Rich Felker [Tue, 20 Sep 2011 15:18:13 +0000 (11:18 -0400)]
fix statvfs.c to match new fsid_t definition
Rich Felker [Tue, 20 Sep 2011 15:16:27 +0000 (11:16 -0400)]
fix the fsid_t structure to match name of __val
this is a case of poorly written man pages not matching the actual
implementation, and why i hate implementing nonstandard interfaces
with no actual documentation of how they're intended to work.
Rich Felker [Tue, 20 Sep 2011 14:58:18 +0000 (10:58 -0400)]
fix broken siginfo_t with _GNU_SOURCE defined
this bug was introduced in a recent patch. the problem we're working
around is that broken GNU software wants to use "struct siginfo"
rather than "siginfo_t", but "siginfo" is not in the reserved
namespace and thus not legal for the standard header to use.
Rich Felker [Tue, 20 Sep 2011 03:35:48 +0000 (23:35 -0400)]
fix the definition of struct statvfs to match lsb abi
at the same time, make struct statfs match the traditional definition
and make it more useful, especially the fsid_t stuff.
Rich Felker [Tue, 20 Sep 2011 02:27:02 +0000 (22:27 -0400)]
add some more siginfo aliases broken software expects...
Rich Felker [Tue, 20 Sep 2011 00:02:12 +0000 (20:02 -0400)]
cleanup redundancy in bits/signal.h versions
Rich Felker [Mon, 19 Sep 2011 21:39:51 +0000 (17:39 -0400)]
fix the type of wchar_t on arm; support wchar_t varying with arch
really wchar_t should never vary, but the ARM EABI defines it as an
unsigned 32-bit int instead of a signed one, and gcc follows this
nonsense. thus, to give a conformant environment, we have to follow
(otherwise L""[0] and L'\0' would be 0U rather than 0, but the
application would be unaware due to a mismatched definition for
WCHAR_MIN and WCHAR_MAX, and Bad Things could happen with respect to
signed/unsigned comparisons, promotions, etc.).
fortunately no rules are imposed by the C standard on the relationship
between wchar_t and wint_t, and WEOF has type wint_t, so we can still
make wint_t always-signed and use -1 for WEOF.
Rich Felker [Mon, 19 Sep 2011 20:59:10 +0000 (16:59 -0400)]
fix incorrect long double parameters on arm (and other future ports)
this was the cause of crashes in printf when attempting to print
floating point values.
Rich Felker [Mon, 19 Sep 2011 15:56:09 +0000 (11:56 -0400)]
fix broken multi-inclusion guard in sys/reg.h
Rich Felker [Sun, 18 Sep 2011 20:44:54 +0000 (16:44 -0400)]
initial commit of the arm port
this port assumes eabi calling conventions, eabi linux syscall
convention, and presence of the kernel helpers at 0xffff0f?0 needed
for threads support. otherwise it makes very few assumptions, and the
code should work even on armv4 without thumb support, as well as on
systems with thumb interworking. the bits headers declare this a
little endian system, but as far as i can tell the code should work
equally well on big endian.
some small details are probably broken; so far, testing has been
limited to qemu/aboriginal linux.
Rich Felker [Sun, 18 Sep 2011 20:42:06 +0000 (16:42 -0400)]
disable dynamic linking/loading code in static libc builds, for now
it does not work, but some configure scripts will falsely detect
support then generate programs that crash when they call dlopen.
Rich Felker [Sun, 18 Sep 2011 20:34:13 +0000 (16:34 -0400)]
cleanup more bits cruft (sysmacros and socket)
Rich Felker [Sun, 18 Sep 2011 19:39:52 +0000 (15:39 -0400)]
more bits junk (tcp.h)
Rich Felker [Sun, 18 Sep 2011 19:31:44 +0000 (15:31 -0400)]
move invariant netinet/in.h stuff out of bits/in.h
Rich Felker [Sun, 18 Sep 2011 14:14:37 +0000 (10:14 -0400)]
overhaul clone syscall wrapping
several things are changed. first, i have removed the old __uniclone
function signature and replaced it with the "standard" linux
__clone/clone signature. this was necessary to expose clone to
applications anyway, and it makes it easier to port __clone to new
archs, since it's now testable independently of pthread_create.
