ATM pthread_cond_signal is unnecessarily slow, because it wakes one waiter
(which at least on UP usually means an immediate context switch to one of
the waiter threads). This waiter wakes up and after a few instructions it
attempts to acquire the cv internal lock, but that lock is still held by
the thread calling pthread_cond_signal. So it goes to sleep and eventually
the signalling thread is scheduled in, unlocks the internal lock and wakes
the waiter again.
Now, before 2003-09-21 NPTL was using FUTEX_REQUEUE in pthread_cond_signal
to avoid this performance issue, but it was removed when locks were
redesigned to the 3 state scheme (unlocked, locked uncontended, locked
contended).
Following scenario shows why simply using FUTEX_REQUEUE in
pthread_cond_signal together with using lll_mutex_unlock_force in place of
lll_mutex_unlock is not enough and probably why it has been disabled at
that time:
The number is value in cv->__data.__lock.
thr1 thr2 thr3
0 pthread_cond_wait
1 lll_mutex_lock (cv->__data.__lock)
0 lll_mutex_unlock (cv->__data.__lock)
0 lll_futex_wait (&cv->__data.__futex, futexval)
0 pthread_cond_signal
1 lll_mutex_lock (cv->__data.__lock)
1 pthread_cond_signal
2 lll_mutex_lock (cv->__data.__lock)
2 lll_futex_wait (&cv->__data.__lock, 2)
2 lll_futex_requeue (&cv->__data.__futex, 0, 1, &cv->__data.__lock)
# FUTEX_REQUEUE, not FUTEX_CMP_REQUEUE
2 lll_mutex_unlock_force (cv->__data.__lock)
0 cv->__data.__lock = 0
0 lll_futex_wake (&cv->__data.__lock, 1)
1 lll_mutex_lock (cv->__data.__lock)
0 lll_mutex_unlock (cv->__data.__lock)
# Here, lll_mutex_unlock doesn't know there are threads waiting
# on the internal cv's lock
Now, I believe it is possible to use FUTEX_REQUEUE in pthread_cond_signal,
but it will cost us not one, but 2 extra syscalls and, what's worse, one of
these extra syscalls will be done for every single waiting loop in
pthread_cond_*wait.
We would need to use lll_mutex_unlock_force in pthread_cond_signal after
requeue and lll_mutex_cond_lock in pthread_cond_*wait after lll_futex_wait.
Another alternative is to do the unlocking pthread_cond_signal needs to do
(the lock can't be unlocked before lll_futex_wake, as that is racy) in the
kernel.
I have implemented both variants, futex-requeue-glibc.patch is the first
one and futex-wake_op{,-glibc}.patch is the unlocking inside of the kernel.
The kernel interface allows userland to specify how exactly an unlocking
operation should look like (some atomic arithmetic operation with optional
constant argument and comparison of the previous futex value with another
constant).
It has been implemented just for ppc*, x86_64 and i?86, for other
architectures I'm including just a stub header which can be used as a
starting point by maintainers to write support for their arches and ATM
will just return -ENOSYS for FUTEX_WAKE_OP. The requeue patch has been
(lightly) tested just on x86_64, the wake_op patch on ppc64 kernel running
32-bit and 64-bit NPTL and x86_64 kernel running 32-bit and 64-bit NPTL.
With the following benchmark on UP x86-64 I get:
for i in nptl-orig nptl-requeue nptl-wake_op; do echo time elf/ld.so --library-path .:$i /tmp/bench; \
for j in 1 2; do echo ( time elf/ld.so --library-path .:$i /tmp/bench ) 2>&1; done; done
time elf/ld.so --library-path .:nptl-orig /tmp/bench
real 0m0.655s user 0m0.253s sys 0m0.403s
real 0m0.657s user 0m0.269s sys 0m0.388s
time elf/ld.so --library-path .:nptl-requeue /tmp/bench
real 0m0.496s user 0m0.225s sys 0m0.271s
real 0m0.531s user 0m0.242s sys 0m0.288s
time elf/ld.so --library-path .:nptl-wake_op /tmp/bench
real 0m0.380s user 0m0.176s sys 0m0.204s
real 0m0.382s user 0m0.175s sys 0m0.207s
The benchmark is at:
http://sourceware.org/ml/libc-alpha/2005-03/txt00001.txt
Older futex-requeue-glibc.patch version is at:
http://sourceware.org/ml/libc-alpha/2005-03/txt00002.txt
Older futex-wake_op-glibc.patch version is at:
http://sourceware.org/ml/libc-alpha/2005-03/txt00003.txt
Will post a new version (just x86-64 fixes so that the patch
applies against pthread_cond_signal.S) to libc-hacker ml soon.
