The visibility control in pthread_atfork.h is incorrect.
It breaks 64bit libc.so by hiding pthread_atfork.
This reverts commit 6df122f852.
Change-Id: I21e4b344d500c6f6de0ccb7420b916c4e233dd34
This doesn't affect code like Chrome that correctly ignores EINTR on
close, makes code that tries TEMP_FAILURE_RETRY work (where before it might
have closed a different fd and appeared to succeed, or had a bogus EBADF),
and makes "goto fail" code work (instead of mistakenly assuming that EINTR
means that the close failed).
Who loses? Anyone actively trying to detect that they caught a signal while
in close(2). I don't think those people exist, and I think they have better
alternatives available.
Bug: https://code.google.com/p/chromium/issues/detail?id=269623
Bug: http://b/20501816
Change-Id: I11e2f66532fe5d1b0082b2433212e24bdda8219b
NVIDIA binary blobs are assuming that __cache_clear, _Unwind_Backtrace,
and _Unwind_GetIP are all in some library that they link, but now we've
cleaned up this leakage, they're no longer getting it. Deliberately leak
the symbols from libc.so until we get new blobs.
Bug: http://b/20065774
Change-Id: I92ef07b2bce8d1ad719bf40dab41d745cd6904d4
This used to be handled by -fvisibility=hidden on libc_cxa, but that
was broken by the resolution of https://llvm.org/PR22419 (introduced
to Android in today's clang update).
Now we just use a version script that prevents these from being
re-exported from our shared libraries.
Change-Id: Ib290e1d0d7426e09ad17a91178162fff6dbdcfa9
This is initial implementations; does not yet handle
dlclose - undefined behavior, needs linker support to
handle it right.
Bug: 19800080
Bug: 16696563
Change-Id: I7a3e21ed7f7ec01e62ea1b7cb2ab253590ea0686
As glibc/netbsd don't protect access to thread struct members by a global
lock, we don't want to do it either. This change reduces the
responsibility of g_thread_list_lock to only protect g_thread_list.
Bug: 19636317
Change-Id: I897890710653dac165d8fa4452c7ecf74abdbf2b
The kernel system call faccessat() does not have any flags arguments,
so passing flags to the kernel is currently ignored.
Fix the kernel system call so that no flags argument is passed in.
Ensure that we don't support AT_SYMLINK_NOFOLLOW. This non-POSIX
(http://pubs.opengroup.org/onlinepubs/9699919799/functions/access.html)
flag is a glibc extension, and has non-intuitive, error prone behavior.
For example, consider the following code:
symlink("foo.is.dangling", "foo");
if (faccessat(AT_FDCWD, "foo", R_OK, AT_SYMLINK_NOFOLLOW) == 0) {
int fd = openat(AT_FDCWD, "foo", O_RDONLY | O_NOFOLLOW);
}
The faccessat() call in glibc will return true, but an attempt to
open the dangling symlink will end up failing. GLIBC documents this
as returning the access mode of the symlink itself, which will
always return true for any symlink on Linux.
Some further discussions of this are at:
* http://lists.landley.net/pipermail/toybox-landley.net/2014-September/003617.html
* http://permalink.gmane.org/gmane.linux.lib.musl.general/6952
AT_SYMLINK_NOFOLLOW seems broken by design. I suspect this is why this
function was never added to POSIX. (note that "access" is pretty much
broken by design too, since it introduces a race condition between
check and action). We shouldn't support this until it's clearly
documented by POSIX or we can have it produce intuitive results.
Don't support AT_EACCESS for now. Implementing it is complicated, and
pretty much useless on Android, since we don't have setuid binaries.
See http://git.musl-libc.org/cgit/musl/commit/?id=0a05eace163cee9b08571d2ff9d90f5e82d9c228
for how an implementation might look.
Bug: 18867827
Change-Id: I25b86c5020f3152ffa3ac3047f6c4152908d0e04
We still have issues with clang coverage in static libraries, so we
need to make sure we follow suit with the rest of libc for now.
Bug: 17574078
Change-Id: I2ab58a84b1caa0d8d08415d240c35adec5b1e150
The replacement new failures present in newer versions are present
here as well, with the following new issues:
XPASS std/localization/locale.categories/category.numeric/locale.nm.put/facet.num.put.members/put_long_double.pass.cpp
This is from the -NaN formatting fix in bionic. We previously had this
wrong, and the upstream test is also wrong. There's currently an XFAIL
for Android in this test because I haven't fixed the upstream test
yet. After that is done, I'll need to teach the test runner how to
XFAIL older Android versions...
