[ Upstream commit a476aae3f1dc78a162a0d2e7945feea7d2b29401 ]
Commit 688eb8191b ("x86/csum: Improve performance of `csum_partial`")
ended up improving the code generation for the IP csum calculations, and
in particular special-casing the 40-byte case that is a hot case for
IPv6 headers.
It then had _another_ special case for the 64-byte unrolled loop, which
did two chains of 32-byte blocks, which allows modern CPU's to improve
performance by doing the chains in parallel thanks to renaming the carry
flag.
This just unifies the special cases and combines them into just one
single helper the 40-byte csum case, and replaces the 64-byte case by a
80-byte case that just does that single helper twice. It avoids having
all these different versions of inline assembly, and actually improved
performance further in my tests.
There was never anything magical about the 64-byte unrolled case, even
though it happens to be a common size (and typically is the cacheline
size).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 5d4acb62853abac1da2deebcb1c1c5b79219bf3b ]
The special case for odd aligned buffers is unnecessary and mostly
just adds overhead. Aligned buffers is the expectations, and even for
unaligned buffer, the only case that was helped is if the buffer was
1-byte from word aligned which is ~1/7 of the cases. Overall it seems
highly unlikely to be worth to extra branch.
It was left in the previous perf improvement patch because I was
erroneously comparing the exact output of `csum_partial(...)`, but
really we only need `csum_fold(csum_partial(...))` to match so its
safe to remove.
All csum kunit tests pass.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Laight <david.laight@aculab.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Clang warns:
arch/x86/lib/csum-partial_64.c:74:20: error: variable 'result' is uninitialized when used here [-Werror,-Wuninitialized]
return csum_tail(result, temp64, odd);
^~~~~~
arch/x86/lib/csum-partial_64.c:48:22: note: initialize the variable 'result' to silence this warning
unsigned odd, result;
^
= 0
1 error generated.
The only initialization and uses of result in csum_partial() were moved
into csum_tail() but result is still being passed by value to
csum_tail() (clang's -Wuninitialized does not do interprocedural
analysis to realize that result is always assigned in csum_tail()
however). Sink the declaration of result into csum_tail() to clear up
the warning.
Closes: https://lore.kernel.org/202305262039.3HUYjWJk-lkp@intel.com/
Fixes: 688eb8191b ("x86/csum: Improve performance of `csum_partial`")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/all/20230526-csum_partial-wuninitialized-v1-1-ebc0108dcec1%40kernel.org
1) Add special case for len == 40 as that is the hottest value. The
nets a ~8-9% latency improvement and a ~30% throughput improvement
in the len == 40 case.
2) Use multiple accumulators in the 64-byte loop. This dramatically
improves ILP and results in up to a 40% latency/throughput
improvement (better for more iterations).
Results from benchmarking on Icelake. Times measured with rdtsc()
len lat_new lat_old r tput_new tput_old r
8 3.58 3.47 1.032 3.58 3.51 1.021
16 4.14 4.02 1.028 3.96 3.78 1.046
24 4.99 5.03 0.992 4.23 4.03 1.050
32 5.09 5.08 1.001 4.68 4.47 1.048
40 5.57 6.08 0.916 3.05 4.43 0.690
48 6.65 6.63 1.003 4.97 4.69 1.059
56 7.74 7.72 1.003 5.22 4.95 1.055
64 6.65 7.22 0.921 6.38 6.42 0.994
96 9.43 9.96 0.946 7.46 7.54 0.990
128 9.39 12.15 0.773 8.90 8.79 1.012
200 12.65 18.08 0.699 11.63 11.60 1.002
272 15.82 23.37 0.677 14.43 14.35 1.005
440 24.12 36.43 0.662 21.57 22.69 0.951
952 46.20 74.01 0.624 42.98 53.12 0.809
1024 47.12 78.24 0.602 46.36 58.83 0.788
1552 72.01 117.30 0.614 71.92 96.78 0.743
2048 93.07 153.25 0.607 93.28 137.20 0.680
2600 114.73 194.30 0.590 114.28 179.32 0.637
3608 156.34 268.41 0.582 154.97 254.02 0.610
4096 175.01 304.03 0.576 175.89 292.08 0.602
There is no such thing as a free lunch, however, and the special case
for len == 40 does add overhead to the len != 40 cases. This seems to
amount to be ~5% throughput and slightly less in terms of latency.
Testing:
Part of this change is a new kunit test. The tests check all
alignment X length pairs in [0, 64) X [0, 512).
There are three cases.
1) Precomputed random inputs/seed. The expected results where
generated use the generic implementation (which is assumed to be
non-buggy).
2) An input of all 1s. The goal of this test is to catch any case
a carry is missing.
3) An input that never carries. The goal of this test si to catch
any case of incorrectly carrying.
More exhaustive tests that test all alignment X length pairs in
[0, 8192) X [0, 8192] on random data are also available here:
https://github.com/goldsteinn/csum-reproduction
The reposity also has the code for reproducing the above benchmark
numbers.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/all/20230511011002.935690-1-goldstein.w.n%40gmail.com
With more NICs supporting CHECKSUM_COMPLETE, and IPv6 being widely
used csum_partial() is heavily used with small amount of bytes, and is
consuming many cycles.
IPv6 header size, for instance, is 40 bytes.
Another thing to consider is that NET_IP_ALIGN is 0 on x86, meaning
that network headers are not word-aligned, unless the driver forces
this.
This means that csum_partial() fetches one u16 to 'align the buffer',
then performs three u64 additions with carry in a loop, then a
remaining u32, then a remaining u16.