Attached is the kernel FUTEX_WAKE_OP patch as well as a simple-minded
testcase that will not test the atomicity of the operation, but at least
check if the threads that should have been woken up are woken up and
whether the arithmetic operation in the kernel gave the expected results.
Acked-by: Ingo Molnar <mingo@redhat.com>
Cc: Ulrich Drepper <drepper@redhat.com>
Cc: Jamie Lokier <jamie@shareable.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Yoichi Yuasa <yuasa@hh.iij4u.or.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This is used only in slab.c and each architecture gets to define whcih
underlying type is to be used.
Seems a bit silly - move it to slab.c and use the same type for all
architectures: unsigned int.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Someone mentioned that almost all the architectures used basically the same
implementation of get_order. This patch consolidates them into
asm-generic/page.h and includes that in the appropriate places. The
exceptions are ia64 and ppc which have their own (presumably optimised)
versions.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This kills warnings when building drivers/ide/ide-iops.c
and puts us in-line with what other platforms do here.
Signed-off-by: David S. Miller <davem@davemloft.net>
We can put the __softirq_pending mask in the cpudata,
no need for the silly NR_CPUS array in kernel/softirq.c
Signed-off-by: David S. Miller <davem@davemloft.net>
It appears that a memory barrier soon after a mispredicted
branch, not just in the delay slot, can cause the hang
condition of this cpu errata.
So move them out-of-line, and explicitly put them into
a "branch always, predict taken" delay slot which should
fully kill this problem.
Signed-off-by: David S. Miller <davem@davemloft.net>
When the spinlock routines were moved out of line into
kernel/spinlock.c this made it so that the debugging
spinlocks record lock acquisition program counts in the
kernel/spinlock.c functions not in their callers.
This makes the debugging info kind of useless.
So record the correct caller's program counter and
now this feature is useful once more.
Signed-off-by: David S. Miller <davem@davemloft.net>
Removed sparc64 architecture specific users of asm/segment.h and
asm-sparc64/segment.h itself
Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Current uncorrectable error handling was poor enough
that the processor could just loop taking the same
trap over and over again. Fix things up so that we
at least get a log message and perhaps even some register
state.
In the process, much consolidation became possible,
particularly with the correctable error handler.
Prefix assembler and C function names with "spitfire"
to indicate that these are for Ultra-I/II/IIi/IIe only.
More work is needed to make these routines robust and
featureful to the level of the Ultra-III error handlers.
Signed-off-by: David S. Miller <davem@davemloft.net>
GCC 4.x really dislikes the games we are playing in
unaligned.c, and the cleanest way to fix this is to
move things into assembler.
Noted by Al Viro.
Signed-off-by: David S. Miller <davem@davemloft.net>
When the kernel is working well and we want to restart cleanly
kernel_restart is the function to use. But in many instances
the kernel wants to reboot when thing are expected to be working
very badly such as from panic or a software watchdog handler.
This patch adds the function emergency_restart() so that
callers can be clear what semantics they expect when calling
restart. emergency_restart() is expected to be callable
from interrupt context and possibly reliable in even more
trying circumstances.
This is an initial generic implementation for all architectures.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
These two bits were accesses non-atomically from assembler
code. So, in order to eliminate any potential races resulting
from that, move these pieces of state into two bytes elsewhere
in struct thread_info.
Signed-off-by: David S. Miller <davem@davemloft.net>
It is only used by some localized code in irq.c, and also
delete enable_prom_timer() as that is totally unused.
Signed-off-by: David S. Miller <davem@davemloft.net>
rwsem_atomic_update and rwsem_atomic_add can be implemented
straightly using atomic_*() routines.
Also, rwsem_cmpxchgw() is totally unused, kill it.
Signed-off-by: David S. Miller <davem@davemloft.net>
Noticed this while comparing sparc64's bitops.h to ppc64's.
We can cast the volatile memory argument to be non-volatile.
While we're here, __inline__ --> inline.
Signed-off-by: David S. Miller <davem@davemloft.net>
Jakub Jelinek
Linus Torvalds
Kyle Moffett
Stephen Rothwell
David S. Miller
Patrick McHardy
Kumar Gala
Eric W. Biederman
Len Brown
David Shaohua Li