OSDN Git Service

android-x86/bionic.git
10 years agoHide dlmalloc symbols.
Elliott Hughes [Wed, 4 Jun 2014 01:47:17 +0000 (18:47 -0700)]
Hide dlmalloc symbols.

We need to leave dlmalloc_trim and dlmalloc_inspect_all exposed for
the VM, but if we're seriously looking at other malloc implementations,
that's something we're going to have to fix.

Bug: 11156955
Change-Id: If85156c280044f1616c09a3c50ba674aaf0e8d3a

10 years agoMerge "Work around "error: variable 'uio' set but not used"."
Elliott Hughes [Tue, 3 Jun 2014 23:41:35 +0000 (23:41 +0000)]
Merge "Work around "error: variable 'uio' set but not used"."

10 years agoWork around "error: variable 'uio' set but not used".
Elliott Hughes [Tue, 3 Jun 2014 23:39:58 +0000 (16:39 -0700)]
Work around "error: variable 'uio' set but not used".

I've sent the real fix upstream, and will merge that when they commit it.

Change-Id: I75e5cab87026f8990c585d364dd178753d727e2a

10 years agoMerge "Remove an unused parameter from the dlopen implementation."
Elliott Hughes [Tue, 3 Jun 2014 23:36:35 +0000 (23:36 +0000)]
Merge "Remove an unused parameter from the dlopen implementation."

10 years agoRemove an unused parameter from the dlopen implementation.
Elliott Hughes [Tue, 3 Jun 2014 23:23:08 +0000 (16:23 -0700)]
Remove an unused parameter from the dlopen implementation.

Change-Id: I9297913a743bb570b48ebbe594ff711a481e89e0

10 years agoMerge "Turn on -Wunused and fix the mistakes it uncovers."
Elliott Hughes [Tue, 3 Jun 2014 23:14:20 +0000 (23:14 +0000)]
Merge "Turn on -Wunused and fix the mistakes it uncovers."

10 years agoTurn on -Wunused and fix the mistakes it uncovers.
Elliott Hughes [Tue, 3 Jun 2014 22:22:34 +0000 (15:22 -0700)]
Turn on -Wunused and fix the mistakes it uncovers.

Change-Id: I023d2d8b547fbc21d4124bb7510d42b06a0dc501

10 years agoMerge "Use __libc_fatal() for failed malloc in new"
Dan Albert [Tue, 3 Jun 2014 16:53:46 +0000 (16:53 +0000)]
Merge "Use __libc_fatal() for failed malloc in new"

10 years agoUse __libc_fatal() for failed malloc in new
Dan Albert [Fri, 30 May 2014 23:00:53 +0000 (16:00 -0700)]
Use __libc_fatal() for failed malloc in new

This way we can print a useful message to the log isntead of just dying
mysteriously.

Change-Id: Ib660c2fd8ce8bb9aa0d0bb634ae08c645d3901e5

10 years agoam 2deef73f: Merge "Avoid a system call in \'gettid\'."
Elliott Hughes [Mon, 2 Jun 2014 17:48:53 +0000 (17:48 +0000)]
am 2deef73f: Merge "Avoid a system call in \'gettid\'."

* commit '2deef73fe1e9dbf57b683a15fd1e0c2922e6ecc0':
  Avoid a system call in 'gettid'.

10 years agoMerge "Avoid a system call in 'gettid'."
Elliott Hughes [Mon, 2 Jun 2014 17:43:27 +0000 (17:43 +0000)]
Merge "Avoid a system call in 'gettid'."

10 years agoAvoid a system call in 'gettid'.
Elliott Hughes [Sat, 31 May 2014 02:00:03 +0000 (19:00 -0700)]
Avoid a system call in 'gettid'.

System calls can be pretty slow. This is mako, which has one of our
lowest latencies:

                       iterations      ns/op
  BM_unistd_getpid       10000000        209
  BM_unistd_gettid      200000000          8

Bug: 15297299 (kernel panic from too many gettid calls)
Bug: 15315766 (excessive gettid overhead in liblogd)
Change-Id: I49656c0fc5b5d092390264a59e4f2c0d8a8b1aeb

10 years agoam 831405b7: Merge "Make atexit.dlclose work for glibc"
Dmitriy Ivanov [Sat, 31 May 2014 01:31:26 +0000 (01:31 +0000)]
am 831405b7: Merge "Make atexit.dlclose work for glibc"