secondly, i have removed all references to the ugly ldt descriptor
structure (i386 only) from the c code and pthread structure. in places
where it is needed, it is now created on the stack just when it's
needed, in assembly code. thus, the i386 __clone function takes the
desired thread pointer as its argument, rather than an ldt descriptor
pointer, just like on all other sane archs. this should not affect
applications since there is really no way an application can use clone
with threads/tls in a way that doesn't horribly conflict with and
clobber the underlying implementation's use. applications are expected
to use clone only for creating actual processes, possibly with new
namespace features and whatnot.
Rich Felker [Sat, 17 Sep 2011 15:41:11 +0000 (11:41 -0400)]
release notes for 0.8.2
Rich Felker [Sat, 17 Sep 2011 15:37:46 +0000 (11:37 -0400)]
dummy implementation of set_thread_area
eventually we may have a working "generic" implementation for archs
that don't need anything special. in any case, the goal of having
stubs like this is to allow early testing of new ports before all the
details needed for threads have been filled in. more functions like
this will follow.
Rich Felker [Fri, 16 Sep 2011 18:39:15 +0000 (14:39 -0400)]
fix assumptions that char is signed
Rich Felker [Fri, 16 Sep 2011 18:33:51 +0000 (14:33 -0400)]
fix more instances of old a_xchg (use new a_swap name)
Rich Felker [Fri, 16 Sep 2011 18:29:24 +0000 (14:29 -0400)]
use a_swap rather than old name a_xchg
Rich Felker [Fri, 16 Sep 2011 16:44:39 +0000 (12:44 -0400)]
fix generic sigsetjmp (unused anyway) pointer signedness error
Rich Felker [Fri, 16 Sep 2011 16:08:20 +0000 (12:08 -0400)]
fix idiotic const-correctness error in lio_listio
i blame this one on posix for using hideous const-qualified double
pointers which are unusable without hideous casts.
Rich Felker [Fri, 16 Sep 2011 14:13:00 +0000 (10:13 -0400)]
fix ptrace (maybe)
Rich Felker [Thu, 15 Sep 2011 17:03:13 +0000 (13:03 -0400)]
implement ptrace syscall wrapper (untested)
Rich Felker [Wed, 14 Sep 2011 23:11:15 +0000 (19:11 -0400)]
0.8.1 release notes
Rich Felker [Wed, 14 Sep 2011 13:37:01 +0000 (09:37 -0400)]
remove incorrectly-made-visible internal dst offset variable
Rich Felker [Wed, 14 Sep 2011 13:36:38 +0000 (09:36 -0400)]
fix inconsistent signature for aio_error
Rich Felker [Wed, 14 Sep 2011 01:21:53 +0000 (21:21 -0400)]
fix return types for aio_read and aio_write again
previous fix was backwards and propagated the wrong type rather than
the right one...
Rich Felker [Wed, 14 Sep 2011 01:09:35 +0000 (21:09 -0400)]
fix various errors in function signatures/prototypes found by nsz
Rich Felker [Tue, 13 Sep 2011 20:12:57 +0000 (16:12 -0400)]
fix cruft left in release notes
Rich Felker [Tue, 13 Sep 2011 19:39:58 +0000 (15:39 -0400)]
fix prototypes for aio_read and aio_write
Rich Felker [Tue, 13 Sep 2011 18:45:59 +0000 (14:45 -0400)]
add missing posix_spawnattr_init/destroy functions
Rich Felker [Tue, 13 Sep 2011 17:41:40 +0000 (13:41 -0400)]
0.8.0 release
Rich Felker [Tue, 13 Sep 2011 13:53:41 +0000 (09:53 -0400)]
remove some stray trailing space characters
Rich Felker [Mon, 12 Sep 2011 04:00:30 +0000 (00:00 -0400)]
implement gnu sigisemptyset
Rich Felker [Mon, 12 Sep 2011 02:50:02 +0000 (22:50 -0400)]
add prototypes for GNU *_unlocked stdio functions
actually these are just weak aliases for the normal locking versions
right now, and they will probably stay that way since making them
lock-free without slowing down the normal versions would require
significant code duplication for no benefit.