With this new version, it performs a loop only for the 64 bytes blocks,
then the remaining is bisected.
Testing on various CPUs, all of them show a big reduction in
csum_partial() cost (by 50 to 80 %)
Before:
4.16% [kernel] [k] csum_partial
After:
0.83% [kernel] [k] csum_partial
If run in a loop 1,000,000 times:
Before:
26,922,913 cycles # 3846130.429 GHz
80,302,961 instructions # 2.98 insn per cycle
21,059,816 branches # 3008545142.857 M/sec
2,896 branch-misses # 0.01% of all branches
After:
17,960,709 cycles # 3592141.800 GHz
41,292,805 instructions # 2.30 insn per cycle
11,058,119 branches # 2211623800.000 M/sec
2,997 branch-misses # 0.03% of all branches
[ bp: Massage, merge in subsequent fixes into a single patch:
- um compilation error due to missing load_unaligned_zeropad():
- Reported-by: kernel test robot <lkp@intel.com>
- Link: https://lkml.kernel.org/r/20211118175239.1525650-1-eric.dumazet@gmail.com
- Fix initial seed for odd buffers
- Reported-by: Noah Goldstein <goldstein.w.n@gmail.com>
- Link: https://lkml.kernel.org/r/20211125141817.3541501-1-eric.dumazet@gmail.com
]
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Link: https://lore.kernel.org/r/20211112161950.528886-1-eric.dumazet@gmail.com
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.
By default all files without license information are under the default
license of the kernel, which is GPL version 2.
Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier. The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.
This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.
How this work was done:
Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
- file had no licensing information it it.
- file was a */uapi/* one with no licensing information in it,
- file was a */uapi/* one with existing licensing information,
Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.
The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne. Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.
The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed. Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.
Criteria used to select files for SPDX license identifier tagging was:
- Files considered eligible had to be source code files.
- Make and config files were included as candidates if they contained >5
lines of source
- File already had some variant of a license header in it (even if <5
lines).
All documentation files were explicitly excluded.
The following heuristics were used to determine which SPDX license
identifiers to apply.
- when both scanners couldn't find any license traces, file was
considered to have no license information in it, and the top level
COPYING file license applied.
For non */uapi/* files that summary was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 11139
and resulted in the first patch in this series.
If that file was a */uapi/* path one, it was "GPL-2.0 WITH
Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 WITH Linux-syscall-note 930
and resulted in the second patch in this series.
- if a file had some form of licensing information in it, and was one
of the */uapi/* ones, it was denoted with the Linux-syscall-note if
any GPL family license was found in the file or had no licensing in
it (per prior point). Results summary:
SPDX license identifier # files
---------------------------------------------------|------
GPL-2.0 WITH Linux-syscall-note 270
GPL-2.0+ WITH Linux-syscall-note 169
((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21
((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17
LGPL-2.1+ WITH Linux-syscall-note 15
GPL-1.0+ WITH Linux-syscall-note 14
((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5
LGPL-2.0+ WITH Linux-syscall-note 4
LGPL-2.1 WITH Linux-syscall-note 3
((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3
((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1
and that resulted in the third patch in this series.
- when the two scanners agreed on the detected license(s), that became
the concluded license(s).
- when there was disagreement between the two scanners (one detected a
license but the other didn't, or they both detected different
licenses) a manual inspection of the file occurred.
- In most cases a manual inspection of the information in the file
resulted in a clear resolution of the license that should apply (and
which scanner probably needed to revisit its heuristics).
- When it was not immediately clear, the license identifier was
confirmed with lawyers working with the Linux Foundation.
- If there was any question as to the appropriate license identifier,
the file was flagged for further research and to be revisited later
in time.
In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.
Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights. The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.
Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.
In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.
Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
- a full scancode scan run, collecting the matched texts, detected
license ids and scores
- reviewing anything where there was a license detected (about 500+
files) to ensure that the applied SPDX license was correct
- reviewing anything where there was no detection but the patch license
was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
SPDX license was correct
This produced a worksheet with 20 files needing minor correction. This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.
These .csv files were then reviewed by Greg. Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected. This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.) Finally Greg ran the script using the .csv files to
generate the patches.
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Historically a lot of these existed because we did not have
a distinction between what was modular code and what was providing
support to modules via EXPORT_SYMBOL and friends. That changed
when we forked out support for the latter into the export.h file.
This means we should be able to reduce the usage of module.h
in code that is obj-y Makefile or bool Kconfig. The advantage
in doing so is that module.h itself sources about 15 other headers;
adding significantly to what we feed cpp, and it can obscure what
headers we are effectively using.
Since module.h was the source for init.h (for __init) and for
export.h (for EXPORT_SYMBOL) we consider each obj-y/bool instance
for the presence of either and replace as needed. Build testing
revealed a couple implicit header usage issues that were fixed.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20160714001901.31603-5-paul.gortmaker@windriver.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
They were generated by 'codespell' and then manually reviewed.
Signed-off-by: Lucas De Marchi <lucas.demarchi@profusion.mobi>
Cc: trivial@kernel.org
LKML-Reference: <1300389856-1099-3-git-send-email-lucas.demarchi@profusion.mobi>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Fix this symbol export problem:
Building modules, stage 2.
MODPOST 193 modules
ERROR: "csum_partial" [fs/reiserfs/reiserfs.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2
This is due to a known weakness of symbol exports: if a symbol's
only in-core user is an EXPORT_SYMBOL from a lib-y section, the
symbol is not linked in.
The solution is to move the export to x8664_ksyms_64.c - but the real
solution would be to fix kbuild.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Linus Torvalds