OSDN Git Service

android-x86/bionic.git
7 years agoMerge "Move seccomp policy to bionic"
Treehugger Robot [Tue, 28 Feb 2017 15:36:07 +0000 (15:36 +0000)]
Merge "Move seccomp policy to bionic"

7 years agoMerge "Fix greylist exception to account for linked namespaces"
Treehugger Robot [Tue, 28 Feb 2017 00:42:54 +0000 (00:42 +0000)]
Merge "Fix greylist exception to account for linked namespaces"

7 years agoMerge "Fix pre-L MB_CUR_MAX."
Treehugger Robot [Mon, 27 Feb 2017 22:22:26 +0000 (22:22 +0000)]
Merge "Fix pre-L MB_CUR_MAX."

7 years agoFix greylist exception to account for linked namespaces
Dimitry Ivanov [Mon, 27 Feb 2017 20:17:47 +0000 (12:17 -0800)]
Fix greylist exception to account for linked namespaces

Do not load second copy of libraries that are supposed to
be provided by linked namespaces. Also do not print
error in the log if caller tries to open shared library
using absolute path for apps targeting N+.

Bug: http://b/35454141
Bug: http://b/26833548
Bug: http://b/35338922
Test: run bionic-unit-tests --gtest_filter=dl*
Change-Id: Icf3aeedff18d287d2ba0b3df3808b100f3ef5f7a

7 years agoMove seccomp policy to bionic
Paul Lawrence [Thu, 16 Feb 2017 17:24:39 +0000 (09:24 -0800)]
Move seccomp policy to bionic

Test: Built and checked booted
Change-Id: Iaec1265fe5a55c4df90ab9e45b010ef36faf6bba

7 years agoMerge "Update to kernel headers v4.10."
Christopher Ferris [Mon, 27 Feb 2017 20:09:54 +0000 (20:09 +0000)]
Merge "Update to kernel headers v4.10."

7 years agoMerge "Revert "Move seccomp policy to bionic""
Paul Lawrence [Mon, 27 Feb 2017 16:39:11 +0000 (16:39 +0000)]
Merge "Revert "Move seccomp policy to bionic""

7 years agoRevert "Move seccomp policy to bionic"
Paul Lawrence [Mon, 27 Feb 2017 16:32:37 +0000 (16:32 +0000)]
Revert "Move seccomp policy to bionic"

This reverts commit 06a32206c5430321dd3fc02b70acbf99383786c1.

Reverting build-breaking change

Change-Id: Ib3698bca8f905033a9c7f22bc2fa9f7e7bf75873

7 years agoMerge "Move seccomp policy to bionic"
Paul Lawrence [Mon, 27 Feb 2017 16:09:39 +0000 (16:09 +0000)]
Merge "Move seccomp policy to bionic"

7 years agoMerge "Add tests for <endian.h>."
Treehugger Robot [Sat, 25 Feb 2017 03:10:18 +0000 (03:10 +0000)]
Merge "Add tests for <endian.h>."

7 years agoMerge "loader: set PT_INTERP to itself"
Treehugger Robot [Sat, 25 Feb 2017 02:24:28 +0000 (02:24 +0000)]
Merge "loader: set PT_INTERP to itself"

7 years agoMerge "libc: __system_property_set uses writev to write atomically"
Treehugger Robot [Sat, 25 Feb 2017 02:06:19 +0000 (02:06 +0000)]
Merge "libc: __system_property_set uses writev to write atomically"

7 years agoMerge "Move __system_property_*_serial back to LIBC"
Treehugger Robot [Sat, 25 Feb 2017 02:02:02 +0000 (02:02 +0000)]
Merge "Move __system_property_*_serial back to LIBC"

7 years agoloader: set PT_INTERP to itself
Dimitry Ivanov [Thu, 23 Feb 2017 19:53:43 +0000 (11:53 -0800)]
loader: set PT_INTERP to itself

Some versions of kernel set AT_BASE to 0
if dynamic loader does not have PT_INTERP
set.

Bug: http://b/30739481
Test: run /system/bin/linker64 and /system/bin/linker
Change-Id: I1b67777166fe917d3ee1a97277045ca6f5db0084

7 years agoAdd tests for <endian.h>.
Elliott Hughes [Sat, 25 Feb 2017 00:19:53 +0000 (16:19 -0800)]
Add tests for <endian.h>.

Also, for the stuff that's also in <netinet/in.h> as real functions,
check that they're there too (and as functions rather than macros,
since that was historically not true).

