OSDN Git Service

android-x86/bionic.git
7 years agoFix <dirent.h> d_ino/d_off types.
Elliott Hughes [Wed, 5 Jul 2017 19:34:29 +0000 (12:34 -0700)]
Fix <dirent.h> d_ino/d_off types.

Bug: http://b/63336806
Test: builds
Change-Id: I781e7c66c41bcb772d49ce7c23aab185ddd5da1e

7 years agoMerge "Work around missing _Exit in old API levels."
Elliott Hughes [Sat, 1 Jul 2017 15:48:58 +0000 (15:48 +0000)]
Merge "Work around missing _Exit in old API levels."

7 years agoWork around missing _Exit in old API levels.
Elliott Hughes [Fri, 30 Jun 2017 23:39:25 +0000 (16:39 -0700)]
Work around missing _Exit in old API levels.

Rather than do this in libandroid_support, we may as well just stick it with
the other historical stdlib workarounds in bionic itself...

Bug: N/A
Test: built new NDK test
Change-Id: Ia5cf4010581eb79d4adf924e87d0bc050b9e2839

7 years agoMerge "Add version info for C++ versions of basename, strchrnul."
Treehugger Robot [Fri, 30 Jun 2017 22:33:27 +0000 (22:33 +0000)]
Merge "Add version info for C++ versions of basename, strchrnul."

7 years agoAdd version info for C++ versions of basename, strchrnul.
Josh Gao [Fri, 30 Jun 2017 20:20:28 +0000 (13:20 -0700)]
Add version info for C++ versions of basename, strchrnul.

These still won't get guards added by the preprocessor, because it
compiles with C-only.

Bug: https://github.com/android-ndk/ndk/issues/440
Test: treehugger
Change-Id: I893b345e528ed1b761e0db00700037411bbb8b78

7 years agoMerge "Revert "arm64: expand CFI shadow to cover 48-bit virtual addresses""
Roozbeh Pournader [Fri, 30 Jun 2017 01:10:10 +0000 (01:10 +0000)]
Merge "Revert "arm64: expand CFI shadow to cover 48-bit virtual addresses""

7 years agoRevert "arm64: expand CFI shadow to cover 48-bit virtual addresses"
Elliott Hughes [Thu, 29 Jun 2017 23:48:43 +0000 (23:48 +0000)]
Revert "arm64: expand CFI shadow to cover 48-bit virtual addresses"

This reverts commit ae735163e51fbc7cac4217ce392c4894efdfc5a7.

QA claims this causes:

  AndroidRuntime: java.lang.NullPointerException: Attempt to invoke virtual method 'void android.media.MediaPlayer.setSurface(android.view.Surface)' on a null object reference
  AndroidRuntime:        at com.android.setupwizardlib....

Bug: http://b/63141434
Change-Id: I05a6849471623d4cde8b254b1020b0ccbd84b699

7 years agoMerge "bionic: fix assorted static analyzer warnings"
Treehugger Robot [Thu, 29 Jun 2017 16:27:52 +0000 (16:27 +0000)]
Merge "bionic: fix assorted static analyzer warnings"

7 years agoMerge "arm64: expand CFI shadow to cover 48-bit virtual addresses"
Elliott Hughes [Thu, 29 Jun 2017 15:23:32 +0000 (15:23 +0000)]
Merge "arm64: expand CFI shadow to cover 48-bit virtual addresses"

7 years agoMerge "Put back more legacy pthread APIs."
Treehugger Robot [Thu, 29 Jun 2017 00:39:30 +0000 (00:39 +0000)]
Merge "Put back more legacy pthread APIs."

7 years agobionic: fix assorted static analyzer warnings
George Burgess IV [Tue, 27 Jun 2017 23:23:45 +0000 (16:23 -0700)]
bionic: fix assorted static analyzer warnings

Warnings:

bionic/libc/bionic/fts.c:722:5: warning: Null passed to a callee that
requires a non-null 1st parameter

bionic/libc/bionic/sched_cpualloc.c:34:25: warning: Result of 'malloc'
is converted to a pointer of type 'cpu_set_t', which is incompatible
with sizeof operand type 'unsigned long'

bionic/linker/linker_main.cpp:315:7: warning: Access to field 'e_type'
results in a dereference of a null pointer (loaded from variable
'elf_hdr')

bionic/linker/linker_main.cpp:493:66: warning: Access to field 'e_phoff'
results in a dereference of a null pointer (loaded from variable
'elf_hdr')

bionic/linker/linker_main.cpp:90:14: warning: Access to field 'next'
results in a dereference of a null pointer (loaded from variable 'prev')

Bug: None
Test: mma; analyzer warnings are gone. CtsBionicTestCases pass.
Change-Id: I699a60c2c6f64c50b9ea06848a680c98a8abb44a

7 years agoarm64: expand CFI shadow to cover 48-bit virtual addresses
Kevin Brodsky [Mon, 26 Jun 2017 10:46:02 +0000 (11:46 +0100)]
arm64: expand CFI shadow to cover 48-bit virtual addresses

Starting from Linux 4.7, arm64's defconfig enables 48-bit VAs, see:
https://git.kernel.org/torvalds/c/211102d8

On arm64, the CFI shadow configuration currently assumes that VAs
are 39-bit long, and as expected this results in a segfault on a
(defconfig) 4.7+ kernel, when linking a CFI-enabled library.

