65810f0ef0 fixed a robust mutex bug but
introduced BZ 21778: if the CAS used to try to acquire a lock fails, the
expected value is not updated, which breaks other cases in the loce
acquisition loop. The fix is to simply update the expected value with
the value returned by the CAS, which ensures that behavior is as if the
first case with the CAS never happened (if the CAS fails).
This is a regression introduced in the last release.
Tested on x86_64, i686, ppc64, ppc64le, s390x, aarch64, armv7hl.
Without this fix, the rwlock can fail to execute the explicit hand-over
in certain cases (e.g., empty critical sections that switch quickly between
read and write phases). This can then lead to errors in how __wrphase_futex
is accessed, which in turn can lead to deadlocks.
Single thread optimization is valid if at thread creation time the
optimization can be disabled. This is in principle true for all
stream objects that user code can access (and thus needs locking),
using the same internal list as fflush(0) uses. However in glibc
open_memstream is not on that list (BZ 21735) so the optimization
has to be disabled.
* libio/memstream.c (__open_memstream): Set _IO_FLAGS2_NEED_LOCK.
* libio/wmemstream.c (open_wmemstream): Likewise.
* nptl/tst-memstream.c: New.
This patch increases the timeouts for some tests that I've seen timing
out on slow systems in my 2.26 release testing. (In the case of
tst-tsearch.c, increasing the timeout means removing a setting of 10
that was put there before the default timeout was increased to 20
seconds, so putting the default into effect.)
* iconvdata/tst-loading.c (TIMEOUT): Define to 30.
* misc/tst-tsearch.c (TIMEOUT): Remove.
* nptl/tst-create-detached.c (TIMEOUT): Define to 100.
* nptl/tst-robust-fork.c (TIMEOUT): Likewise.
* nptl/tst-rwlock19.c (TIMEOUT): Likewise.
* string/tst-cmp.c (TIMEOUT): Define to 600.
Locking overhead can be significant in some stdio operations
that are common in single threaded applications.
This patch adds the _IO_FLAGS2_NEED_LOCK flag to indicate if
an _IO_FILE object needs to be locked and some of the stdio
functions just jump to their _unlocked variant when not. The
flag is set on all _IO_FILE objects when the first thread is
created. A new GLIBC_PRIVATE libc symbol, _IO_enable_locks,
was added to do this from libpthread.
The optimization can be applied to more stdio functions,
currently it is only applied to single flag check or single
non-wide-char standard operations. The flag should probably
be never set for files with _IO_USER_LOCK, but that's just a
further optimization, not a correctness requirement.
The optimization is valid in a single thread because stdio
operations are non-as-safe (so lock state is not observable
from a signal handler) and stdio locks are recursive (so lock
state is not observable via deadlock). The optimization is not
valid if a thread may be created while an stdio lock is taken
and thus it should be disabled if any user code may run during
an stdio operation (interposed malloc, printf hooks, etc).
This makes the optimization more complicated for some stdio
operations (e.g. printf), but those are bigger and thus less
important to optimize so this patch does not try to do that.
* libio/libio.h (_IO_FLAGS2_NEED_LOCK, _IO_need_lock): Define.
* libio/libioP.h (_IO_enable_locks): Declare.
* libio/Versions (_IO_enable_locks): New symbol.
* libio/genops.c (_IO_enable_locks): Define.
(_IO_old_init): Initialize flags2.
* libio/feof.c.c (_IO_feof): Avoid locking when not needed.
* libio/ferror.c (_IO_ferror): Likewise.
* libio/fputc.c (fputc): Likewise.
* libio/putc.c (_IO_putc): Likewise.
* libio/getc.c (_IO_getc): Likewise.
* libio/getchar.c (getchar): Likewise.
* libio/ioungetc.c (_IO_ungetc): Likewise.
* nptl/pthread_create.c (__pthread_create_2_1): Enable stdio locks.
* libio/iofopncook.c (_IO_fopencookie): Enable locking for the file.
* sysdeps/pthread/flockfile.c (__flockfile): Likewise.
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 optimizes the generic spinlock code.
The type pthread_spinlock_t is a typedef to volatile int on all archs.
Passing a volatile pointer to the atomic macros which are not mapped to the
C11 atomic builtins can lead to extra stores and loads to stack if such
a macro creates a temporary variable by using "__typeof (*(mem)) tmp;".
Thus, those macros which are used by spinlock code - atomic_exchange_acquire,
atomic_load_relaxed, atomic_compare_exchange_weak - have to be adjusted.
According to the comment from Szabolcs Nagy, the type of a cast expression is
unqualified (see http://www.open-std.org/jtc1/sc22/wg14/www/docs/dr_423.htm):
__typeof ((__typeof (*(mem)) *(mem)) tmp;
Thus from spinlock perspective the variable tmp is of type int instead of
type volatile int. This patch adjusts those macros in include/atomic.h.
With this construct GCC >= 5 omits the extra stores and loads.