Bug: http://b/28432448
Test: ran tests
Change-Id: I7e4ae926f7e02de3b6dd38d1953e5b3b43d44f74

7 years agoMove __system_property_*_serial back to LIBC
Dimitry Ivanov [Fri, 24 Feb 2017 23:36:17 +0000 (15:36 -0800)]
Move __system_property_*_serial back to LIBC

Bug: http://b/35764972
Bug: http://b/34114501
Test: make
Change-Id: Ibbda0107d07c432110a0633de03259e39480fde4

7 years agolibc: __system_property_set uses writev to write atomically
Dimitry Ivanov [Fri, 24 Feb 2017 01:57:14 +0000 (17:57 -0800)]
libc: __system_property_set uses writev to write atomically

__system_property_set sometimes produces broken_pipe error
when trying to write a property.

This change improves error messages and uses writev() instead
of sequence of send() calls.

Bug: http://b/35381074
Test: bionic-unit-tests --gtest_filter=prop*
Change-Id: I7a5b169c015db4e6b720370e58662de8206d1086

7 years agoMove seccomp policy to bionic
Paul Lawrence [Thu, 16 Feb 2017 17:24:39 +0000 (09:24 -0800)]
Move seccomp policy to bionic

Test: Built and checked booted

Change-Id: If777eed75d5280c7a390399261e97125c04767b2

7 years agoMerge "Hide internal __system_property_* functions"
Dimitry Ivanov [Fri, 24 Feb 2017 19:11:12 +0000 (19:11 +0000)]
Merge "Hide internal __system_property_* functions"

7 years agoMerge "Replace cxa_guard fences with acquire loads"
Hans Boehm [Fri, 24 Feb 2017 18:50:34 +0000 (18:50 +0000)]
Merge "Replace cxa_guard fences with acquire loads"

7 years agoMerge "Use <linux/eventpoll.h> to implement <sys/epoll.h>."
Elliott Hughes [Fri, 24 Feb 2017 16:55:25 +0000 (16:55 +0000)]
Merge "Use <linux/eventpoll.h> to implement <sys/epoll.h>."

7 years agoMerge "Silence a compiler warning due to unused variable on mips"
Pavel Labath [Fri, 24 Feb 2017 10:22:40 +0000 (10:22 +0000)]
Merge "Silence a compiler warning due to unused variable on mips"

7 years agoSilence a compiler warning due to unused variable on mips
Pavel Labath [Fri, 24 Feb 2017 10:14:13 +0000 (10:14 +0000)]
Silence a compiler warning due to unused variable on mips

oops, I did it again.

Test: lunch mips && cd bionic && mma
Change-Id: I1fe2cb847d0698b34766869f9626398dbcf81960

7 years agoMerge "Add "imprecise" watchpoint ptrace test"
Pavel Labath [Fri, 24 Feb 2017 09:16:19 +0000 (09:16 +0000)]
Merge "Add "imprecise" watchpoint ptrace test"

7 years agoAdd "imprecise" watchpoint ptrace test
Pavel Labath [Wed, 22 Feb 2017 18:22:46 +0000 (18:22 +0000)]
Add "imprecise" watchpoint ptrace test

This tests for the presence of a kernel bug that meant that the kernel
would sometimes fail to report the watchpoint hit if the hardware
reported a address which did not exactly match the address range being
watched (which it is allowed to do per ARM spec if the instruction
accesses a larger block of memory than the region being watched). This
bug was fixed in linux kernel 4.9, and has been backported to older
android kernels.

Bug: 30802222
Bug: 30919905
Test: bionic-unit-tests --gtest_filter="sys_ptrace.*"
Change-Id: I80c35b29eaf28e2dbacb9e8ee5317fdea653fc87

7 years agoUse <linux/eventpoll.h> to implement <sys/epoll.h>.
Elliott Hughes [Thu, 23 Feb 2017 07:22:51 +0000 (23:22 -0800)]
Use <linux/eventpoll.h> to implement <sys/epoll.h>.

Bug: https://github.com/android-ndk/ndk/issues/302
Test: builds
Change-Id: Ia3074326a128c38f2488e342c028cc030801cfd9

7 years agoMerge "Expand whitelist"
Treehugger Robot [Thu, 23 Feb 2017 23:54:34 +0000 (23:54 +0000)]
Merge "Expand whitelist"

7 years agoMerge "versioner: fix darwin build."
Treehugger Robot [Thu, 23 Feb 2017 23:51:14 +0000 (23:51 +0000)]
Merge "versioner: fix darwin build."

7 years agoHide internal __system_property_* functions
Dimitry Ivanov [Thu, 16 Feb 2017 23:34:21 +0000 (15:34 -0800)]
Hide internal __system_property_* functions

Bug: http://b/34114501
Test: bionic-unit-tests --gtest_filter=prop*
Change-Id: I1fc57b4ced6aaf841aad64e12e7696d25c2e027b

7 years agoMerge "Add legacy inlines for more termios stuff."
Treehugger Robot [Thu, 23 Feb 2017 22:01:10 +0000 (22:01 +0000)]
Merge "Add legacy inlines for more termios stuff."

7 years agoversioner: fix darwin build.
Josh Gao [Thu, 23 Feb 2017 21:52:49 +0000 (13:52 -0800)]
versioner: fix darwin build.