* commit '831405b749d15a11fb947a40d61fd858e952d860':
  Make atexit.dlclose work for glibc

10 years agoMerge "Make atexit.dlclose work for glibc"
Dmitriy Ivanov [Sat, 31 May 2014 01:28:32 +0000 (01:28 +0000)]
Merge "Make atexit.dlclose work for glibc"

10 years agoMake atexit.dlclose work for glibc
Dmitriy Ivanov [Sat, 31 May 2014 01:03:26 +0000 (18:03 -0700)]
Make atexit.dlclose work for glibc

 * build libtest_atexit.so
 * remove atexit(NULL) - glibc segfaults on it

Bug: 15350494
Change-Id: I27d79130c32c5691b0e8790a57d92f9710f5bf4a

10 years agoam 66b5e6b6: Merge "Do not run dlclose test on glibc."
Christopher Ferris [Sat, 31 May 2014 00:43:52 +0000 (00:43 +0000)]
am 66b5e6b6: Merge "Do not run dlclose test on glibc."

* commit '66b5e6b609f16062262b777905bdd2305a941596':
  Do not run dlclose test on glibc.

10 years agoMerge "Do not run dlclose test on glibc."
Christopher Ferris [Sat, 31 May 2014 00:41:37 +0000 (00:41 +0000)]
Merge "Do not run dlclose test on glibc."

10 years agoam 7a5732bd: Merge "Add some missing __noreturn attributes."
Elliott Hughes [Fri, 30 May 2014 23:01:01 +0000 (23:01 +0000)]
am 7a5732bd: Merge "Add some missing __noreturn attributes."

* commit '7a5732bd83c64489b9bd953f49fb6ad3b420126c':
  Add some missing __noreturn attributes.

10 years agoMerge "Add some missing __noreturn attributes."
Elliott Hughes [Fri, 30 May 2014 22:58:41 +0000 (22:58 +0000)]
Merge "Add some missing __noreturn attributes."

10 years agoAdd some missing __noreturn attributes.
Elliott Hughes [Fri, 30 May 2014 21:49:00 +0000 (14:49 -0700)]
Add some missing __noreturn attributes.

This is more honest, and lets us remove a hack.

Change-Id: I309f064743ded5248573ccafc379f78d417db08d

10 years agoam 3a3b27e0: Merge "x86: stop unwinding past __bionic_clone."
Elliott Hughes [Fri, 30 May 2014 21:43:47 +0000 (21:43 +0000)]
am 3a3b27e0: Merge "x86: stop unwinding past __bionic_clone."

* commit '3a3b27e04df395b789b92af3d0ed05d2a0f3bf04':
  x86: stop unwinding past __bionic_clone.

10 years agoMerge "x86: stop unwinding past __bionic_clone."
Elliott Hughes [Fri, 30 May 2014 21:40:28 +0000 (21:40 +0000)]
Merge "x86: stop unwinding past __bionic_clone."

10 years agoDo not run dlclose test on glibc.
Christopher Ferris [Fri, 30 May 2014 18:23:21 +0000 (11:23 -0700)]
Do not run dlclose test on glibc.

The support library for this test is not built for host by default.
Even if the support library is built, the test segfaults on glibc.

Change-Id: I9cb7a364c59b55d4bf5d8634293037cd9bae020b

10 years agox86: stop unwinding past __bionic_clone.
Elliott Hughes [Fri, 30 May 2014 18:15:32 +0000 (11:15 -0700)]
x86: stop unwinding past __bionic_clone.

The apparent "previous" frame is on another thread's stack.

Bug: 14270816
Bug: 15195760
Change-Id: I884f370e116203c5fa3b05a75fb8f356537261c4

10 years agoam 1672b056: Merge "Fix unwinding_through_signal_frame test."
Christopher Ferris [Fri, 30 May 2014 03:55:51 +0000 (03:55 +0000)]
am 1672b056: Merge "Fix unwinding_through_signal_frame test."

* commit '1672b0563693eaaf92db4f304308d456a6102e64':
  Fix unwinding_through_signal_frame test.

10 years agoMerge "Fix unwinding_through_signal_frame test."
Christopher Ferris [Fri, 30 May 2014 03:53:00 +0000 (03:53 +0000)]
Merge "Fix unwinding_through_signal_frame test."

10 years agoFix unwinding_through_signal_frame test.
Christopher Ferris [Fri, 30 May 2014 03:00:12 +0000 (20:00 -0700)]
Fix unwinding_through_signal_frame test.

