due to moving waiters from the cond var to the mutex in bcast, these
waiters upon wakeup would steal slots in the count from newer waiters
that had not yet been signaled, preventing the signal function from
taking any action.
to solve the problem, we simply use two separate waiter counts, and so
that the original "total" waiters count is undisturbed by broadcast
and still available for signal.
the changes to syscall_ret are mostly no-ops in the generated code,
just cleanup of type issues and removal of some implementation-defined
behavior. the one exception is the change in the comparison value,
which is fixed so that 0xf...f000 (which in principle could be a valid
return value for mmap, although probably never in reality) is not
treated as an error return.
testing revealed that the old implementation, while correct, was
giving way too many spurious wakeups due to races changing the value
of the condition futex. in a test program with 5 threads receiving
broadcast signals, the number of returns from pthread_cond_wait was
roughly 3 times what it should have been (2 spurious wakeups for every
legitimate wakeup). moreover, the magnitude of this effect seems to
grow with the number of threads.
the old implementation may also have had some nasty race conditions
with reuse of the cond var with a new mutex.
the new implementation is based on incrementing a sequence number with
each signal event. this sequence number has nothing to do with the
number of threads intended to be woken; it's only used to provide a
value for the futex wait to avoid deadlock. in theory there is a
danger of race conditions due to the value wrapping around after 2^32
signals. it would be nice to eliminate that, if there's a way.
testing showed no spurious wakeups (though they are of course
possible) with the new implementation, as well as slightly improved
performance.
this avoids the "stampede effect" where pthread_cond_broadcast would
result in all waiters waking up simultaneously, only to immediately
contend for the mutex and go back to sleep.
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.
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.
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!)
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.
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.
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.
previously, stdio used spinlocks, which would be unacceptable if we
ever add support for thread priorities, and which yielded
pathologically bad performance if an application attempted to use
flockfile on a key file as a major/primary locking mechanism.
i had held off on making this change for fear that it would hurt
performance in the non-threaded case, but actually support for
recursive locking had already inflicted that cost. by having the
internal locking functions store a flag indicating whether they need
to perform unlocking, rather than using the actual recursive lock
counter, i was able to combine the conditionals at unlock time,
eliminating any additional cost, and also avoid a nasty corner case
where a huge number of calls to ftrylockfile could cause deadlock
later at the point of internal locking.
this commit also fixes some issues with usage of pthread_self
conflicting with __attribute__((const)) which resulted in crashes with
some compiler versions/optimizations, mainly in flockfile prior to
pthread_create.
changing credentials in a multi-threaded program is extremely
difficult on linux because it requires synchronizing the change
between all threads, which have their own thread-local credentials on
the kernel side. this is further complicated by the fact that changing
the real uid can fail due to exceeding RLIMIT_NPROC, making it
possible that the syscall will succeed in some threads but fail in
others.
the old __rsyscall approach being replaced was robust in that it would
report failure if any one thread failed, but in this case, the program
would be left in an inconsistent state where individual threads might
have different uid. (this was not as bad as glibc, which would
sometimes even fail to report the failure entirely!)
the new approach being committed refuses to change real user id when
it cannot temporarily set the rlimit to infinity. this is completely
POSIX conformant since POSIX does not require an implementation to
allow real-user-id changes for non-privileged processes whatsoever.
still, setting the real uid can fail due to memory allocation in the
kernel, but this can only happen if there is not already a cached
object for the target user. thus, we forcibly serialize the syscalls
attempts, and fail the entire operation on the first failure. this
*should* lead to an all-or-nothing success/failure result, but it's
still fragile and highly dependent on kernel developers not breaking
things worse than they're already broken.
ideally linux will eventually add a CLONE_USERCRED flag that would
give POSIX conformant credential changes without any hacks from
userspace, and all of this code would become redundant and could be
removed ~10 years down the line when everyone has abandoned the old
broken kernels. i'm not holding my breath...
1. my interpretation of subject sequence definition was wrong. adjust
parser to conform to the standard.
2. some code for handling tail overflow case was missing (forgot to
finish writing it).