The atomic macros are replaced by the C11 like atomic macros and thus
the code is aligned to it. The pthread_spin_unlock implementation is now
using release memory order instead of sequentially consistent memory order.
The issue with passed volatile int pointers applies to the C11 like atomic
macros as well as the ones used before.
I've added a glibc_likely hint to the first atomic exchange in
pthread_spin_lock in order to return immediately to the caller if the lock is
free. Without the hint, there is an additional jump if the lock is free.
I've added the atomic_spin_nop macro within the loop of plain reads.
The plain reads are also realized by C11 like atomic_load_relaxed macro.
The new define ATOMIC_EXCHANGE_USES_CAS determines if the first try to acquire
the spinlock in pthread_spin_lock or pthread_spin_trylock is an exchange
or a CAS. This is defined in atomic-machine.h for all architectures.
The define SPIN_LOCK_READS_BETWEEN_CMPXCHG is now removed.
There is no technical reason for throwing in a CAS every now and then,
and so far we have no evidence that it can improve performance.
If that would be the case, we have to adjust other spin-waiting loops
elsewhere, too! Using a CAS loop without plain reads is not a good idea
on many targets and wasn't used by one. Thus there is now no option to
do so.
Architectures are now using the generic spinlock automatically if they
do not provide an own implementation. Thus the pthread_spin_lock.c files
in sysdeps folder are deleted.
ChangeLog:
* NEWS: Mention new spinlock implementation.
* include/atomic.h:
(__atomic_val_bysize): Cast type to omit volatile qualifier.
(atomic_exchange_acq): Likewise.
(atomic_load_relaxed): Likewise.
(ATOMIC_EXCHANGE_USES_CAS): Check definition.
* nptl/pthread_spin_init.c (pthread_spin_init):
Use atomic_store_relaxed.
* nptl/pthread_spin_lock.c (pthread_spin_lock):
Use C11-like atomic macros.
* nptl/pthread_spin_trylock.c (pthread_spin_trylock):
Likewise.
* nptl/pthread_spin_unlock.c (pthread_spin_unlock):
Use atomic_store_release.
* sysdeps/aarch64/nptl/pthread_spin_lock.c: Delete File.
* sysdeps/arm/nptl/pthread_spin_lock.c: Likewise.
* sysdeps/hppa/nptl/pthread_spin_lock.c: Likewise.
* sysdeps/m68k/nptl/pthread_spin_lock.c: Likewise.
* sysdeps/microblaze/nptl/pthread_spin_lock.c: Likewise.
* sysdeps/mips/nptl/pthread_spin_lock.c: Likewise.
* sysdeps/nios2/nptl/pthread_spin_lock.c: Likewise.
* sysdeps/aarch64/atomic-machine.h (ATOMIC_EXCHANGE_USES_CAS): Define.
* sysdeps/alpha/atomic-machine.h: Likewise.
* sysdeps/arm/atomic-machine.h: Likewise.
* sysdeps/i386/atomic-machine.h: Likewise.
* sysdeps/ia64/atomic-machine.h: Likewise.
* sysdeps/m68k/coldfire/atomic-machine.h: Likewise.
* sysdeps/m68k/m680x0/m68020/atomic-machine.h: Likewise.
* sysdeps/microblaze/atomic-machine.h: Likewise.
* sysdeps/mips/atomic-machine.h: Likewise.
* sysdeps/powerpc/powerpc32/atomic-machine.h: Likewise.
* sysdeps/powerpc/powerpc64/atomic-machine.h: Likewise.
* sysdeps/s390/atomic-machine.h: Likewise.
* sysdeps/sparc/sparc32/atomic-machine.h: Likewise.
* sysdeps/sparc/sparc32/sparcv9/atomic-machine.h: Likewise.
* sysdeps/sparc/sparc64/atomic-machine.h: Likewise.
* sysdeps/tile/tilegx/atomic-machine.h: Likewise.
* sysdeps/tile/tilepro/atomic-machine.h: Likewise.
* sysdeps/unix/sysv/linux/hppa/atomic-machine.h: Likewise.
* sysdeps/unix/sysv/linux/m68k/coldfire/atomic-machine.h: Likewise.
* sysdeps/unix/sysv/linux/nios2/atomic-machine.h: Likewise.
* sysdeps/unix/sysv/linux/sh/atomic-machine.h: Likewise.
* sysdeps/x86_64/atomic-machine.h: Likewise.
This patch adds support of preadv2 and pwritev2 which are similar to
preadv/pwritev but with an extra flag argument. As for preadv/pwritev
both interfaces are added a non-standard GNU API.
For default 'posix' implementation trying to emulate the Linux supported
flags is troublesome:
* We can not temporary change the file state of the O_DSYNC and O_SYNC
flags to emulate RWF_{D}SYNC (attempts to change the state of using
fcntl are silently ignored).
* IOCB_HIPRI requires the file opened in O_DIRECT and uses an internal
semantic not provided by any other flag (O_NONBLOCK for instance).