I accidentally removed the compilcation of the test implementation file
with special flags needed for the test to work. This change creates the
impl as a library with those flags back.

Bug: 14819262
Change-Id: Ib84fd26a7f4d40a0267d3ed686185b0abc5a3706

10 years agoam 94170742: Merge "Fix x86 cfi directives for syscalls."
Christopher Ferris [Fri, 30 May 2014 02:22:34 +0000 (02:22 +0000)]
am 94170742: Merge "Fix x86 cfi directives for syscalls."

* commit '941707429d8577e90580277c51ad8a354f664586':
  Fix x86 cfi directives for syscalls.

10 years agoMerge "Fix x86 cfi directives for syscalls."
Christopher Ferris [Fri, 30 May 2014 02:20:27 +0000 (02:20 +0000)]
Merge "Fix x86 cfi directives for syscalls."

10 years agoFix x86 cfi directives for syscalls.
Christopher Ferris [Fri, 30 May 2014 01:17:09 +0000 (18:17 -0700)]
Fix x86 cfi directives for syscalls.

The syscall generation always used 4 bytes for each push cfi directive.
However, the first push should always use an offset of 8 bytes, each
subsequent push after that is only 4 bytes though.

Change-Id: Ibaabd107f399ef67010b9a08213783957c2f74a9

10 years agoam 264d1b83: Merge "Remove issetugid."
Elliott Hughes [Fri, 30 May 2014 00:10:42 +0000 (00:10 +0000)]
am 264d1b83: Merge "Remove issetugid."

* commit '264d1b832510b746c1d45b5efaa33c164d6f8b2b':
  Remove issetugid.

10 years agoMerge "Remove issetugid."
Elliott Hughes [Fri, 30 May 2014 00:07:52 +0000 (00:07 +0000)]
Merge "Remove issetugid."

10 years agoam 5800aab3: Merge "Fix wmemmove test."
Christopher Ferris [Thu, 29 May 2014 23:43:06 +0000 (23:43 +0000)]
am 5800aab3: Merge "Fix wmemmove test."

* commit '5800aab347a3764f5edb3db0da83b7400ed8426e':
  Fix wmemmove test.

10 years agoRemove issetugid.
Elliott Hughes [Thu, 29 May 2014 23:04:53 +0000 (16:04 -0700)]
Remove issetugid.

Bug: 14569474
Change-Id: I752c77ed53c724f144f0b618443eb95e87f2929a

10 years agoMerge "Fix wmemmove test."
Christopher Ferris [Thu, 29 May 2014 23:40:39 +0000 (23:40 +0000)]
Merge "Fix wmemmove test."

10 years agoam fd9f07a2: Merge "Workaround the GCC "(long)fn -> lfn" optimization bug."
Ben Cheng [Thu, 29 May 2014 23:38:27 +0000 (23:38 +0000)]
am fd9f07a2: Merge "Workaround the GCC "(long)fn -> lfn" optimization bug."

* commit 'fd9f07a2a2aa6c17b99c4bf9815e380eeac6aa0f':
  Workaround the GCC "(long)fn -> lfn" optimization bug.

10 years agoMerge "Workaround the GCC "(long)fn -> lfn" optimization bug."
Ben Cheng [Thu, 29 May 2014 23:33:04 +0000 (23:33 +0000)]
Merge "Workaround the GCC "(long)fn -> lfn" optimization bug."

10 years agoWorkaround the GCC "(long)fn -> lfn" optimization bug.
Ben Cheng [Thu, 29 May 2014 23:02:09 +0000 (16:02 -0700)]
Workaround the GCC "(long)fn -> lfn" optimization bug.

Pass "-fno-builtin-rint -fno-builtin-rintl -fno-builtin-rintf" to avoid
generating self recursions in lrint, lrintl, and lrintf.

BUG: 14225968
Change-Id: Iffc4d8f0692580d636a529f4c8aa9a7a10d7acd8

10 years agoFix wmemmove test.
Christopher Ferris [Thu, 29 May 2014 22:42:42 +0000 (15:42 -0700)]
Fix wmemmove test.

I accidentally copied over the nul terminator with the test.

Change-Id: I24a9aa05d4fba4f383fa38a3041bb6a6b179130c

10 years agoam d4fbb1a8: Merge "Fix dprintf test."
Christopher Ferris [Thu, 29 May 2014 22:44:25 +0000 (22:44 +0000)]
am d4fbb1a8: Merge "Fix dprintf test."