Use struct stat::st_mtime instead of the underlying st_mtim, which is
called something different on Darwin.

Test: mma on linux, darwin
Change-Id: I2695a6c83ebb7d08ec56b1355e0f4bc0993a0acb

7 years agoMerge "Revert "Run the versioner as part of the build.""
Dan Albert [Thu, 23 Feb 2017 21:36:48 +0000 (21:36 +0000)]
Merge "Revert "Run the versioner as part of the build.""

7 years agoRevert "Run the versioner as part of the build."
Dan Albert [Thu, 23 Feb 2017 21:30:35 +0000 (21:30 +0000)]
Revert "Run the versioner as part of the build."

This reverts commit 45715b2a88c8daf5a721ce9b28391120f5507764.

Reason for revert: Darwin sucks

Change-Id: I004dd7fa5e14a43ffe04ace2f5a06341df61b8cd

7 years agoMerge "Run the versioner as part of the build."
Dan Albert [Thu, 23 Feb 2017 21:06:47 +0000 (21:06 +0000)]
Merge "Run the versioner as part of the build."

7 years agoExpand whitelist
Paul Lawrence [Wed, 15 Feb 2017 21:40:22 +0000 (13:40 -0800)]
Expand whitelist

Bug: 35217603
Test: App no longer triggers seccomp exceptions when launched
Change-Id: I8ea904640a2f14c67a075e593067327407766220

7 years agoMerge "Autogenerate single policy from syscalls and whitelist"
Treehugger Robot [Thu, 23 Feb 2017 20:27:45 +0000 (20:27 +0000)]
Merge "Autogenerate single policy from syscalls and whitelist"

7 years agoReplace cxa_guard fences with acquire loads
Hans Boehm [Wed, 22 Feb 2017 23:34:29 +0000 (15:34 -0800)]
Replace cxa_guard fences with acquire loads

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

7 years agoAdd legacy inlines for more termios stuff.
Dan Albert [Thu, 23 Feb 2017 19:09:32 +0000 (11:09 -0800)]
Add legacy inlines for more termios stuff.

Test: make checkbuild # with my versioner-in-build patches
Bug: https://github.com/android-ndk/ndk/issues/302
Change-Id: Ib00b5dadf23592d101486b4f2188285ec03c9e2a

7 years agoRun the versioner as part of the build.
Dan Albert [Thu, 16 Feb 2017 01:18:01 +0000 (17:18 -0800)]
Run the versioner as part of the build.

Running this periodically and checking in the results is not working
out, since the result is just that I'm having to spend a lot of time
cleaning up the headers every time I need to update them in the NDK.

Run the versioner as part of the build instead. This way bionic
header changes behave like the rest of the NDK headers and will
affect NDK builds in the platform *immediately*.

Remove the preupload hook for the versioner since it's part of the
normal build now.

The versioner's dependencies directory needs to be moved because
soong won't let us try to do things outside our module's directory
(in this case libc).

Unfortunately this means we need to build the versioner for Darwin,
because we now need it to perform a platform build.

Test: make checkbuild
Bug: None
Change-Id: Icdab8a962354d9e945072dc3f806baea376c8db4

7 years agoAutogenerate single policy from syscalls and whitelist
Paul Lawrence [Tue, 14 Feb 2017 21:32:23 +0000 (13:32 -0800)]
Autogenerate single policy from syscalls and whitelist

Bug: 35392119
Bug: 34465958
Test: Check boots and same syscalls are blocked as before

Change-Id: I9efa97032c59aebbbfd32e6f0d2d491f6254f0a2

7 years agoMerge "Add test for a greylisted library"
Dimitry Ivanov [Thu, 23 Feb 2017 17:11:18 +0000 (17:11 +0000)]
Merge "Add test for a greylisted library"

7 years agoAdd test for a greylisted library
Dimitry Ivanov [Tue, 21 Feb 2017 21:41:08 +0000 (13:41 -0800)]
Add test for a greylisted library

Test: bionic-unit-tests --gtest_filter=dlext.ns*
Bug: http://b/35338922
Change-Id: I2f4895cb1ec458c2f565a7e4d06f7a1f6200e7aa

7 years agoMerge "Cope with argv[0] being null in the dynamic linker."
Treehugger Robot [Thu, 23 Feb 2017 03:29:42 +0000 (03:29 +0000)]
Merge "Cope with argv[0] being null in the dynamic linker."

7 years agoMerge "Allocate thread local buffers in __init_tls."
Josh Gao [Thu, 23 Feb 2017 03:24:44 +0000 (03:24 +0000)]
Merge "Allocate thread local buffers in __init_tls."

7 years agoCope with argv[0] being null in the dynamic linker.
Elliott Hughes [Thu, 23 Feb 2017 01:31:41 +0000 (17:31 -0800)]
Cope with argv[0] being null in the dynamic linker.

Somewhat unsurprisingly, very few commands are happy to be run like this,
in particular multiplexed commands like toybox. But that's no reason for
the linker to get in the way too.