So default sysdeps/posix implementations fails with EOPNOTSUPP for any non
supported flag (which are none currently) calls generic preadv/pwritev.
Basically this implementation supports only preadv2 called as preadv (with
flags sets to 0).
The Linux one uses the preadv2/pwritev2 syscall if defined, otherwise it
call preadv/writev. Instead of using the previous __ASSUME_* to
unconditionally issue the syscall (and avoid building the fallback routine),
it call pread/write if the preadv2/pwritev2 syscalls fails. The idea
is just avoid adding another __ASSUME_* and checking each architecture
on every kernel bump and simplify code conditionals.
Checked on x86_64-linux-gnu and on i686-linux-gnu and a check with
run-built-tests=no on aarch64-linux-gnu, alpha-linux-gnu, arm-linux-gnueabihf,
ia64-linux-gnu, m68k-linux-gnu, microblaze-linux-gnu, mips{64,64n32}-linux-gnu,
nios2-linux-gnu, powerpc{64,64le}-linux-gnu, s390{x}-linux-gnu,
sparc{64,v9}-linux-gnu, tile{gx,pro}-linux-gnu, and sh4-linux-gnu (all using
gcc 6.3).
* NEWS: Add note about pwritev2 and preadv2 inclusion.
* misc/Makefile (routines): Add preadv2, preadv64v2, pwritev2, and
pwritev64v2.
(tests): Add tst-preadvwritev2 and tst-preadvwritev64v2.
* misc/Versions (GLIBC_2.26): Add preadv2, preadv64v2, pwritev2, and
pwritev64v2.
* misc/preadv2.c: New file.
* misc/preadv64v2.c: Likewise.
* misc/pwritev2.c: Likewise.
* misc/pwritev64v2.c: Likewise.
* misc/tst-preadvwritev2.c: Likewise.
* misc/tst-preadvwritev64v2.c: Likewise.
* manual/llio.texi: Add preadv2 and pwritev2 documentation.
* misc/sys/uio.h [__USE_GNU && !__USE_FILE_OFFSET64] (preadv2): New
prototype.
[__USE_GNU && !__USE_FILE_OFFSET64] (pwritev2): Likewise.
[__USE_GNU && __USE_FILE_OFFSET64] (preadv64v2): Likewise.
[__USE_GNU && __USE_FILE_OFFSET64] (pwritev64v2): Likewise.
* misc/tst-preadvwritev-common.c (PREADV): Define if not defined.
(PWRITEV): Likewise.
(do_test_with_offset): Use PREADV and PWRITEV macros and check for
ENOSYS.
* nptl/tst-cancel4.c (tf_pwritev2): New test.
(tf_preadv2): Likewise.
(tf_fsync): Add tf_pwritev2 and tf_preadv2.
* sysdeps/posix/preadv2.c: Likewise.
* sysdeps/posix/preadv64v2.c: Likewise.
* sysdeps/posix/pwritev2.c: Likewise.
* sysdeps/posix/pwritev64v2.c: Likewise.
* sysdeps/unix/sysv/linux/kernel-features.h: Add comment for syscall
support in kernel.
* sysdeps/unix/sysv/linux/preadv2.c: Likewise.
* sysdeps/unix/sysv/linux/preadv64v2.c: Likewise.
* sysdeps/unix/sysv/linux/pwritev2.c: Likewise.
* sysdeps/unix/sysv/linux/pwritev64v2.c: Likewise.
* sysdeps/unix/sysv/linux/preadv.c (preadv): Add libc_hidden_def.
* sysdeps/unix/sysv/linux/preadv64.c (preadv64): Likewise.
* sysdeps/unix/sysv/linux/pwritev.c (pwritev): Likewise.
* sysdeps/unix/sysv/linux/pwritev64.c (pwritev64): Likewise.
* sysdeps/unix/sysv/linux/bits/uio.h: Add supported preadv2/pwritev2
support flags on Linux.
* sysdeps/unix/sysv/linux/aarch64/libc.abilist (GLIBC_2.26): Add
preadv2, preadv64v2, pwritev2, pwritev64v2.
* sysdeps/unix/sysv/linux/alpha/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/arm/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/hppa/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/i386/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/ia64/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/m68k/coldfire/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/m68k/m680x0/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/microblaze/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/fpu/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/nofpu/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/mips/mips64/n32/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/mips/mips64/n64/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/nios2/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libc.abilist
(GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libc.abilist
(GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/powerpc/powerpc64/libc-le.abilist
(GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/s390/s390-64/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/sh/libc.abilist (GLIBC_2.26): Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc32/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/sparc/sparc64/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/sysv/linux/tile/tilegx/tilegx32/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/sysv/linux/tile/tilegx/tilegx64/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/tile/tilepro/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/x86_64/64/libc.abilist (GLIBC_2.26):
Likewise.
* sysdeps/unix/sysv/linux/x86_64/x32/libc.abilist (GLIBC_2.26):
Likewise.
