ABE benchmarking automation - TCWG-360 [8/10]
* Pursued sneaky bugs through twisty mazes of bash
* Learned that ControlMaster does not play nice with my scripts
* Learned more about IPC pitfalls
* Cleaned up build behaviour of scripts somewhat
* Implemented support for benchmarking against sysroot (needed by Will
and by me)
* Generally tried to help Charles
* A bit of Jenkins testing
Misc [2/10]
* Featuring part 1 of an office move
=Plan=
Holiday Wednesday, 1/2 of Thursday
Part 2 of office move
Support benchmark users
Look more at cache effects on Juno
Get Jenkins working
Benchmark catomic patches (as much as I can without putting even more
pressure on the Junos)
Write up/think about where to go with libm exercising
Final preparation for Connect
The Linaro Toolchain Working Group (TCWG) is pleased to announce the 2015.02
stable release of the Linaro GCC 4.9 source package.
Linaro GCC 4.9 2015.02 is the eleventh Linaro GCC source package release in the
4.9 series. It is based on FSF GCC 4.9.3-pre+svn220525 and includes performance
improvements and bug fixes.
With the imminent release of ARMv8 hardware and the recent release of the
GCC 4.9 compiler the Linaro TCWG will be focusing on stabilization and
performance of the compiler as the FSF GCC compiler. The Linaro TCWG provides
stable[1] quarterly releases and monthly engineering[2] releases.
Interesting changes in this GCC source package release include
* Backport of [AArch32] PR63676, exit tree fold when node be TREE_CLOBBER_P
* Backport of [AArch32] M0/M1 small multiply
* Backport of [AArch64] Fix unsafe access to deallocated stack
* Backport of Int div by constant compilation enhancement
* Backport of [AArch32] Fix PR target/64460: Set 'shift' attr properly on some patterns
* Backport of [AArch32] Fix typo in *<arith_shift_insn>_shiftsi
* Backport of [AArch64] Let LR register allocable
* Backport of Improve warning message for bitfields
* Backport of [AArch32] PR rtl-optimization/64011
* Backport of Fix checking on MAX_PENDING_LIST_LENGTH
* Backport of Remove VEC_LSHIFT_EXPR and vec_shl_optab
* Backport of [AArch64] Use new reduc_[us](min|max)_scal optabs, inc. for builtins
* Backport of [AArch64] Use new reduc_plus_scal optabs, inc. for __builtins
* Backport of Add new optabs for reducing vectors to scalars
* Backport of [Vectorizer] Make REDUC_xxx_EXPR tree codes produce a scalar result
* Backport of [AArch32] new cortex-m7 tune option
* Backport of [AArch32] cortex-m7 scheduling
* Backport of Improve Neon intrinsics testing
* Backport of [AArch32] [testsuite] gnu11 cleanup for aapcs testcases
Linaro GCC 4.8 2015.02 is the sixteenth release in the 4.8 series and is in
maintenance. Based off the latest GCC 4.8.4+svn220525 release, it includes
performance improvements and bug fixes.
The Linaro GCC 4.8 maintenance branch will be retired when Linaro GCC 5.1 is delivered.
Linaro GCC 4.9 will become the new maintenance release. Interesting changes in this GCC source
package release include:
* Updates to GCC 4.8.4+svn220525
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 Launchpad against "Linaro GCC project":
http://bugs.launchpad.net/gcc-linaro/+filebug.
* Questions? "ask Linaro":
http://ask.linaro.org/.
* Interested in commercial support? inquire at "Linaro support":mailto:
support(a)linaro.org
[1] Stable source package releases are defined as releases where the full Linaro
Toolchain validation plan is executed.
[2] Engineering source package releases are defined as releases where the
compiler is only put through unit-testing and full validation is not
performed.
--
Michael Collison
Linaro Toolchain Working Group
michael.collison(a)linaro.org
== This week ==
* Linaro 4.8 and 4.9 Releases (9/10)
- Validated backports, followed release process and staged tarballs
for release
* Misc (1/10)
- Conference calls
== Next week ==
- Backports
- Work on bug fixes
* TSAN support for Aarch64 (3/10)
Working on memory map layout for TSAN in Aarch64 (TCWG-581) .
Experimenting switching on 64 bit allocator as default.
* Bug 869 - Analyzing tree dumps on how VECT_CONVERT_EXPR works (2/10)
* Emails, meetings. (3/10)
* Linaro remote hack sessions.
* internal AMD meetings/event, 1-1 with AMD manager, .
* GCC mailing list.
Leave on 10-Feb-2015 (2/10)
== Plan ==
* TSAN support for Aarch64 work on memory layout.
