* Two half day off. [2/10]
# Progress #
* TCWG-332, fails in gdb.threads/multiple-step-overs.exp. [1/10]
Testing the simpler approach suggested during the review.
* TCWG-387, done. [1/10] GDB patches are pushed in.
* TCWG-422, GNU vector extension support in ARM GDB. [2/10]
Patches are done, and being tested.
* TCWG-423, GNU vector extension support in AArch64 GDB. [2/10]
Writing patches. Find more issues for AArch64 that GDB doesn't
fully understand the AArch64 calling convention. Need more work here.
* Review ARM GDBserver software single step patch. [1/10]
* Misc, meeting, email, [1/10]
# Plan #
* Off on Wed afternoon.
* TCWG-422, post patches
* TCWG-423, continue.
--
Yao
The Linaro Toolchain Working Group is pleased to announce the availability
of the Linaro Stable Binary Toolchain GCC 5.2-2015.11-rc1
Release-Candidate Archives.
http://snapshots.linaro.org/components/toolchain/binaries/5.2-2015.11-rc1/http://snapshots.linaro.org/components/toolchain/gcc-linaro/5.2-2015.11-rc1/
These archives provide cross-toolchain executables (compiler, debugger,
linker, etc.) and shared libraries (libstdc++, libc, etc.) that target ARM
or Aarch64 GNU/Linux and bare-metal environments. The cross-toolchain
binaries execute on a Linux or MS Windows (under mingw32) host
operating-system.
Please evaluate this release-candidate for correctness. Linaro will
shortly spin the Linaro GCC 5.2-2015.11 release if this release-candidate
passes stakeholder validation.
For bugs related to this release-candidate please email
linaro-toolchain(a)lists.linaro.org or file a bug at
https://bugs.linaro.org/enter_bug.cgi?product=Linux%20Binary%20toolchain
NEWS
* GCC 5.2 2015.11-rc1
The Linaro GCC 5.2 2015.11-rc1 binary toolchain release-candidate is
built from the Linaro GCC-5.2-2015.11 release-candidate source archive.
The Linaro GCC-5.2-2015.11 release source archive is derived from the same
sources as the Linaro GCC-5.2-2015.10 snapshot source archive.
--
Ryan S. Arnold
Linaro Toolchain Working Group - Engineering Manager
www.linaro.org
Dear List,
I'm new to this list and have some questions.
Looking at the created code of GCC on ARMv8, we noticed some areas where there is room for performance improvements.
I assume that these items might already be noticed by you guys.
For example:
1) We noticed that when writing typical DGEMM like code, GCC includes unnecessary DUP instruction
2) GCC seems unwilling to use LDP loads
3) For optimal FPU performance on some A57 its needed to interleave instruction working on ODD and EVEN registers
GCC seem not properly support this. Here sometimes 100% performance increase could be reached by different instruction interleaving.
4) Some work loops highly benefit of interleaving of FPU instructinons and loads.
GCC seems to likes to re-arrange the code so that most or all loads are put on top of the loop.
This can reduce the performance of a well written workloop significantly.
I have no patches to fix this.
But I can produce C- code and ASM output which will show these performance issues.
Please tell me what the next recommended step will be now.
Are all these items known already, or shall I provide code examples to further explain them?
Kind regards
Gunnar von Boehn
== Progress ==
* Validation
- comparing results and build times between the 2 labs
- tuning jobs scheduling to avoid deadlocks
* GCC trunk monitoring
- lots of validation results to check after 1 week of holidays :-)
- a few regressions/new failures/wrong tests reported
* Backports
- a few reviews
== Next ==
* Infrastructure/Validation
* GCC dev: try to fix vqtbl intrinsics for aarch64_be before e/o stage1
The Linaro Toolchain Working Group (TCWG) is pleased to announce the
2015.10 snapshot of the Linaro GCC 4.9 source package.
This snapshot[1] is based on FSF GCC 4.9.4-pre+svn229467 and includes
performance improvements and bug fixes backported from mainline GCC.
This snapshot contents will be part of the 2015.11 stable [1]
quarterly release.
This snapshot tarball is available on:
http://snapshots.linaro.org/components/toolchain/gcc-linaro/4.9-2015.10/
Interesting changes in this GCC source package snapshot include:
* Updates to GCC 4.9.4-pre+svn229467
* Backport of [Bugfix] PR tree-optimization/65735
* Backport of [Bugfix] PR tree-optimization/65177
* Backport of [Bugfix] PR tree-optimization/65048
Feedback and Support
Subscribe to the important Linaro mailing lists and join our IRC
channels to stay on top of Linaro development.