* commit 'd4fbb1a8617285ab1d29324308aa016377fc36f3':
  Fix dprintf test.

10 years agoMerge "Fix dprintf test."
Christopher Ferris [Thu, 29 May 2014 22:41:50 +0000 (22:41 +0000)]
Merge "Fix dprintf test."

10 years agoam aa5f32ea: Merge "Add Aarch64 optimized bzero based on memset"
Christopher Ferris [Thu, 29 May 2014 21:23:16 +0000 (21:23 +0000)]
am aa5f32ea: Merge "Add Aarch64 optimized bzero based on memset"

* commit 'aa5f32ea1c8af9776a1c26a027495954c835be55':
  Add Aarch64 optimized bzero based on memset

10 years agoMerge "Add Aarch64 optimized bzero based on memset"
Christopher Ferris [Thu, 29 May 2014 21:19:44 +0000 (21:19 +0000)]
Merge "Add Aarch64 optimized bzero based on memset"

10 years agoAdd Aarch64 optimized bzero based on memset
Bernhard Rosenkraenzer [Mon, 19 May 2014 11:16:41 +0000 (13:16 +0200)]
Add Aarch64 optimized bzero based on memset

Change-Id: I62cf25bfe23b9d811e00af7307bbd19d89937792
Signed-off-by: Bernhard Rosenkraenzer <Bernhard.Rosenkranzer@linaro.org>
10 years agoFix dprintf test.
Christopher Ferris [Thu, 29 May 2014 19:49:35 +0000 (12:49 -0700)]
Fix dprintf test.

Change-Id: Ied783d545b6a35a70a158249f3957a9d00971627

10 years agoam 4dd0b72d: Merge "Return the actual success result from __futex."
Elliott Hughes [Thu, 29 May 2014 20:14:35 +0000 (20:14 +0000)]
am 4dd0b72d: Merge "Return the actual success result from __futex."

* commit '4dd0b72dd36f3d9afd6bf0c184866f0d9242084f':
  Return the actual success result from __futex.

10 years agoMerge "Return the actual success result from __futex."
Elliott Hughes [Thu, 29 May 2014 20:12:06 +0000 (20:12 +0000)]
Merge "Return the actual success result from __futex."

10 years agoam 1a57de71: Merge "Fix overflow testing in sbrk."
Christopher Ferris [Thu, 29 May 2014 18:06:46 +0000 (18:06 +0000)]
am 1a57de71: Merge "Fix overflow testing in sbrk."

* commit '1a57de713adfb81842cb2ebdfc051ecb9afa55f1':
  Fix overflow testing in sbrk.

10 years agoMerge "Fix overflow testing in sbrk."
Christopher Ferris [Thu, 29 May 2014 18:03:27 +0000 (18:03 +0000)]
Merge "Fix overflow testing in sbrk."

10 years agoFix overflow testing in sbrk.
Christopher Ferris [Thu, 22 May 2014 02:03:34 +0000 (19:03 -0700)]
Fix overflow testing in sbrk.

Modify the overflow testing for sbrk.

Bug: 15188366

Change-Id: Ia83f85f7c1789454d872279bd41f38f1ce6b8a34

10 years agoReturn the actual success result from __futex.
Elliott Hughes [Thu, 29 May 2014 17:27:55 +0000 (10:27 -0700)]
Return the actual success result from __futex.

futex(2) can return non-zero successes.

Bug: 15195455
Change-Id: I7818bc922a5a2df31228ff72c169320b5e69a544

10 years agoam ef023873: Merge "Minor style cleanup of some code I had to look at."
Elliott Hughes [Thu, 29 May 2014 16:15:52 +0000 (16:15 +0000)]
am ef023873: Merge "Minor style cleanup of some code I had to look at."

* commit 'ef023873e2bbd83f4fde69fd427e04ef1eb6e769':
  Minor style cleanup of some code I had to look at.

10 years agoam 0bdd668f: Merge "Upgrade to tzdata2014d."
Elliott Hughes [Thu, 29 May 2014 16:15:51 +0000 (16:15 +0000)]
am 0bdd668f: Merge "Upgrade to tzdata2014d."

* commit '0bdd668f05b22b44547a2b9f05e506166b22f02a':
  Upgrade to tzdata2014d.