* Bug869
* Review/Watch connect slides.
== This week ==
* Created dummy release in preparation for February Linaro release (2/10)
* Backports (2/10)
- Backported 216220 and 217229
- Additional merge issues indicates another backport is required.
- I will abandon the backports for this release.
* Bug fixing (5/10)
- Bug 535 - Could not reproduce after triage.
- Bug 539 - Additional triage.
* Misc (1/10)
- Conference calls
== Next week ==
- Commit validated backports and create release
== Progress ==
* AArch64 ILP32 toolchain (5/10)
- Got SPEC2k results for LP64 and ILP32
* Committed patch for AArch64 -Bsymbolic ld issue (1/10)
* Other stuff (2/10)
- Working on handing over various projects
- Email, meetings, etc.
* Friday travelling to Connect (2/10)
== Issues ==
* Juno boards have been busy
== Plan ==
* Connect
--
Will Newton
Toolchain Working Group, Linaro
== Progress ==
* Automation Framework (CARD-1378 1/10)
- Restarting build-02
- Trying new way to setup juno-01, failed
- Moving gateway-config to new private git server
* Release 3.6 (CARD-1431 4/10)
- Found NEON regression, reverting patches
- Setting up more new NEON buildbots
* Libraries (CARD-1831 1/10)
- Opening discussion on libunwinder move, getting consensus
* Background (2/10)
- Code review, meetings, discussions, etc.
- FOSDEM 2015 (http://llvm.org/devmtg/2015-02/)
* Flying to Connect (2/10)
== Issues ==
After working on Sunday and all nights since, I'll be off tomorrow and
will get ready to Connect on Friday. Email/HO/IRC as usual.
== Plan ==
Connect
* TSAN support for Aarch64 (5/10)
Working on memory map layout for TSAN in Aarch64 (TCWG-581) .
* Emails, meetings. (3/10)
* 1-1 with maxim, linaro status call.
* internal AMD meetings, 1-1 with AMD manager.
* GCC mailing list.
Leave on 26-Jan-2014 (India Holiday) (2/10)
Planned Leave on 04-Feb-2014
== Plan ==
* TSAN support for Aarch64 work on memory layout.
== This week ==
* Backports (10/10)
- 216734 - Temporarily remove aarch64_gimple_fold_builtin code for
reduction operations
- 216736 - [Vectorizer] Make REDUC_xxx_EXPR tree codes produce a
scalar result
- 216737 - Add new optabs for reducing vectors to scalars
- 216738 - Use new reduc_plus_scal optabs, inc. for __builtins
- 216741 - Use new reduc_[us](min|max)_scal optabs, inc. for builtins
- 216742 - Restore gimple_folding of reduction intrinsics
- 216779 - Remove VEC_LSHIFT_EXPR and vec_shl_optab
- 217331 - Fix checking on MAX_PENDING_LIST_LENGTH
- 217430 - Fix typo in *<arith_shift_insn>_shiftsi
- 217431 - Let LR register allocable
- 217533 - Pair load store instructions using a generic scheduling
fusion pass
- 219717 - PR rtl-optimization/64011
- 219718 - Improve warning message
== Next week ==
- Prepare for release by creating "dummy" release
- Work on bug fixes
Hi,
We've noticed that in the 14.11 toolchain binary release that the .asc files
contain md5sums instead of PGP signatures. Would it be possible to either
switch these files to contain signatures or change the filename to indicate
that they contain md5sums?
I also noticed that an md5sum is missing:
http://releases.linaro.org/14.11/components/toolchain/binaries/armeb-linux-…
Finally, there appears to be some inconsistent usage of tuples in the download
directories. Would it be possible to use the same tuple for a particular
toolchain everywhere that tuple is used? For example, for the arm-none-eabi
toolchain, the directory is "arm-none-eabi", but the compiler tarball has only
"arm-eabi".
Thanks,
Chris
--
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project
== Progress ==
* Automation Framework (CARD-1378 1/10)
- Setting up new box
* Release 3.6 (CARD-1431 5/10)
- Looking for the patch that introduced a major NEON regression since December
* Background (4/10)
- Code review, meetings, discussions, etc.
- More Jira farming...
- Some EuroLLVM paper reviews
- LLVM/Android planning
== Plan ==
* Keep trying to find the bug in release 3.6
* Try to have some time to fix the NEON unwinder issue
* Preparing for Connect
== Issues ==
I cocked up the LLVM 3.6 ARM release last December when I
inadvertently ignored a bug report about clang not self-hosting on
NEON machines. To make matters worse, I was the one that decided to
move *all* our buildbots to non-NEON machines, expecting that the
test-suite on a Chromebook would catch all NEON problems.
