As a follow up to Ibba98f5d88be1c306d14e9b9366302ecbef6d534, where we
added a work around to convert the CLOCK_REALTIME timeouts to
CLOCK_MONOTONIC for pthread and semaphore timed wait functions, we're
introducing a set of _monotonic_np versions of each of these functions
that wait on CLOCK_MONOTONIC directly.
The primary motivation here is that while the above work around helps
for 3rd party code, it creates a dilemma when implementing new code
that would use these functions: either one implements code with these
functions knowing there is a race condition possible or one avoids
these functions and reinvent their own waiting/signaling mechanisms.
Neither are satisfactory, so we create a third option to use these
Android specific _monotonic_np functions that completely remove the
race condition while keeping the rest of the interface.
Specifically this adds the below functions:
pthread_mutex_timedlock_monotonic_np()
pthread_cond_timedwait_monotonic_np()
pthread_rwlock_timedrdlock_monotonic_np()
pthread_rwlock_timedwrlock_monotonic_np()
sem_timedwait_monotonic_np()
Note that pthread_cond_timedwait_monotonic_np() previously existed and
was removed since it's possible to initialize a condition variable to
use CLOCK_MONOTONIC. It is added back for a mix of reasons,
1) Symmetry with the rest of the functions we're adding
2) libc++ cannot easily take advantage of the new initializer, but
will be able to use this function in order to wait on
std::steady_clock
3) Frankly, it's a better API to specify the clock in the waiter function
than to specify the clock when the condition variable is
initialized.
Bug: 73951740
Test: new unit tests
Change-Id: I23aa5c204e36a194237d41e064c5c8ccaa4204e3
For apps built for Android < P, return EBUSY.
For apps built for Android >= P, abort.
This is to keep old apps work, and help debugging
apps built for >= P.
Bug: http://b/74632097
Test: run bionic-unit-tests.
Test: run bionic-benchmark.
Change-Id: I5271565a1a6ad12678f85d558a7f862a2b7aab4b
Add fast path calling PIMutexTryLock() in pthread_mutex_lock.
Add trace for pi mutex waiting.
Bug: http://b/29177606
Test: run bionic-unit-tests.
Test: run bionic-benchmarks.
Change-Id: I30b6436692d5ea6b63ca9905df745edb843b5528
Bug: http://b/29177606
Test: run bionic-unit-tests on walleye.
Test: run bionic-unit-tests-glibc on host.
Change-Id: Iac349284aa73515f384e7509445f87434757f59e
http://clang.llvm.org/docs/AttributeReference.html#nonnull
_Nonnull is similar to the nonnull attribute in that it will instruct
compilers to warn the user if it can prove that a null argument is
being passed. Unlike the nonnull attribute, this annotation indicated
that a value *should not* be null, not that it *cannot* be null, or
even that the behavior is undefined. The important distinction is that
the optimizer will perform surprising optimizations like the
following:
void foo(void*) __attribute__(nonnull, 1);
int bar(int* p) {
foo(p);
// The following null check will be elided because nonnull
// attribute means that, since we call foo with p, p can be
// assumed to not be null. Thus this will crash if we are called
// with a null pointer.
if (src != NULL) {
return *p;
}
return 0;
}
int main() {
return bar(NULL);
}
Note that by doing this we are no longer attaching any sort of
attribute for GCC (GCC doesn't support attaching nonnull directly to a
parameter, only to the function and naming the arguments
positionally). This means we won't be getting a warning for this case
from GCC any more. People that listen to warnings tend to use clang
anyway, and we're quickly moving toward that as the default, so this
seems to be an acceptable tradeoff.
Change-Id: Ie05fe7cec2f19a082c1defb303f82bcf9241b88d
It actually means "crash immediately". Well, it's an error. And callers are
much more likely to realize their mistake if we crash immediately rather
than return EINVAL. Historically, glibc has crashed and bionic -- before
the recent changes -- returned EINVAL, so this is a behavior change.