FAIL std/localization/locale.categories/category.ctype/facet.ctype.special/facet.ctype.char.dtor/dtor.pass.cpp
dtor.pass.cpp:39: int main(): assertion "globalMemCounter.checkDeleteArrayCalledEq(1)" failed
Haven't investigated this one yet. http://b/19412688
Note that this also needs the libgcc link ordering to be fixed in the
build system, as we'll otherwise depend on libgcc symbols from libc
that may or may not have been there.
The build fix can't be submitted because the proper link order causes
the libgcc unwinder to be used instead of the EHABI one:
http://b/18471342
Bug: 18471532
Change-Id: Icf560485a9b8f5ebbe01e4458703e62ec94df5e1
Now passes all libc++ tests for these targets, with the exception of
the usual failing replacement new tests since libc uses new/delete for
things. I don't know if we can ever really fix these.
Bug: 18471532
Change-Id: Ibc0a15f26b0e4613249b5e15ecf3cf80e523467c
The .note.android.ident section is only used by GDB, which doesn't
care what section type the section is, but it would be convenient
for readelf -n to be able to find the section too.
The old way of getting the .note.android.ident section to be of type
SH_NOTE involved compiling from .c to .s using gcc, running sed to
change progbits to note, and then compiling from .s to .o using gcc.
Since crtbrand.c only contains a section containing data, a
crtbrand.S can be checked in that will compile on all platforms,
avoiding the need for sed.
Also add crtbrand.o to crtbegin_so.o so that libraries also get
the note, and to the crt workaround in arm libc.so.
Change-Id: Ica71942a6af4553b56978ceaa288b3f4c15ebfa2
* changes:
Use LOCAL_LDFLAGS_64 instead of enumerating 64-bit architectures
Fix typo in cpu variant makefile depenendency for arm64
Remove libc_static_common_src_files
Share LP32 makefile settings between arches
libc_static_common_src_files is never set after
c3f114037d, remove the remaining
references to it.
Change-Id: I66364a5c1b031ad69d608f6f44244049192944f6
Add <var>_32 to patch-up-arch-specific-flags, and move the LP32
cruft varaibles from the 32-bit arch specific makefiles into the
top level Android.mk.
Change-Id: Id3fcf6805d4af048c2524c94b1295416ebe7d057
Many libc functions have an option to not follow symbolic
links. This is useful to avoid security sensitive code
from inadvertantly following attacker supplied symlinks
and taking inappropriate action on files it shouldn't.
For example, open() has O_NOFOLLOW, chown() has
lchown(), stat() has lstat(), etc.
There is no such equivalent function for chmod(), such as lchmod().
To address this, POSIX introduced fchmodat(AT_SYMLINK_NOFOLLOW),
which is intended to provide a way to perform a chmod operation
which doesn't follow symlinks.
Currently, the Linux kernel doesn't implement AT_SYMLINK_NOFOLLOW.
In GLIBC, attempting to use the AT_SYMLINK_NOFOLLOW flag causes
fchmodat to return ENOTSUP. Details are in "man fchmodat".
Bionic currently differs from GLIBC in that AT_SYMLINK_NOFOLLOW
is silently ignored and treated as if the flag wasn't present.
This patch provides a userspace implementation of
AT_SYMLINK_NOFOLLOW for bionic. Using open(O_PATH | O_NOFOLLOW),
we can provide a way to atomically change the permissions on
files without worrying about race conditions.
As part of this change, we add support for fchmod on O_PATH
file descriptors, because it's relatively straight forward
and could be useful in the future.
The basic idea behind this implementation comes from
https://sourceware.org/bugzilla/show_bug.cgi?id=14578 , specifically
comment #10.
Change-Id: I1eba0cdb2c509d9193ceecf28f13118188a3cfa7
Interestingly, this mostly involves cleaning up our implementation of
various <string.h> functions.
Change-Id: Ifaef49b5cb997134f7bc0cc31bdac844bdb9e089
Static libraries are painful. Details are in the comment. I'll try to
get prebuilts in to the tree so I can fix this on Monday. Until then,
this isn't actually a regression because we've not had coverage
available for this library until now anyway.
Bug: 17574078
Change-Id: I7505c8a94007203e15a6cf192caa06004849d7d9
We know we can safely statically link `libm`, since it doesn't have
any dependencies on the OS or the layout of a data type that has
changed between releases (like `pthread_t`).
We can safely statically link `libc_syscalls` because the user can
check for and handle `ENOSYS`.
Update `ndk_missing_symbols.py` to account for symbols that are in the
compatibility library.
Improve `symbols.py` to be able to pull symbols from a static library.
Change-Id: Ifb0ede1e8b4a8f0f33865d9fed72fb8b4d443fbc
This is correctness rather than performance, but found while investigating
performance.