3. typo (= instead of ==) caused ERANGE to wrongly behave like EINVAL
this fixes a number of bugs in integer parsing due to lazy haphazard
wrapping, as well as some misinterpretations of the standard. the new
parser is able to work character-at-a-time or on whole strings, making
it easy to support the wide functions without unbounded space for
conversion. it will also be possible to update scanf to use the new
parser.
this seems to be necessary to make the linker accept the functions in
a shared library (perhaps to generate PLT entries?)
strictly speaking libc-internal asm should not need it. i might clean
that up later.
if thread id was reused by the kernel between the time pthread_kill
read it from the userspace pthread_t object and the time of the tgkill
syscall, a signal could be sent to the wrong thread. the tgkill
syscall was supposed to prevent this race (versus the old tkill
syscall) but it can't; it can only help in the case where the tid is
reused in a different process, but not when the tid is reused in the
same process.
the only solution i can see is an extra lock to prevent threads from
exiting while another thread is trying to pthread_kill them. it should
be very very cheap in the non-contended case.
the new approach relies on the fact that the only ways to create
sigset_t objects without invoking UB are to use the sig*set()
functions, or from the masks returned by sigprocmask, sigaction, etc.
or in the ucontext_t argument to a signal handler. thus, as long as
sigfillset and sigaddset avoid adding the "protected" signals, there
is no way the application will ever obtain a sigset_t including these
bits, and thus no need to add the overhead of checking/clearing them
when sigprocmask or sigaction is called.
note that the old code actually *failed* to remove the bits from
sa_mask when sigaction was called.
the new implementations are also significantly smaller, simpler, and
faster due to ignoring the useless "GNU HURD signals" 65-1024, which
are not used and, if there's any sanity in the world, never will be
used.
the previous implementation had at least 2 problems:
1. the case where additional threads reached the barrier before the
first wave was finished leaving the barrier was untested and seemed
not to be working.
2. threads leaving the barrier continued to access memory within the
barrier object after other threads had successfully returned from
pthread_barrier_wait. this could lead to memory corruption or crashes
if the barrier object had automatic storage in one of the waiting
threads and went out of scope before all threads finished returning,
or if one thread unmapped the memory in which the barrier object
lived.
the new implementation avoids both problems by making the barrier
state essentially local to the first thread which enters the barrier
wait, and forces that thread to be the last to return.
with this patch, musl compiles and mostly works with pcc 1.0.0. a few
tests are still failing and i'm uncertain whether they are due to
portability problems in musl, or bugs in pcc, but i suspect the
latter.
some functions that should have been testing whether pthread_self()
had been called and initialized the thread pointer were instead
testing whether pthread_create() had been called and actually made the
program "threaded". while it's unlikely any mismatch would occur in
real-world problems, this could have introduced subtle bugs. now, we
store the address of the main thread's thread descriptor in the libc
structure and use its presence as a flag that the thread register is
initialized. note that after fork, the calling thread (not necessarily
the original main thread) is the new main thread.
the goal is to be able to use pthread_setcancelstate internally in
the implementation, whenever a function might want to use functions
which are cancellation points but avoid becoming a cancellation point
itself. i could have just used a separate internal function for
temporarily inhibiting cancellation, but the solution in this commit
is better because (1) it's one less implementation-specific detail in
functions that need to use it, and (2) application code can also get
the same benefit.
previously, pthread_setcancelstate dependend on pthread_self, which
would pull in unwanted thread setup overhead for non-threaded
programs. now, it temporarily stores the state in the global libc
struct if threads have not been initialized, and later moves it if
needed. this way we can instead use __pthread_self, which has no
dependencies and assumes that the thread register is already valid.
this patch improves the correctness, simplicity, and size of
cancellation-related code. modulo any small errors, it should now be
completely conformant, safe, and resource-leak free.
the notion of entering and exiting cancellation-point context has been
completely eliminated and replaced with alternative syscall assembly
code for cancellable syscalls. the assembly is responsible for setting
up execution context information (stack pointer and address of the
syscall instruction) which the cancellation signal handler can use to
determine whether the interrupted code was in a cancellable state.
these changes eliminate race conditions in the previous generation of
cancellation handling code (whereby a cancellation request received
just prior to the syscall would not be processed, leaving the syscall
to block, potentially indefinitely), and remedy an issue where
non-cancellable syscalls made from signal handlers became cancellable
if the signal handler interrupted a cancellation point.
x86_64 asm is untested and may need a second try to get it right.
otherwise we cannot support an application's desire to use
asynchronous cancellation within the callback function. this change
also slightly debloats pthread_create.c.
this is something of a tradeoff, as now set*id() functions, rather
than pthread_create, are what pull in the code overhead for dealing
with linux's refusal to implement proper POSIX thread-vs-process
semantics. my motivations are:
1. it's cleaner this way, especially cleaner to optimize out the
rsyscall locking overhead from pthread_create when it's not needed.