Bug: http://b/33276926
Test: new test
Change-Id: I6dd71ea0183f4da83571039c2198ebb6ed38520e

7 years agoAllocate thread local buffers in __init_tls.
Josh Gao [Wed, 22 Feb 2017 20:19:05 +0000 (12:19 -0800)]
Allocate thread local buffers in __init_tls.

Thread local buffers were using pthread_setspecific for storage with
lazy initialization. pthread_setspecific shares TLS slots between the
linker and libc.so, so thread local buffers being initialized in a
different order between libc.so and the linker meant that bad things
would happen (manifesting as snprintf not working because the
locale was mangled)

Bug: http://b/20464031
Test: /data/nativetest64/bionic-unit-tests/bionic-unit-tests
      everything passes
Test: /data/nativetest/bionic-unit-tests/bionic-unit-tests
      thread_local tests are failing both before and after (KUSER_HELPERS?)
Test: /data/nativetest64/bionic-unit-tests-static/bionic-unit-tests-static
      no additional failures
Change-Id: I9f445a77c6e86979f3fa49c4a5feecf6ec2b0c3f

7 years agoMerge "Add timeout support to __system_property_wait."
Treehugger Robot [Wed, 22 Feb 2017 01:43:30 +0000 (01:43 +0000)]
Merge "Add timeout support to __system_property_wait."

7 years agoMerge "Downgrade the special case of pthread_t(0) to a warning."
Treehugger Robot [Wed, 22 Feb 2017 00:37:08 +0000 (00:37 +0000)]
Merge "Downgrade the special case of pthread_t(0) to a warning."

7 years agoFix pre-L MB_CUR_MAX.
Dan Albert [Tue, 21 Feb 2017 23:54:58 +0000 (15:54 -0800)]
Fix pre-L MB_CUR_MAX.

The default locale is POSIX, not C.UTF-8. POSIX explicitly states
that MB_CUR_MAX is 1 for the POSIX locale:

http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/stdlib.h.html

Test: Made change in the NDK, built libc++ against it, ran NDK libc++
      tests.
Bug: None

Change-Id: Ic2f6f96aa4a7f20d619030f41323831d01002715

7 years agoUpdate to kernel headers v4.10.
Christopher Ferris [Tue, 21 Feb 2017 20:35:09 +0000 (12:35 -0800)]
Update to kernel headers v4.10.

Test: Built angler, booted on angler, ran bionic unit tests.
Change-Id: Ia24511e74106116ea84b44ab724865ec492de8f9

7 years agoAdd timeout support to __system_property_wait.
Elliott Hughes [Fri, 17 Feb 2017 01:13:04 +0000 (17:13 -0800)]
Add timeout support to __system_property_wait.

Bug: http://b/35201172
Test: ran tests
Change-Id: I3a78813bf3cd50d1b308ecb3c742f923606c0cc4

7 years agoDowngrade the special case of pthread_t(0) to a warning.
Elliott Hughes [Tue, 21 Feb 2017 21:15:20 +0000 (13:15 -0800)]
Downgrade the special case of pthread_t(0) to a warning.

So far this is the only issue we've hit in vendor code, and we've hit
it several times already. Rather than try to fix bullhead (the current
problem), let's just admit that the special case of 0 is a lot less
worrying.

Also fix the test expectations to correspond to the new abort message.

Bug: http://b/35455349 (crashes on 0)
Bug: http://b/35622944 (tests)
Test: ran tests
Change-Id: Iec57011fa699a954ebeaec151db2193e36d1ef35

7 years agoMerge "Fix a comment in the pthread_cond_t implementation."
Treehugger Robot [Sun, 19 Feb 2017 05:19:54 +0000 (05:19 +0000)]
Merge "Fix a comment in the pthread_cond_t implementation."

7 years agoFix a comment in the pthread_cond_t implementation.
Elliott Hughes [Sun, 19 Feb 2017 00:35:36 +0000 (16:35 -0800)]
Fix a comment in the pthread_cond_t implementation.

Bug: http://b/34592766
Test: N/A
Change-Id: I341f837dca52865ae5f8fe9ff50ba8999e297179

7 years agoMerge "Fix debug.ld for apps with long names and services"
Treehugger Robot [Sat, 18 Feb 2017 00:50:47 +0000 (00:50 +0000)]
Merge "Fix debug.ld for apps with long names and services"

7 years agoMerge "Revert "Revert "Make libc.so global"""
Treehugger Robot [Sat, 18 Feb 2017 00:18:26 +0000 (00:18 +0000)]
Merge "Revert "Revert "Make libc.so global"""

7 years agoFix debug.ld for apps with long names and services
Dimitry Ivanov [Fri, 17 Feb 2017 22:41:30 +0000 (14:41 -0800)]
Fix debug.ld for apps with long names and services

1. There is no longer limit on property names - remove
   the trimming the name of the property.
2. Make debug.ld work for processes with names ending with ":something"
   This is naming convention for services:
   https://developer.android.com/guide/components/services.html

Bug: http://b/35338922
Bug: http://b/33926793
Test: manual - set ld.debug.app property for the app
Test: from http://b/35338922 and see that it works
Test: for the service as well.
Change-Id: Ic7c6d4edce4a5a22f144496d5c0a3e458217c6e4

7 years agoRevert "Revert "Make libc.so global""
Dimitry Ivanov [Thu, 16 Feb 2017 20:03:21 +0000 (12:03 -0800)]
Revert "Revert "Make libc.so global""

This reverts commit 65236d43e8e8ea2155c0d64e35f00b3c30b21147.