Consequently, this change increases the max target address to
account for the new max VA size.

Change-Id: I3fb808563fa77a457c65e9663da0613117332072

7 years agoMerge "Enable dlfcn.dt_runpath_absolute_path in glibc tests"
Treehugger Robot [Tue, 27 Jun 2017 19:46:28 +0000 (19:46 +0000)]
Merge "Enable dlfcn.dt_runpath_absolute_path in glibc tests"

7 years agoEnable dlfcn.dt_runpath_absolute_path in glibc tests
Dimitry Ivanov [Tue, 27 Jun 2017 18:02:51 +0000 (11:02 -0700)]
Enable dlfcn.dt_runpath_absolute_path in glibc tests

dlfcn.dt_runpath_absolute_path no longer fails when
running bionic-unit-tests on glibc

Bug: http://b/25320359
Test: bionic-unit-tests-glibc --gtest_filter=dlfcn.dt*
Change-Id: Ia0ecf43293cb0fed28516fd8cb87e7f5f2d6e648

7 years agoMerge "Expand Seccomp whitelist"
Treehugger Robot [Fri, 23 Jun 2017 17:39:05 +0000 (17:39 +0000)]
Merge "Expand Seccomp whitelist"

7 years agoExpand Seccomp whitelist
Paul Lawrence [Fri, 23 Jun 2017 15:10:18 +0000 (08:10 -0700)]
Expand Seccomp whitelist

This seccomp failure is in the fault handler:

05-25 12:03:25.042 10201 27425 27425 F DEBUG   : backtrace:
05-25 12:03:25.042 10201 27425 27425 F DEBUG   :     #00 pc 00015380
/data/data/redacted/files/storage/lib/libcrashsdk.so

So whenever an app using this crash sdk crashes it looks like a seccomp
problem.  Fixing this won't stop the apps crashing, but will make the
crash reports accurate and useful.

So yes, the bug below is already fixed, but this issue has come back 2
or 3 times with different apps (latest is b/62874867). This change
doesn't fix that crash either, but again it improves the reporting.

Bug: 62090571
Test: Device boots, app still fails but no longer with SECCOMP error
Change-Id: Ie0f8dc965001c8bc43f6a545b35bdcd38f006213

7 years agoMerge "Whitelist __NR__newselect."
Treehugger Robot [Fri, 23 Jun 2017 01:39:07 +0000 (01:39 +0000)]
Merge "Whitelist __NR__newselect."

7 years agoWhitelist __NR__newselect.
Josh Gao [Thu, 22 Jun 2017 23:46:19 +0000 (16:46 -0700)]
Whitelist __NR__newselect.

Bug: http://b/62779795
Test: treehugger
Change-Id: I3e0279385641d4f06fcdeb955819a01c12c5229e

7 years agoPut back more legacy pthread APIs.
Dan Albert [Thu, 22 Jun 2017 19:40:55 +0000 (12:40 -0700)]
Put back more legacy pthread APIs.

Test: make checkbuild
Bug: https://github.com/android-ndk/ndk/issues/423
Bug: https://stackoverflow.com/q/44580542/632035
Change-Id: Ibf52a969afffbfcdf6793a0bf8b0e10bbdd1f32c

7 years agoMerge "Disable stack protector for __libc_preinit"
Treehugger Robot [Thu, 22 Jun 2017 01:40:04 +0000 (01:40 +0000)]
Merge "Disable stack protector for __libc_preinit"

7 years agoDisable stack protector for __libc_preinit
Stephen Crane [Tue, 9 May 2017 21:27:43 +0000 (14:27 -0700)]
Disable stack protector for __libc_preinit

__libc_preinit sets up the stack protector global cookie value, and thus
cannot intialize a stack protector cookie for itself in the function
prologue. LTO compilation can inline functions requiring a stack
protector into __libc_preinit. This patch disables stack protection for
__libc_preinit and forces all potentially inlined functions into a
helper that can have a stack protector.

Test: run bionic-unit-tests

Change-Id: I45911611190f216c91eb6feff722967214c5f99f

7 years agoMerge "Shave a stack frame off asserts."
Elliott Hughes [Wed, 21 Jun 2017 15:09:53 +0000 (15:09 +0000)]
Merge "Shave a stack frame off asserts."

7 years agoShave a stack frame off asserts.
Elliott Hughes [Wed, 21 Jun 2017 00:28:42 +0000 (17:28 -0700)]
Shave a stack frame off asserts.

No-one cares about seeing "async_safe_fatal" (which you have to admit is a
pretty confusing name for an app developer anyway).

