== Progress ==
* Releases 4.7 and 4.8 2013.10
* Tried cbuildv2: difficult because of our company proxy, making some
changes necessary to cbuildv2. Tutorial sessions in Connect will be
welcome.
* Started chasing a regression in armv5 targets, handed over to Kugan.
* Noticed regressions in some cases with Charles' testsuite patch.
== Next ==
Short week: Monday/Tuesday off; Friday: travelling to Connect
* Look at trunks' new vectorizer cost model.
* Prepare Connect
== Progress ==
* Short week Hujj and Eid-Al-Adha Public Holidays in Pakistan 15-17
October 2013.
(TCWG-280)
ARM GDB Process Record Syscall support:
* Performed testing with all fixes incorporated and debugged failures
in remote configuration.
== Plan ==
(TCWG-266), (TCWG-270), (TCWG-250), (TCWG-280) and (TCWG-288)
* Arm process record: Submit all fixes and updates in three patches.
* Prepare for connect and travel to Santa Clara.
Very short week was on leave 14th and 16th October (India Holidays).
== Progress ==
* libssp glibc
Glibc tests are Segmenting at dl_close_worker, when I set pointer
guard. Debuging shows segfaults while expanding THREAD_GSCOPE_WAIT.
My assumption is thread pointer points to TCB and after allocating 16
Bytes before that, we can set pointer gaurd at tp -16. But not sure,
why it affects the pthread structure header which holds gscope_flag.
Also looked at TLS_MULTIPLE_THREADS_IN_TCB, enabling it lead to
__libc_multiple_threads undefined errors.
== Plan ==
* Continue libssp support.
* PGO for aarch64
== Progress ==
* 64bit division for targets without 32bit div instructions
- Finished the implementation
- Benchmarked it
- Getting ready to post RFC patch
* Connect Slides
- Worked on slides for connect
* Vector regression in 4.8-2013.09
- Started bisecting to find the issue
== Plan ==
* post RFC patch for 64bit division
* Vector regression in 4.8-2013.09
* Connect preparation
== Progress ==
* Improved the format of the emails that get sent to the cbuildv2
list from commits.
* Get Odroid U2 running stably, did native build on an external
USB disk. Upgraded to saucy. Did full toolchain build and test
run. 65 minutes per build.
* Got Odroid XU board to boot, then upgraded to saucy, not very
stable yet. Unlike the XU, it has a horrible lag when typing to
it via ssh.
* Updated chromebook to saucy, did native build and test run on
external USB drive. The U2 was faster, even with an older USB 2
only drive. 192 minutes for a build.
* Worked at verifying Cbuildv2 produced tarballs again since there
have been many little tweaks. The few differences between
Cbuildv2 produced tarballs and the official Linaro releases is
documentation related. For example, Cbuildv2 doesn't build the java
docs.
* Did slides for Connect presentation.
* Sat bagged an 900 foot ice route at over 13,000 feet. 8 mile
approach.
== Plan ==
* Try to work around problem with importing test results. (need to
work with LAVA team for this one)
* Copy tarballs to snapshots.linaro.org. (need to work with LAVA
team for this one)
* Get remote toolchain testing working on the newly functioning
Odroid board(s).
* Continued improvements to Cbuildv2.
== Issues ==
* My network and landline was down to upstream repairs for a day
and a half, bad timing during release week... :-(
Hi Bharath. Have a look under contrib/linaro/sysroot - there's a
script in there that you can fiddle with to pick a later distro.
-- Michael
On 16 October 2013 17:52, bhs <bharath.vegito(a)gmail.com> wrote:
> Hi Michael,
>
> I see that linaro's build system is using pre-built sysroot as an input
> to build cross-toolchain for ARM. I have a use case were we want
> eglibc-2.17(sysroot from raring) along with gcc-4.7.3 crosstoolchain. I
> could not any reference to the method in README file.