** Linaro Toolchain Development "mailing list":
http://lists.linaro.org/mailman/listinfo/linaro-toolchain
** Linaro Toolchain IRC channel on irc.freenode.net at @#linaro-tcwg@
* Bug reports should be filed in bugzilla against GCC product:
http://bugs.linaro.org/enter_bug.cgi?product=GCC
* Interested in commercial support? inquire at "Linaro support":
mailto:support@linaro.org
[1]. Stable source package releases are defined as releases where the
full Linaro Toolchain validation plan is executed.
[2]. Source package snapshots are defined when the compiler is only
put through unit-testing and full validation is not performed.
== Progress ==
o Linaro GCC (9/10)
* Backports and reviews for our GCC 5 2015.11 snapshot
* FSF branch merge and needed backports for our GCC 4.9 2015.10 snapshot
o Misc (1/10)
* Various meetings
== Plan ==
o Complete 4.9 snapshot
Noise control experiments - TCWG-358 [3/10]
* Some analysis of data to date
Debian filesystem - TCWG-360 [3/10]
* Got stuck on LAVA interactions
* Now booting-to-LAVA-usability, needs some cleanup and testing with
real benchmark runs
Benchmarking-via-Jenkins - TCWG-348 [1/10]
* Picked back up on understanding that LAVA uinstance is a-coming
* Hacked pbl.py (post-build-lava) to generate suitable JSON
** As a bonus, this can work as a CLI job-submission tool
=Plan=
Holiday Friday (pending approval)
Set up crashdumping on my Juno, try to learn why it crashes
Finish Debian filesystem
Get Jenkins generating and submitting jobs suitable for uinstance
Write up noise control report (probably will get bumped to next week)
== This week ==
* TCWG-317 - Exploit wide add operations when appropriate for Aarch32 (3/10)
- Could not reproduce test failure in qemu
- Waiting on feedback from Maxim who is running patch on Linaro
validation
* TCWG-369 - Exploit wide add operations when appropriate for Aarch64 (4/10)
- Debugging tree vectorizer to determine why loops with no wide add
operations are no longer being vectorized
* TCWG 146 - Debugging regression on arm big endian with Linaro 5 branch
(2/10)
* Misc (1/10)
- Conference calls
== Next week ==
- TCWG-317 - Resolve lto big endian failures
- TCWG-369 - Identify why loops are not being vectorized
- TCWG-146 - Resolve big endian failures
== This Week ==
* TCWG-72 / PR43721 (8/10)
- Completed writing expand_DIVMOD() and submitted patch upstream
- Reworked patch according to Richard Biener's comments.
(http://people.linaro.org/~prathamesh.kulkarni/pr43721-patch-v2.diff)
- Wrote test-cases.
* TCWG-319 (1/10)
- Benchmark run without patch complete for "int" benchmarks on a53, a57
- Benchmark run with patch in progress for "int" benchmarks on a53, a57
* Misc (1/10)
- Meetings
== Next Week ==
- Continue with TCWG-72, TCWG-310li, TCWG-319 benchmarking
# Progress #
* TCWG-332, [1/10], patch is posted for review.
* TCWG-187, [1/10], looks like a kernel (~3.4) bug on setting VFP
registers through ptrace. Fails go away after I upgrade the kernel.
* TCWG-387, [2/10], binutils patch is committed, while GDB
patches are being reviewed.
* TCWG-422, [3/10], Read AAPCS and gcc source code to understand the
calling convention. On going.
* FSF patch review, [2/10]
** Review patch set "all-stop on top of non-stop for remote".
** Review ARM fast tracepoint patches.
** Reopen PR 15564, as the fail isn't fixed.
* Misc, meeting, [1/10]
# Plan #
* TCWG-422
* Support ST on using AArch64 multi-arch GDB if needed.
* Review ARM GDBserver software single step patch.
--
Yao
== This week ==
* TCWG-317 - Exploit wide add operations when appropriate for Aarch32 (5/10)
- Continued to debug big endian/lto failure of four torture tests
- Differences in lto vs. non-lto executables showed no significant
differences
* TCWG-369 - Exploit wide add operations when appropriate for Aarch64 (4/10)
- Resolved three of six test suite failures
* Misc (1/10)
- Conference calls
== Next week ==
- TCWG-317 - Resolve lto big endian failures by debugging test cases
with qemu
- TCWG-369 - Resolve remaining test suite failures
--
Michael Collison
Linaro Toolchain Working Group
michael.collison(a)linaro.org
Hello all,
When upgrading meta-linaro/meta-linaro-toolchain from daisy to fido, we faced drastic performance downgrade (-34.06%) on linux ipv4 forwarding. The board is ls1021atwr (ARM Cortex-A7 MPCore compliant with ARMv7-A architecture).