Rich Felker [Mon, 12 Sep 2011 02:45:56 +0000 (22:45 -0400)]
add dummied strverscmp (obnoxious GNU function)
programs that use this tend to horribly botch international text
support, so it's questionable whether we want to support it even in
the long term... for now, it's just a dummy that calls strcmp.
Rich Felker [Mon, 12 Sep 2011 02:43:57 +0000 (22:43 -0400)]
declare alloca in stdlib.h when _GNU_SOURCE is defined
Rich Felker [Sun, 11 Sep 2011 16:35:41 +0000 (12:35 -0400)]
fix serious bug in pthread_join
on spurious wakeups/returns from __timedwait, pthread_join would
"succeed" and unmap the thread's stack while it was still running. at
best this would lead to SIGSEGV when the thread resumed execution, but
in the worst case, the thread would later resume executing on top of
another new thread's stack mapped at the same address.
spent about 4 hours tracking this bug down, chasing rare
difficult-to-reproduce stack corruption in a stress test program.
still no idea *what* caused the spurious wakeups; i suspect it's a
kernel bug.
Rich Felker [Sat, 10 Sep 2011 03:10:27 +0000 (23:10 -0400)]
fix pthread_join wait call: thread termination tid futex is not private
this seeme to be the bug that prevented enabling of private futex
support. i'm going to hold off on switching to private futexes until
after the next release, and until i get a chance to audit all
wait/wake calls to make sure they're using the correct private
argument, but with this change it should be safe to enable private
futex support.
Rich Felker [Fri, 9 Sep 2011 05:07:38 +0000 (01:07 -0400)]
implement POSIX asynchronous io
some features are not yet supported, and only minimal testing has been
performed. should be considered experimental at this point.
Rich Felker [Tue, 6 Sep 2011 02:23:06 +0000 (22:23 -0400)]
fix incorrect overflow errors on strtoul, etc.
Rich Felker [Mon, 5 Sep 2011 19:39:36 +0000 (15:39 -0400)]
strptime: fix use of uninitialized dest field in converting integer
Rich Felker [Mon, 5 Sep 2011 01:53:20 +0000 (21:53 -0400)]
more fmemopen null termination fixes
null termination is only added when current size grows.
in update modes, null termination is not added if it does not fit
(i.e. it is not allowed to clobber data).
these rules make very little sense, but that's how it goes..
Rich Felker [Mon, 5 Sep 2011 01:40:42 +0000 (21:40 -0400)]
fix some fmemopen behaviors
read should not be allowed past "current size".
append mode should write at "current size", not buffer size.
null termination should not be written except when "current size" grows.
Rich Felker [Mon, 5 Sep 2011 01:37:07 +0000 (21:37 -0400)]
handle pending cancellation when enabling async cancellation
this is not strictly required by the standard, but without it, there
is a race condition where cancellation arriving just before async
cancellation is enabled might not be acted upon. it is impossible for
a conforming application to work around this race condition since
calling pthread_testcancel after setting async cancellation mode is
not allowed (pthread_testcancel is not specified to be
async-cancel-safe). thus the implementation should be responsible for
eliminating the race, from a quality-of-implementation standpoint.
Rich Felker [Sun, 4 Sep 2011 20:06:38 +0000 (16:06 -0400)]
fmemopen: fix eof handling, hopefully right this time
Rich Felker [Sun, 4 Sep 2011 20:04:28 +0000 (16:04 -0400)]
fmemopen fixes
disallow seek past end of buffer (per posix)
fix position accounting to include data buffered for read
don't set eof flag when no data was requested
Rich Felker [Sun, 4 Sep 2011 14:29:04 +0000 (10:29 -0400)]
memstreams: fix incorrect handling of file pos > current size
the addition is safe and cannot overflow because both operands are
positive when considered as signed quantities.