2. it's expected that only a tiny number of core system programs will
ever use set*id() functions, whereas many programs may want to use
threads, and making thread overhead tiny is an incentive for "light"
programs to try threads.
with these small changes, libc functions which need to call functions
which are cancellation points, but which themselves must not be
cancellation points, can use the CANCELPT_INHIBIT and CANCELPT_RESUME
macros to temporarily inhibit all cancellation.
we want to keep atomically updated fields (locks and thread count) and
really anything writable far away from frequently-needed function
pointers. stuff some rarely-needed function pointers in between to
pad, hopefully up to a cache line boundary.
the major idea of this patch is not to depend on having the timer
pointer delivered to the signal handler, and instead use the thread
pointer to get the callback function address and argument. this way,
the parent thread can make the timer_create syscall while the child
thread is starting, and it should never have to block waiting for the
barrier.
- there is no longer any risk of spoofing cancellation requests, since
the cancel flag is set in pthread_cancel rather than in the signal
handler.
- cancellation signal is no longer unblocked when running the
cancellation handlers. instead, pthread_create will cause any new
threads created from a cancellation handler to unblock their own
cancellation signal.
- various tweaks in preparation for POSIX timer support.
the biggest change in this commit is that stdio now uses readv to fill
the caller's buffer and the FILE buffer with a single syscall, and
likewise writev to flush the FILE buffer and write out the caller's
buffer in a single syscall.
making this change required fundamental architectural changes to
stdio, so i also made a number of other improvements in the process:
- the implementation no longer assumes that further io will fail
following errors, and no longer blocks io when the error flag is set
(though the latter could easily be changed back if desired)
- unbuffered mode is no longer implemented as a one-byte buffer. as a
consequence, scanf unreading has to use ungetc, to the unget buffer
has been enlarged to hold at least 2 wide characters.
- the FILE structure has been rearranged to maintain the locations of
the fields that might be used in glibc getc/putc type macros, while
shrinking the structure to save some space.
- error cases for fflush, fseek, etc. should be more correct.
- library-internal macros are used for getc_unlocked and putc_unlocked
now, eliminating some ugly code duplication. __uflow and __overflow
are no longer used anywhere but these macros. switch to read or
write mode is also separated so the code can be better shared, e.g.
with ungetc.
- lots of other small things.
this commit addresses two issues:
1. a race condition, whereby a cancellation request occurring after a
syscall returned from kernelspace but before the subsequent
CANCELPT_END would cause cancellable resource-allocating syscalls
(like open) to leak resources.
2. signal handlers invoked while the thread was blocked at a
cancellation point behaved as if asynchronous cancellation mode wer in
effect, resulting in potentially dangerous state corruption if a
cancellation request occurs.
the glibc/nptl implementation of threads shares both of these issues.
with this commit, both are fixed. however, cancellation points
encountered in a signal handler will not be acted upon if the signal
was received while the thread was already at a cancellation point.
they will of course be acted upon after the signal handler returns, so
in real-world usage where signal handlers quickly return, it should
not be a problem. it's possible to solve this problem too by having
sigaction() wrap all signal handlers with a function that uses a
pthread_cleanup handler to catch cancellation, patch up the saved
context, and return into the cancellable function that will catch and
act upon the cancellation. however that would be a lot of complexity
for minimal if any benefit...
with this patch, the syscallN() functions are no longer needed; a
variadic syscall() macro allows syscalls with anywhere from 0 to 6
arguments to be made with a single macro name. also, manually casting
each non-integer argument with (long) is no longer necessary; the
casts are hidden in the macros.
some source files which depended on being able to define the old macro
SYSCALL_RETURNS_ERRNO have been modified to directly use __syscall()
instead of syscall(). references to SYSCALL_SIGSET_SIZE and SYSCALL_LL
have also been changed.
x86_64 has not been tested, and may need a follow-up commit to fix any
minor bugs/oversights.
this commit shuffles around the location of syscall definitions so
that we can make a syscall() library function with both SYS_* and
__NR_* style syscall names available to user applications, provides
the syscall() library function, and optimizes the code that performs
the actual inline syscalls in the library itself.
previously on i386 when built as PIC (shared library), syscalls were
incurring bus lock (lock prefix) overhead at entry and exit, due to
the way the ebx register was being loaded (xchg instruction with a
memory operand). now the xchg takes place between two registers.
further cleanup to arch/$(ARCH)/syscall.h is planned.
Rich Felker