Hi,
With respect to the MPFR build error,
(http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58578), is movs appropraite
as shown below? When _err1 is negative the condition is evaluating it to
be true.
/* C Codeoes */
if (_err1 > 0)
{
/* working code */
.loc 1 67 0
cmp r3, #0
ble .L6
/* not working code*/
.loc 1 67 0
movs r3, r3, asl #1
ble .L6
When I looked at the ARM documents, I found the following
Condition flags
-------------------
If S is specified, for MOV instructions:
1. update the N and Z flags according to the result
2. can update the C flag during the calculation of Operand2 (see
Flexible second operand)
3. do not affect the V flag.
And also
le Signed less than or equal. (Z==1) || (N!=V)
Does this means we cant movs at comparision.
http://gcc.gnu.org/ml/gcc-patches/2013-02/msg00861.html added this and I
guess it is intentional.
Am I missing anything here?
Thanks,
Kugan
== Progress ==
* AArch64 frame grows downward: no feedback yet.
* Disable-peeling: trying to tune the vectorizer cost model so that it
is less aggressive.
* AArch64 bootstrap failure: build it still on-going, so still hoping
to reproduce it.
* GCC trunk cross-validation:
- aarch64 added
- looking at how to handle compute farm jobs timeouts nicely
* Backports:
- launched several merge requests for a few individual patches.
== Next ==
* AArch64 frame grows downward: continue
* Disable-peeling: continue
* Backports : spend little time on checking the results
Progress
. worked on cleaning up arm config in gcc (TCWG-201)
. not really convinced I have a way to properly test the result
. a bit of bug triage
. a bit more on the 4.7 segfault bug (lp1215925)
Plans
. ping the arm*gnueabihf testsuite patch
. finish the config clean up
. revisit 4.7 segfault, now I have more suggestions from Matt
== Issues ==
* None.
== Progress ==
* Misc:
- Installed Saucy on Chromebook.
* LRA on AArch64:
- Committed RTL analyser patch.
- Committed patch which switch on LRA on AArch64.
* LRA on AArch32:
- Committed patch which fix failing assembly scan test with LRA.
- Committed LRA register notes DEAD and UNSUSED handling in LRA.
- Bootstrapped GCC with LRA on Chromebook.
- Analysis of testsuite results ongoing.
== Next ==
* Continue on LRA.
Short week (Tuesday and Wednesday went to US consulate Chennai for
VISA interview and document processing).
== Progress ==
* gprof support for aarch64
Re-based and tested. Asked permission for upstreaming in trunk.
Gibc patches are upstreamed. Thanks to Marcus.
* Aarch64 Boot strap failure.
Tried to reproduce it in V8 model. Build failing in stage 1 for me.
* 1-1 Meet with Christophe.
== Plan ==
* Back to libssp support.
* Aarch64 boot strap issue.
== Issues ==
* LTO/PGO work stopped now since libssp support priority is more.
Misc
------
October 2 India Holiday.
== Progress ==
* MPFR and GMP Build error with Latest 4.8 Release
- Reproduced MPFR error with a simplified testcase
- Found the buggy codesequence and the patch that introduced it.
- Looking at solution for this
- Reprdouced GMP error with -march=armv7-a and -mthumb
- continuing with the invetigation
* 64bit division
- Looked at current libgcc implementaion and studied the alorithm.
* Addressing patch review
- Posted second patch for zero/sign extension
* spec2k comparison between ARM and x86
- Ran spec2k with various options to validate code analysis
== Plan ==
* Continue with spec2k comparison between ARM and x86
* Continue 64bit division
* Fix GMP and MPFR build error
== Progress ==
(TCWG-270)
* Tested GDB Process Record/Replay and reverse debug feature for arm.
* Thumb32 support patch along with a couple of bug fixes made huge
improvement in testsuite results.
(TCWG-267) and (TCWG-269)
* Reviewed all previous submitted patches and wrote response patch updates.
* Implemented gdb/testsuite/gdb.dwaf2/dw2-error.exp using dwarf assembler.
(TCWG-268)
* Submitted patch for failing gdb/testsuite/gdb.dwaf2/dw2-case-insensitive.exp
(TCWG-272)
* Re-viewed and verified gdb bugs reported on launchpad and created
JIRA cards where required.
(TCWG-265)
* Generated Test data to test instruction decoding in arm specific
code of gdb process record/replay.