On arm:

     #00 pc 0001a43c  /system/lib/libc.so (abort+63)
     #01 pc 0001a627  /system/lib/libc.so (__assert+14)

And aarch64:

     #00 pc 000000000001d75c  /system/lib64/libc.so (abort+120)
     #01 pc 000000000001dad0  /system/lib64/libc.so (__assert+44)

Bug: N/A
Test: ran `crasher assert` and `crasher64 assert`
Change-Id: I00be71c566c74cdb00f8e95d634777155bc3da03

7 years agoMerge "Shave another uninteresting stack frame off aborts."
Treehugger Robot [Wed, 21 Jun 2017 00:20:33 +0000 (00:20 +0000)]
Merge "Shave another uninteresting stack frame off aborts."

7 years agoShave another uninteresting stack frame off aborts.
Elliott Hughes [Tue, 20 Jun 2017 21:26:56 +0000 (14:26 -0700)]
Shave another uninteresting stack frame off aborts.

With this, stack frame 0 is the abort, not tgkill.

arm:

     #00 pc 0001a41c  /system/lib/libc.so (abort+63)

arm64:

     #00 pc 000000000001d75c  /system/lib64/libc.so (abort+120)

Also "include what you use" for <sys/syscall.h>.

Bug: N/A
Test: ran `crasher abort` and `crasher64 abort`
Change-Id: I6517ac67b39b4133e890d52efc115071c812958b

7 years agoMerge "make libasync_safe vendor_available:true"
Treehugger Robot [Tue, 20 Jun 2017 10:35:52 +0000 (10:35 +0000)]
Merge "make libasync_safe vendor_available:true"

7 years agoMerge "linker_config_test: allow search paths of systems without a vendor partition"
Jaesung Chung [Mon, 19 Jun 2017 00:13:07 +0000 (00:13 +0000)]
Merge "linker_config_test: allow search paths of systems without a vendor partition"

7 years agolinker_config_test: allow search paths of systems without a vendor partition
Jaesung Chung [Fri, 16 Jun 2017 05:43:16 +0000 (14:43 +0900)]
linker_config_test: allow search paths of systems without a vendor partition

linker_config#smoke and linker_config.asan_smoke are trying to find
paths under the /vendor directory. If there is no vendor partition,
the real path of them is started with /system/vendor.

This CL allows those paths in the tests by getting the resolved paths
for systems without a vendor partition.

Bug: http://b/62562515
Test: linker_config_test passes without a vendor partition.
Change-Id: Id6d16ef623efd81ab9083c3e819da2ad22a28bf8

7 years agoMerge "Add sys/stat.h compatibility macros for old NDKs."
Treehugger Robot [Fri, 16 Jun 2017 22:51:58 +0000 (22:51 +0000)]
Merge "Add sys/stat.h compatibility macros for old NDKs."

7 years agoMerge "Remove obsolete <machine/cpu-features.h>."
Treehugger Robot [Fri, 16 Jun 2017 21:04:10 +0000 (21:04 +0000)]
Merge "Remove obsolete <machine/cpu-features.h>."

7 years agoMerge "Add #defines for some key enums."
Elliott Hughes [Fri, 16 Jun 2017 20:56:03 +0000 (20:56 +0000)]
Merge "Add #defines for some key enums."

7 years agoRemove obsolete <machine/cpu-features.h>.
Elliott Hughes [Fri, 16 Jun 2017 18:05:16 +0000 (11:05 -0700)]
Remove obsolete <machine/cpu-features.h>.

This is no longer used in the platform, and shouldn't be used in NDK. Apps
should use the NDK's cpu-features module, which supports (a) more specific
queries and (b) all Android architectures, not just 32-bit ARM.

Bug: http://b/18556103
Test: builds
Change-Id: I544ef267a6d7d887223186180c77d9ad0321e605

7 years agoMerge "Add benchmarks for strcpy and strcmp."
Christopher Ferris [Fri, 16 Jun 2017 16:40:01 +0000 (16:40 +0000)]
Merge "Add benchmarks for strcpy and strcmp."

7 years agoMerge "libc: warn that cpu-features.h is deprecated"
Elliott Hughes [Fri, 16 Jun 2017 14:40:08 +0000 (14:40 +0000)]
Merge "libc: warn that cpu-features.h is deprecated"

7 years agoMerge "Update posix_memalign testing."
Christopher Ferris [Thu, 15 Jun 2017 23:11:48 +0000 (23:11 +0000)]
Merge "Update posix_memalign testing."

7 years agoAdd #defines for some key enums.
Elliott Hughes [Thu, 15 Jun 2017 22:12:29 +0000 (15:12 -0700)]
Add #defines for some key enums.

Building ruby actually trips over both of these:

* if the RTLD_ constants aren't #defined, it uses its own incorrect values.

* if the REG_ constants aren't #defined, it confuses x86 with x86-64.

In all other places where we have enums in our headers, we already match
existing glibc practice.

Bug: http://b/62531921
Test: builds
Change-Id: I5b3aab25a1a24611bdc58f2eda4104a78e9f841c

7 years agoMerge "Move a <wchar.h> implementation detail into the implementation."
Elliott Hughes [Thu, 15 Jun 2017 21:30:06 +0000 (21:30 +0000)]
Merge "Move a <wchar.h> implementation detail into the implementation."

7 years agoUpdate posix_memalign testing.
Christopher Ferris [Thu, 15 Jun 2017 20:30:44 +0000 (13:30 -0700)]
Update posix_memalign testing.

