Mathieu Desnoyers [Sat, 9 Jul 2011 19:02:26 +0000 (15:02 -0400)]
rculfhash: shrink size of dummy node
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Sat, 9 Jul 2011 17:28:01 +0000 (13:28 -0400)]
rculfhash: wait for call_rcu quiescence before destroy
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 16:27:19 +0000 (12:27 -0400)]
rculfhash: add c++ ifdef
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 16:19:14 +0000 (12:19 -0400)]
rculfhash test: add options
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 15:49:09 +0000 (11:49 -0400)]
rculfhash test: use per-cpu call-rcu threads
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 15:48:34 +0000 (11:48 -0400)]
rculfhash: update add_unique api
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 15:13:06 +0000 (11:13 -0400)]
rculfhash: fix add unique
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 14:43:51 +0000 (10:43 -0400)]
rculfhash test: increases random pool size
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 14:36:39 +0000 (10:36 -0400)]
rculfhash: remove hash from nodes
Shrink node size, can get it quickly by reversing the reverse_hash.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 14:24:00 +0000 (10:24 -0400)]
rculfhash: change order of node fields for locality
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 13:48:44 +0000 (09:48 -0400)]
rculfhash: merge node gc into add loop
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 13:43:50 +0000 (09:43 -0400)]
rculfhash: cleanup
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 13:20:39 +0000 (09:20 -0400)]
rculfhash test: add nr_leaked count
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 7 Jul 2011 13:09:04 +0000 (09:09 -0400)]
rculfhash: garbage collection scheme
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 20:05:48 +0000 (16:05 -0400)]
rculfhash: flip order of removal
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 19:00:14 +0000 (15:00 -0400)]
rculfhash: Remove "value" pointer
Use container_of instead.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 17:37:19 +0000 (13:37 -0400)]
rculfhash test: add missing free
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 15:19:08 +0000 (11:19 -0400)]
rculfhash: add unique
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 14:50:09 +0000 (10:50 -0400)]
Update resize thresholds
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 14:45:04 +0000 (10:45 -0400)]
rculfhash updates
- Add and remove now completely lock-free (no busy-wait loop).
- Add resize_initiated flag to skip multiple call_rcu calls.
- Identical reverse hashes now only count as one chain entry in the
chain length accounting. We cannot split identical hashes by growing
the hash table anyway, so it was leading to large tables uselessly
when we had a lot of similar hashes.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 05:07:33 +0000 (01:07 -0400)]
Comment list behavior wrt lock-freedom
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 04:52:07 +0000 (00:52 -0400)]
Fix reverse hash comparison
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 04:44:43 +0000 (00:44 -0400)]
rculfhash: use threshold and target
Accept to target bucket len greater than 1 to diminish the memory
consumption without too much performance impact.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 03:47:32 +0000 (23:47 -0400)]
rculfhash: implement real hash function
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 02:43:17 +0000 (22:43 -0400)]
rculfhash: fix resize (use log2 of chain length)
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 6 Jul 2011 01:02:11 +0000 (21:02 -0400)]
Merge remote branch 'origin/urcu/ht' into urcu/ht
Conflicts:
urcu-ht.h
Mathieu Desnoyers [Wed, 6 Jul 2011 01:00:42 +0000 (21:00 -0400)]
rcu lf hash table runtime fixes
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 5 Jul 2011 23:35:08 +0000 (19:35 -0400)]
split-ordered hash table
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Sat, 2 Jul 2011 15:31:09 +0000 (11:31 -0400)]
Rename urcu-ht to rculfhash
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Sat, 2 Jul 2011 15:24:18 +0000 (11:24 -0400)]
Merge branch 'master' into urcu/ht
Conflicts:
Makefile.inc
tests/Makefile.inc
urcu-defer.h
Mathieu Desnoyers [Fri, 1 Jul 2011 20:17:55 +0000 (16:17 -0400)]
rculfstack: header implementation should be defined as static inline
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 1 Jul 2011 20:17:14 +0000 (16:17 -0400)]
wfstack: header implementation should be defined as static inline
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 30 Jun 2011 14:45:33 +0000 (10:45 -0400)]
urcu tests: hold mutex across use of custom allocator
A thread preempted for a long period of time could race, when scheduled
again, with another thread that would have been allocating/freeing
entries (thus wrapping-around the available buffer), which would trigger
this race only when overcommitting the number of threads compared to the
number of available CPUs.