Bug: http://b/35417197
Bug: http://b/35338922
Test: vogar --classpath out/target/common/obj/JAVA_LIBRARIES/core-tests_intermediates/classes.jack  --mode=device --variant=X32 dalvik.system.JniTest#testGetSuperclass
Change-Id: Ibf3bb7396e7d11ebe95e2f9267d0fc4af3fbe8c5

7 years agoMerge "Cleanup now that we don't need __NDK_FPABI__."
Treehugger Robot [Fri, 17 Feb 2017 21:11:26 +0000 (21:11 +0000)]
Merge "Cleanup now that we don't need __NDK_FPABI__."

7 years agoMerge "Include the pthread_t in the "bad pthread_t" fatal abort."
Treehugger Robot [Fri, 17 Feb 2017 20:44:43 +0000 (20:44 +0000)]
Merge "Include the pthread_t in the "bad pthread_t" fatal abort."

7 years agoMerge "Add tests for ptrace resumption behavior."
Josh Gao [Fri, 17 Feb 2017 19:35:58 +0000 (19:35 +0000)]
Merge "Add tests for ptrace resumption behavior."

7 years agoCleanup now that we don't need __NDK_FPABI__.
Dan Albert [Fri, 17 Feb 2017 19:08:53 +0000 (11:08 -0800)]
Cleanup now that we don't need __NDK_FPABI__.

Test: tools/update_headers.sh && make checkbuild
Bug: None
Change-Id: I1b9986c4d7d5da349a8ad394e8fcf64b87c89fdf

7 years agoInclude the pthread_t in the "bad pthread_t" fatal abort.
Elliott Hughes [Fri, 17 Feb 2017 18:27:45 +0000 (10:27 -0800)]
Include the pthread_t in the "bad pthread_t" fatal abort.

Also reword the message to be stronger.

Bug: http://b/35455349
Test: manual
Change-Id: I8f34fd42f3b635c95a7b921645a016fb303ce3ad

7 years agoMerge changes from topic 'debuggerd_inproc'
Josh Gao [Thu, 16 Feb 2017 23:11:51 +0000 (23:11 +0000)]
Merge changes from topic 'debuggerd_inproc'

* changes:
  linker: use fallback crash handler dumping.
  linker: add android_use_fallback_allocator.

7 years agoAdd tests for ptrace resumption behavior.
Josh Gao [Thu, 16 Feb 2017 22:12:41 +0000 (14:12 -0800)]
Add tests for ptrace resumption behavior.

Add tests that ensure that the kernel behaves properly w.r.t.
resumption of ptraced processes when the tracer dies.

Bug: http://b/34516140
Test: /data/nativetest/bionic-unit-tests/bionic-unit-tests --gtest_filter="PtraceResumption*"
Change-Id: Id35e069a7e5edd6964637dd3f6358ad59db19792

7 years agoMerge "Add __libc_format_buffer_va_list."
Josh Gao [Thu, 16 Feb 2017 03:42:50 +0000 (03:42 +0000)]
Merge "Add __libc_format_buffer_va_list."

7 years agoMerge "Unify linker files under one license (BSD)"
Treehugger Robot [Thu, 16 Feb 2017 02:15:54 +0000 (02:15 +0000)]
Merge "Unify linker files under one license (BSD)"

7 years agolinker: use fallback crash handler dumping.
Josh Gao [Thu, 9 Feb 2017 01:27:20 +0000 (17:27 -0800)]
linker: use fallback crash handler dumping.

Bug: http://b/34684590
Test: mma
Change-Id: I0801a83768f172308339ab2118e492b5df96acd0

7 years agolinker: add android_use_fallback_allocator.
Josh Gao [Thu, 9 Feb 2017 18:54:44 +0000 (10:54 -0800)]
linker: add android_use_fallback_allocator.

Add a function to enable a fallback allocator to use for crash handling
in a signal handler.

Bug: http://b/34684590
Test: crasher PR_SET_NO_NEW_PRIVS
Change-Id: Ifa5de636164f34b8cb2fdec4471c20f8516b6dbe

7 years agoAdd __libc_format_buffer_va_list.
Josh Gao [Wed, 15 Feb 2017 19:46:55 +0000 (11:46 -0800)]
Add __libc_format_buffer_va_list.