We did some investigation, and suspected that it is caused by the toolchain upgrade. Did someone meet similar issue?
The detailed toolchain info:
https://git.linaro.org/openembedded/meta-linaro.git/shortlog/refs/heads/fido
gcc-linaro-4.9.3-2015.03, glibc-linaro-2.20-2014.11, binutils-linaro-2.25-2015.01
https://git.linaro.org/openembedded/meta-linaro.git/shortlog/refs/heads/dai…
gcc-linaro-4.8.3-2014.04, eglibc-linaro-2.19-r2014.04, binutils-linaro-2.24-r2014.03
thanks.
-Ting
o Week 42 off (10/10)
== Progress ==
o Linaro GCC (9/10)
* Completed backports and reviews
* Delivered GCC 5 2015.10 snapshot
* Completed TCWG 389 (deploy snapshot job)
o Misc (1/10)
* Various meetings
== Plan ==
o Back to work
Mustang benchmarking bringup - (no ticket) [2/10]
Controlled Debian build - TCWG-360 [1/10]
* Constructed filesystem, works OK as chroot
* Next step is to boot it
Investigate Workload Automation framework - TCWG-361 [1/10]
* Seems to basically work
* SPEC support dubious
Misc
* LAVA uinstance background (TCWG-396, < 1/10)
* Noise control experiments (TCWG-358, < 1/10)
* ARM management duties [2/10]
== Progress ==
* Maintenance (5/10)
- Investigating some LivermoreLoops that don't vectorise
- Reducing cases to single specific causes
* Background (5/10)
- Code review, meetings, discussions, general support, etc.
- More code of conduct stuff, license change to Apache 2
- Following up on sanitizers VMA discussion
* Two days (Tue and Thu) off. [4/10]
# Progress #
* TCWG-180, patch is committed by Andrew Stubs. [1/10]
I review and approve it.
* Review patch set "all-stop on top of non-stop for remote". [3/10]
Test them on aarch64 and arm. Triage and fix one regression. Review
is still on going.
* Review patch set "ARM gdbserver software breakpoint". [2/10]
Review and approve them. They are preparatory to ARM tracepoint.
# Plan #
* Patch review.
* TCWG-387, Move GDB using libopcodes to decode instructions.
--
Yao
The Linaro Toolchain Working Group is pleased to announce the availability
of the Linaro Stable Binary Toolchain GCC 5.1-2015.08 Archives.
http://releases.linaro.org/components/toolchain/binaries/5.1-2015.08/
These archives provide cross-toolchain executables (compiler, debugger,
linker, etc.) and shared libraries (libstdc++, libc, etc.) that target ARM
or Aarch64 GNU/Linux and bare-metal environments. The cross-toolchain
binaries execute on a Linux or MS Windows (under mingw32) host
operating-system.
For bugs related to this release please email
linaro-toolchain(a)lists.linaro.org or file a bug at
https://bugs.linaro.org/enter_bug.cgi?product=Linux%20Binary%20toolchain
NEWS
* 2015.08
* The Linaro GCC 5.1 2015.08 binary toolchain release is based on the Linaro
GCC-5.1-2015.08 source archive release
<http://releases.linaro.org/components/toolchain/gcc-linaro/5.1-2015.08/>.
The major changes contained in 2015.08 over 2015.05 are described in the
sections below on 2015.08-rc1 and 2015.08-rc2 .
* 2015.08-rc2
* The Linaro 2015.08-rc2 release-candidate binary toolchain is based on
the Linaro GCC-5.1-2015.08-rc1 release-candidate source archive. The only
changes between 2015.08-rc1 and 2015.08-rc2 were the following changes in
how the binary toolchains were built. The compiler itself was not changed.
* x86_64 hosted, armv8l-linux-gnueabihf targetted cross toolchains are
now correctly configured. This was broken in 2015.08-rc1. The
cross-compiler targetting armv8l-linux-gnueabihf is now correctly
configured with --with-arch=armv8-a --with-fpu=neon-fp-armv8
--with-float=hard --disable-multilib --enable-multiarch.
* Glibc’s slibdir and libdir were once again modified to address
Linaro Bugzilla
1717 – Linaro-4.9-2015.05 moved system libs from /libc/lib/ to
/libc/usr/lib/ which breaks things. The following are now the correct
locations:
libdir=lib/ (linker-scripts and static archives)
slibdir=usr/lib/ (shared objects)
rtlddir=lib/ (dynamic linker)
* 2015.08-rc1
* x86_64 hosted, armv8l-linux-gnueabihf targetted cross toolchain now
provided.