Move all tests into stdlib_test.cpp since that's where the definition lives
in bionic.

Add a sweep test and a various size test.

Test: Run new unit tests on glibc and angler.
Change-Id: Ief1301f402bea82ce90240500dd6a01636dbdbae

7 years agoMerge "Add copy-only, seek-only, and copy-and-seek benchmarks for strcat"
Christopher Ferris [Thu, 15 Jun 2017 20:34:54 +0000 (20:34 +0000)]
Merge "Add copy-only, seek-only, and copy-and-seek benchmarks for strcat"

7 years agoMerge "MIPS64: move fstat64/fstatat64/_flush_cache syscalls to SECCOMP_WHITELIST...
Treehugger Robot [Thu, 15 Jun 2017 18:21:12 +0000 (18:21 +0000)]
Merge "MIPS64: move fstat64/fstatat64/_flush_cache syscalls to SECCOMP_WHITELIST.TXT"

7 years agoMove a <wchar.h> implementation detail into the implementation.
Elliott Hughes [Thu, 15 Jun 2017 17:36:43 +0000 (10:36 -0700)]
Move a <wchar.h> implementation detail into the implementation.

No other C library expose these, and I couldn't find any callers.

Bug: http://b/62531921
Test: builds
Change-Id: I4a3505bc0897286a4036c48066b98d16665b573a

7 years agoMerge "Expose pthread_mutex_lock_timeout_np to the NDK."
Treehugger Robot [Thu, 15 Jun 2017 16:15:22 +0000 (16:15 +0000)]
Merge "Expose pthread_mutex_lock_timeout_np to the NDK."

7 years agoMerge "Don't use the overloadable attribute if we're not fortified."
Elliott Hughes [Thu, 15 Jun 2017 15:06:34 +0000 (15:06 +0000)]
Merge "Don't use the overloadable attribute if we're not fortified."

7 years agoMerge "Move libc/zoneinfo/ files to system/timezone"
Nicolas Geoffray [Thu, 15 Jun 2017 09:05:31 +0000 (09:05 +0000)]
Merge "Move libc/zoneinfo/ files to system/timezone"

7 years agoMIPS64: move fstat64/fstatat64/_flush_cache syscalls to SECCOMP_WHITELIST.TXT
Miodrag Dinic [Mon, 12 Jun 2017 08:52:12 +0000 (10:52 +0200)]
MIPS64: move fstat64/fstatat64/_flush_cache syscalls to SECCOMP_WHITELIST.TXT

fstat64/fstatat64/_flush_cache were accidentally put in SYSCALLS.TXT in:
    https://android-review.googlesource.com/#/c/390454/

This patch just moves them to SECCOMP_WHITELIST.TXT because we do not
want stubs accidenatally generated for the mentioned syscalls using
gensyscalls.py script.

This commit does not introduce any functional changes to mips64_policy.cpp.

Test: Run genseccomp.py -> File seccomp/mips64_policy.cpp not changed.
Test: Run gensyscalls.py -> INFO:root:no changes detected!

Change-Id: I3b527b3d9f18715c44a4e6ddc6db6e49f48f4890
Signed-off-by: Miodrag Dinic <miodrag.dinic@imgtec.com>
7 years agoMerge "Add <sys/mtio.h>."
Treehugger Robot [Thu, 15 Jun 2017 02:40:33 +0000 (02:40 +0000)]
Merge "Add <sys/mtio.h>."

7 years agoAdd benchmarks for strcpy and strcmp.
Anders Lewis [Mon, 12 Jun 2017 19:33:06 +0000 (12:33 -0700)]
Add benchmarks for strcpy and strcmp.

Test: Ran benchmarks and verified that runtimes were sensible.
Change-Id: I609e3261503a015bfd4982767fef3ce56141aebc

7 years agoAdd <sys/mtio.h>.
Elliott Hughes [Thu, 15 Jun 2017 00:16:05 +0000 (17:16 -0700)]
Add <sys/mtio.h>.

In a similar style to some of our other "not really, but enough" headers
like <sys/vt.h>.

Bug: N/A
Test: build GNU dd or BSD dd with a standalone toolchain
Change-Id: I8fbd1aac1d97e24b05e7aae8a55666300b5bf1ed

7 years agoDon't use the overloadable attribute if we're not fortified.
Elliott Hughes [Wed, 14 Jun 2017 23:54:32 +0000 (16:54 -0700)]
Don't use the overloadable attribute if we're not fortified.

See code comment.

Bug: N/A
Test: building bash and binutils with a standalone toolchain
Change-Id: I73b5d55716072ca1b03b1b0157d65cb1ceb22841

7 years agoAdd copy-only, seek-only, and copy-and-seek benchmarks for strcat
Anders Lewis [Mon, 12 Jun 2017 18:24:01 +0000 (11:24 -0700)]
Add copy-only, seek-only, and copy-and-seek benchmarks for strcat

Test: Ran benchmarks and verified that runtimes were sensible.
Change-Id: I6da3aaea0ba4817662534b24873993589385e6c5

7 years agoAdd sys/stat.h compatibility macros for old NDKs.
Dan Albert [Mon, 12 Jun 2017 21:59:38 +0000 (14:59 -0700)]
Add sys/stat.h compatibility macros for old NDKs.

