Before aosp/1086558, this was needed, since we would have
_USING_LIBCXX defined, yet <atomic> was not available, so we'd run
into compiler issues. After that change, we're using
__has_include(<atomic>) so this is no longer an issue.
Test: build
Change-Id: Ia81d343000e7d90f07422120fc1e1eb5becb01aa
pthread's and semaphore's default behavior is to use CLOCK_REALTIME,
however this behavior is essentially never intended, as that clock is
prone to change discontinuously.
What users really intend is to use CLOCK_MONOTONIC, however only
pthread_cond_timedwait() provides this as an option and even there, a
large amount of existing code does not opt into CLOCK_MONOTONIC.
We have seen numerous bugs directly attributable to this difference.
Therefore, we provide this general workaround to always use
CLOCK_MONOTONIC for waiting, regardless of what the input timespec
is.
Specifically this impacts the below APIs:
pthread_mutex_timedlock()
pthread_cond_timedwait()
pthread_rwlock_timedrdlock()
pthread_rwlock_timedwrlock()
sem_timedwait()
Test: boot bullhead, boot sailfish
Test: bionic pthread/semaphore unit tests
Test: check that pthread_cond_timedwait() timeouts are uneffected by
CLOCK_REALTIME time changes
Bug: 64694413
Bug: 64623895
Bug: 35756266
Bug: 35678943
Change-Id: Ibba98f5d88be1c306d14e9b9366302ecbef6d534
This seemed to be the only place in bionic where a fence on a
performance-critical path could be easily replaced by a stronger
load/store order constraint. Do so.
On x86 this should generate the same code either way. Based on
microbenchmarks of the relevant ARM instructions, this is currently
performance-neutral in this kind of context. But in the future, the
newly generated acquire loads should give us a performance benefit.
Test: Booted AOSP
Change-Id: I7823e11d6ae4fd58e0425244c293262e2320fd81
The Android build system always links against libstdc++.so anyway. Having
operator new and operator delete in a separate library means we can't use
constructors and destructors on heap-allocated objects inside the C library,
which is quite an unfortunate limitation.
This will be cheaper too; on LP64 we can stop linking against the [now empty]
libstdc++.so giving the dynamic linker one less library to worry about for
every process.
There's precedent too --- we already have no libpthread or librt.
For now I'm leaving the include files where they are, and I'm generating a
dummy libstdc++.so and libstdc++.a. We can come back and clean that up later
if all goes well.
Bug: 13367666
Change-Id: I6f3e27ea7c30d03d6394965d0400c9dc87fa83db
2014-05-14 18:18:55 -07:00
Renamed from libstdc++/src/one_time_construction.cpp (Browse further)