(TCWG-280)
* Reviewed previously submitted patches related to gdb process record
syscall support for arm targets.
* Submitted US Visa application for processing and got appointment for
Visa Interview.
* Reviewed all open JIRA cards and created new ones for future progress on GDB
== Plan ==
(TCWG-270)
* Submitt patch for bug fixes and show testsuite improvement with bug
fixes and thumb32 patch.
* Do more testing and analyse failures and try to fix them if possible.
* Verify issues reported by maintainers in old arm instruction decoding code.
(TCWG-267), (TCWG-268) and (TCWG-269)
* Update gdb/testsuite/gdb.dwaf2/dw2-error.exp as asked by gdb maintainers.
* Ping or Send response to comments.
(TCWG-280)
* Test syscall support with old patch and make changes if required.
* Travel to Islamabad on Tuesday to attend US visa interview early
morning Wednesday.
== Issues ==
* None
== Progress ==
* Update Conditional compare (CCMP) patch according to community comments.
* Investigate how to generate CCMP to support more than two compares.
- Investigate ifcombine to create more CCMP opportunities.
- Enhancing phiopt to handle CCMP case like:
if (a == 0 && b > c)
return 0
return a;
== Plan ==
* Continue on CCMP to support more than two compares.
== Panned leaves ==
* Oct. 1-7: National Day of China holiday
== Progress ==
* Profiling ImageMagick
- Clang 3.4-svn is around 10% less performing than GCC 4.7.3
- Mainly due to recursive octree-style map/reduce of most functions,
spending a lot of time
before/after branches
- Any optimization on that front won't fit the "Incubation" tag
* Profiling EEMBC
- Found 2 cases where Clang is *a lot* better than GCC 4.7.3 (pruning
bmark.c, producing much
smaller objects, and vectorizing telecom/autcor, which GCC can't)
- Most of the time, GCC is slightly better (~5% faster, ~10% smaller,
geomean)
- Wrote some scripts to run any two compilers against each other, compare
the results and
dump the ratios in a TSV file.
== Plans ==
* Update EEMBC Linaro image with Clang toolchain changes
* Look for low-hanging fruits in EEMBC
* Try comparison with GCC 4.8 and 4.9, too
* Find a way to extract Phoronix results in TSV format
== Progress ==
* Checked testsuite for glibc and gcc bootstrap on aarch64 native
* Released Linaro gdb 7.6.1 2013.09-1
* Submitted glibc pointer guard patch using TLS
* Submitted respin of pointer guard patch using a global variable
* Submitted binutils aarch64 ifunc patches for a second time
* A little bit of time on malloc
* Built office furniture!
== Issues ==
* None
== Plan ==
* Finish off pointer guard
* Hopefully back to malloc
--
Will Newton
Toolchain Working Group, Linaro
The Linaro Toolchain and Platform Working Groups are pleased to announce
the 2013.09 release of the Linaro Toolchain Binaries, a pre-built version
of Linaro GCC and Linaro GDB that runs on generic Linux or Windows and
targets the glibc Linaro Evaluation Build.
This release introduces a set of experimental (pre-alpha) crosscompilers to
big-endian ARM architectures.
Uses include:
* Cross compiling ARM applications from your laptop
* Remote debugging
* Build the Linux kernel for your board
What's included:
* Linaro GCC 4.8 2013.09
* Linaro Newlib 2.0 2013.06
* Linaro Binutils 2.23 2013.06
* Linaro Eglibc 2.17-2013.07-2
* Linaro GDB 7.6 2013.05
* A statically linked gdbserver
* A system root
* Manuals under share/doc/
The system root contains the basic header files and libraries to link your
programs against.
Interesting changes include:
* gcc updated to 4.8-2013.09
* Initial big-endian pre-alpha versions added
The Linux version is supported on Ubuntu 10.04.3 and 12.04, Debian 6.0.2,
Fedora 16, openSUSE 12.1, Red Hat Enterprise Linux Workstation 5.7 and
later, and should run on any Linux Standard Base 3.0 compatible
distribution. Please see the README about running on x86_64 hosts.
The Windows version is supported on Windows XP Pro SP3, Windows Vista
Business SP2, and Windows 7 Pro SP1.
The binaries and build scripts are available from:
http://releases.linaro.org/13.09/components/toolchain/binaries
Need help? Ask a question on https://ask.linaro.org/
Already on Launchpad? Submit a bug at
https://bugs.launchpad.net/linaro-toolchain-binaries
On IRC? See us on #linaro on Freenode.