Bug: http://b/35367169
Test: m
Change-Id: I133f231d3b93bdef56d06497679320a89c7188a9

7 years agoUnify linker files under one license (BSD)
Dimitry Ivanov [Wed, 15 Feb 2017 23:31:13 +0000 (15:31 -0800)]
Unify linker files under one license (BSD)

Historically we had part of the linker licensed under BSD and
another part under Apache 2 license. This commit makes all the
linker code licensed under BSD license.

Test: m
Change-Id: I11b8163ae75966b5768d3fe992679de376106515

7 years agoMerge changes I9d06ea8a,I2d157024
Treehugger Robot [Wed, 15 Feb 2017 19:14:49 +0000 (19:14 +0000)]
Merge changes I9d06ea8a,I2d157024

* changes:
  Fixup API level guards for new fortify functions.
  Guard the GNU strerror_r with an API check.

7 years agoMerge "add fortified implementations of send/sendto"
Treehugger Robot [Wed, 15 Feb 2017 18:26:49 +0000 (18:26 +0000)]
Merge "add fortified implementations of send/sendto"

7 years agoFixup API level guards for new fortify functions.
Dan Albert [Wed, 15 Feb 2017 03:28:18 +0000 (19:28 -0800)]
Fixup API level guards for new fortify functions.

Test: tools/update_headers.sh && make checkbuild
Bug: None
Change-Id: I9d06ea8a5ee9dc27d957f59e6e84150651ed1c76

7 years agoGuard the GNU strerror_r with an API check.
Dan Albert [Wed, 15 Feb 2017 00:33:06 +0000 (16:33 -0800)]
Guard the GNU strerror_r with an API check.

The deprecated headers have always had only the POSIX definition
available (and it's always been available). With the unified headers
as they are now, we actually make it unavailable for C++ users (C++
implies _GNU_SOURCE) targeting below M. Adding this guard means that
pre-M users will still at least get the POSIX one.

It's not great that moving to M as your target API will actually
change the signature of your strerror_r, but I don't see a better
option here (not until we have the compatibility library, anyway).

Test: make checkbuild
Bug: None
Change-Id: I2d15702467533a826c4ec10fd973ee929d2b562a

7 years agoadd fortified implementations of send/sendto
Daniel Micay [Tue, 14 Feb 2017 01:27:59 +0000 (17:27 -0800)]
add fortified implementations of send/sendto

Bug: None
Test: Bullhead builds+boots; CtsBionicTestCases passes.
Change-Id: I2f137a100f679f7f2145d84b2f29ddd3e96a36ae

7 years agoMerge "Add details on why AT_SYMLINK_NOFOLLOW is dangerous"
Treehugger Robot [Tue, 14 Feb 2017 23:08:12 +0000 (23:08 +0000)]
Merge "Add details on why AT_SYMLINK_NOFOLLOW is dangerous"

7 years agoMerge "Revert "Make libc.so global""
Treehugger Robot [Tue, 14 Feb 2017 22:40:58 +0000 (22:40 +0000)]
Merge "Revert "Make libc.so global""

7 years agoAdd details on why AT_SYMLINK_NOFOLLOW is dangerous
Nick Kralevich [Tue, 14 Feb 2017 17:49:30 +0000 (09:49 -0800)]
Add details on why AT_SYMLINK_NOFOLLOW is dangerous

The comment isn't helpful as-is. Provide some clarifying information.

Test: code compiles. No functional changes.
Change-Id: I5267e0bc68857fdc8a4b3384a2a1b0d37693ee6e

7 years agoRevert "Make libc.so global"
Dimitry Ivanov [Tue, 14 Feb 2017 19:03:26 +0000 (19:03 +0000)]
Revert "Make libc.so global"

This reverts commit 879177c9e12df4efe9f17ff3536005c258f625e8.

Bug: http://b/26833548
Change-Id: I23e0b34777d18e369063185da14c57994807da29

7 years agoMerge "Be more strict about using invalid `pthread_t`s."
Elliott Hughes [Tue, 14 Feb 2017 18:32:18 +0000 (18:32 +0000)]
Merge "Be more strict about using invalid `pthread_t`s."

7 years agoMerge "Add __system_property_wait and return the serial in __system_property_read_cal...
Treehugger Robot [Tue, 14 Feb 2017 03:48:23 +0000 (03:48 +0000)]
Merge "Add __system_property_wait and return the serial in __system_property_read_callback."

7 years agoBe more strict about using invalid `pthread_t`s.
Elliott Hughes [Tue, 14 Feb 2017 01:59:29 +0000 (17:59 -0800)]
Be more strict about using invalid `pthread_t`s.

Another release, another attempt to remove the global thread list.

But this time, let's admit that it's not going away. We can switch to using
a read/write lock for the global thread list, and to aborting rather than
quietly returning ESRCH if we're given an invalid pthread_t.