Test: make checkbuild
Bug: https://github.com/android-ndk/ndk/issues/417
Change-Id: I04fdb2f31131e87907adab9f3264b783917de542

7 years agoMove libc/zoneinfo/ files to system/timezone
Neil Fuller [Mon, 12 Jun 2017 15:57:29 +0000 (16:57 +0100)]
Move libc/zoneinfo/ files to system/timezone

Move libc/zoneinfo/ files to system/timezone.

Bug: 31008728
Bug: 36882778
Test: make -j30 droid / boot device
Merged-In: I1d75b0be12b41258f89614b92c1ef68aa7ebfaf9
Change-Id: I1d75b0be12b41258f89614b92c1ef68aa7ebfaf9

7 years agomake libasync_safe vendor_available:true
Jiyong Park [Mon, 15 May 2017 06:22:12 +0000 (15:22 +0900)]
make libasync_safe vendor_available:true

This, when used with BOARD_VNDK_VERSION flag, enables the
module to be built and installed twice, one for /system,
other for /vendor.

libasync_safe is used by libbacktrace which is in turn used by libs in
/vendor and thus marked as vendor_available. Therefore, libasync_safe
is also marked as vendor_available.

Details: https://android-review.googlesource.com/368372

Bug: 33241851
Test: BOARD_VNDK_VERSION=current m libasync_safe.vendor successful

Merged-In: I4b4de3b99a024d52612109cee3d66b4e5fc12dec
Change-Id: I4b4de3b99a024d52612109cee3d66b4e5fc12dec
(cherry picked from commit ec0d6b416e6f9c7044c79fdc8a4760f23ade8eb4)

7 years agoMerge "Remove `volatile` from the `__errno` declaration."
Treehugger Robot [Wed, 14 Jun 2017 00:44:31 +0000 (00:44 +0000)]
Merge "Remove `volatile` from the `__errno` declaration."

7 years agoMerge "The bsd_signal declaration shouldn't be visible if API > 21."
Treehugger Robot [Tue, 13 Jun 2017 23:20:38 +0000 (23:20 +0000)]
Merge "The bsd_signal declaration shouldn't be visible if API > 21."

7 years agoRemove `volatile` from the `__errno` declaration.
Elliott Hughes [Tue, 13 Jun 2017 21:43:53 +0000 (14:43 -0700)]
Remove `volatile` from the `__errno` declaration.

This breaks code that declares `errno` itself for whatever crazy reason:

  b.c:22:12: error: conflicting types for '__errno'
  extern int errno;
             ^
  /usr/local/google/ndkports/toolchain/bin/../sysroot/usr/include/errno.h:47:20: note: expanded from macro 'errno'
  #define  errno   (*__errno())
                     ^
  /usr/local/google/ndkports/toolchain/bin/../sysroot/usr/include/errno.h:44:15: note: previous declaration is here
  volatile int* __errno(void) __attribute_const__;
                ^

Bug: N/A
Test: built various bits of GNU source
Change-Id: I27c03bf3bde419a001f98f1ea6c267c847f31271

7 years agoThe bsd_signal declaration shouldn't be visible if API > 21.
Elliott Hughes [Tue, 13 Jun 2017 21:29:15 +0000 (14:29 -0700)]
The bsd_signal declaration shouldn't be visible if API > 21.

This causes GNU make to fail to build:

  main.o:main.c:(.text+0x18c): more undefined references to `bsd_signal' follow

Bug: N/A
Test: built GNU make with a hacked standalone toolchain
Change-Id: Icdfb4cf674df8abed44f6671454bed9f8d68e74b

7 years agoMerge "libc: Sync qsort(3) with FreeBSD revision r318515."
Treehugger Robot [Tue, 13 Jun 2017 20:33:48 +0000 (20:33 +0000)]
Merge "libc: Sync qsort(3) with FreeBSD revision r318515."

7 years agolibc: Sync qsort(3) with FreeBSD revision r318515.
Xin Li [Fri, 19 May 2017 17:13:08 +0000 (10:13 -0700)]
libc: Sync qsort(3) with FreeBSD revision r318515.

r279663: enhance to handle 32-bit aligned data on 64-bit systems

r279666: qsort(3): small style(9) cleanups.

r288030: Use ANSI C prototypes.  Eliminates -Wold-style-definition
warnings.

r318514: Use size_t.

r318515: Reduce the possible recursion level to log2(N) by always
recursing on the smaller partition instead.

Bug: 38450217
Test: builds, ./tests/run-on-host.sh 64
Change-Id: Ieb622e6f54f8f6d6f62e1bdb491d3b5c8e2f9965

7 years agoMerge "Split memmove benchmark into separate overlapping (src before dst and dst...
Christopher Ferris [Tue, 13 Jun 2017 18:05:01 +0000 (18:05 +0000)]
Merge "Split memmove benchmark into separate overlapping (src before dst and dst before src) and non-overlapping tests."