Change-Id: I0c2373a6703b20b8a97aacc1e66368a5885e8c51
If calling pthread_mutex_trylock from pthread_mutex_destroy, tsan
warns about an attempt to destroy a locked mutex.
Bug: 25392375
Change-Id: I5feee20e7a0d0915adad24da874ec1ccce241381
The pthread_mutex_lock and pthread_mutex_unlock were allowed to
fail silently on L 32 bit devices when passed a NULL. We changed
this to a crash on 32 bit devices, but there are still games that make
these calls and are not likely to be updated. Therefore, once again
allow NULL to be passed in on 32 bit devices.
Bug: 19995172
Change-Id: If7e8860075ecd63c0064d80f64e226fad7bd3c26
This library calls pthread_mutex_lock and pthread_mutex_unlock with a NULL
pthread_mutex_t*. This gives them (and their users) one release to fix things.
Bug: 17443936
Change-Id: I3b63c9a3dd63db0833f21073e323b3236a13b47a
bionic_systrace.h contains an implementation of tracing that
can be used with systrace.py and its associated viewer. pthread_mutex
now uses this tracing to track pthread_mutex contention, which can be
enabled by using the "bionic" command line option to systrace.
Bug: 15116468
Change-Id: I30ed5b377c91ca4c36568a0e647ddf95d4e4a61a
The comment used the phrase "normal case" when it more specifically refers to the "recursive case," so I changed it to that.
Change-Id: I8335cce4dee933c6a463aee653b28bd986b5b5e4
This hasn't built in over one release cycle and no one even noticed.
art does this the right way and other projects should do the same.
Change-Id: I7d1fb84c4080e008f329ee73e209ce85a36e6d55
GCC is removing these checks anyway because it knows the arguments
must be non-null, so leaving this code around is just confusing.
We know from experience that people were shipping code with locking
bugs because they weren't checking for error returns. Failing hard
like glibc does seems the better choice. (And it's what the checked
in code was already doing; this patch doesn't change that. It just
makes it more obvious that that's what's going on.)
Change-Id: I167c6d7c0a296822baf0cb9b43b97821eba7ab35
This replaces the non-standard pthread_mutex_lock_timeout_np, which we have
to keep around on LP32 for binary compatibility.
Change-Id: I098dc7cd38369f0c1bec1fac35687fbd27392e00
Previously we were checking against a positive errno which
would not be returned from a system call.
Change-Id: I8e3a36f6fbf5ccc2191a152a1def37e2d6f93124
Most callers won't check for EINVAL, so it's best to fail early.
GCC takes the nonnull attribute as a guarantee that an argument
won't be NULL, so these hacks were already ineffective, which is
how we found that at least one commercial game was using NULL
as if it's a mutex, but actually getting no-op behavior.
Bug: 11971278
Change-Id: I89646e043d931778805a8b692e07a34d076ee6bf
The kernel now maintains the pthread_internal_t::tid field for us,
and __clone was only used in one place so let's inline it so we don't
have to leave such a dangerous function lying around. Also rename
files to match their content and remove some useless #includes.
Change-Id: I24299fb4a940e394de75f864ee36fdabbd9438f9
<pthread.h> was missing nonnull attributes, noreturn on pthread_exit,
and had incorrect cv qualifiers for several standard functions.
I've also marked the non-standard stuff (where I count glibc rather
than POSIX as "standard") so we can revisit this cruft for LP64 and
try to ensure we're compatible with glibc.
I've also broken out the pthread_cond* functions into a new file.
I've made the remaining pthread files (plus ptrace) part of the bionic code
and fixed all the warnings.
I've added a few more smoke tests for chunks of untested pthread functionality.
We no longer need the libc_static_common_src_files hack for any of the
pthread implementation because we long since stripped out the rest of
the armv5 support, and this hack was just to ensure that __get_tls in libc.a
went via the kernel if necessary.
This patch also finishes the job of breaking up the pthread.c monolith, and
adds a handful of new tests.
Change-Id: Idc0ae7f5d8aa65989598acd4c01a874fe21582c7
