← Back to team overview

group.of.nepali.translators team mailing list archive

[Bug 1667407] Re: improve 2x-3x sha256sum performance on ppc64le due to current gcc optimization bug

 

This bug was fixed in the package coreutils - 8.25-2ubuntu3~16.10

---------------
coreutils (8.25-2ubuntu3~16.10) yakkety-proposed; urgency=medium

  * SRU: LP: #1667407. Add -fno-schedule-insns to CFLAGS for sha256.o
    on ppc64el to fix performance regression.

 -- Matthias Klose <doko@xxxxxxxxxx>  Tue, 28 Feb 2017 10:18:46 +0100

** Changed in: coreutils (Ubuntu Yakkety)
       Status: Fix Committed => Fix Released

-- 
You received this bug notification because you are a member of नेपाली
भाषा समायोजकहरुको समूह, which is subscribed to Xenial.
Matching subscriptions: Ubuntu 16.04 Bugs
https://bugs.launchpad.net/bugs/1667407

Title:
   improve 2x-3x sha256sum performance on ppc64le due to current gcc
  optimization bug

Status in coreutils package in Ubuntu:
  Fix Released
Status in coreutils source package in Xenial:
  Fix Committed
Status in coreutils source package in Yakkety:
  Fix Released
Status in coreutils package in Debian:
  Fix Released

Bug description:
  [Impact]

   * Performance drop of 2x-3x on ppc64le when using sha256sum

   * Please backport this bug to last LTS as this application is widely
  used.

   * This fix is due to a optimization issue found on gcc v4.9 to gcc
  v7.0.

  [Test Case]

   * Run sha256sum with a big file and measure the time it takes. The
  patch improves this timing up to 3 times faster.

  [Regression Potential]

   * This patch is specifically for the sha256.o object, affecting only
  this binary.

  [Other Info]

  Michael Stone's improved patch:

  Index: coreutils-8.26/Makefile.in
  ===================================================================
  --- coreutils-8.26.orig/Makefile.in     2016-11-30 13:34:55.000000000 -0500
  +++ coreutils-8.26/Makefile.in  2017-02-22 07:18:55.352394058 -0500
  @@ -14661,6 +14661,10 @@

  $(TEST_LOGS): $(PROGRAMS)

  +ifeq ($(DEB_TARGET_ARCH), ppc64el)
  +lib/sha256.o: CFLAGS+=-fno-schedule-insns
  +endif
  +
  # Tell versions [3.59,3.63) of GNU make to not export all variables.
  # Otherwise a system limit (for SysV at least) may be exceeded.
  .NOEXPORT:

  Original bug description:
   
  The sha256sum provided by coreutils (without openssl) is performing
  poorly with gcc versions >= 4.9 until 7.0 (currently under development).
  The reason for that is the -fschedule-insns optimization that is used
  with -O2. By simply deactivating it, there is a performance improvement
  of 2 to 3 times.

  I'm using Ubuntu 16.10 and the coreutils package version
  8.25-2ubuntu2.

  Please check the following closed debian bug report:
  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=854053

  Be aware of the following conditions that are required:
  * If ppc64le
  * If gcc being used is >= 4.9 and < 7.0

  Notes:
  1) gcc-7 is not affected by this bug (verified on 20170129 snapshot).
  2) clang is not affected by this bug (verified on v3.8 and v3.9).
  3) strangely the sha512 is not affected by this.

  Below a demonstration of how it performs on my POWER8 machine:

  ===================================================
  $ (./configure && make -j9) > /dev/null && time src/sha256sum ~/ubuntu-16.10-server-ppc64el.iso
  configure: WARNING: libacl development library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without ACL support.
  configure: WARNING: libattr development library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without xattr support.
  configure: WARNING: libcap library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without capability support.
  configure: WARNING: libgmp development library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without GMP support.
  src/who.c: In function 'print_user':
  src/who.c:454:20: warning: initialization discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers]
           int   *a = utmp_ent->ut_addr_v6;
                      ^~~~~~~~
  d14bdb413ea6cdc8d9354fcbc37a834b7de0c23f992deb0c6764d0fd5d65408e  /home/gut/ubuntu-16.10-server-ppc64el.iso

  real    0m18.670s
  user    0m16.566s
  sys     0m0.745s

  $ # now with the following patch: ## Check Michael Stone's patch for an improved version.
  $ diff Makefile.in ../Makefile.in
  8989c8989
  < @am__fastdepCC_TRUE@  $(COMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
  ---
  > @am__fastdepCC_TRUE@  $(COMPILE) $$([ "$@" == "lib/sha256.o" ] && echo "-fno-schedule-insns") -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
  $ cp ../Makefile.in Makefile.in
  $ (./configure && make -j9) > /dev/null && time src/sha256sum ~/ubuntu-16.10-server-ppc64el.iso
  configure: WARNING: libacl development library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without ACL support.
  configure: WARNING: libattr development library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without xattr support.
  configure: WARNING: libcap library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without capability support.
  configure: WARNING: libgmp development library was not found or not usable.
  configure: WARNING: GNU coreutils will be built without GMP support.
  src/who.c: In function 'print_user':
  src/who.c:454:20: warning: initialization discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers]
           int   *a = utmp_ent->ut_addr_v6;
                      ^~~~~~~~
  d14bdb413ea6cdc8d9354fcbc37a834b7de0c23f992deb0c6764d0fd5d65408e  /home/gut/ubuntu-16.10-server-ppc64el.iso

  real    0m5.903s
  user    0m5.560s
  sys     0m0.255s

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/coreutils/+bug/1667407/+subscriptions