Rich Felker [Sun, 4 Sep 2011 04:08:32 +0000 (00:08 -0400)]
optimize seek function for memory streams
Rich Felker [Sun, 4 Sep 2011 04:06:01 +0000 (00:06 -0400)]
fix twos complement overflow bug in mem streams boundary check
the expression -off is not safe in case off is the most-negative
value. instead apply - to base which is known to be non-negative and
bounded within sanity.
Rich Felker [Sun, 4 Sep 2011 03:26:17 +0000 (23:26 -0400)]
implement fmemopen
testing so far has been minimal. may need further work.
Rich Felker [Sun, 4 Sep 2011 00:19:51 +0000 (20:19 -0400)]
fix some length calculations in memory streams
Rich Felker [Sat, 3 Sep 2011 23:49:46 +0000 (19:49 -0400)]
implement open_wmemstream
not heavily tested, but it seems to be correct, including the odd
behavior that seeking is in terms of wide character count. this
precludes any simple buffering, so we just make the stream unbuffered.
Rich Felker [Sat, 3 Sep 2011 20:10:30 +0000 (16:10 -0400)]
fix RTLD_NEXT on x86_64
the return address was being truncated to 32 bits, preventing the
dlsym code from determining which module contains the calling code.
Rich Felker [Sat, 3 Sep 2011 04:45:21 +0000 (00:45 -0400)]
implement open_memstream
this is the first attempt, and may have bugs. only minimal testing has
been performed.
Rich Felker [Fri, 2 Sep 2011 18:25:43 +0000 (14:25 -0400)]
fix missing prototypes/wrong signature for psiginfo, psignal
Rich Felker [Sat, 27 Aug 2011 23:42:13 +0000 (19:42 -0400)]
fix broken FD_* macros on 64-bit targets
1 is too small if int is 32-bit but unsigned long is 64-bit. be
explicit and use 1UL.
Rich Felker [Tue, 23 Aug 2011 16:55:47 +0000 (12:55 -0400)]
bring back ___environ symbol (3 underscores)
its existence doesn't hurt anything, and dynamic-linked binaries using
previous versions of musl were wrongly binding to it instead of
__environ.
Rich Felker [Tue, 23 Aug 2011 13:43:45 +0000 (09:43 -0400)]
use new a_crash() asm to optimize double-free handler.
gcc generates extremely bad code (7 byte immediate mov) for the old
null pointer write approach. it should be generating something like
"xor %eax,%eax ; mov %al,(%eax)". in any case, using a dedicated
crashing opcode accomplishes the same thing in one byte.
Rich Felker [Tue, 23 Aug 2011 13:37:39 +0000 (09:37 -0400)]
security hardening: ensure suid programs have valid stdin/out/err
this behavior (opening fds 0-2 for a suid program) is explicitly
allowed (but not required) by POSIX to protect badly-written suid
programs from clobbering files they later open.
this commit does add some cost in startup code, but the availability
of auxv and the security flag will be useful elsewhere in the future.
in particular auxv is needed for static-linked vdso support, which is
still waiting to be committed (sorry nik!)
Rich Felker [Wed, 17 Aug 2011 03:23:15 +0000 (23:23 -0400)]
in pathconf, -1, not 0, means unsupported.. syncio presumably works, too.
Rich Felker [Wed, 17 Aug 2011 03:18:30 +0000 (23:18 -0400)]
fix bogus pathconf result for file size bits
Rich Felker [Tue, 16 Aug 2011 14:38:33 +0000 (10:38 -0400)]
partially working strptime
it's missing at least:
- derived fields
- week numbers
- short year (without century) support
- locale modifiers
Rich Felker [Tue, 16 Aug 2011 12:50:03 +0000 (08:50 -0400)]
ldso: move the suid/secure check code closer to env/auxv processing
this does not change behavior, but the idea is to avoid letting other
code build up between these two points, whereby the environment
variables might get used before security it checked.
Rich Felker [Tue, 16 Aug 2011 11:46:42 +0000 (07:46 -0400)]
honor AT_SECURE aux vector flag
Rich Felker [Tue, 16 Aug 2011 04:42:13 +0000 (00:42 -0400)]
RTLD_NEXT support
the asm wrapper is needed to get the return address without
compiler-specific extensions.