7 years agoExpose pthread_mutex_lock_timeout_np to the NDK.
Dan Albert [Mon, 12 Jun 2017 21:17:49 +0000 (14:17 -0700)]
Expose pthread_mutex_lock_timeout_np to the NDK.

The proper API for this isn't available until L, so expose this for
API levels earlier than that.

Test: make checkbuild
Bug: https://github.com/android-ndk/ndk/issues/420
Change-Id: I382b8f557be9530f3e13aaae353b4a6e7f9301ab

7 years agoSplit memmove benchmark into separate overlapping (src before dst and dst before...
Anders Lewis [Wed, 7 Jun 2017 20:00:38 +0000 (13:00 -0700)]
Split memmove benchmark into separate overlapping (src before dst and dst before src) and non-overlapping tests.

Test: Ran benchmarks and verified that runtimes were sensible.
Change-Id: Ia7fbc5596dfdc162fff2364f4ef6dafe5549b907

7 years agoMerge "Explain why we have a test that FALLOC_FL_PUNCH_HOLE is missing."
Josh Gao [Fri, 9 Jun 2017 06:01:24 +0000 (06:01 +0000)]
Merge "Explain why we have a test that FALLOC_FL_PUNCH_HOLE is missing."

7 years agoMerge "Use libdemangle for function names."
Christopher Ferris [Fri, 2 Jun 2017 22:02:53 +0000 (22:02 +0000)]
Merge "Use libdemangle for function names."

7 years agoExplain why we have a test that FALLOC_FL_PUNCH_HOLE is missing.
Josh Gao [Fri, 2 Jun 2017 21:57:49 +0000 (14:57 -0700)]
Explain why we have a test that FALLOC_FL_PUNCH_HOLE is missing.

Change-Id: I7388f2e0bc419f195247a1d3e73195cb4d199d5c
Test: none

7 years agoUse libdemangle for function names.
Christopher Ferris [Thu, 1 Jun 2017 21:16:32 +0000 (14:16 -0700)]
Use libdemangle for function names.

Bug: 62141808

Test: Ran unit tests, ran malloc debug with leaks and verified functions
Test: are demangled.
Change-Id: I1900632e9a55bb818b1aabc798224ff3eebc1fca

7 years agolibc: warn that cpu-features.h is deprecated
Jake Weinstein [Tue, 11 Apr 2017 05:24:22 +0000 (01:24 -0400)]
libc: warn that cpu-features.h is deprecated

Test: mm

Change-Id: I39668fc75e72004224e35ac9e7d6ffc858df2169

7 years agoMerge "Fix boost build with unified headers."
Treehugger Robot [Thu, 1 Jun 2017 23:30:56 +0000 (23:30 +0000)]
Merge "Fix boost build with unified headers."

7 years agoFix boost build with unified headers.
Elliott Hughes [Thu, 1 Jun 2017 21:08:58 +0000 (14:08 -0700)]
Fix boost build with unified headers.

boost (1.64.0 at least) assumes that if you have POSIX_MADV_NORMAL, you have
posix_madvise. With unified headers, this isn't true. Rather than make life
harder for projects that don't use configure, just make it so. We already
applied similar workarounds for epoll_create1 and inotify_init1.

Bug: https://github.com/android-ndk/ndk/issues/395
Test: built boost (long story!)
Change-Id: I5d2d8de7b30921dde913251d35dcd249a2876f94

7 years agoMerge "MIPS: 32/64-bit VDSO support"
Treehugger Robot [Thu, 1 Jun 2017 17:44:40 +0000 (17:44 +0000)]
Merge "MIPS: 32/64-bit VDSO support"

7 years agoMIPS: 32/64-bit VDSO support
Goran Ferenc [Tue, 30 May 2017 14:00:41 +0000 (16:00 +0200)]
MIPS: 32/64-bit VDSO support

File libc/SYSCALLS.TXT is updated to generate bionic's system call wrappers
for clock_gettime() & gettimeofday() that will be called if kernel vdso
implementations fail to execute.
The system call wrappers are generated using a python script gensyscalls.py.

Since all architectures support vdso now, there is no more need for conditional
statements regarding supported architectures in libc/bionic/vdso.cpp &
libc/private/bionic_vdso.h files.

Test: builds
Change-Id: I7213f29c179a7929851499d78a72900638ae861a
Signed-off-by: Goran Ferenc <goran.ferenc@imgtec.com>
7 years agoMerge "Make dl_iterate_phdr return correct name for first entry"
Treehugger Robot [Thu, 1 Jun 2017 07:16:11 +0000 (07:16 +0000)]
Merge "Make dl_iterate_phdr return correct name for first entry"

7 years agoMake dl_iterate_phdr return correct name for first entry
Dimitry Ivanov [Wed, 31 May 2017 22:07:41 +0000 (15:07 -0700)]
Make dl_iterate_phdr return correct name for first entry

Test: bionic-unit-tests --gtest_filter=link*
Change-Id: Ib3f1e0fbc76fed9a5f27ffdd9bbf847a3e8d3665

7 years agoMerge "Move mallopt to correct section."
Christopher Ferris [Wed, 31 May 2017 20:07:29 +0000 (20:07 +0000)]
Merge "Move mallopt to correct section."