>
> How is the precise-sysroot used in building linaro toolchain binaries
> created?
> Thanks
> Bharath
> --
> This message was sent from Launchpad by
> bhs (https://launchpad.net/~bharath-vegito)
> using the "Contact this user" link on your profile page
> (https://launchpad.net/~michaelh1).
> For more information see
> https://help.launchpad.net/YourAccount/ContactingPeople
I accidentally sent without completing my 'Progress' section.
== Progress ==
eglibc -> glibc transition:
* Investigated upstream plan to retire eglibc and the implication of
losing options groups.
Cbuildv2 - Random patches
* [PATCH][Cbuildv2/get_source 17/17] lib/common.sh: get_source - Fix
for partial match in snapshots with <package>.conf:latest set.
* [PATCH][Cbuildv2/get_source 16/17] testsuite/test.sh: Add get_URL
test for <repo>.git@revision.
* [PATCH][Cbuildv2/get_source 15/17] lib/common.sh: Fix so that
get_URL works for <repo>.git@revision.
* [PATCH][Cbuildv2/get_source 14/17]
testsuite:test_sources.conf|test.sh: reserve nomatch.git for testing.
* [PATCH][Cbuildv2/get_source 13/17] testsuite/test.sh: Add tests
for user git repos not ending in .git.
* [PATCH][Cbuildv2/get_source 12/17] [test_]sources.conf: Unify
behind <label>.git identifiers for all git repos.
* [PATCH][Cbuildb2/get_source 11/17] New variable ${sources_conf} to
replace hardcoded config/sources.conf.
* [PATCH][Cbuildv2/get_source 10/17] testsuite/test.sh: Add test for
tag matching an svn repo in sources.conf.
* [PATCH][Cbuildv2/get_source 09/17] lib/common.sh: echo ${url} in
leg that processes direct non .git repositories.
* [PATCH][Cbuildv2/get_source 08/17] lib/common.sh: Fix for
unknown/unmatched get_source request.
* [PATCH][Cbuildv2/get_source 07/17] lib/common.sh: echo ${url} in
get_source when finding a successful snapshot match.
* [PATCH][Cbuildv2/get_source 06/17] lib/common.sh: find_snapshot
should return list of found snapshots even on error.
* [PATCH][Cbuildv2/get_source 05/17] lib/common.sh: Invoke
find_snapshot as a subshell in get_source.
* [PATCH][Cbuildv2/get_source 04/17] lib/common.sh: Suppress
extraneous trailing whitespace in get_URL and get_source output.
* [PATCH][Cbuildv2/get_source 03/17] testsuite/test.sh: Suppress
stderr noise and add new failing get_source tests.
* [PATCH][Cbuildv2/get_source 02/17] lib/checkout.sh: checkout
should use get_srcdir to reuse code.
* [PATCH][Cbuilv2/get_source 01/17] lib/make.sh: Don't continue with
build if source isn't found.
* [PATCH][Cbuildv2/get_source 00/17] Patchset to exhaustively test
get_source code legs and fix corner cases.
* [PATCH] cbuild2/lib/common.sh: redirect trace/error/warning, etc. to stderr.
* [PATCH] cbuild/lib/configure.sh: Fix stamp-configure* to prevent
spurious configure running.
* [COMMITTED][PATCH] cbuild/lib/make.sh: Add ${builddir}/install.log
from make install output.
* [COMMITTED][PATCH] lib/common.sh (list_URL): correct bash keyword
"locals" -> "local".
* [COMMITTED][PATCH] config/gcc.conf: Add --disable-shared to stage
1 default_configure_flags.
* [PATCH][test_fetch_md5sums_fix] testsuite/test.sh: Add second
fetch md5sums test and always use wget -q in testsuite.
* [PATCH][test_fetch_md5sums_fix] testsuite/test.sh: Always override
${local_snapshots} for tests.