Other ways that you can contact us or get involved are listed at
https://wiki.linaro.org/GettingInvolved.
### About Linaro gdb
Linaro gdb is a release of the GNU debugger with bug fixes and
enhancements for ARM platforms. gdb allows interactive debugging of
C and C++ programs natively and remotely.
### Linaro gdb 7.6.1 2013.09-1
The Linaro Toolchain Working Group is pleased to announce the 2013.09-1
release of Linaro gdb 7.6.1, the first release in the 7.6.1 series.
This release is based on the latest upstream gdb 7.6 stable branch,
but with additional features and bug fixes.
#### Additional Features
* None
#### Bug Fixes
* Fixes for debugging on AArch64 with hardware breakpoints.
### Source
#### Release Tarball
* https://releases.linaro.org/13.09/components/toolchain/gdb-linaro
#### Development Tree
* git://git.linaro.org/toolchain/gdb.git
This release was built from the `linaro_gdb-7_6_1-2013_09-1_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
== Progress ==
* Finally got GCC 4.7 automatic updates working from launchpad.
* Continuing with Jenkins/Cbuildv2/Git integration.
* Added support for using specific commit IDs with Git and
Cbuildv2. Also many improvements to how git branches and
revisions are handled.
* Worked on importing testing results from Jenkins builds, there
seems to be a problem finding the .sum files when the build is
complete.
* Started getting new Odroid boards online running Ubuntu.
== Plan ==
* Work on getting Canadian Cross builds working in Cbuildv2.
* Finish getting Odroid boards setup.
* Compare automated tarballs with the current release tarballs.
* Continuing with Jenkins/Cbuildv2/Git integration.
* Try to work around problem with importing test results.
== Issues ==
* Phone and network access has been randomly down for extended
periods. I assume it's flood repairs related.
== Progress ==
* GCC trunk cross-validation:
- stabilizing reporting (dealing with thread tests giving unstable
results when run under qemu, timeouts caused by overload of the
compute farm, ....)
- adding aarch64
* Backports:
- updated list (57 candidates for 4.8-2013.03), discussed
prioritization and collaboration with ARM.
- sent 1st branch-merge candidate for testing, as a preparation to
partially backport a bug fix from Richard.
* Aarch64 frame grows downward:
- modified test-case shows that code can be better with frame grows
downward than upward, depending on variables order and layout.
== Next ==
* Backports
- start backporting to prepare 2013.10 release
* AArch64 frame grows downward: continue
* Disable-peeling: resume work
== Progress ==
* Visit to Cambridge (Tue - Thu) and annual leave (Fri).
* Applied gas fix for neon vector load/store constraints.
* Applied glibc memcpy performance tweak.
* Applied gdbserver AArch64 fix to trunk and branch.
* Updated gdb release process and built a release candidate.
* Started glibc ARM pointer guard work
== Issues ==
* None
== Plan ==
* gdb 2013.09 release
* Various small items arising out of meetings last week
* Complete glibc ARM pointer guard
--
Will Newton
Toolchain Working Group, Linaro
== Issues ==
* None.
== Progress ==
* LRA on AArch64:
- Validate Vladimir's patch to support
HARD_REGNO_CALL_PART_CLOBBERED macro in LRA.
- Reduced and analysed libstdc++ failures : they were due to PCH
usage, but the issue was fixed with a rebase on trunk's head.
- Testsuite is now clean.
- Waiting for my rtl analyser patch to be accepted.
* LRA on AArch32:
- No more libstdc++ failures, they were PCH related too.
- One more issue in ARM mode testsuite, related bad REG_NOTES info
after LRA pass.
- Reduced failing Thumb testcase.
== Next ==
* Continue on LRA.
== Progress ==
* Spent week in Linaro's Cambridge office, had meetings with team mates.
* Visited ARM office in Cambridge and met toolchain team there.
* Submitted patch for GDB Record and Replay thumb32 support upstream.
* Reviewed and reproduced gdb bugs reported on launchpad and changed
their status according.
* Reviewed previously submitted gdb patches and replied to comments if required.
* Reviewed x86 record syscall code in GDB for adding support for
recording ARM syscall record.
* Updated JIRA cards or added new issues against bugs or features.
== Plan ==
* Progress on record and replay feature by testing instruction set support.