7 years agoMerge "Revert "Make mallopt available in the future.""
Christopher Ferris [Wed, 31 May 2017 18:15:09 +0000 (18:15 +0000)]
Merge "Revert "Make mallopt available in the future.""

7 years agoMove mallopt to correct section.
Christopher Ferris [Wed, 31 May 2017 18:00:18 +0000 (11:00 -0700)]
Move mallopt to correct section.

Bug: 62202823

Test: Built and built unit tests.
Change-Id: Icefc152eb756fd60ce8f51d694f22923a41a9cd1

7 years agoMerge "Add mallopt to canonical list of exports."
Christopher Ferris [Wed, 31 May 2017 17:40:31 +0000 (17:40 +0000)]
Merge "Add mallopt to canonical list of exports."

7 years agoAdd mallopt to canonical list of exports.
Christopher Ferris [Wed, 31 May 2017 00:35:28 +0000 (17:35 -0700)]
Add mallopt to canonical list of exports.

Bug: 62202823

Test: Ran script and verified that none of the other map files are modified.
Change-Id: Ib890ce77c60b877434fbb65a5add8f6c4603d3d0

7 years agoRevert "Make mallopt available in the future."
Christopher Ferris [Wed, 31 May 2017 00:14:15 +0000 (00:14 +0000)]
Revert "Make mallopt available in the future."

This reverts commit 083923d96acb7cfe4e1a1c22100047fa87feb347.

Reason for revert: No longer correct, this will be in API 26.

Change-Id: Iaf82d797405135fcd30455fd750b23fc57e9623a

7 years agoMerge "Enable exporting the non-uapi scsi headers."
Elliott Hughes [Fri, 26 May 2017 18:04:26 +0000 (18:04 +0000)]
Merge "Enable exporting the non-uapi scsi headers."

7 years agoMerge "Add offset to backtrace_string printing."
Christopher Ferris [Fri, 26 May 2017 16:57:58 +0000 (16:57 +0000)]
Merge "Add offset to backtrace_string printing."

7 years agoEnable exporting the non-uapi scsi headers.
Elliott Hughes [Fri, 26 May 2017 00:13:32 +0000 (17:13 -0700)]
Enable exporting the non-uapi scsi headers.

Callers are supposed to #include <scsi/sg.h> but if we tell soong to add
bionic/libc/kernel/android/ to the include path, the uapi headers in there
would be (unintentionally) accessible as either <linux/name.h> or
<uapi/linux/name.h>.

Bug: N/A (hit while upgrading strace to 4.17)
Test: builds
Change-Id: I8d47dd51da688c38f747a255d401dfb2c209c805

7 years agoAdd offset to backtrace_string printing.
Christopher Ferris [Fri, 26 May 2017 01:26:37 +0000 (18:26 -0700)]
Add offset to backtrace_string printing.

The libmemunreachable code uses backtrace_string to print out the backtrace
information. However, when the offset of the map that a frame is in is
non-zero, no information is printed. This means that when a frame comes from
a shared library loaded from an apk, it's impossible to tell which
shared library the frame is really coming from.

Add the offset display when it's non-zero. This means this can now be fed
to development/scripts/stack to decode the frame.

Bug: 37276041

Test: Ran dumpsys meminfo --unreachable on an app that leaks from a
Test: shared library and verified the offset output is present.
Change-Id: I61d34ae3f617622d354cc099eff520a64782b6e2

7 years agoMerge "Relax flaky _SC_AVPHYS_PAGES check in sys_sysinfo_test"
Treehugger Robot [Fri, 26 May 2017 01:30:01 +0000 (01:30 +0000)]
Merge "Relax flaky _SC_AVPHYS_PAGES check in sys_sysinfo_test"

7 years agoMerge "Expand whitelist"
Treehugger Robot [Fri, 26 May 2017 01:01:54 +0000 (01:01 +0000)]
Merge "Expand whitelist"

7 years agoMerge "Update malloc debug documentation."
Christopher Ferris [Fri, 26 May 2017 00:29:59 +0000 (00:29 +0000)]
Merge "Update malloc debug documentation."

7 years agoUpdate malloc debug documentation.
Christopher Ferris [Thu, 25 May 2017 02:04:33 +0000 (19:04 -0700)]
Update malloc debug documentation.

Test: NA
Change-Id: I8ab73bfc28c7a673efb5b0cec6067989153f092a

7 years agoExpand whitelist
Paul Lawrence [Thu, 25 May 2017 21:15:23 +0000 (14:15 -0700)]
Expand whitelist

Bug: 37769298
Test: Boots. Cannot test app behavior without account
Change-Id: Iebb7616f100368bf2e702ec51f637df1f3727885

7 years agoMerge "Remove the repetitive warnings from the uapi headers."
Treehugger Robot [Thu, 25 May 2017 22:50:05 +0000 (22:50 +0000)]
Merge "Remove the repetitive warnings from the uapi headers."

7 years agoRelax flaky _SC_AVPHYS_PAGES check in sys_sysinfo_test
Dimitry Ivanov [Thu, 25 May 2017 21:02:03 +0000 (14:02 -0700)]
Relax flaky _SC_AVPHYS_PAGES check in sys_sysinfo_test

The number of available physical pages is not constant and can
potentially change between calls to get_avphys_pages and
sysconf(_SC_AVPHYS_PAGES) calls.