10 years agoMerge "Minor style cleanup of some code I had to look at."
Elliott Hughes [Thu, 29 May 2014 16:12:01 +0000 (16:12 +0000)]
Merge "Minor style cleanup of some code I had to look at."

10 years agoMerge "Upgrade to tzdata2014d."
Elliott Hughes [Thu, 29 May 2014 16:09:39 +0000 (16:09 +0000)]
Merge "Upgrade to tzdata2014d."

10 years agoUpgrade to tzdata2014d.
Elliott Hughes [Thu, 29 May 2014 03:33:33 +0000 (20:33 -0700)]
Upgrade to tzdata2014d.

There are no substantive changes to the data in this release. I'm applying
this just for completeness.

Change-Id: Ibe953969cb16db4d8d1ddb7dd3c943316e6e46b4

10 years agoMinor style cleanup of some code I had to look at.
Elliott Hughes [Thu, 29 May 2014 03:30:40 +0000 (20:30 -0700)]
Minor style cleanup of some code I had to look at.

(It turns out that this is the only place we're saying __inline in C++.)

Change-Id: I8095e67a385087817c47caab9a621f82f8e0cfc8

10 years agoam fae42a83: Merge "Revert "Revert "Lose the hand-written futex assembler."""
Elliott Hughes [Thu, 29 May 2014 03:16:18 +0000 (03:16 +0000)]
am fae42a83: Merge "Revert "Revert "Lose the hand-written futex assembler."""

* commit 'fae42a837ae7f880de9561c7b3c2b18e4a531e90':
  Revert "Revert "Lose the hand-written futex assembler.""

10 years agoMerge "Revert "Revert "Lose the hand-written futex assembler."""
Elliott Hughes [Thu, 29 May 2014 03:13:23 +0000 (03:13 +0000)]
Merge "Revert "Revert "Lose the hand-written futex assembler."""

10 years agoRevert "Revert "Lose the hand-written futex assembler.""
Elliott Hughes [Wed, 28 May 2014 19:35:33 +0000 (19:35 +0000)]
Revert "Revert "Lose the hand-written futex assembler.""

The problem with the original patch was that using syscall(3) means that
errno can be set, but pthread_create(3) was abusing the TLS errno slot as
a pthread_mutex_t for the thread startup handshake.

There was also a mistake in the check for syscall failures --- it should
have checked against -1 instead of 0 (not just because that's the default
idiom, but also here because futex(2) can legitimately return values > 0).

This patch stops abusing the TLS errno slot and adds a pthread_mutex_t to
pthread_internal_t instead. (Note that for LP64 sizeof(pthread_mutex_t) >
sizeof(uintptr_t), so we could potentially clobber other TLS slots too.)

I've also rewritten the LP32 compatibility stubs to directly reuse the
code from the .h file.

This reverts commit 75c55ff84ebfa686c7ae2cc8ee431c6a33bd46b4.

Bug: 15195455
Change-Id: I6ffb13e5cf6a35d8f59f692d94192aae9ab4593d

10 years agoam 32c8ae9d: Merge "Fixes stdatomic.h test to match C11 spec"
Dan Albert [Thu, 29 May 2014 00:52:18 +0000 (00:52 +0000)]
am 32c8ae9d: Merge "Fixes stdatomic.h test to match C11 spec"

* commit '32c8ae9de0e51ac64efcfd2e5da34382deee959e':
  Fixes stdatomic.h test to match C11 spec

10 years agoMerge "Fixes stdatomic.h test to match C11 spec"
Dan Albert [Thu, 29 May 2014 00:50:26 +0000 (00:50 +0000)]
Merge "Fixes stdatomic.h test to match C11 spec"

10 years agoFixes stdatomic.h test to match C11 spec
Dan Albert [Wed, 28 May 2014 23:27:32 +0000 (16:27 -0700)]
Fixes stdatomic.h test to match C11 spec

C11 defines the expected value to atomic_compare_exchange_* as being non-atomic
types. Using an atomic type is a syntax error in clang.

http://en.cppreference.com/w/c/atomic/atomic_compare_exchange

Change-Id: I74de1061fa1fc50d835451792d902000f368200e

10 years agoam dc3fb119: Merge "Updated ambiguous comment in pthread_mutex_lock."
Dan Albert [Wed, 28 May 2014 22:10:22 +0000 (22:10 +0000)]
am dc3fb119: Merge "Updated ambiguous comment in pthread_mutex_lock."