This change affects pthread_detach, pthread_getcpuclockid,
pthread_getschedparam/pthread_setschedparam, pthread_join, and pthread_kill:
instead of returning ESRCH when passed an invalid pthread_t, if you're
targeting O or above, they'll abort with the message "attempt to use
invalid pthread_t".

Note that this doesn't change behavior as much as you might think: the old
lookup only held the global thread list lock for the duration of the lookup,
so there was still a race between that and the dereference in the caller,
given that callers actually need the tid to pass to some syscall or other,
and sometimes update fields in the pthread_internal_t struct too.

(This patch replaces such users with calls to pthread_gettid_np, which
at least makes the TOCTOU window smaller.)

We can't check thread->tid against 0 to see whether a pthread_t is still
valid because a dead thread gets its thread struct unmapped along with its
stack, so the dereference isn't safe.

Taking the affected functions one by one:

    * pthread_getcpuclockid and pthread_getschedparam/pthread_setschedparam
      should be fine. Unsafe calls to those seem highly unlikely.

    * Unsafe pthread_detach callers probably want to switch to
      pthread_attr_setdetachstate instead, or using
      pthread_detach(pthread_self()) from the new thread's start routine
      rather than doing the detach in the parent.

    * pthread_join calls should be safe anyway, because a joinable thread
      won't actually exit and unmap until it's joined. If you're joining an
      unjoinable thread, the fix is to stop marking it detached. If you're
      joining an already-joined thread, you need to rethink your design.

    * Unsafe pthread_kill calls aren't portably fixable. (And are obviously
      inherently non-portable as-is.) The best alternative on Android is to
      use pthread_gettid_np at some point that you know the thread to be
      alive, and then call kill/tgkill directly.

      That's still not completely safe because if you're too late, the tid
      may have been reused, but then your code is inherently unsafe anyway.

Bug: http://b/19636317
Test: ran tests
Change-Id: I0372c4428e8a7f1c3af5c9334f5d9c25f2c73f21

7 years agoMerge "Do not use std::vector in android_namespace_t::is_accessible"
Treehugger Robot [Mon, 13 Feb 2017 21:43:48 +0000 (21:43 +0000)]
Merge "Do not use std::vector in android_namespace_t::is_accessible"

7 years agoAdd __system_property_wait and return the serial in __system_property_read_callback.
Elliott Hughes [Sat, 11 Feb 2017 02:13:46 +0000 (18:13 -0800)]
Add __system_property_wait and return the serial in __system_property_read_callback.

In order to implement android::base::WaitForProperty well, we need a way to
wait not for *any* property to change (__system_property_wait_any), but to
specifically wait for the property represented by a given `prop_info` to
change.

The android::base::WaitForProperty implementation, like attempts to cache
system properties in the past, also needs a way to keep serials and values
in sync, but the existing functions don't provide a cheap way to get a
consistent snapshot. Change the __system_property_read_callback callback's
type to include the serial corresponding to the given value.

Add a test, slightly clean up some of the existing tests (and name them to
include the names of the functions they're testing, in our usual style).

Bug: http://b/35201172
Test: ran tests
Change-Id: Ibc8ebe2e88eef1e333a1bd3dd7f68135f1ba7fb5

7 years agoMerge "Make libc.so global"
Treehugger Robot [Mon, 13 Feb 2017 20:29:51 +0000 (20:29 +0000)]
Merge "Make libc.so global"

7 years agoMerge "Match __bos0 to __pass_object_size0 in FORTIFY"
Treehugger Robot [Mon, 13 Feb 2017 19:39:28 +0000 (19:39 +0000)]
Merge "Match __bos0 to __pass_object_size0 in FORTIFY"

7 years agoDo not use std::vector in android_namespace_t::is_accessible
Dimitry Ivanov [Mon, 13 Feb 2017 18:49:40 +0000 (10:49 -0800)]
Do not use std::vector in android_namespace_t::is_accessible

Avoid constructing vector and walking all the parents of a soinfo
to check if it is accessible. The most likely scenario that the
very first check returns true.

Bug: http://b/35313368
Test: bionic-unit-tests --gtest_filter=dl*:Dl*
Change-Id: I06c65cf61ed1c30e5e454a169de4c41038863587

7 years agoMake libc.so global
Dimitry Ivanov [Mon, 13 Feb 2017 17:17:21 +0000 (09:17 -0800)]
Make libc.so global

This is a way to avoid loading multiple libc.so
when non-default namespace search path includes
/system/lib. This is used by some art tests.

