- 2014.01 releases: (2/10)
* 4.7 ready, 4.8 needed a short daily to include our multilib patch
to support our binary releases.
* retried cbuildv2 and sent feedback. Should be mostly OK by now.
- cross-validations: (2/10)
* followup-up, bugzilla
* thinking about short term Neon intrinsics testsuite use without dejagnu
- libsanitizer (2/10)
* resumed work for aarch64
- misc (2/10): conf-call and meetings
- internal support (2/10)
* issue:
boards used for benchmarking have been off for several weeks. Pending
jobs disappeared from the list, but I got no log.
== Progress ==
* Implement fork over HW breaks support for arm native targets.
[TCWG-177] [6/10]
* ARM Process Record:
- All patches accepted and upstream.
- Spent time making changes to VFP, SIMD etc patch. [2/10]
- Configuration and Git ssh setup for gdb commit rights . [1/10]
* Submitted a new fix and worked with community for the fix. [1/10]
== Plan ==
* Continue to work on fork over HW breaks support for arm native
targets. [TCWG-177]
* Patch updates and submission to all in progress upstream submissions.
* Travel to Islamabad for Macau visa application.
== Progress ==
* Libc probes support for Aarch64.
Testing the patch for "libc" probes support to "__longjump".
Native compiling in openemebedded.
Obtained a Linaro open embedded image with system tap enabled.
Getting Assembler Error during "malloc" module linking in "glibc"
configured with --enable-systemtap.
In libc_mallopt invalid "asm", invalid expression as operand for
LIBC_PROBE call.
Cross compiling.
Tried copying header files from system tap source to SYSROOT for cross
compiling glibc. The cross compiled GCC does not recognize the header
files.
* Investigate "PGO" for Aarch64.
Started bootstap testing GCC with PGO for GCC.
Stage 1 completed. stage profile in progress.
== Plan ==
- Run EMBCC benchmark with -PGO enabled.
- Work on "glibc" probe.
== Issues ==
Waiting for feedback on "libssp" machine descriptions changes.
Misc
------
14-Jan-2014 Local holiday.
== Issues ==
* None.
== Progress ==
* Test Linaro cbuildv2.
- Write a reference script to support windows installer in cbuildv2.
- Do some basic validation work against the binaries built from cbuildv2.
* Reproduce and work out a patch for "ICE when building
linux-linaro-tracking" (lp:1268893, PR59837) .
* Test to merge Linaro local armv4t multilib patch to Linaro 4.8.
* Investigate CCMP chance by taking a Bool_ var as var != 0.
== Plans ==
* cbuildv2 validation
* Tuning ccmp performance
== Planed leaves ==
* Jan. 31 - Feb. 9.
== This week ==
- Completed merge of crypto backports 206128, 206130, 206131, 206132 and
206149 from trunk to Linaro 4.8 branch
- Out sick Friday
== Next week ==
- Build and test crypto backports
- Begin merging aarch64 crypto backport to Linaro 4.8 branch
== Future ==
== Progress ==
* Cbuildv2 fix for *_release() functions which were changing out of
the build directory but then not changing back to the builddir, which
caused problems with paths relative to the builddir. (1/10)
* TCWG-340 Enable Glibc remote testsuite integration - (7/10)
- Working with upstream community on scoping and designing remote
testsuite integration.
- Setting up arm chroot environment on chromebook with documentation
on how to make this reproduceable.
* TCWG-352 AArch64 clone is missing some CFI markup - (2/10)
- Setup aarch64 environment for testing and built static clone tests
to test the patch.
== Plan ==
* Start implementing (code) for glibc remote testsuite integration.
== Issues ==
None
== Progress ==
* Buildbots
- Odroid U2 died after a few days on USB stick, too. (I guess we're stuck
with Chromebooks)
- Upgrading chromebooks (to Saucy) due to a new GCC/libstdc++ 4.7+
restriction
- Shuffling chromebooks, we now have three buildbots:
- llvm-chrome-01: quick-build+check
- llvm-chrome-02: self-host+check
- llvm-chrome-03: test-suite
* MCJIT Remote
- Re-factored the communication protocol of MCJIT for remote processes
- Still seeing one random failure on ARM, investigating
* Compiler-rt
- Investigating how to get it compiling all libraries and sanitizers on ARM
* Background
- Patch reviews, especially IAS and EH
* Time
- CARD-862 8/10
- Others 2/10
== Plan ==
* Continue MCJIT and Compiler-RT investigations
* Continue reviewing IAS and EH patches
* Try to run the EH tests on the test-suite with the current changes
== Progress ==
* Add support for AArch32 ARMv8 VRINT to QEMU (4/10, TCWG-50)
* Add support for AArch32 ARMv8 VCVT to QEMU (4/10, TCWG-49)
* Add support for AArch32 ARMv8 16<->64bit VCVT to QEMU (1/10, TCWG-51)
* Fixed gas ARM VCVT encoding bug (1/10)
== Issues ==
* None
== Plan ==
* Complete outstanding QEMU work
* ARM setjmp/longjmp SystemTap probes
* malloc
--
Will Newton
Toolchain Working Group, Linaro
Hi,
I've been playing with linux-linaro-tracking and the 2013.12 (and
2013.10, 2013.11) toolchain release and when turning on the NEON in
kernel option I get:
CC lib/raid6/neon4.o
lib/raid6/neon4.c: In function 'raid6_neon4_gen_syndrome_real':
lib/raid6/neon4.c:113:1: internal compiler error: in
dwarf2out_frame_debug_adjust_cfa, at dwarf2cfi.c:1078
}
^
Use the attached .txt as .config and do 'make zImage' to replicate the
problem. You will probably need to attached patch as well since some of
the linaro patches break namespace support :(
regards,
--
Koen Kooi
Builds and Baselines | Release Manager
Linaro.org | Open source software for ARM SoCs