Relax the assert to check that sconf(_SC_AVPHYS_PAGES) <= phys_pages.

Bug: http://b/31502852
Test: build and run bionic-unit-tests
Change-Id: Ibf8873d2151d93239391d2638dfbf055b70cde3a

7 years agoRemove the repetitive warnings from the uapi headers.
Elliott Hughes [Thu, 25 May 2017 20:48:01 +0000 (13:48 -0700)]
Remove the repetitive warnings from the uapi headers.

Having

  WARNING: DO NOT EDIT, AUTO-GENERATED CODE - SEE TOP FOR INSTRUCTIONS

every four lines made the headers harder to read, made the diffs much worse
each time we upgraded, and wasn't really providing any benefit. Before the
next uapi update, let's just stop doing this.

Bug: N/A
Test: builds, manually inspected files look right
Change-Id: Id7088cf750894c9d24950f3d53587fe3156c4f7d

7 years agoMerge "Stop including <sys/sysmacros.h> from <sys/types.h>."
Treehugger Robot [Thu, 25 May 2017 01:43:05 +0000 (01:43 +0000)]
Merge "Stop including <sys/sysmacros.h> from <sys/types.h>."

7 years agoMerge "Fix assembler warnings."
Christopher Ferris [Thu, 25 May 2017 01:26:17 +0000 (01:26 +0000)]
Merge "Fix assembler warnings."

7 years agoFix assembler warnings.
Christopher Ferris [Wed, 24 May 2017 23:00:37 +0000 (16:00 -0700)]
Fix assembler warnings.

There are a few instructions deprecated on armv8 that result in lots
of warnings. Add an arch directive so that these warnings go away.

This doesn't cause any problems because the instructions still
execute properly.

Bug: 38319728

Test: Built all of these assembler files and verified the warning are gone.
Change-Id: If063defdd16f290c01975233c8d257d1b2005e76

7 years agoMerge "Remove obsolete __stack_chk_fail_local."
Elliott Hughes [Wed, 24 May 2017 15:05:59 +0000 (15:05 +0000)]
Merge "Remove obsolete __stack_chk_fail_local."

7 years agoMerge "Fix _POSIX_THREAD_PROCESS_SHARED."
Treehugger Robot [Wed, 24 May 2017 07:13:34 +0000 (07:13 +0000)]
Merge "Fix _POSIX_THREAD_PROCESS_SHARED."

7 years agoRemove obsolete __stack_chk_fail_local.
Elliott Hughes [Tue, 23 May 2017 23:43:30 +0000 (16:43 -0700)]
Remove obsolete __stack_chk_fail_local.

Clang doesn't use this.

Bug: N/A
Test: x86 emulator builds and boots
Change-Id: I2865c0d568a644f61f34bdea539daff5224896bc

7 years agoMerge "Switch to inline assembler in crtbegin."
Elliott Hughes [Tue, 23 May 2017 23:39:59 +0000 (23:39 +0000)]
Merge "Switch to inline assembler in crtbegin."

7 years agoStop including <sys/sysmacros.h> from <sys/types.h>.
Elliott Hughes [Tue, 23 May 2017 22:53:36 +0000 (15:53 -0700)]
Stop including <sys/sysmacros.h> from <sys/types.h>.

<sys/types.h> unconditionally includes <sys/sysmacros.h>.

<sys/sysmacros.h> defines major, minor, and makedev. In the deprecated NDK
headers, these were inline functions. In the unified headers, they are
function-like macros.

The inline functions would only collide with another function called major,
minor, or makedev but the macros will replace anything with a function-like
form such as definitions of class methods called major, minor, or makedev.
This causes code which has such definitions to fail to compile when moving
from the deprecated headers to the unified headers.

Resolve this by removing the transitive dependency. Folks who want
<sys/sysmacros.h> can ask for it explicitly (it's been in both bionic and
glibc for years, though I don't think macOS has it [as usual]).

Bug: https://github.com/android-ndk/ndk/issues/398
Test: builds
Change-Id: If2921c7b5d979c3066f199f22c64d4d2f7bf6632

7 years agoSwitch to inline assembler in crtbegin.
Elliott Hughes [Tue, 23 May 2017 18:03:58 +0000 (11:03 -0700)]
Switch to inline assembler in crtbegin.

Using __builtin_frame_address was clever, but didn't work for arm64 (for
reasons which were never investigated) and the ChromeOS folks claim it
causes trouble for x86 with ARC++ (though without a reproduceable test case).

Naked functions turn out to be quite unevenly supported: some architectures
do the right thing, others don't; some architectures warn, others don't (and
the warnings don't always match the platforms that _actually_ have problems).

Inline assembler also removes the guessing games: everyone knows what the
couple of instructions _ought_ to be, and now we don't have to reason about
what the compiler will actually do (yet still keep the majority of the code
in C).

Bug: N/A
Test: builds, boots
Change-Id: I14207ef50ca46b6eca273c3cb7509c311146a3ca