The Documentation/memory-barriers.txt document requires that atomic
operations that return a value act as a memory barrier both before
and after the actual atomic operation.
Our current implementation doesn't guarantee this. More specifically,
while a load following the isync can not be issued before stwcx. has
completed, that completion doesn't architecturally means that the
result of stwcx. is visible to other processors (or any previous stores
for that matter) (typically, the other processors L1 caches can still
hold the old value).
This has caused an actual crash in RCU torture testing on Power 7
This fixes it by changing those atomic ops to use new macros instead
of RELEASE/ACQUIRE barriers, called ATOMIC_ENTRY and ATMOIC_EXIT barriers,
which are then defined respectively to lwsync and sync.
I haven't had a chance to measure the performance impact (or rather
what I measured with kernel compiles is in the noise, I yet have to
find a more precise benchmark)
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
kdump fails because we try to execute an HV only instruction. Feature
fixups are being applied after we copy the exception vectors down to 0
so they miss out on any updates.
We have always had this issue but it only became critical in v3.0
when we added CFAR support (breaks POWER5) and v3.1 when we added
POWERNV (breaks everyone).
Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: <stable@kernel.org> [v3.0+]
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Nick Piggin discovered that lwsync barriers around locks were faster than isync
on 970. That was a long time ago and I completely dropped the ball in testing
his patches across other ppc64 processors.
Turns out the idea helps on other chips. Using a microbenchmark that
uses a lot of threads to contend on a global pthread mutex (and therefore a
global futex), POWER6 improves 8% and POWER7 improves 2%. I checked POWER5
and while I couldn't measure an improvement, there was no regression.
This patch uses the lwsync patching code to replace the isyncs with lwsyncs
on CPUs that support the instruction. We were marking POWER3 and RS64 as lwsync
capable but in reality they treat it as a full sync (ie slow). Remove the
CPU_FTR_LWSYNC bit from these CPUs so they continue to use the faster isync
method.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
For performance reasons we are about to change ISYNC_ON_SMP to sometimes be
lwsync. Now that the macro name doesn't make sense, change it and LWSYNC_ON_SMP
to better explain what the barriers are doing.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Change 2d1b202762 ("powerpc: Fixup
lwsync at runtime") removed __SUBARCH_HAS_LWSYNC, causing smp_wmb to
revert back to eieio for all CPUs. This restores the behaviour
intorduced in 74f0609526 ("powerpc:
Optimise smp_wmb on 64-bit processors").
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
from include/asm-powerpc. This is the result of a
mkdir arch/powerpc/include/asm
git mv include/asm-powerpc/* arch/powerpc/include/asm
Followed by a few documentation/comment fixups and a couple of places
where <asm-powepc/...> was being used explicitly. Of the latter only
one was outside the arch code and it is a driver only built for powerpc.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Benjamin Herrenschmidt