* commit 'dc3fb119497382fc8d9c3b87c5ef8321141330dc':
  Updated ambiguous comment in pthread_mutex_lock.

10 years agoMerge "Updated ambiguous comment in pthread_mutex_lock."
Dan Albert [Wed, 28 May 2014 22:00:50 +0000 (22:00 +0000)]
Merge "Updated ambiguous comment in pthread_mutex_lock."

10 years agoam 8a405b02: Merge "Add optimized AArch64 versions of bcopy and wmemmove based on...
Christopher Ferris [Wed, 28 May 2014 21:38:01 +0000 (21:38 +0000)]
am 8a405b02: Merge "Add optimized AArch64 versions of bcopy and wmemmove based on memmove"

* commit '8a405b027a9c8b3a614016f4824a03fa252e9e1e':
  Add optimized AArch64 versions of bcopy and wmemmove based on memmove

10 years agoMerge "Add optimized AArch64 versions of bcopy and wmemmove based on memmove"
Christopher Ferris [Wed, 28 May 2014 21:35:04 +0000 (21:35 +0000)]
Merge "Add optimized AArch64 versions of bcopy and wmemmove based on memmove"

10 years agoUpdated ambiguous comment in pthread_mutex_lock.
Brigid Smith [Wed, 28 May 2014 18:45:48 +0000 (11:45 -0700)]
Updated ambiguous comment in pthread_mutex_lock.

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

10 years agoam 52f74322: Merge "Revert "Lose the hand-written futex assembler.""
Narayan Kamath [Wed, 28 May 2014 18:06:05 +0000 (18:06 +0000)]
am 52f74322: Merge "Revert "Lose the hand-written futex assembler.""

* commit '52f74322b1d72d57146107f32ee2c76c421bf4b1':
  Revert "Lose the hand-written futex assembler."

10 years agoMerge "Revert "Lose the hand-written futex assembler.""
Narayan Kamath [Wed, 28 May 2014 18:04:26 +0000 (18:04 +0000)]
Merge "Revert "Lose the hand-written futex assembler.""

10 years agoRevert "Lose the hand-written futex assembler."
Narayan Kamath [Wed, 28 May 2014 18:02:33 +0000 (18:02 +0000)]
Revert "Lose the hand-written futex assembler."

This reverts commit ced906c849704f379d7191822f6d74993d4fa296.

Causes issues on art / dalvik due to a broken return value
check and other undiagnosed issues.

bug: 15195455

Change-Id: I5d6bbb389ecefb0e33a5237421a9d56d32a9317c

10 years agoam 53e70fe3: Merge "Remove unnecessary padding in ucontext_t"
Calin Juravle [Wed, 28 May 2014 17:50:43 +0000 (17:50 +0000)]
am 53e70fe3: Merge "Remove unnecessary padding in ucontext_t"

* commit '53e70fe30646453a5deba0730337cb9674cd6fc8':
  Remove unnecessary padding in ucontext_t

10 years agoMerge "Remove unnecessary padding in ucontext_t"
Calin Juravle [Wed, 28 May 2014 17:48:05 +0000 (17:48 +0000)]
Merge "Remove unnecessary padding in ucontext_t"

10 years agoRemove unnecessary padding in ucontext_t
Calin Juravle [Wed, 28 May 2014 14:56:46 +0000 (15:56 +0100)]
Remove unnecessary padding in ucontext_t

- removed extra padding from ucontex_t on x86, x86_64 and mips
- force ucontext_t struct to use 64 bits for the sigmask on all arches
(previsouly wrong since sigset_t on arm and x86 is only 32 bits)
- arm and arm64 continue to use padding to match glibc sigset_t

This brings Bionic on par with the kernel w.r.t ucontext padding.

Bug: 12828904
Change-Id: Ia8915ace694ecb4695603a334e697985f4c3e7aa

10 years agoam 7fdde4a5: Merge "Expose __srefill for LP64 until we can fix the NDK."
Elliott Hughes [Wed, 28 May 2014 05:22:38 +0000 (05:22 +0000)]
am 7fdde4a5: Merge "Expose __srefill for LP64 until we can fix the NDK."

* commit '7fdde4a58b304a7c4f87987f3fde16bdded63a76':
  Expose __srefill for LP64 until we can fix the NDK.

10 years agoMerge "Expose __srefill for LP64 until we can fix the NDK."
Elliott Hughes [Wed, 28 May 2014 05:19:45 +0000 (05:19 +0000)]
Merge "Expose __srefill for LP64 until we can fix the NDK."