== Issues ==
* None
== Plan ==
Objective before Linaro Connect:
* Wiki/README on using cbuildv2 sources.conf and <package>.conf:latest=
* git workflow document: collaborative use of git in Linaro
* wiki/README on using cbuildv2 local_snapshots for
specific/non-remote archives or git repositories.
General Cbuildv2:
* Audit usage of "local foo=`bar`; if test $?". Change to: "local
foo; foo=`bar`; if test $?".
- This is a known issue and very nefarious.
* Investigate dryrun to use stderr 2>&1?
* Port get_URL, get_source, and find_snapshot to unique behind
<repo>/<branch>@<revision> return type.
* Test local_snapshot overrides
== Progress ==
eglibc -> glibc transition:
* Investigated upstream plan to retire eglibc and the implication of
losing options groups.
Cbuildv2
*
[PATCH][test_fetch_md5sums_fix] testsuite/test.sh: Always override
${local_snapshots} for tests.
[PATCH][test_fetch_md5sums_fix] testsuite/test.sh: Add second fetch
md5sums test and always use wget -q in testsuite.
== Issues ==
* None
== Plan ==
Objective before Linaro Connect:
* Wiki/README on using cbuildv2 sources.conf and <package>.conf:latest=
* git workflow document: collaborative use of git in Linaro
* wiki/README on using cbuildv2 local_snapshots for
specific/non-remote archives or git repositories.
General Cbuildv2:
* Audit usage of "local foo=`bar`; if test $?". Change to: "local
foo; foo=`bar`; if test $?".
- This is a known issue and very nefarious.
* Investigate dryrun to use stderr 2>&1?
* Port get_URL, get_source, and find_snapshot to unique behind
<repo>/<branch>@<revision> return type.
* Test local_snapshot overrides
Hi,
We are working on userspace aarch64 and aarch32 apps and we are trying
to run 32/64 mode app simultaneously.
Is there any multi-arch support for mixed rootfs (has linkers/libs of
both arm and arm64)?
--
Thanks,
-Zhou
Progress:
* back from sabbatical!
* caught up on admin, email, etc
* code review: Christoffer's GIC related patches; VSEL support in 32 bit v8;
support for VBAR; integrator bugfixes; others
* working on a pullreq for ARM stuff to go into QEMU 1.7
(we're now in softfreeze)
Plans:
* attend KVM Forum, Connect
* review the SuSE aarch64-linux-user patches
-- PMM
== Issues ==
* None
== Progress ==
* Prepare Linaro toolchain binaries 2013.10 release
- Update EGLIBC version to 2013.10 and enable-profile for AARCH64.
- Generate Linaro prebuilt sysroot for arm-linux-gnueabihf.
* Commit patch to Reassociate X == CST1 || X == CST2 if popcount (CST2
- CST1) == 1 into ((X - CST1) & ~(CST2 - CST1)) == 0
* Send ifcombine patch (to generate more conditional compare) for review.
* Root cause pr58775.
* Test conditional compare changes.
- Bootstrap and no make check regression on THUMB mode.
- ARM mode test is ongoing.
== Plan ==
* Send out the CCMP patch for review.
== Issues ==
* Short week
- school closed on Thursday
- holiday Friday
- several visits to the dentist
- feeling poorly, I think I have shellfish allergy
== Progress ==
* EEMBC
- Identified the problem on test not vectorizing
- Some ways to fix it: SLP, stride-access, unrolling
- Started implementation of GEP-PHI detection for reduction variables
* Odroid XU
- Bot died after a few hours, not really stable
- Trying again, at home, both odroids in parallel, they seem ok
* Background
- General reviews, etc.
The Linaro Toolchain Working Group is pleased to announce the
release of both Linaro GCC 4.8 and Linaro GCC 4.7.
Linaro GCC 4.7 2013.10 is the nineteenth release in the 4.7
series. Based off the latest GCC 4.7.4+svn203509 release, this is the
sixth release after entering maintenance.