Bug: 18593728
Change-Id: Idbdfed89d1931fcfae65db29d662108d4bbd9b65
change to behaviour the same as glibc for the check about buflen
Change-Id: I98265a8fe441df6fed2527686f89b087364ca53d
Signed-off-by: Yongqin Liu <yongqin.liu@linaro.org>
Strictly speaking, this only implements the _l variants of the functions
we actually have. We're still missing nl_langinfo_l, for example, but we
don't have nl_langinfo either.
Change-Id: Ie711c7b04e7b9100932a13f5a5d5b28847eb4c12
Unless we completely redo how we build bionic (so that the object
files for libc.so get built separately from libc.a), we can't enable
ASAN here, as libc.a gets linked into static executables.
Change-Id: I2ce4f51248bd51c4213a555ff481b6faabbf53f8
All we're actually interested in is the unwinder. Since that's now a
separate library, just use that.
Change-Id: If86071a0d850da961336a58147b70369ace7bd12
Add the missing prototypes, fix the existing prototypes to use clockid_t
rather than int, fix clock_nanosleep's failure behavior, and add simple
tests.
Bug: 17644443
Bug: https://code.google.com/p/android/issues/detail?id=77372
Change-Id: I03fba369939403918abcabae9551a7123953d780
Signed-off-by: Haruki Hasegawa <h6a.h4i.0@gmail.com>
This means all our stdio implementation is now the OpenBSD implementation.
The only thing we lose is the STDIO_THREAD_LOCK calls but they were no-ops
anyway.
We should probably talk to upstream about this. Either fix the locking or,
preferably, encourage them to move away from this pooling (especially since
there's no eviction policy).
Bug: 17154680
Change-Id: Ie2523e444a7d0965b8d141d57e3e11f6432d5b9a
fpathconf(3) and pathconf(3) can share code. There's no such
header file as <pathconf.h>. glibc/POSIX and BSD disagree about where
the _POSIX_* definitions should go.
Change-Id: I4a67f1595c9f5fbb26700a131178eedebd6bf712
A lot of third-party code calls the private __get_thread symbol,
often as part of a backport of bionic's pthread_rwlock implementation.
Hopefully this will go away for LP64 (since you're guaranteed the
real implementation there), but there are still APIs that take a tid
and no way to convert between a pthread_t and a tid. pthread_gettid_np
is a public API for that. To aid the transition, make __get_thread
available again for LP32.
(cherry-pick of 27efc48814b8153c55cbcd0af5d9add824816e69.)
Bug: 14079438
Change-Id: I43fabc7f1918250d31d4665ffa4ca352d0dbeac1
Enable the -std=gnu++11 flag for libstdc++ static and
dynamic libs.
ScopeGuard uses DISABLE_ macros instead of '= delete';
Change-Id: I07e21b306f95fffd49345f7fa136cfdac61e0225
Clang is still disabled for x86 and x86_64 long double code,
for x86_64 special assembly instruction, and the linker module.
BUG: 17163651
BUG: 17302991
BUG: 17403674
Change-Id: I43c5f5f0ddc0c2a31918f00b57150dadd8f3f26b
This speeds up the debug malloc code by using the original unwinding code.
The only catch is that it has to link in the libc++ arm unwind code or
there will be crashes when attempting to unwind through libc++ compiled
code.
Bug: 16874447
(cherry picked from commit 3f7635f490)
Change-Id: If8a3821cdd95ed481bb496bf2daab449d13790f8
Create a method of disabling the debug allocation code paths so that
it's possible to use the libunwindbacktrace library without any
modifications.
Use this path to create and destroy the maps for the process. It's not
stricly necessary in the init code since the symbols are not modified
until after the initialize calls.
Also, remove the debug_XXX source files that doesn't need to be in libc.so.
Fix the maps reading code since it was completely broken for 64 bit.
Bug: 16408686
Change-Id: I6b02ef6ce26fdb7a59ad1029e7cbba9accceb704
Also clean up the implementation of all the pty functions, add tests,
and fix the stub implementations of ttyname(3) and ttyname_r(3).
Bug: https://code.google.com/p/android/issues/detail?id=58888
Change-Id: I0fb36438cd1abf8d4e87c29415f03db9ba13c3c2
bionic_systrace.h contains an implementation of tracing that
can be used with systrace.py and its associated viewer. pthread_mutex
now uses this tracing to track pthread_mutex contention, which can be
enabled by using the "bionic" command line option to systrace.
Bug: 15116468
Change-Id: I30ed5b377c91ca4c36568a0e647ddf95d4e4a61a
I've also added insque(3) and remque(3) (from NetBSD because the OpenBSD
ones are currently broken for non-circular lists).
I've not added the three hash table functions that should be in this header
because they operate on a single global hash table and thus aren't likely
to be useful.
