Recent probing at the Linux Kernel Memory Model uncovered a
'surprise'. Strongly ordered architectures where the atomic RmW
primitive implies full memory ordering and
smp_mb__{before,after}_atomic() are a simple barrier() (such as MIPS
without WEAK_REORDERING_BEYOND_LLSC) fail for:
*x = 1;
atomic_inc(u);
smp_mb__after_atomic();
r0 = *y;
Because, while the atomic_inc() implies memory order, it
(surprisingly) does not provide a compiler barrier. This then allows
the compiler to re-order like so:
atomic_inc(u);
*x = 1;
smp_mb__after_atomic();
r0 = *y;
Which the CPU is then allowed to re-order (under TSO rules) like:
atomic_inc(u);
r0 = *y;
*x = 1;
And this very much was not intended. Therefore strengthen the atomic
RmW ops to include a compiler barrier.
Reported-by: Andrea Parri <andrea.parri@amarulasolutions.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Paul Burton <paul.burton@mips.com>
The comment describing the loongson_llsc_mb() reorder case doesn't
make any sense what so ever. Instruction re-ordering is not an SMP
artifact, but rather a CPU local phenomenon. Clarify the comment by
explaining that these issue cause a coherence fail.
For the branch speculation case; if futex_atomic_cmpxchg_inatomic()
needs one at the bne branch target, then surely the normal
__cmpxch_asm() implementation does too. We cannot rely on the
barriers from cmpxchg() because cmpxchg_local() is implemented with
the same macro, and branch prediction and speculation are, too, CPU
local.
Fixes: e02e07e312 ("MIPS: Loongson: Introduce and use loongson_llsc_mb()")
Cc: Huacai Chen <chenhc@lemote.com>
Cc: Huang Pei <huangpei@loongson.cn>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Paul Burton <paul.burton@mips.com>
- Support for the MIPSr6 MemoryMapID register & Global INValidate TLB
(GINVT) instructions, allowing for more efficient TLB maintenance when
running on a CPU such as the I6500 that supports these.
- Enable huge page support for MIPS64r6.
- Optimize post-DMA cache sync by removing that code entirely for kernel
configurations in which we know it won't be needed.
- The number of pages allocated for interrupt stacks is now calculated
correctly, where before we would wastefully allocate too much memory
in some configurations.
- The ath79 platform migrates to devicetree.
- The bcm47xx platform sees fixes for the Buffalo WHR-G54S board.
- The ingenic/jz4740 platform gains support for appended devicetrees.
- The cavium_octeon, lantiq, loongson32 & sgi-ip27 platforms all see
cleanups as do various pieces of core architecture code.
-----BEGIN PGP SIGNATURE-----
iIsEABYIADMWIQRgLjeFAZEXQzy86/s+p5+stXUA3QUCXH3BQxUccGF1bC5idXJ0
b25AbWlwcy5jb20ACgkQPqefrLV1AN1+4wD+Oh4JTfZN/NEOQMlrSkXxjEHqjX3u
1Y6CiiPCs+q2UnYBANb+ic+ZH5MnvJxxmcvlYI2q3rIh4b8TDriip4KMUTUP
=Sw9X
-----END PGP SIGNATURE-----
Merge tag 'mips_5.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux
Pull MIPS updates from Paul Burton:
- Support for the MIPSr6 MemoryMapID register & Global INValidate TLB
(GINVT) instructions, allowing for more efficient TLB maintenance
when running on a CPU such as the I6500 that supports these.
- Enable huge page support for MIPS64r6.
- Optimize post-DMA cache sync by removing that code entirely for
kernel configurations in which we know it won't be needed.
- The number of pages allocated for interrupt stacks is now calculated
correctly, where before we would wastefully allocate too much memory
in some configurations.
- The ath79 platform migrates to devicetree.
- The bcm47xx platform sees fixes for the Buffalo WHR-G54S board.
- The ingenic/jz4740 platform gains support for appended devicetrees.
- The cavium_octeon, lantiq, loongson32 & sgi-ip27 platforms all see
cleanups as do various pieces of core architecture code.