Interesting changes include:
* Updates to GCC 4.7.4+svn203509
* Fix for LP #1229170
Linaro GCC 4.8 2013.10 is the seventh release in the 4.8 series. Based
off the latest GCC 4.8.1+svn203510 release, it includes performance
improvements and bug fixes.
Interesting changes include:
* Updates to GCC 4.8.1+svn203510
* Improved AArch64 support (CRC extension, improved intrinsics, gprof
support)
* Improved Aarch32 support (bug fixes, better code generation,
improved multilib)
* Backports for bug fixes (PR58578
We are aware that FSF GCC 4.8.2 was released earlier this week.
However, this release happened too late in our cycle for us to use it
as the basis for Linaro GCC 4.8 2013.10. The only functional
difference contained in FSF GCC 4.8.2 that is not in Linaro GCC 4.8
2013.10 is a correction to a x86 test case.
The source tarballs are available from:
https://launchpad.net/gcc-linaro/+milestone/4.7-2013.10https://launchpad.net/gcc-linaro/+milestone/4.8-2013.10
Downloads are available from the Linaro GCC page on Launchpad:
https://launchpad.net/gcc-linaro
More information on the features and issues are available from the
release pages:
https://launchpad.net/gcc-linaro/4.7/4.7-2013.10https://launchpad.net/gcc-linaro/4.8/4.8-2013.10
Mailing list: http://lists.linaro.org/mailman/listinfo/linaro-toolchain
Bugs: https://bugs.launchpad.net/gcc-linaro/
Questions? https://ask.linaro.org/
Interested in commercial support? Inquire at support(a)linaro.org
### About Linaro binutils
Linaro binutils is a release of the GNU binutils with bug fixes and
enhancements for ARM platforms. GNU binutils is a collection of tools
including the `ld` linker and `as` assembler.
### Linaro binutils 2.23.2 2013.10-4
The Linaro Toolchain Working Group is pleased to announce the 2013.10-4
release of Linaro binutils 2.23.2, the third release in the 2.23 series.
This release is based on the latest GNU binutils 2.23 stable branch, but
with additional features and bug fixes.
#### Additional Features
* AArch64 GC sections support
* Support passing -mcpu=cortex-a53 and -mcpu=cortex-a57 to AArch64 gas
* Support passing -mcpu=cortex-a53 and -mcpu=cortex-a57 to ARM gas
* Support passing -mcpu=cortex-a12 to ARM gas
* Support AArch64 CRC instructions
* Support ARM CRC instructions
#### Bug Fixes
* Bug fix for AArch64 GOT layout
* Fixed source tarball structure
### Source
#### Release Tarball
* https://releases.linaro.org/13.10/components/toolchain/binutils-linaro
#### Development Tree
* git://git.linaro.org/toolchain/binutils.git
This release was built from the `linaro_binutils-2_23_2-2013_10-4_release`
tag.
### 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`
* Questions? [ask Linaro](http://ask.linaro.org/).
* Interested in commercial support? inquire at [Linaro
support](mailto:support@linaro.org)
--
Will Newton
Toolchain Working Group, Linaro
### About Linaro eglibc
Linaro eglibc is a release of the eglibc C library with bug fixes and
enhancements for ARM platforms. eglibc is a variant of the GNU libc
designed to work well on embedded systems.
### Linaro eglibc 2.18 2013.10
The Linaro Toolchain Working Group is pleased to announce the 2013.10
release of Linaro eglibc 2.18, the second release in the 2.18 series.
This release is based on the latest upstream eglibc 2.18 stable branch,
but with additional features and bug fixes.
#### Additional Features
* AArch64 profiling support
* Optimized strlen for ARMv6T2
* Small enhancement to memcpy performance
* ARM pointer encryption support
#### Bug Fixes
* Fix for integer overflows in memalign/pvalloc/valloc
### Source
#### Release Tarball
* https://releases.linaro.org/13.10/components/toolchain/eglibc-linaro
#### Development Tree
* git://git.linaro.org/toolchain/eglibc.git
This release was built from the `linaro_eglibc-2_18-2013_10_release` tag.