Delivering on REQ-477 – Enable x86_64 to Aarch32 (32-bit ARMv8-A)
cross binary toolchain product release and CARD-1637 – Enable Aarch32
(32-bit ARMv8-A) cross binary toolchain product releases ,
armv8l-linux-gnueabihf targetted toolchains are now available as part of
this release-candidate.
* Python support in GDB for both Linux and Mingw32 (32-bit windows).
Delivered as requested in the linaro-toolchain mailing list post
title – windows binary builds with gdb-python enabled?.
* Added missing expat support to GDB.
This addresses the following linaro-toolchain mailing list post –
"Missing expat support in GDB 7.8 multi-lib enablement for arm bare-metal
targets."
* A fix for multilib enablement in baremetal toolchains (as described
in ABE Review 6862).
* Library Paths are now congruent with older Linaro Toolchain path
layouts. This addresses Linaro Bugzilla 1717 – Linaro-4.9-2015.05 moved
system libs from /libc/lib/ to /libc/usr/lib/ which breaks things.
libdir=lib/ (shared objects)
slibdir=usr/lib/ (static libraries)
rtlddir=lib/ (dynamic linker)
--
Ryan S. Arnold
Linaro Toolchain Working Group - Engineering Manager
www.linaro.org
Hi,
The default gcc version of Yocto 2.0(Jethro) is 5.2.0, the gcc version in meta-linaro is gcc-4.9.4, may I know if gcc-linaro-4.9.4 will be used for Yocto 2.0 release? when will the linaro gcc be upgraded to gcc-5.x?
BTW, can someone please help point out where the Linaro toolchain roadmap can be got?
Best Regards,
Zhenhua
== Progress ==
LLDB development
-- Verify LLDB watchpoints on Android Arm and AArch64 targets. [TCWG-405] [4/10]
- Ran LLDB testsuite on Nexus5, Nexus 7, Nexus S and Nexus 9 devices.
- Started into failing watchpoint tests on all devices.
-- Investigate and fix LLDB watchpoints tests that are failing on
Android Nexus 9 AArch64 target. [TCWG-405] [5/10]
- Debugging of ptrace watchpoint installation code.
- Prepared a fix and tested on Nexus 9.
Miscellaneous [1/10]
-- Meetings, emails, discussions etc.
== Plan ==
Further progress on investigating LLDB watchpoint issues on Android
Nexus devices.
Local testbots setup using new Android Nexus and Arm linux devices.
== Progress ==
* Libraries (3/10)
- Back investigating libc++ issues in AArch64 buildbot
- http://buildmaster.tcwglab.linaro.org/builders/clang-cmake-aarch64-prototype
- Trying to solve the rt/unw/c++ library entanglement in the driver
* Buildbots (1/10)
- Trying to perf an application on APM, but it crashed while
- Rebooted it for now, but on the list to migrate to a newer kernel
* Background (6/10)
- Code review, meetings, discussions, general support, etc.
- Laptop playing stupid again, lost a day trying to fix things
- New LLVM "code of conduct" sparked a huge discussion
Investigate effectiveness of noise-control measures - TCWG-358 [4/10]
* Recovered from a couple more crashes, existing logging sheds no light
* Explored the existing data a bit, need more to draw any conclusions
Controlled image builds - TCWG-360 [2/10]
* Looked into controlled, repeatable debian builds, B&B-style
* Got the general shape of it, checking my understanding with Fathi
LAVA microinstance - TCWG-396 [1/10]
* Sent doc for review, revised it a bit
Misc [3/10]
* Exported benchmark/results handling rules to a place other
Linaro-ans can read them
* Updated example LAVA jobs to use tcwg tag (for load balancing purposes)
* Meetings, mail, etc
=Plan=
Generate more noise data, watching out for crashes
Produce a candidate Debian image, document how I did it
If time, learn about Workload Automation
== This week ==
* TCWG-317 - Exploit wide add operations when appropriate for Aarch32 (6/10)
- Debugging failures on -flto/big endian combination
- Wrote code for big endian lane support; failures still exist
- Debugging lto to see if optimization is causing failures
* TCWG-77 - Transform end of loop conditions to min_expr (1/10)
- patch approved, checked in upstream
* Misc (1/10)
- Conference calls
* US Holiday, Monday October 12th (2/10)
== Next week ==
- TCWG-317 - Debug to determine if lto is causing failures
== Progress ==
- Run spec2006 with DS5/Streamline Performance Analyzer (3/10)
- Setup DS5
- Setting up client for chromebook which requires kernel rebuild
- Widening pass (TCWG-547) - 5/10
* iterated based on review comments
* re-wrote GIMPLE_DEBUG handling
- Misc (2/10)
* gcc/bug list
== Plan ==
* perf with spec2000
* continue with widening pass based on feedback