Rich Felker [Tue, 16 Aug 2011 04:24:36 +0000 (00:24 -0400)]
LD_PRELOAD support
Rich Felker [Mon, 15 Aug 2011 05:59:15 +0000 (01:59 -0400)]
simplify and improve double-free check
a valid mmapped block will have an even (actually aligned) "extra"
field, whereas a freed chunk on the heap will always have an in-use
neighbor.
this fixes a potential bug if mmap ever allocated memory below the
main program/brk (in which case it would be wrongly-detected as a
double-free by the old code) and allows the double-free check to work
for donated memory outside of the brk area (or, in the future,
secondary heap zones if support for their creation is added).
Rich Felker [Sun, 14 Aug 2011 19:19:17 +0000 (15:19 -0400)]
typo in macro definitions for x86_64
Rich Felker [Sun, 14 Aug 2011 19:17:36 +0000 (15:17 -0400)]
macro for pthread_equal
no sense bloating apps with a function call for an equality comparison...
Rich Felker [Sat, 13 Aug 2011 12:42:15 +0000 (08:42 -0400)]
fix missing include in last commit
Rich Felker [Sat, 13 Aug 2011 12:40:11 +0000 (08:40 -0400)]
fix clock() function
it previously was returning the pseudo-monotonic-realtime clock
returned by times() rather than process cputime. it also violated C
namespace by pulling in times().
we now use clock_gettime() if available because times() has
ridiculously bad resolution. still provide a fallback for ancient
kernels without clock_gettime.
Rich Felker [Fri, 12 Aug 2011 14:37:12 +0000 (10:37 -0400)]
implement forkall
this is a "nonstandard" function that was "rejected" by POSIX, but
nonetheless had its behavior documented in the POSIX rationale for
fork. it's present on solaris and possibly some other systems, and
duplicates the whole calling process, not just a single thread. glibc
does not have this function. it should not be used in programs
intending to be portable, but may be useful for testing,
checkpointing, etc. and it's an interesting (and quite small) example
of the usefulness of the __synccall framework originally written to
work around deficiencies in linux's setuid syscall.
Rich Felker [Fri, 12 Aug 2011 14:32:22 +0000 (10:32 -0400)]
pthread and synccall cleanup, new __synccall_wait op
fix up clone signature to match the actual behavior. the new
__syncall_wait function allows a __synccall callback to wait for other
threads to continue without returning, so that it can resume action
after the caller finishes. this interface could be made significantly
more general/powerful with minimal effort, but i'll wait to do that
until it's actually useful for something.
Rich Felker [Fri, 12 Aug 2011 05:11:28 +0000 (01:11 -0400)]
more efficient signal blocking for timer threads
due to the barrier, it's safe just to block signals in the new thread,
rather than blocking and unblocking in the parent thread.
Rich Felker [Fri, 12 Aug 2011 02:34:09 +0000 (22:34 -0400)]
normal exit from timer thread should run dtors, restore cancel state
Rich Felker [Thu, 11 Aug 2011 19:54:06 +0000 (15:54 -0400)]
block signals in timer threads
if a timer thread leaves signals unblocked, any future attempt by the
main thread to prevent the process from being terminated by blocking
signals will fail, since the signal can still be delivered to the
timer thread.
Rich Felker [Sun, 7 Aug 2011 22:31:06 +0000 (18:31 -0400)]
condition variable signal/bcast need not wake unless there are waiters
Rich Felker [Sun, 7 Aug 2011 19:48:16 +0000 (15:48 -0400)]
use weak aliase rather than weak reference for vdso clock_gettime
this works around pcc's lack of working support for weak references,
and in principle is nice because it gets us back to the stage where
the only weak symbol feature we use is weak aliases, nothing else.
having fewer dependencies on fancy linker features is a good thing.
Rich Felker [Sun, 7 Aug 2011 15:14:32 +0000 (11:14 -0400)]
simplify unified timed wait code, drop support for newer method
the new absolute-time-based wait kernelside was hard to get right and
basically just code duplication. it could only improve "performance"
when waiting, and even then, the improvement was just slight drop in
cpu usage during a wait.
actually, with vdso clock_gettime, the "old" way will be even faster
than the "new" way if the time has already expired, since it will not
invoke any syscalls. it can determine entirely in userspace that it
needs to return ETIMEDOUT.