10 years agoExpose __srefill for LP64 until we can fix the NDK.
Elliott Hughes [Wed, 28 May 2014 04:23:36 +0000 (21:23 -0700)]
Expose __srefill for LP64 until we can fix the NDK.

The LP64 has a duplicate copy of part of stdio, and relies
on bionic supplying this part. We should remove the hack from
the NDK, at least for LP64, and then revert this.

Bug: 15291317
Change-Id: I75e06e130188ca0aeb9d50dfe3a3e48a1d3968b7

10 years agoam 1454ff25: Merge "Clear the return address register in __bionic_clone."
Elliott Hughes [Tue, 27 May 2014 22:55:34 +0000 (22:55 +0000)]
am 1454ff25: Merge "Clear the return address register in __bionic_clone."

* commit '1454ff2574f0bee9fb3646c9c6dd30443e0dbd38':
  Clear the return address register in __bionic_clone.

10 years agoMerge "Clear the return address register in __bionic_clone."
Elliott Hughes [Tue, 27 May 2014 22:53:52 +0000 (22:53 +0000)]
Merge "Clear the return address register in __bionic_clone."

10 years agoClear the return address register in __bionic_clone.
Elliott Hughes [Tue, 27 May 2014 22:47:32 +0000 (15:47 -0700)]
Clear the return address register in __bionic_clone.

The unwinder will end up on the parent thread's stack if the return
address register is not cleared.

Bug: 14270816
Bug: 15195760
Change-Id: Iebcdf5cc881cad245643bb193760de35eb7f8b19

10 years agoam 134911d7: Merge "Use __typeof__() instead of typeof()"
Dan Albert [Tue, 27 May 2014 21:59:33 +0000 (21:59 +0000)]
am 134911d7: Merge "Use __typeof__() instead of typeof()"

* commit '134911d75deab2b5babc175c894c002e875dfed6':
  Use __typeof__() instead of typeof()

10 years agoMerge "Use __typeof__() instead of typeof()"
Dan Albert [Tue, 27 May 2014 21:57:24 +0000 (21:57 +0000)]
Merge "Use __typeof__() instead of typeof()"

10 years agoUse __typeof__() instead of typeof()
Dan Albert [Thu, 22 May 2014 03:33:28 +0000 (20:33 -0700)]
Use __typeof__() instead of typeof()

typeof() is a gcc builtin and is not always available (such as clang when
-std=c*). Both gcc and clang always expose __typeof__(), so just use that.

Change-Id: I75e2a990acc35f6f39b2c353f6548100ec03c67f

10 years agoam 8e85b4f0: Merge "Restore __srefill for LP32."
Elliott Hughes [Tue, 27 May 2014 18:41:52 +0000 (18:41 +0000)]
am 8e85b4f0: Merge "Restore __srefill for LP32."

* commit '8e85b4f00e8816f22bf5b5f20b1f978432f2ae08':
  Restore __srefill for LP32.

10 years agoMerge "Restore __srefill for LP32."
Elliott Hughes [Tue, 27 May 2014 18:38:56 +0000 (18:38 +0000)]
Merge "Restore __srefill for LP32."

10 years agoRestore __srefill for LP32.
Elliott Hughes [Tue, 27 May 2014 18:22:39 +0000 (11:22 -0700)]
Restore __srefill for LP32.

The NDK apparently includes an android_support.a library that
refers to __srefill in its copy of the vsnprintf implementation.

Bug: 15249361
Change-Id: Ic2cf6f21290b3146c42fbe0624f5e4d54f6194b4

10 years agoam b3c75a1e: Merge "Restore __srget and __swbuf for LP32."
Elliott Hughes [Mon, 26 May 2014 05:10:46 +0000 (05:10 +0000)]
am b3c75a1e: Merge "Restore __srget and __swbuf for LP32."

* commit 'b3c75a1e8ffbf575c2a425b34fd67ff8077da299':
  Restore __srget and __swbuf for LP32.

10 years agoMerge "Restore __srget and __swbuf for LP32."
Elliott Hughes [Mon, 26 May 2014 05:07:20 +0000 (05:07 +0000)]
Merge "Restore __srget and __swbuf for LP32."

10 years agoam bc904896: Merge "Add C11 <stdatomic.h>."
Elliott Hughes [Sun, 25 May 2014 17:51:05 +0000 (17:51 +0000)]
am bc904896: Merge "Add C11 <stdatomic.h>."