* tag 'mips_5.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux: (66 commits)
MIPS: lantiq: Remove separate GPHY Firmware loader
MIPS: ingenic: Add support for appended devicetree
MIPS: SGI-IP27: rework HUB interrupts
MIPS: SGI-IP27: do boot CPU init later
MIPS: SGI-IP27: do xtalk scanning later
MIPS: SGI-IP27: use pr_info/pr_emerg and pr_cont to fix output
MIPS: SGI-IP27: clean up bridge access and header files
MIPS: SGI-IP27: get rid of volatile and hubreg_t
MIPS: irq: Allocate accurate order pages for irq stack
MIPS: dma-noncoherent: Remove bogus condition in dma_sync_phys()
MIPS: eBPF: Remove REG_32BIT_ZERO_EX
MIPS: eBPF: Always return sign extended 32b values
MIPS: CM: Fix indentation
MIPS: BCM47XX: Fix/improve Buffalo WHR-G54S support
MIPS: OCTEON: program rx/tx-delay always from DT
MIPS: OCTEON: delete board-specific link status
MIPS: OCTEON: don't lie about interface type of CN3005 board
MIPS: OCTEON: warn if deprecated link status is being used
MIPS: OCTEON: add fixed-link nodes to in-kernel device tree
MIPS: Delete unused flush_cache_sigtramp()
...
Add a family of ginvt_* functions making it easy to emit a GINVT
instruction to globally invalidate TLB entries. We make use of the
_ASM_MACRO infrastructure to support emitting the instructions even if
the assembler isn't new enough to support them natively.
An associated STYPE_GINV definition & sync_ginv() function are added to
emit a sync instruction of type 0x14, which operates as a completion
barrier for these new GINVT (and GINVI) instructions.
Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: linux-mips@vger.kernel.org
On the Loongson-2G/2H/3A/3B there is a hardware flaw that ll/sc and
lld/scd is very weak ordering. We should add sync instructions "before
each ll/lld" and "at the branch-target between ll/sc" to workaround.
Otherwise, this flaw will cause deadlock occasionally (e.g. when doing
heavy load test with LTP).
Below is the explaination of CPU designer:
"For Loongson 3 family, when a memory access instruction (load, store,
or prefetch)'s executing occurs between the execution of LL and SC, the
success or failure of SC is not predictable. Although programmer would
not insert memory access instructions between LL and SC, the memory
instructions before LL in program-order, may dynamically executed
between the execution of LL/SC, so a memory fence (SYNC) is needed
before LL/LLD to avoid this situation.
Since Loongson-3A R2 (3A2000), we have improved our hardware design to
handle this case. But we later deduce a rarely circumstance that some
speculatively executed memory instructions due to branch misprediction
between LL/SC still fall into the above case, so a memory fence (SYNC)
at branch-target (if its target is not between LL/SC) is needed for
Loongson 3A1000, 3B1500, 3A2000 and 3A3000.
Our processor is continually evolving and we aim to to remove all these
workaround-SYNCs around LL/SC for new-come processor."
Here is an example:
Both cpu1 and cpu2 simutaneously run atomic_add by 1 on same atomic var,
this bug cause both 'sc' run by two cpus (in atomic_add) succeed at same
time('sc' return 1), and the variable is only *added by 1*, sometimes,
which is wrong and unacceptable(it should be added by 2).
Why disable fix-loongson3-llsc in compiler?
Because compiler fix will cause problems in kernel's __ex_table section.
This patch fix all the cases in kernel, but:
+. the fix at the end of futex_atomic_cmpxchg_inatomic is for branch-target
of 'bne', there other cases which smp_mb__before_llsc() and smp_llsc_mb() fix
the ll and branch-target coincidently such as atomic_sub_if_positive/
cmpxchg/xchg, just like this one.
+. Loongson 3 does support CONFIG_EDAC_ATOMIC_SCRUB, so no need to touch
edac.h
+. local_ops and cmpxchg_local should not be affected by this bug since
only the owner can write.
+. mips_atomic_set for syscall.c is deprecated and rarely used, just let
it go
Signed-off-by: Huacai Chen <chenhc@lemote.com>
Signed-off-by: Huang Pei <huangpei@loongson.cn>
[paul.burton@mips.com:
- Simplify the addition of -mno-fix-loongson3-llsc to cflags, and add
a comment describing why it's there.
- Make loongson_llsc_mb() a no-op when
CONFIG_CPU_LOONGSON3_WORKAROUNDS=n, rather than a compiler memory
barrier.
- Add a comment describing the bug & how loongson_llsc_mb() helps
in asm/barrier.h.]
Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: ambrosehua@gmail.com
Cc: Steven J . Hill <Steven.Hill@cavium.com>
Cc: linux-mips@linux-mips.org
Cc: Fuxin Zhang <zhangfx@lemote.com>
Cc: Zhangjin Wu <wuzhangjin@gmail.com>
Cc: Li Xuefeng <lixuefeng@loongson.cn>
Cc: Xu Chenghua <xuchenghua@loongson.cn>
Add the definitions of sync stype 0 (global completion barrier) and sync
stype 0x10 (local ordering barrier) to barrier.h for use with the sync
instruction.
These types are defined by the MIPS Instruction Set since R2 of the
architecture and are documented in document MD00087 table 6.5.
Signed-off-by: Matt Redfearn <matt.redfearn@imgtec.com>
Reviewed-by: Paul Burton <paul.burton@imgtec.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/14222/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
This defines __smp_xxx barriers for mips,
for use by virtualization.
smp_xxx barriers are removed as they are
defined correctly by asm-generic/barriers.h
Note: the only exception is smp_mb__before_llsc which is mips-specific.
We define both the __smp_mb__before_llsc variant (for use in
asm/barriers.h) and smp_mb__before_llsc (for use elsewhere on this
architecture).
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
On mips dma_rmb, dma_wmb, smp_store_mb, read_barrier_depends,
smp_read_barrier_depends, smp_store_release and smp_load_acquire match
the asm-generic variants exactly. Drop the local definitions and pull in
asm-generic/barrier.h instead.
This is in preparation to refactoring this code area.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Replace ACCESS_ONCE() macro in smp_store_release() and smp_load_acquire()
with WRITE_ONCE() and READ_ONCE() on x86, arm, arm64, ia64, metag, mips,
powerpc, s390, sparc and asm-generic since ACCESS_ONCE() does not work
reliably on non-scalar types.
WRITE_ONCE() and READ_ONCE() were introduced in the following commits:
230fa253df ("kernel: Provide READ_ONCE and ASSIGN_ONCE")
43239cbe79 ("kernel: Change ASSIGN_ONCE(val, x) to WRITE_ONCE(x, val)")
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc)
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Alexander Duyck <alexander.h.duyck@redhat.com>
Cc: Andre Przywara <andre.przywara@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arch@vger.kernel.org
Link: http://lkml.kernel.org/r/1438528264-714-1-git-send-email-andreyknvl@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Since set_mb() is really about an smp_mb() -- not a IO/DMA barrier
like mb() rename it to match the recent smp_load_acquire() and
smp_store_release().
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Since we assume set_mb() to result in a single store followed by a
full memory barrier, employ WRITE_ONCE().
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
There are a number of situations where the mandatory barriers rmb() and
wmb() are used to order memory/memory operations in the device drivers
and those barriers are much heavier than they actually need to be. For
example in the case of PowerPC wmb() calls the heavy-weight sync
instruction when for coherent memory operations all that is really needed
is an lsync or eieio instruction.
This commit adds a coherent only version of the mandatory memory barriers
rmb() and wmb(). In most cases this should result in the barrier being the
same as the SMP barriers for the SMP case, however in some cases we use a
barrier that is somewhere in between rmb() and smp_rmb(). For example on
ARM the rmb barriers break down as follows:
Barrier Call Explanation
--------- -------- ----------------------------------
rmb() dsb() Data synchronization barrier - system
dma_rmb() dmb(osh) data memory barrier - outer sharable
smp_rmb() dmb(ish) data memory barrier - inner sharable
These new barriers are not as safe as the standard rmb() and wmb().
Specifically they do not guarantee ordering between coherent and incoherent
memories. The primary use case for these would be to enforce ordering of
reads and writes when accessing coherent memory that is shared between the
CPU and a device.
It may also be noted that there is no dma_mb(). Most architectures don't
provide a good mechanism for performing a coherent only full barrier without
resorting to the same mechanism used in mb(). As such there isn't much to
be gained in trying to define such a function.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Michael Ellerman <michael@ellerman.id.au>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: David Miller <davem@davemloft.net>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is meant to cleanup the handling of read_barrier_depends and
smp_read_barrier_depends. In multiple spots in the kernel headers
read_barrier_depends is defined as "do {} while (0)", however we then go
into the SMP vs non-SMP sections and have the SMP version reference
read_barrier_depends, and the non-SMP define it as yet another empty
do/while.