* Review all open JIRA cards for GDB
* Work on getting previous patches committed or resubmit new ones.
* Submit US visa application for LCU13.
== Issues ==
* None
== Progress ==
* 2 day off on Chinese Mid-Autumn Festival
* Update conditional compare (CCMP) patch according to community comments.
* Work out a patch to fix PR 58423.
* Prepare LCU'13 Visa and hotel.
== Plan ==
* Continue on CCMP.
== Panned leaves ==
* Sept. 24 AM: US Visa.
* Oct. 1-7: National Day of China holiday
== Progress ==
* Addressing patch review
- zero/sign extension preparation patch accepted.
- Christophe is helping to commit.
* spec2k comparison between ARM and x86
- Looked at conditional code generation related code in gcc
- looked at ifcvt and code in arm.md and arm.c to understand the
heuristics that dictate how much to conditionalize and what is causing
some of the problems I am seeing.
== Plan ==
* Continue with spec2k comparison between ARM and x86
* Start looking at 64bit division
== Progress ==
* gprof mcount support in glibc
Tested the patch and posted in libcports.
* Fix signal handler issue in glibc for profiling
Found issue with time profiling. Time profiling is obtained by setting up
SIG_PROF interrupt and kernal calls signal handler with interrupted pc
address. But in glibc, aarch64 port does not register the signal
handler properly.
Fixed and sent a patch to team for review. Tested the patch and
posted in libcports.
* Linaro connect - VISA application.
Applied VISA online and scheduled interview with US consulate.
* Libssp support for AArch64 TCWG 23:
Timeout issues disappears after increasing TIMEOUTFACTOR
Base glibc test results are captured in wiki
https://wiki.linaro.org/WorkingGroups/ToolChain/ReleaseAndInfrastructure/Cr…
* 1-1 meeting with Matt and Christophe.
== Plan ==
* Post RFC patch for Libssp support for AArch64 in GCC
* Continue testing the patch for Libssp support for AArch64 in GCC
and glibc TCWG-23.
== Issues ==
* LTO/PGO work stopped now since libssp support priority is more.
Misc
------
Tuesday and Wednesday in Chennai to attend US VISA interview.
== Progress ==
* Trying to profile ImageMagick on A9
- perf is broken on Exynos 4, used Panda
- Linaro image doesn't have all PPAs
- Clang code 10% worse than GCC on some operations
- Code doesn't look that bad, but Clang on x86 is faster
- Will look into that next week
Background
- Meetings at ARM, Linaro
- Omair in Cambridge
- Patch review, fixing bots, discussions, etc
- Moved from Ubuntu to Mint 15, using Gnome Shell
== Plan ==
* Finally look into the profiler output
* Increase the profiling to other image operations
* Profile Phoronix results?
Hi,
In building some code (CRIU) that includes linux/if_packet.h for
x86_64-linux-gnu, arm-linux-gnueabihf, and aarch64-linux-gnu, some
discrepancies popped up relating to whether the TPACKET_V3 structures were
defined. It looks like the headers shipped with the 13.08 arm-linux-gnueabihf
toolchain do not define them, implying kernel headers older than October 2012.
The aarch64-linux-gnu toolchain on the other hand does have them defined. If
this analysis is correct, can I expect the kernel headers in the 32-bit
toolchain to be updated in the near future or should I really be generating
and referencing my own kernel headers?
Thanks,
Christopher
--
Employee of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by the Linux Foundation.
== Progress ==
* GDB Record and Replay (TCWG-197)
Debugged code for an exception coming up in gdb code after applying
Thumb32 patch.
* Travel to UK on Wednesday night stopped in UAE for Thursday and Friday.
* Holiday on Thursday and Friday.
== Plan ==
* Working from Linaro Cambridge from Monday till Friday
* GDB Record and Replay (TCWG-197)
Fix issues Submit patch upstream.
* Meetings with team mates in Cambridge and visit to ARM.
== Progress ==
* 4.8 2013.09 backports:
- Finished ongoing reviews.
* LRA:
- Fixed and refactored rtl analyser patch, waiting fro upstream green light
- Reduced AArch64 testcase, and find the root cause of the issue:
Only 64 bits of callee saved registered v8-v15 are preserved, larger values
has to be handle by the caller, which not the case with LRA. Working on
the issue with LRA maintainer.
* LCU'13
- Hotel and flight booked
== Next ==
* Continue on LRA
