This was originally motivated by noticing that we were setting the
wrong bits for the well-known tls entries. That was a harmless bug
because none of the well-known tls entries has a destructor, but
it's best not to leave land mines lying around.
Also add some missing POSIX constants, a new test, and fix
pthread_key_create's return value when we hit the limit.
Change-Id: Ife26ea2f4b40865308e8410ec803b20bcc3e0ed1
Also update the x86 asm.h to support this; we need it for libm assembler
anyway.
Also clean up the _FBSDID hack in <sys/cdefs.h>.
Change-Id: Iababd977b8110ec022bf7c93f4d62ece47630e7c
This brings us up to date with FreeBSD HEAD, fixes various bugs, unifies
the set of functions we support on ARM, MIPS, and x86, fixes "long double",
adds ISO C99 support, and adds basic unit tests.
It turns out that our "long double" functions have always been broken
for non-normal numbers. This patch fixes that by not using the upstream
implementations and just forwarding to the regular "double" implementation
instead (since "long double" on Android is just "double" anyway, which is
what BSD doesn't support).
All the tests pass on ARM, MIPS, and x86, plus glibc on x86-64.
Bug: 3169850
Bug: 8012787
Bug: https://code.google.com/p/android/issues/detail?id=6697
Change-Id: If0c343030959c24bfc50d4d21c9530052c581837
Replace a kernel header file dependency with files from NetBSD.
They're more complete, and ELF is ELF, whether you're on Linux or a BSD.
Bug: 7973611
Change-Id: I83ee719e7efdf432ec2ddbe8be271d05b2f558d7
prctl.h uses __BEGIN_DECLS but fails to include sys/cdefs.h
(where it's defined). Code which includes prctl.h without
previously including sys/cdefs.h will fail to compile.
Fixed.
Change-Id: If4c9f3308f08b93596dcd00e351ae786807e9320
Currently, system properties are passed via the environment
variable ANDROID_PROPERTY_WORKSPACE and a file descriptor passed
from parent to child. This is insecure for setuid executables,
as the environment variable can be changed by the caller.
Modify system property handling so that we get the properties
from a root owned properties file, rather than using an
environment variable. Fall back to the environment variable
if the file doesn't exist.
Bug: 8045561
Change-Id: I54f3efa98cf7d63d88788da5ce0d19e34fd7851a
Don't do the fortify_source checks if we can determine, at
compile time, that the provided operation is safe.
This avoids silliness like calling fortify source on things like:
size_t len = strlen("asdf");
printf("%d\n", len);
and allows the compiler to optimize this code to:
printf("%d\n", 4);
Defer to gcc's builtin functions instead of pointing our code
to the libc implementation.
Change-Id: I5e1dcb61946461c4afaaaa983e39f07c7a0df0ae
Add signalfd() call to bionic.
Adding the signalfd call was done in 3 steps:
- add signalfd4 system call (function name and syscall
number) to libc/SYSCALLS.TXT
- generate all necessary headers by calling
libc/tools/gensyscalls.py. This patch is adding
the generated files since the build system
does not call gensyscalls.py.
- create the signalfd wrapper in signalfd.cpp and add
the function prototype to sys/signalfd.h
(cherry-pick of 0c11611c11, modified to
work with older versions of GCC still in use on some branches.)
Change-Id: I4c6c3f12199559af8be63f93a5336851b7e63355
Spotted while running the tests on MIPS, where sigset_t is
actually large enough. The bits in sigset_t are used such that
signal 1 is represented by bit 0, so the range of signals is
actually [1, 8*sizeof(sigset_t)]; it seems clearer to reword
the code in terms of valid bit offsets [0, 8*sizeof(sigset_t)),
which leads to the usual bounds checking idiom.
Change-Id: Id899c288e15ff71c85dd2fd33c47f8e97aa1956f
Previously we'd been relying on getting the machine-specific <endian.h>
instead of the top-level <endian.h>, and <sys/endian.h> was basically broken.
Now, with this patch and the previous patch we should have <endian.h>
and <sys/endian.h> behaving the same. This is basically how NetBSD's endian.h
works, and was probably how ours was originally intended to work.
Bug: http://code.google.com/p/android/issues/detail?id=39824
Change-Id: I71de5a507e633de166013a658b5764df9e1aa09c
These checks haven't been as useful as I hoped, and it's
causing a false positive finding. Remove the overlap
compile time checks.
Change-Id: I5d45dde10ae4663d728230d41fa904adf20acaea
You could argue that this is hurting people smart enough to have manually
allocated a large-enough sigset_t, but those people are smart enough to
implement their own sigset functions too.
I wonder whether our least unpleasant way out of our self-inflicted 32-bit
cesspool is to have equivalents of _FILE_OFFSET_BITS such as _SIGSET_T_BITS,
so calling code could opt in? You'd have to be careful passing sigset_t
arguments between code compiled with different options.