### 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`
* Questions? [ask Linaro](http://ask.linaro.org/).
* Interested in commercial support? inquire at [Linaro
support](mailto:support@linaro.org)
--
Will Newton
Toolchain Working Group, Linaro
== Issue ==
* AArch64 frame grows downward: no feedback yet.
== Progress ==
* Disable-peeling: Vectorizer cost model has been changed upstream,
quick look at benchmark results show little difference between before
and after the change. Need to analyze further, since I expected the
regressions previously identified with my patch to appear here too.
* AArch64 bootstrap failure: succesfully bootstrapped in the model.
* GCC trunk validation:
- improved configurability (default cpu, fpu, mode, testsuite flags
are selectable)
- requested feedback to build the list of desirable combinations to validate
* Backports/release 2013.10
- a few more backports completed
- re-trying to backport the ILP32 and refactoring series proves
again to be tricky, but helped identify backports which had escaped
from the list.
- cbuild has disk full again
* GCC Neon intrinsics testsuite:
-- posted a proposition to add my executable tests to the existing
framework. No feedback yet.
== Next ==
* Make 2013.10 releases, both with the historical process and using
cbuildv2, in parallel
* Disable-peeling/new vectorizer cost model: analyze bench results
== Progress ==
* Got cbuildv2 google group setup.
(https://groups.google.com/a/linaro.org/forum/#!forum/cbuildv2)
* Got email notifications working for commits to cbuildv2 repository.
* Continued work on release building via Cbuildv2. Source tarballs
checked out OK, binaries too. Probably a few tweaks left to cover
the minor differences, which are all documentation files.
- Rebuild GCC dynamically so the shared libraries for libgcc and
libstdc++ can go in the runtime binary tarball.
- Build runtime binary tarball.
- Added install script to sysroot tarball.
- Make overriding the default release tag with a user supplied
one work fully.
* Got Odroid U2 board working again, did complete native build
using cbuildv2.
* Reviewed more cbuild2 patches from Ryan.
* Got Jenkins building snapshot tarballs.
== Plan ==
* Do slides for Connect presentation.
* Work on duplicating the 2013.10 release via cbuildv2.
* Get Odroid X2 board to boot, which it currently refuses to do.
* Try to work around problem with importing test results.
* Copy tarballs to snapshots.linaro.org.
== Progress ==
* 4 Day week (October 7th Public Holiday)
* 64bit division
- Looked at libgcc division implementations
- Implemented in C "Align divisor shift dividend method" for
procssors without div instrusttion
- gcc regression now pass with above
* spec2k comparison between ARM and x86
- Continue working on slides for connect
- ran benchmarks to gather required results
== Plan ==
* Continue with spec2k comparison between ARM and x86
* Continye with 64bit division
== Progress ==
(TCWG-266), (TCWG-270), (TCWG-250), (TCWG-280) and (TCWG-288)
* Added arm process record syscall support changes and some bug fixes.
* More improvement to gdb.reverse on arm now 2214 tests are passing in
native mode (144 failures left)
(TCWG-267), (TCWG-268) and (TCWG-269)
* No progress this week.
* Received US visa from US visa passport collection center.
* Booked travel for LCU13.
== Plan ==
* Short week Hujj and Eid-Al-Adha Public Holidays in Pakistan 15-17
October 2013.
(TCWG-250)
* Update/test arm process record thumb32 support patch incorporating
changes suggested by Will in his review.
(TCWG-280)
* Improve code changes and find ways to avoid hard coded constants
imported from Linux Kernel.
== Issues ==
* Develop machine was broken. Have to resetup the system.
== Progress ==
* Commit fix for pr58423
* Enhance Linaro crosstool-ng
- Fix lp:1232678
- Get EGLIBC version from config files automatically.