Bug: http://b/26833548
Test: m -j32 test-art-target-run-test-004-JniTest
Change-Id: I919d3a0560bd3c9ac19df21a235641a667f0f017

7 years agoMerge "Replace public library list with shared lib sonames (part 2/2)"
Dimitry Ivanov [Sat, 11 Feb 2017 19:26:48 +0000 (19:26 +0000)]
Merge "Replace public library list with shared lib sonames (part 2/2)"

7 years agoMerge "loader: fix d-tor call order"
Dimitry Ivanov [Sat, 11 Feb 2017 05:50:53 +0000 (05:50 +0000)]
Merge "loader: fix d-tor call order"

7 years agoMatch __bos0 to __pass_object_size0 in FORTIFY
George Burgess IV [Fri, 10 Feb 2017 21:56:22 +0000 (13:56 -0800)]
Match __bos0 to __pass_object_size0 in FORTIFY

pass_object_size(N) forwards the result of __builtin_object_size(param,
N) to a function. So, a function that looks like:

  size_t foo(void *const p __pass_object_size) { return __bos0(p); }
  int bar = foo(baz);

would effectively be turned into

  size_t foo(void *const p, size_t sz) { return sz; }
  int bar = foo(baz, __bos(baz)); // note that this is not __bos0

This is bad, since if we're using __bos0, we want more relaxed
objectsize checks.

__bos0 should be more permissive than __bos in all cases, so this
change Should Be Fineā„¢.

This change also makes GCC and clang share another function's
implementation (recv). I just realized we need to add special
diagnostic-related overloads bits for clang to it, but I can do that in
another patch.

Bug: None
Test: Bullhead builds and boots; CtsBionicTestCases passes.
Change-Id: I6818d0041328ab5fd0946a1e57321a977c1e1250

7 years agoloader: fix d-tor call order
Dimitry Ivanov [Fri, 10 Feb 2017 19:04:20 +0000 (11:04 -0800)]
loader: fix d-tor call order

In the case when there are multiple dependencies on
the same library in the local_group the unload may
in some situations (covered now by tests) result
calling d-tors for some libraries prematurely.

In order to have correct call order loader checks if this
is last dependency in local group before adding it to BFS
queue.

Bug: http://b/35201832
Test: bionic-unit-tests --gtest_filter=dl*:Dl*
Test: bionic-unit-tests-glibc --gtest_filter=dl*
Change-Id: I4c6955b9032acc7147a51d9f09b61d9e0818700c

7 years agoMerge "Replace public library list with shared lib sonames (part 1/2)"
Treehugger Robot [Fri, 10 Feb 2017 19:01:18 +0000 (19:01 +0000)]
Merge "Replace public library list with shared lib sonames (part 1/2)"

7 years agoReplace public library list with shared lib sonames (part 2/2)
Dimitry Ivanov [Fri, 3 Feb 2017 22:07:34 +0000 (14:07 -0800)]
Replace public library list with shared lib sonames (part 2/2)

This commit updates interface of libdl.c.

1. android_init_namespaces is replaces with android_init_anonymous_namespace
2. added 2 arguments to android_create_namespace to specify linked namespace
   and the list of shared libraries sonames.
3. symbol lookup does not get past boundary libraries (added check and test for it).

Bug: http://b/26833548
Bug: http://b/21879602
Test: bionic-unit-tests --gtest_filter=dl*:Dl*
Change-Id: I32921da487a02e5bd0d2fc528904d1228394bfb9

7 years agoReplace public library list with shared lib sonames (part 1/2)
Dimitry Ivanov [Wed, 1 Feb 2017 23:28:52 +0000 (15:28 -0800)]
Replace public library list with shared lib sonames (part 1/2)

Replace public library list with shared lib sonames
which are property of a link between namespaces

This change does not touch any external interfaces
so from outside it behaves almost as it was before

One significant difference is that there is no longer
need to preload public libraries.

Bug: http://b/26833548
Test: bionic-unit-tests --gtest_filter=dl*:Dl*
Change-Id: I57e44e18a9b4f07dcd6556436346be52f52b79d7

7 years agoMerge "Removing the kuser_helper elf note from building"
Treehugger Robot [Fri, 10 Feb 2017 06:24:31 +0000 (06:24 +0000)]
Merge "Removing the kuser_helper elf note from building"

7 years agoRemoving the kuser_helper elf note from building
mspector [Thu, 9 Feb 2017 01:01:27 +0000 (17:01 -0800)]
Removing the kuser_helper elf note from building

This change removes the kuser_helper note from building automatically
with any binary but leaves the note.
Also fixes a typo in the note.

Original patch adding the note:
Ib8366e2a0810092b71381d57dee4bdaa56369a24

Bug: 34815073
Test: Manual - Note is no longer added to binaries
Change-Id: Ieb81f9d9127d1f8a522434a31c696d743238e2a5

7 years agoMerge "Remove unnecessary memsets in linker allocators"
Treehugger Robot [Fri, 10 Feb 2017 02:32:16 +0000 (02:32 +0000)]
Merge "Remove unnecessary memsets in linker allocators"

7 years agoMerge "libc: add clang FORTIFY support"
Treehugger Robot [Fri, 10 Feb 2017 02:07:28 +0000 (02:07 +0000)]
Merge "libc: add clang FORTIFY support"