Bug: https://code.google.com/p/android/issues/detail?id=73719
Change-Id: I97397a7b921e2e860fd9c8032cafd9097380498a
The getentropy_linux.c is lightly modified to build on Android, but we're now
completely in sync with upstream OpenBSD's arc4random implementation.
Change-Id: If32229fc28aba908035fb38703190d41ddcabc95
Since we don't have syslogd on Android and you can't run one on a non-rooted
device, it's more useful if syslog output just goes to the regular Android
logging system.
Bug: 14292866
Change-Id: Icee7f088b97f88ccbdaf471b98cbac7f19f9210a
Also remove __bionic_name_mem which has exactly one caller, and is only
ever expected to be used in this one place.
Change-Id: I833744f91e887639f5b2d1269f966ee9032af207
Code developed for glibc or older versions of bionic might expect more
randomness than the BSD implementation provides.
Bug: 15829381
Change-Id: Ia5a908a816e0a5f0639f514107a6384a51ec157e
The inclusion of the static libc_common library in the malloc_debug_XXX.so
shared libraries causes constructors to be called twice. This doesn't seem
to have caused any issues when setting the libc.debug.malloc property.
However, jemalloc crashes because there are two jemalloc implementations,
one in the static libc_common library and one in the shared library. Each
implementation has created overlapping thread keys that are not the same.
The crash comes because one of the jemalloc keys is actually used by the
locale setting code. Thus if someone sets the locale, the jemalloc code
crashes trying to access the same key.
Change-Id: Iaac650a82d69064db148a6333e9403744f68b4a4
The res_init.c changes bring us a bit closer to upstream too, though
there's still work to be done there. Some of the remaining differences
look like bugs we'd want to fix, so we should definitely try to come
back to that.
Change-Id: I50baa148e967c90d55d711e9904ad54c7d724d4d
Almost all of our stdio is actually OpenBSD, so although this isn't
really a core part of stdio (it doesn't touch struct FILE, for example)
it probably makes sense for it to come from the same upstream. My
actual motivation though is that it's the only FreeBSD file we have
compiler warnings from.
This patch moves us over to -Werror by default, with only the DNS code
having -Wno-error.
Change-Id: Id244a5b445cba41b0a1ca30298ca7b1ed177810c
These symbols should be public (and Firefox uses them), and we'd also probably
rather have the upstream thread-safe implementation.
Bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1030899
Change-Id: I2a5888fbb3198546848398f576fd2195ff3fe00c
This is actually revision 1.33, which is no longer the latest, but it's
as close to head as we can currently reasonably get. I've also switched
to the OpenBSD getentropy_linux.c implementation of getentropy, lightly
modified to try to report an error on failure.
Bug: 14499627
Change-Id: Ia7c561184b1f366c9bf66f248aa60f0d53535fcb
Since this was not done earlier, there are binary compatibility concerns
that prevent us from being able to apply this to LP32.
Bug: 11156955
Change-Id: Ie717c3ae4b81c749548a45a993c834e109700b27
In practice, with this implementation we never need to make a system call.
We get the main thread's tid (which is the same as our pid) back from
the set_tid_address system call we have to make during initialization.
A new pthread will have the same pid as its parent, and a fork child's
main (and only) thread will have a pid equal to its tid, which we get for
free from the kernel before clone returns.
The only time we'd actually have to make a getpid system call now is if
we take a signal during fork and the signal handler calls getpid. (That,
or we call getpid in the dynamic linker while it's still dealing with its
own relocations and hasn't even set up the main thread yet.)
Bug: 15387103
Change-Id: I6d4718ed0a5c912fc75b5f738c49a023dbed5189
Parts of this are just getting us in sync with upstream, but the
'const' stuff is our own mess. We should kill the *_tz functions
and lose this difference from upstream.
Change-Id: I17d26534ed3f54667143d78147a8c53be56d7b33
Rename jemalloc.cpp to jemalloc_wrapper.cpp to avoid problems with
the libc library having two jemalloc.o files that clobber each other.
Change-Id: I9a2d966dbf414b1367ee0ef1f0d73fca6f25b518
getdtablesize(3) was removed fro POSIX 2004. Keep the symbol around in LP32 for
binary compatibility, but remove the declaration from unistd.h.
Bug: 13935372
Change-Id: I1f96cd290bf9176f922dad58bd5a7ab2cae7ef0f
These were both removed from POSIX 2004, and we don't define an
implementation for getw(3). Keep the definition of put(3) on LP32 for
binary compatibility.
Bug: 13935372
Change-Id: Iba384b45093ac6d2d7c2d81f7980cd7701dd6f56
vfork() was removed from POSIX 2008, so this replaces its implementation
with a call to fork().
Bug: 13935372
Change-Id: I6d99ac9e52a2efc5ee9bda1cab908774b830cedc