Rich Felker [Sun, 7 Aug 2011 08:50:02 +0000 (04:50 -0400)]
add fast path for normal mutexes back to pthread_mutex_lock
Rich Felker [Sun, 7 Aug 2011 04:05:01 +0000 (00:05 -0400)]
close should not be cancellable after "failing" with EINTR
normally we allow cancellation to be acted upon when a syscall fails
with EINTR, since there is no useful status to report to the caller in
this case, and the signal that caused the interruption was almost
surely the cancellation request, anyway.
however, unlike all other syscalls, close has actually performed its
resource-deallocation function whenever it returns, even when it
returned an error. if we allow cancellation at this point, the caller
has no way of informing the program that the file descriptor was
closed, and the program may later try to close the file descriptor
again, possibly closing a different, newly-opened file.
the workaround looks ugly (special-casing one syscall), but it's
actually the case that close is the one and only syscall (at least
among cancellation points) with this ugly property.
Rich Felker [Sun, 7 Aug 2011 00:45:30 +0000 (20:45 -0400)]
ensure the compiler does not move around thread-register-based reads
if gcc decided to move this across a conditional that checks validity
of the thread register, an invalid thread-register-based read could be
performed and raise sigsegv.
Rich Felker [Sun, 7 Aug 2011 00:20:23 +0000 (20:20 -0400)]
simplify multi-threaded errno, eliminate useless function pointer
Rich Felker [Sun, 7 Aug 2011 00:09:51 +0000 (20:09 -0400)]
use weak aliases rather than function pointers to simplify some code
Rich Felker [Fri, 5 Aug 2011 10:43:45 +0000 (06:43 -0400)]
fix off-by-one bug in siglongjmp that caused unpredictable behavior
if saved, signal mask would not be restored unless some low signals
were masked. if not saved, signal mask could be wrongly restored to
uninitialized values. in any, wrong mask would be restored.
i believe this function was written for a very old version of the
jmp_buf structure which did not contain a final 0 field for
compatibility with siglongjmp, and never updated...
Rich Felker [Wed, 3 Aug 2011 23:57:46 +0000 (19:57 -0400)]
further debloat cancellation handlers
cleanup push and pop are also no-ops if pthread_exit is not reachable.
this can make a big difference for library code which needs to protect
itself against cancellation, but which is unlikely to actually be used
in programs with threads/cancellation.
Rich Felker [Wed, 3 Aug 2011 23:50:35 +0000 (19:50 -0400)]
missed detail in cancellation bloat fix
Rich Felker [Wed, 3 Aug 2011 23:45:21 +0000 (19:45 -0400)]
fix static linking dependency bloat with cancellation
previously, pthread_cleanup_push/pop were pulling in all of
pthread_create due to dependency on the __pthread_unwind_next
function. this was not needed, as cancellation cleanup handlers can
never be called unless pthread_exit or pthread_cancel is reachable.
Rich Felker [Wed, 3 Aug 2011 16:13:13 +0000 (12:13 -0400)]
implement if_nameindex and if_freenameindex
Rich Felker [Wed, 3 Aug 2011 14:21:32 +0000 (10:21 -0400)]
overhaul rwlocks to address several issues
like mutexes and semaphores, rwlocks suffered from a race condition
where the unlock operation could access the lock memory after another
thread successfully obtained the lock (and possibly destroyed or
unmapped the object). this has been fixed in the same way it was fixed
for other lock types.
in addition, the previous implementation favored writers over readers.
in the absence of other considerations, that is the best behavior for
rwlocks, and posix explicitly allows it. however posix also requires
read locks to be recursive. if writers are favored, any attempt to
obtain a read lock while a writer is waiting for the lock will fail,
causing "recursive" read locks to deadlock. this can be avoided by
keeping track of which threads already hold read locks, but doing so
requires unbounded memory usage, and there must be a fallback case
that favors readers in case memory allocation failed. and all of this
must be synchronized. the cost, complexity, and risk of errors in
getting it right is too great, so we simply favor readers.