There are limits to human intelligence, not human stupidity.
== Progress ==
* LLD (1/10, TCWG-561, TCWG-563, TCWG-570)
- Updated and added more cards, adding more detail and loose priorities
- Rebased, fixed up and resubmitted relocation test patch
* AArch64 ILP32 toolchain (4/10)
- Attempted various SPEC2k runs and fixed issues found
* Submitted patch for AArch64 -Bsymbolic ld issue (1/10)
* Connect slides (2/10)
* Other stuff (1/10)
- Email, meetings, etc.
* Tuesday looking after sick child
== Issues ==
* Juno access has been patchy
== Plan ==
* Get SPEC2k running ok on Juno and with ILP32
* Push remaining LLD patches
--
Will Newton
Toolchain Working Group, Linaro
Hello,
I have a question about the status of overo support in qemu-linaro vs upstream qemu; from a quick look on the Git commits for qemu-linaro, most things seemed to have been ported back upstream into qemu as of v2.0.0, although there does appear to be some recent new commits. However, I'm looking for overo board support, and although I see that OMAP3 chip support has been merged upstream, there aren't any board configuration files for overo. What would be the easiest way to get working qemu overo board support?
Additionally, I don't see any recent binary packages available for qemu-linaro. The qemu-linaro packages on the Ubuntu PPA haven't been built for almost a year now, and the releases.linaro.org page doesn't seem to have recent qemu-linaro builds either.
Thanks,
Dominic
* TSAN support for Aarch64 (5/10)
Up streamed make file changes for building non x86_targets (TCWG-579).
Working on memory map layout for TSAN in Aarch64 (TCWG-581) .
* Fix Bug 863(Linaro)/ 63949(FSF) (2/10).
Tested patches and fixed few aarch64 regressions.
got feedback that approach is good. planning to send a revised patch
for stage 1.
* Emails, meetings. (3/10)
* 1-1 with maxim,christophe,Ryan, linaro status call.
* internal AMD meetings, 1-1 with AMD manager.
* GCC mailing list.
Leave on 26-Jan-2014 (India Holiday)
== Plan ==
* TSAN support for Aarch64 work on memory layout.
* Send revised patch for Bug 863
== Progress ==
LLDB vs GDB investigation [TCWG-583] [4/10]
-- Scoring of previous patches and mailing list discussions
-- Tried and was able to build+test lldb on x86_64 linux
-- Tried and was able to build (but not test) lldb on arm linux
-- Tried lldb AArch64 build but couldnt get it going successfully.
Some gdb bug fixing and wrap up open gdb stuff [4/10]
Miscellaneous [2/10]
-- Meetings, Emails etc
-- Connect Preparation: Flights Confirmation, Visa pick up etc.
== Plan ==
Try fixing LLDB testsuite on arm and try out aarch64 build if possible.
ABE benchmarking automation - TCWG-360 [6/10]
* Fixed an awkward escaped process
* Struggled with LAVA timeouts & discovered that some Junos
accidentally have a baked-in timeout
* More discussion/thinking about where to store benchmark source
* Some trouble with intermittent flakiness when copying large chunks
of data over the lab network
Investigating cache effects on Juno - LDTS-1238 [2/10]
Misc - [2/10]
* Featuring catomics (TCWG-436) - implemented a few missing catomics
(well under 1/10)
=Plan=
Pack for office move
Fix intermittent copying problem
Look more at cache effects on Juno
Test Jenkins dispatch (Rob says it works now)
* Look into triggering coremark on every backport
Carry on, hopefully close, storage discussion
Benchmark catomic patches
Write up/think about where to go with libm exercising
== Progress ==
* Automation Framework (CARD-1378 2/10)
- Rebooting machines, again...
- Fiddling with bugzilla, wiki
* Release 3.6 (TCWG-575 5/10)
- Investigating ARMv7 bootstrapping failures
- Rebuilding manually with cmake, as autoconf build is bogus
- Turns out wasn't autoconf, but a NEON bug reported 14th Dec
- Benchmarking AArch64 3.6-RC1 against 3.5.1 - all good
* Background (3/10)
- Code review, meetings, discussions, etc.
- More Jira farming
- Investigating some weird llvm-objdump issues
- Fixing ninja output issue on bots with an ugly hack
== Plan ==
* Bisect NEON bug, revert the patch, back-port to 4.6 RC2
* Add a NEON+T2 full buildbot (dragonboard) to the list of public bots
* Benchmark ARMv7 3.6 RC1 or RC2 against 3.5.1