- Build aarch64 sysroot to support profile
(https://launchpad.net/linaro-toolchain-binaries/support/01/).
* Update the patch "Reassociate X == CST1 || X == CST2 if popcount
(CST2 - CST1) == 1 into ((X - CST1) & ~(CST2 - CST1)) == 0" according
to comments.
* Continue on CCMP
- Send the phiopt patch to handle CCMP for community review. Jeff
will follow-up it.
- Enhance ifcombine to generate more CCMP.
- Handle more than 2 compares in expand.
- Add new instruction patterns to generate CCMP for ARM. Test is ongoing.
== Plan ==
* Send ifcombine patch for community review.
* Continue on CCMP testing.
== Progress ==
* CbuildV2 Fixes to add fetch md5sums to testsuite/test.sh:
- Subject: [PATCH][test_fetch_md5sums 1/2] lib/fetch.sh: Add
missing ${local_snapshots} to md5sum path.
- Subject: [PATCH][test_fetch_md5sums 2/2] testsuite/test.sh: Add
fetch tests section.
* CbuildV2 Fixes to get testsuite/test.sh working in the source directory:
- Subject: [PATCH][cbuild2 1/3] testsuite/test.sh: Don't use
"warning" function before it is sourced.
- Subject: [PATCH][cbuild2 2/3] testsuite/test.sh: Remove
redundant sourcing of host.conf.
- Subject: [PATCH][cbuild2 3/3] testsuite/test.sh: Re-enable
running tests from src dir.
CbuildV2 random fixes:
- Subject: [PATCH][cbuild2 2/4] cbuild2.sh: correct --disable "boostrap"
- Subject: [PATCH][cbuild2 3/4] config/ffmpeg.conf: separate options
* CbuildV2 fixes to remove ${dir}/md5sums usage (it is redunant with
snapshots/md5sums).
- Subject: [PATCH][cbuild2 1/4] Remove unneeded
infrastructure/md5sums fragment
- Subject: [PATCH][cbuild2 4/4] Remove ${dir} specific md5sums files.
* Fixed error where gcc stage 2 configure returns early (erroneously)
on success.
- [PATCH] Only return early on error following configure
* Fixed error in downloading branch named git repos in lib/checkout.sh.
- [PATCH] Fix git clone to get repo without branch name as part of the url.
== Issues ==
Contacted its@linaro requesting permission to be able to create
personal branches in rsavoye’s cbuild2 git tree. They created a group
for cbuildv2.
Got work laptop installed (dualboot) with Fedora 19 and ran into some
errors with secure boot.
== Plan ==
* Spend some time looking at our options with regard to EGLIBC options
groups when forced to move to glibc proper.
*
== Progress ==
* Backports and release for binutils 2013.10 (x4!)
* Backports and release for eglibc 2013.10
* Submitted a couple more glibc memalign patches
* Committed aarch64 gdb fix
* Bits of malloc work in between
== Issues ==
* None
== Plan ==
* Try and get some uninterrupted time on malloc
--
Will Newton
Toolchain Working Group, Linaro
== Progress ==
* EEMBC
- Updating Linaro image with Clang toolchain changes
- Updating compare_all script to rank boards
- Investigating differences in performance against GCC
* ODroid XU
- Setting up server image (Linaro 13.09 + odroid specific changes)
- CPU scheduler is a mess, forcing "performance" governor at boot seems to
work
- No journal in root fs, remounts ro all the time (bad image?)
- LLVM builds are slow (as fast as Chromebook), maybe it's the SDCard
- Setting it as self-host buildbot in the lab, going well
* Background
- Massive patch review (Sub-Register Live Range Analysis for NEON/VFP et
al)
- LLVM Sync and EuroLLVM meetings
== Plan ==
* Continue investigating performance problems in EEMBC
* Check Odroid XU internal buildbot for stability