Bug: 5828899
Change-Id: I0ae60ee8544835b069a2b20568f38ec142e0737b
The near duplicates upset fussier compilers that insist that
typedefs be exactly the same, but the fix isn't to make all
copies identical...
Change-Id: Icfdace41726f36ec33c9ae919dbb5a54d3529cc9
Define the macros ACCESSPERMS, ALLPERMS and DEFFILEMODE.
These macros originates from BSD but has been available in glibc
for quite some time.
Change-Id: I429cd30aa4e73f53b153ee7740070cebba166c57
__WINT_TYPE__ type provided by gcc. It references to unsigned int
type for android and linux. Patch corrects wint_t typedef to
__WINT_TYPE__.
Signed-off-by: Sergey Melnikov <sergey.melnikov@intel.com>
Change-Id: Iabeb9fcb0b7bb303a8b220043e339126f125dd68
The declaration for alphasort() in <dirent.h> used the deprecated:
int alphasort(const void*, const void*);
while both Posix and GLibc use instead:
int alphasort(const struct dirent** a, const struct dirent** b);
See: http://pubs.opengroup.org/onlinepubs/9699919799/functions/alphasort.html
This patch does the following:
- Update the declaration to match Posix/GLibc
- Get rid of the upstream BSD code which isn't compatible with the new
signature.
- Implement a new trivial alphasort() with the right signature, and
ensure that it uses strcoll() instead of strcmp().
- Remove Bionic-specific #ifdef .. #else .. #endif block in
dirent_test.cpp which uses alphasort().
Even through strcoll() currently uses strcmp(), this does the right
thing in the case where we decide to update strcoll() to properly
implement locale-specific ordered comparison.
Change-Id: I4fd45604d8a940aaf2eb0ecd7d73e2f11c9bca96
This patch updates the C library headers to provide ucontext_t
definitions for three architectures.
+ Fix <signal.h> to always define 'struct sigcontext'.
The new declarations are announced with new macros defined in
<sys/cdefs.h> in order to make it easier to adapt client code
that already defines its own, incompatible, versions of the
structures seen here.
http://code.google.com/p/android/issues/detail?id=34784
Change-Id: Ie78c48690a4ce61c50593f6c39639be7fead3596
LONG_LONG_MIN, LONG_LONG_MAX and ULONG_LONG_MAX are
GLibc-specific macros that are better defined in <limits.h>
instead of the current exotic location (<pthread.h>).
Note that GCC's <limits.h> only defines these macros
when __GNU_LIBRARY__ is also defined. This is only the
case when building against GLibc, so manually redefine
the macros here.
Note that using LLONG_MIN/LLONG_MAX/ULLONG_MAX is the
C99-compliant way to get these values, but it's easier
to define these compatibility macros for the sake of
porting existing code.
Change-Id: I8023918d73b4685238054932f94a4006c1ca7d03
perf_event_open syscall has a different syscall number for
the 3 supported architectures: arm, x86 and mips. Currenlty
only the arm syscall number is defined for all architectures.
Tracing tools like perf will not work on other architectures
than arm.
Add the different values for perf_event_open on x86 and mips
and run gensyscalls.py to update generated headers.
Change-Id: I2ed78bd42c0e5df8dbc51d784be49cccda5fab30
Author: Irina Tirdea <irina.tirdea@intel.com>
Signed-off-by: Irina Tirdea <irina.tirdea@intel.com>
Signed-off-by: Shuo Gao <shuo.gao@intel.com>
Signed-off-by: Bruce Beare <bruce.j.beare@intel.com>
Signed-off-by: Jack Ren <jack.ren@intel.com>
There's a (bad) definition of NULL in <linux/stddef.h>, and this
definition of NULL, and the One True definition in the <stddef.h> provided
by the compiler. This change at least kills one of the bad duplicates.
Killing the <linux/stddef.h> one is harder, because it's in a generated
file.
Change-Id: Iea4ccb12d6758199f312ea9cd753b84322d5c471
Some userspace programs (e.g. perf) need getline.
Changes:
() add getdelim.c, getline.c from NetBSD (http://netbsd.org/) under the
NetBSD Foundation's (TNF) license ("2 clause" Berkeley-style license).
() add stub for reentrant.h header that is needed by getdelim.c
() add tests for getdelim(3) and getline(3).
() update NOTICE file.
Change-Id: I22ed82dd5904b9d7a3695535c04f502be3c27c5d
Signed-off-by: Irina Tirdea <irina.tirdea@intel.com>
This makes the constants correspond to those in the Linux kernel's
include/linux/syslog.h, but keeping our old badly-named constants for
source compatibility.
Change-Id: Ia47d1299205754cbfffd29ed48b497b53d1edaae
Please see "man 3 ftw" for a description of the
ftw / nftw functions.
This code is taken directly from netbsd unmodified.
Change-Id: Ia4879ac57212b424adf5281b5e92858e216d0f14