With this commit I went through and cleaned out the duplicate definitions
and reduced the number of definitions down to 2 per header. In addition I
moved the 50 line comments for the macro from the x86 and mips headers that
defined it as an empty do/while to those that were actually defining the
macro, alpha and blackfin.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
MIPS is interesting and has hardware variants that reorder over ll/sc
as well as those that do not.
Implement the 2 new barrier functions as per the old barriers.
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/n/tip-9ph49jbae3hol9v721sbc2g6@git.kernel.org
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Maciej W. Rozycki" <macro@codesourcery.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
A number of situations currently require the heavyweight smp_mb(),
even though there is no need to order prior stores against later
loads. Many architectures have much cheaper ways to handle these
situations, but the Linux kernel currently has no portable way
to make use of them.
This commit therefore supplies smp_load_acquire() and
smp_store_release() to remedy this situation. The new
smp_load_acquire() primitive orders the specified load against
any subsequent reads or writes, while the new smp_store_release()
primitive orders the specifed store against any prior reads or
writes. These primitives allow array-based circular FIFOs to be
implemented without an smp_mb(), and also allow a theoretical
hole in rcu_assign_pointer() to be closed at no additional
expense on most architectures.
In addition, the RCU experience transitioning from explicit
smp_read_barrier_depends() and smp_wmb() to rcu_dereference()
and rcu_assign_pointer(), respectively resulted in substantial
improvements in readability. It therefore seems likely that
replacing other explicit barriers with smp_load_acquire() and
smp_store_release() will provide similar benefits. It appears
that roughly half of the explicit barriers in core kernel code
might be so replaced.
[Changelog by PaulMck]
Reviewed-by: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Michael Ellerman <michael@ellerman.id.au>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Victor Kaplansky <VICTORK@il.ibm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Link: http://lkml.kernel.org/r/20131213150640.908486364@infradead.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Another whitespace clean-up, this removes tabs from between sentences in
some comments.
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6103/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Having received another series of whitespace patches I decided to do this
once and for all rather than dealing with this kind of patches trickling
in forever.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The current locking mechanism uses a ll/sc sequence to release a
spinlock. This is slower than a wmb() followed by a store to unlock.
The branching forward to .subsection 2 on sc failure slows down the
contended case. So we get rid of that part too.
Since we are now working on naturally aligned u16 values, we can get
rid of a masking operation as the LHU already does the right thing.
The ANDI are reversed for better scheduling on multi-issue CPUs
On a 12 CPU 750MHz Octeon cn5750 this patch improves ipv4 UDP packet
forwarding rates from 3.58*10^6 PPS to 3.99*10^6 PPS, or about 11%.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/937/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
In order to achieve correct synchronization semantics, the Octeon port
had defined CONFIG_WEAK_REORDERING_BEYOND_LLSC. This resulted in code
that looks like:
sync
ll ...
.
.
.
sc ...
.
.
sync
The second SYNC was redundant, but harmless.
Octeon has a SYNCW instruction that acts as a write-memory-barrier
(due to an erratum in some parts two SYNCW are used). It is much
faster than SYNC because it imposes ordering on the writes, but
doesn't otherwise stall the execution pipeline. On Octeon, SYNC
stalls execution until all preceeding writes are committed to the
coherent memory system.
Using:
syncw;syncw
ll
.
.
.
sc
.
.
Has identical semantics to the first sequence, but is much faster.
The SYNCW orders the writes, and the SC will not complete successfully
until the write is committed to the coherent memory system. So at the
end all preceeding writes have been committed. Since Octeon does not
do speculative reads, this functions as a full barrier.
The patch removes CONFIG_WEAK_REORDERING_BEYOND_LLSC, and substitutes
SYNCW for SYNC in write-memory-barriers.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/850/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Replace some instances of smp_llsc_mb() with a new macro
smp_mb__before_llsc(). It is used before ll/sc sequences that are
documented as needing write barrier semantics.
The default implementation of smp_mb__before_llsc() is just smp_llsc_mb(),
so there are no changes in semantics.
Also simplify definition of smp_mb(), smp_rmb(), and smp_wmb() to be just
barrier() in the non-SMP case.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/851/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The smp_llsc_rmb() and smp_llsc_wmb() macros are not used in the tree,
remove them.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/848/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Peter Zijlstra