* commit 'bc904896a6f846e2648922d46324ea626ae9e1d1':
  Add C11 <stdatomic.h>.

10 years agoMerge "Add C11 <stdatomic.h>."
Elliott Hughes [Sun, 25 May 2014 17:48:18 +0000 (17:48 +0000)]
Merge "Add C11 <stdatomic.h>."

10 years agoRestore __srget and __swbuf for LP32.
Elliott Hughes [Sun, 25 May 2014 17:38:25 +0000 (10:38 -0700)]
Restore __srget and __swbuf for LP32.

Anthony King <anthonydking@slimroms.net> reports that for Grouper the
Nvidia GL blobs need access to __swbuf. This is because the old <stdio.h>
had inline getc and putc implementations that directly referred to these
symbols.

Change-Id: I11a7b5550018ecc93d8f195c99857759669b2906

10 years agoam 255789e4: Merge "Restore tkill for LP32."
Elliott Hughes [Sun, 25 May 2014 03:18:14 +0000 (03:18 +0000)]
am 255789e4: Merge "Restore tkill for LP32."

* commit '255789e4d12d95dbb8e34cfae7fdc58514843ba5':
  Restore tkill for LP32.

10 years agoMerge "Restore tkill for LP32."
Elliott Hughes [Sun, 25 May 2014 03:14:38 +0000 (03:14 +0000)]
Merge "Restore tkill for LP32."

10 years agoRestore tkill for LP32.
Anthony King [Sat, 24 May 2014 16:47:14 +0000 (16:47 +0000)]
Restore tkill for LP32.

libmono from Unity still requires tkill(2).

Change-Id: I37a1994b08086c7fedb5c78ea0dadf2d72bc1463
Signed-off-by: Anthony King <anthonydking@slimroms.net>
10 years agoam d4f37110: Merge "Restore __futex_wake and __futex_wait for LP32."
Elliott Hughes [Sun, 25 May 2014 03:11:02 +0000 (03:11 +0000)]
am d4f37110: Merge "Restore __futex_wake and __futex_wait for LP32."

* commit 'd4f37110ce4b8439dd30540565f7c73e00cfe2fb':
  Restore __futex_wake and __futex_wait for LP32.

10 years agoMerge "Restore __futex_wake and __futex_wait for LP32."
Elliott Hughes [Sun, 25 May 2014 03:09:00 +0000 (03:09 +0000)]
Merge "Restore __futex_wake and __futex_wait for LP32."

10 years agoRestore __futex_wake and __futex_wait for LP32.
Elliott Hughes [Sun, 25 May 2014 00:19:36 +0000 (17:19 -0700)]
Restore __futex_wake and __futex_wait for LP32.

Should fix Skype:

  java.lang.UnsatisfiedLinkError: dlopen failed: cannot locate symbol "__futex_wake" referenced by "libsliq.so"...

Bug: 15196718
Change-Id: I8a18e18d830f0436e820cbde577121bd92710803

10 years agoAdd C11 <stdatomic.h>.
Elliott Hughes [Sat, 24 May 2014 03:06:03 +0000 (20:06 -0700)]
Add C11 <stdatomic.h>.

Bug: 14903517
Change-Id: I631dbfdaa698cf7fea8e3b5e18a32586383e62a5

10 years agoAdd optimized AArch64 versions of bcopy and wmemmove based on memmove
Bernhard Rosenkraenzer [Fri, 23 May 2014 15:44:18 +0000 (17:44 +0200)]
Add optimized AArch64 versions of bcopy and wmemmove based on memmove

Add optimized versions of bcopy and wmemmove for AArch64 based on the
memmove implementation

Change-Id: I82fbe8a7221ce224c567ffcfed7a94a53640fca8
Signed-off-by: Bernhard Rosenkraenzer <Bernhard.Rosenkranzer@linaro.org>
10 years agoam 1c4d8301: Merge "Revert "Add optimized AArch64 versions of bcopy and wmemmove...
Christopher Ferris [Sat, 24 May 2014 01:05:42 +0000 (01:05 +0000)]
am 1c4d8301: Merge "Revert "Add optimized AArch64 versions of bcopy and wmemmove based on memmove""

* commit '1c4d83012fea4e6d011215bef1665cf68d2ac47c':
  Revert "Add optimized AArch64 versions of bcopy and wmemmove based on memmove"