As suggested by Peter Anvin.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: H . Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
While hard to measure, reducing the number of possibly/likely
mis-predicted branches can generally be expected to be slightly
better.
Other than apparent at the first glance, this also doesn't grow
the function size (the alignment gap to the next function just
gets smaller).
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/4F218584020000780006F422@nat28.tlf.novell.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
While currently there doesn't appear to be any reachable in-tree
case where such large memory blocks may be passed to memcpy(),
we already had hit the problem in our Xen kernels. Just like
done recently for mmeset(), rather than working around it,
prevent others from falling into the same trap by fixing this
long standing limitation.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/4F21846F020000780006F3FA@nat28.tlf.novell.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Support memcpy() with enhanced rep movsb. On processors supporting enhanced
rep movsb, the alternative memcpy() function using enhanced rep movsb overrides the original function and the fast string
function.
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Link: http://lkml.kernel.org/r/1305671358-14478-8-git-send-email-fenghua.yu@intel.com
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
All read operations after allocation stage can run speculatively,
all write operation will run in program order, and if addresses are
different read may run before older write operation, otherwise wait
until write commit. However CPU don't check each address bit,
so read could fail to recognize different address even they
are in different page.For example if rsi is 0xf004, rdi is 0xe008,
in following operation there will generate big performance latency.
1. movq (%rsi), %rax
2. movq %rax, (%rdi)
3. movq 8(%rsi), %rax
4. movq %rax, 8(%rdi)
If %rsi and rdi were in really the same meory page, there are TRUE
read-after-write dependence because instruction 2 write 0x008 and
instruction 3 read 0x00c, the two address are overlap partially.
Actually there are in different page and no any issues,
but without checking each address bit CPU could think they are
in the same page, and instruction 3 have to wait for instruction 2
to write data into cache from write buffer, then load data from cache,
the cost time read spent is equal to mfence instruction. We may avoid it by
tuning operation sequence as follow.
1. movq 8(%rsi), %rax
2. movq %rax, 8(%rdi)
3. movq (%rsi), %rax
4. movq %rax, (%rdi)
Instruction 3 read 0x004, instruction 2 write address 0x010, no any
dependence. At last on Core2 we gain 1.83x speedup compared with
original instruction sequence. In this patch we first handle small
size(less 20bytes), then jump to different copy mode. Based on our
micro-benchmark small bytes from 1 to 127 bytes, we got up to 2X
improvement, and up to 1.5X improvement for 1024 bytes on Corei7. (We
use our micro-benchmark, and will do further test according to your
requirment)
Signed-off-by: Ma Ling <ling.ma@intel.com>
LKML-Reference: <1277753065-18610-1-git-send-email-ling.ma@intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
We already have cpufeature indicies above 255, so use a 16-bit number
for the alternatives index. This consumes a padding field and so
doesn't add any size, but it means that abusing the padding field to
create assembly errors on overflow no longer works. We can retain the
test simply by redirecting it to the .discard section, however.
[ v3: updated to include open-coded locations ]
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
LKML-Reference: <tip-f88731e3068f9d1392ba71cc9f50f035d26a0d4f@git.kernel.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
In order to avoid unnecessary chains of branches, rather than
implementing memcpy()/memset()'s access to their alternative
implementations via a jump, patch the (larger) original function
directly.
The memcpy() part of this is slightly subtle: while alternative
instruction patching does itself use memcpy(), with the
replacement block being less than 64-bytes in size the main loop
of the original function doesn't get used for copying memcpy_c()
over memcpy(), and hence we can safely write over its beginning.
Also note that the CFI annotations are fine for both variants of
each of the functions.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
LKML-Reference: <4B2BB8D30200007800026AF2@vpn.id2.novell.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: micro-optimization
This should slightly improve its performance.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
LKML-Reference: <49B8F641.76E4.0078.0@novell.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Andy Shevchenko