Taking the mutex across alloc and free to fix this.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 20:39:19 +0000 (16:39 -0400)]
urcu-bp: mremap wrapper fix
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 20:38:06 +0000 (16:38 -0400)]
urcu-bp: add mremap wrapper for non-linux systems
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 20:31:16 +0000 (16:31 -0400)]
urcu-bp: don't copy old region upon mremap
mremap keeps the same virtual pages for the old/new mappings. So
explicitly copying from the old mapping is not needed, and probably
buggy, since the old mapping might have been unmapped.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 18:35:08 +0000 (14:35 -0400)]
api.h: fix type warning
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 18:20:31 +0000 (14:20 -0400)]
non-linux system membarrier fallback update
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 18:12:23 +0000 (14:12 -0400)]
pthread_self should be cast to unsigned long
Used as part of the random seed along with time().
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 17:47:28 +0000 (13:47 -0400)]
Add MAP_ANONYMOUS mapping to MAP_ANON for BSD build
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 17:43:22 +0000 (13:43 -0400)]
tests: only include syscall.h on linux systems
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 17:34:48 +0000 (13:34 -0400)]
urcu-bp, urcu-qsbr: remove unneeded syscall.h include
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 17:30:42 +0000 (13:30 -0400)]
urcu-defer: remove unneeded syscall.h include
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 17:29:38 +0000 (13:29 -0400)]
call_rcu: per_cpu_call_rcu_data should be non-const
On FreeBSD:
In file included from urcu.c:438:
urcu-call-rcu-impl.h: In function 'get_cpu_call_rcu_data_mb':
urcu-call-rcu-impl.h:325: warning: return discards qualifiers from pointer target type
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 17:02:45 +0000 (13:02 -0400)]
Update README info about gcc compability
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 16:51:41 +0000 (12:51 -0400)]
Fix urcu-call-rcu-impl.h missing call_rcu_wait for non-linux systems
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 16:49:35 +0000 (12:49 -0400)]
compat_futex_async: fix missing return
compat_futex.c: In function `compat_futex_async':
compat_futex.c:114: warning: control reaches end of non-void function
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 16:46:52 +0000 (12:46 -0400)]
Check for __linux__ before including syscall.h
Reported-by: Jean-François Brousseau <jfb@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 16:30:30 +0000 (12:30 -0400)]
Remove unused local variables from compat_futex_async
Reported-by: Jean-François Brousseau <jfb@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 29 Jun 2011 16:29:39 +0000 (12:29 -0400)]
Remove unneeded syscall.h include from urcu-call-rcu-impl.h
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 27 Jun 2011 21:52:41 +0000 (17:52 -0400)]
Add ChangeLog file to Makefile.am
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 27 Jun 2011 21:49:38 +0000 (17:49 -0400)]
Add changelog, update version to 0.6.3
Mathieu Desnoyers [Mon, 27 Jun 2011 21:07:25 +0000 (17:07 -0400)]
uatomic: fix warning about unused variable
Fix:
compat_arch_x86.c: In function '_compat_uatomic_set':
compat_arch_x86.c:104:16: warning: variable 'result' set but not used [-Wunused-but-set-variable]
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 27 Jun 2011 20:54:56 +0000 (16:54 -0400)]
uatomic: fix i386 support
Fix:
* Incorrect prototype for uatomic_and and uatomic_or in i386
compatibility code.
* Missing $(COMPAT) code inclusion in wfq/lfq tests.
* Silence gcc warnings about compat code (branch volountarily causing a
linker error, which can never return).