tracking of the owner of write locks has been removed, as it was not
useful for anything. it could allow deadlock detection, but it's not
clear to me that returning EDEADLK (which a buggy program is likely to
ignore) is better than deadlocking; at least the latter behavior
prevents further data corruption. a correct program cannot invoke this
situation anyway.
the reader count and write lock state, as well as the "last minute"
waiter flag have all been combined into a single atomic lock. this
means all state transitions for the lock are atomic compare-and-swap
operations. this makes establishing correctness much easier and may
improve performance.
finally, some code duplication has been cleaned up. more is called
for, especially the standard __timedwait idiom repeated in all locks.
Rich Felker [Wed, 3 Aug 2011 12:36:13 +0000 (08:36 -0400)]
timedwait: play it safe for now
it's unclear whether EINVAL or ENOSYS is used when the operation is
not supported, so check for both...
Rich Felker [Wed, 3 Aug 2011 01:22:56 +0000 (21:22 -0400)]
fix stubbed-out reboot call
Rich Felker [Wed, 3 Aug 2011 01:18:43 +0000 (21:18 -0400)]
correctly handle old kernels without FUTEX_WAIT_BITSET
futex returns EINVAL, not ENOSYS, when op is not supported.
unfortunately this looks just like EINVAL from other causes, and we
end up running the fallback code and getting EINVAL again. fortunately
this case should be rare since correct code should not generate EINVAL
anyway.
Rich Felker [Wed, 3 Aug 2011 01:15:20 +0000 (21:15 -0400)]
fix sem_timedwait bug introduced in timedwait unification
this dec used to be performed by the cancellation handler, which was
called when popped.
Rich Felker [Wed, 3 Aug 2011 01:11:36 +0000 (21:11 -0400)]
unify and overhaul timed futex waits
new features:
- FUTEX_WAIT_BITSET op will be used for timed waits if available. this
saves a call to clock_gettime.
- error checking for the timespec struct is now inside __timedwait so
it doesn't need to be duplicated everywhere. cond_timedwait still
needs to duplicate it to avoid unlocking the mutex, though.
- pushing and popping the cancellation handler is delegated to
__timedwait, and cancellable/non-cancellable waits are unified.
Rich Felker [Wed, 3 Aug 2011 00:31:15 +0000 (20:31 -0400)]
avoid accessing mutex memory after atomic unlock
this change is needed to fix a race condition and ensure that it's
possible to unlock and destroy or unmap the mutex as soon as
pthread_mutex_lock succeeds. POSIX explicitly gives such an example in
the rationale and requires an implementation to allow such usage.
Rich Felker [Tue, 2 Aug 2011 23:59:56 +0000 (19:59 -0400)]
fix breakage in cancellation due to signal functions overhaul
sigaddset was not accepting SIGCANCEL as a valid signal number.
Rich Felker [Tue, 2 Aug 2011 23:19:09 +0000 (19:19 -0400)]
overhaul posix semaphores to fix destructability race
the race condition these changes address is described in glibc bug
report number 12674:
http://sourceware.org/bugzilla/show_bug.cgi?id=12674
up until now, musl has shared the bug, and i had not been able to
figure out how to eliminate it. in short, the problem is that it's not
valid for sem_post to inspect the waiters count after incrementing the
semaphore value, because another thread may have already successfully
returned from sem_wait, (rightly) deemed itself the only remaining
user of the semaphore, and chosen to destroy and free it (or unmap the
shared memory it's stored in). POSIX is not explicit in blessing this
usage, but it gives a very explicit analogous example with mutexes
(which, in musl and glibc, also suffer from the same race condition
bug) in the rationale for pthread_mutex_destroy.
the new semaphore implementation augments the waiter count with a
redundant waiter indication in the semaphore value itself,
representing the presence of "last minute" waiters that may have
arrived after sem_post read the waiter count. this allows sem_post to
read the waiter count prior to incrementing the semaphore value,
rather than after incrementing it, so as to avoid accessing the
semaphore memory whatsoever after the increment takes place.
a similar, but much simpler, fix should be possible for mutexes and
other locking primitives whose usage rules are stricter than
semaphores.