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 14 Jun 2011 13:30:19 +0000 (09:30 -0400)]
Update version to 0.6.2
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 14 Jun 2011 13:15:20 +0000 (09:15 -0400)]
uatomic generic: map uatomic_and/or to correct symbols
Leading underscore was missing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 14 Jun 2011 12:55:44 +0000 (08:55 -0400)]
Fix generic atomic ops and/or: add missing return
Fixes build on powerpc:
./urcu/uatomic/generic.h: In function '_uatomic_and':
./urcu/uatomic/generic.h:310:2: warning: 'return' with a value, in
function returning void
./urcu/uatomic/generic.h: In function '_uatomic_or':
./urcu/uatomic/generic.h:374:2: warning: 'return' with a value, in
function returning void
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 13 Jun 2011 20:06:52 +0000 (16:06 -0400)]
Fix test warning
Patch from:
https://build.opensuse.org/package/view_file?file=liburcu0-cleanup.patch&package=liburcu0&project=devel%3Atools%3Alttng&srcmd5=
fa3ab9fae0401b4c0f73a9cf474eea16
fixes a warning in the liburcu tests.
Reported-by: Yannick Brosseau <yannick.brosseau@polymtl.ca>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 13 Jun 2011 14:01:18 +0000 (10:01 -0400)]
Add missing headers into tarball, version 0.6.1
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Sat, 11 Jun 2011 00:46:03 +0000 (20:46 -0400)]
Update version to 0.6.0
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Sat, 11 Jun 2011 00:16:28 +0000 (20:16 -0400)]
Futex: turn "int" into "int32_t" for portability
Even though int is 32-bit on all architectures supported by liburcu so
far, make it future-proof by uint a int32_t, which enforces the same
type width used by the system call in the kernel.
Using int32_t and not uint32_t to make comparison with 0 more
straightforward.
Reported-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 23:20:50 +0000 (19:20 -0400)]
Add back uatomic_arch.h for slow deprecation process
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 23:17:25 +0000 (19:17 -0400)]
Add back urcu-futex.h and urcu_ref.h for slow deprecation
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 22:36:59 +0000 (18:36 -0400)]
Add ldconfig step to README
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Yannick Brosseau [Fri, 10 Jun 2011 21:28:50 +0000 (17:28 -0400)]
Add a pkg-config file for the new liburcu-cds
Signed-off-by: Yannick Brosseau <yannick.brosseau@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Alexandre Montplaisir [Fri, 10 Jun 2011 21:14:23 +0000 (17:14 -0400)]
Consolidate _LIBADD and _DEPENDENCIES in Makefile.am
Signed-off-by: Alexandre Montplaisir <alexandre.montplaisir@polymtl.ca>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 21:11:04 +0000 (17:11 -0400)]
Add urcu/cds.h placeholder for all CDS headers
Using CDS headers becomes as simple as:
#include <urcu/cds.h>
and link with
-lurcu-cds
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Yannick Brosseau [Fri, 10 Jun 2011 20:58:15 +0000 (16:58 -0400)]
Move the version-info to the AM_LDFLAGS variable
Since all the lib builded in this directory use the same version
we can put this command in the global variable
Signed-off-by: Yannick Brosseau <yannick.brosseau@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 20:11:30 +0000 (16:11 -0400)]
Installed headers should only include other headers with < >
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 20:06:00 +0000 (16:06 -0400)]
Rename urcu/urcu-futex.h to urcu/futex.h
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 20:02:31 +0000 (16:02 -0400)]
Rename urcu/urcu_ref.h to urcu/ref.h
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 19:58:34 +0000 (15:58 -0400)]
Headers: move uatomic_*.h to urcu/uatomic/*.h, rename uatomic_arch.h to uatomic.h
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 19:48:01 +0000 (15:48 -0400)]
Headers: move arch_*.h to urcu/arch/*.h
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 19:32:31 +0000 (15:32 -0400)]
Headers: move *-static.h headers to urcu/static/
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 19:20:46 +0000 (15:20 -0400)]
Install urcu/map/*.h into system
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 19:16:36 +0000 (15:16 -0400)]
Headers: move *-map.h headers to urcu/map/
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Yannick Brosseau [Fri, 10 Jun 2011 15:35:49 +0000 (11:35 -0400)]
Add library version information
Following the guidelines from libtool
(http://www.gnu.org/software/libtool/manual/html_node/Updating-version-info.htm)
this patch add version information to the distributed libraries.
For the next release, the version will be 1:0:0.
It will need to be updated before each release.
Signed-off-by: Yannick Brosseau <yannick.brosseau@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 00:42:26 +0000 (20:42 -0400)]
lfqueue: update comments, cleanup
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 00:38:17 +0000 (20:38 -0400)]
lfqueue: don't use defer_rcu anymore
lfqueue now uses call_rcu instead.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 00:33:05 +0000 (20:33 -0400)]
Combine stack and queues info liburcu-cds.so
Pretty much all libraries use a single .so for their data types. Do the
same here.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jun 2011 00:31:29 +0000 (20:31 -0400)]
cds containers: lfqueue and lfstack: don't depend on a particular rcu flavor
Remove rcu_read lock/unlock from the code, require the caller to ensure
protection. First move towards a single .so for all data containers.
This changes the lfqueue API, which is not very much used yet AFAIK.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Yannick Brosseau [Thu, 9 Jun 2011 18:04:48 +0000 (14:04 -0400)]
Add missing pkgconfig files to Makefile.am
Signed-off-by: Yannick Brosseau <yannick.brosseau@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 16:54:38 +0000 (12:54 -0400)]
arm: remove useless declarations
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 9 Jun 2011 14:50:01 +0000 (10:50 -0400)]
call_rcu: add back delay in list non-empty case
commit
bc94ca9bada25f7403e3e859caa241146ae8e338 changed the !RT behavior
slightly: when the list is not empty, it does not wait for a delay
anymore. Add this delay back, to ensure we don't flood the system with
frequent synchronize_rcu() calls, which would slow down readers.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 9 Jun 2011 14:39:56 +0000 (10:39 -0400)]
call_rcu: handle retry without wait correctly
The wait scheme has an implementation problem: if the list is not empty
when the !RT scheme checks for it, it will restart the loop and
decrement the futex (again) without calling call_rcu_wait() (which would
wait until it is set back to 0). So in this case, we can end up
decrementing "futex" to values well below -1.
Fix this by moving the decrement before the loop, and duplicate it after
return from call_rcu_wait() + poll() delay. Also move the "set futex to
0 upon stopping" outside of the loop: this is the only way the loop can
be stopped anyway.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 14:16:40 +0000 (10:16 -0400)]
urcu: fix dependencies on libwfqueue.la
Without this patch, "make -jN" usually fails for N as low as 4.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 14:15:28 +0000 (10:15 -0400)]
urcu_ref: do error checking on urcu_ref_put
Underflow is what you really want to trap. When you reach urcu_ref_get,
it is already too late.
[Edit: whitespaces -> tab cleanup]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 14:13:13 +0000 (10:13 -0400)]
use generic-size macros for common implementation of atomic ops
The definition of _uatomic_cmpxchg is different in x86 and other
architectures. For x86 it is a 4-argument macro, for other
architectures it is a 3-argument function. This patch makes it easier
to implement atomic operations incrementally (first as a generic version
and then in machine-specific code), which aids testing and
bisectability.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 13:32:58 +0000 (09:32 -0400)]
call_rcu: drop mutex
The mutex is being used only to protect OR accesses to the flags.
Just use atomic operations for that.
[ Edit: this also fixes busy-looping on flags that were previously read
without volatile access, which could lead to never-ending loop given the
appropriate set of compiler optimisations. ]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 13:30:40 +0000 (09:30 -0400)]
uatomic: add uatomic_and
Only x86 support is arch-specific for now, the rest uses either cmpxchg
or gcc __sync_ builtins.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Thu, 9 Jun 2011 13:29:18 +0000 (09:29 -0400)]
uatomic: add uatomic_or
For now, only the version returning void is used. Also, I am not
providing PPC versions because I would be unable to test it right
now.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 9 Jun 2011 00:39:47 +0000 (20:39 -0400)]
Add missing -lwfqueue to add rcu flavors
Otherwise, linking with the flavor .so leads to:
/usr/local/lib/liburcu-bp.so: undefined reference to `cds_wfq_enqueue'
/usr/local/lib/liburcu-bp.so: undefined reference to `cds_wfq_init'
/usr/local/lib/liburcu-bp.so: undefined reference to `cds_wfq_node_init'
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paul E. McKenney [Wed, 8 Jun 2011 23:32:23 +0000 (19:32 -0400)]
Add a cds_list_del_init()
Needed by some perfbook example code.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paul E. McKenney [Wed, 8 Jun 2011 23:31:04 +0000 (19:31 -0400)]
Add comment to flag purpose of the ->qlen field
It is write-only, but is needed for debugging purposes.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 8 Jun 2011 22:28:46 +0000 (18:28 -0400)]
call_rcu: fix futex-based wakeup
The initial decrement was missing from the implementation. It now
behaves exactly like the urcu.c wait/wakeup code.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Wed, 8 Jun 2011 22:11:34 +0000 (18:11 -0400)]
atomic: fix typo in x86 compat implementation
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Paolo Bonzini [Wed, 8 Jun 2011 22:09:19 +0000 (18:09 -0400)]
rcutorture: make goflag volatile
Even cmm_barrier may not be enough as a compiler barrier in the presence
of static variables, and cmm_mb would not be either! The compiler
assumes that calling a function does not clobber static variables if it
can prove that they do not escape (which also implies the correct phase
of the moon).
Rather than sprinkling the code with compiler barriers, I am taking the
easy way out and declaring the variable volatile. Without this patch,
rcutorture_qsbr never finishes for me (GCC 4.5.0).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 8 Jun 2011 22:05:53 +0000 (18:05 -0400)]
urcu-qsbr: fix typo
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 7 Jun 2011 04:16:00 +0000 (00:16 -0400)]
call_rcu: use futex for wakeup scheme
If I remove the URCU_CALL_RCU_RT flag from the rbtree single writer
test, thus using the pthread_cond_signal mechanism, there is a huge
slowdown: without cpu affinity for the worker threads, it crawls to 129
updates/s (looks like mutex contention between the thread calling
call_rcu and the call_rcu thread). Adding CPU affinity to the per-cpu
call_rcu threads, I get 546 updates/s, which is slightly better (better
cache locality, and maybe the mutex contention is not as bad thanks to
the two threads sharing the same CPU).
So I decided to try replacing pthread_cond_wait/signal with my
futex-based implementation I use for the rest of the urcu lib: it has
the advantage of removing the mutex from the call_rcu() execution
entirely, sampling the "futex" variable without any mutex whatsoever for
the case where no wakeup is needed.
Disabling URCU_CALL_RCU_RT flag, with per-cpu affined call_rcu threads,
with my futex-based wakeup implementation, I get 55754 updates/s (even
better than with URCU_CALL_RCU_RT flag!).
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 6 Jun 2011 21:24:42 +0000 (17:24 -0400)]
call_rcu: use cpu affinity for per-cpu call_rcu threads
I played a bit with the call_rcu() implementation alongside with my
rbtree tests, and noticed the following:
If I use per-cpu call_rcu threads with URCU_CALL_RCU_RT flag, with one
updater thread only for my rbtree (no reader), I get 38365 updates/s.
If I add cpu affinity to these per-cpu call_rcu threads (I have prepared
a patch that does this), it jumps to 54219 updates/s. So it looks like
keeping per-cpu affinity for the call_rcu thread is a good thing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
This page took 0.046127 seconds and 4 git commands to generate.