Hi all,
I've spent a little while porting an optimization from Python 3 to
Python 2.7 (http://bugs.python.org/issue4753). The idea of the patch is
to improve performance by dispatching opcodes on computed labels rather
than a big switch -- and so confusing the branch predictor less.
The problem with this is that the last bit of code for each opcode ends
up being the same, so common subexpression elimination wants to coalesce
all these bits, which neatly and completely nullifies the point of the
optimization. Playing around just building from source directly, it
seems that -fno-gcse prevents gcc from doing this, and the resulting
interpreter shows a small performance improvement over a build that does
not include the patch.
However, when I build a debian package containing the patch, I see no
improvement at all. My theory, and I'd like you guys to tell me if this
makes sense, is that this is because the Debian package uses link time
optimization, and so even though I carefully compile ceval.c with
-fno-gcse, the common subexpression elimination happens anyway at link
time. I've tried staring at disassembly to confirm or deny this but I
don't know ARM assembly very well and the compiled function is roughtly
10k instructions long so I didn't get very far with this (I can supply
the disassembly if someone wants to see it!).
Is there some way I can tell GCC to not compile perform CSE on a section
of code? I guess I can make sure that the whole program, linker step
and all, is compiled with -fno-gcse but that seems a bit of a blunt
hammer.
I'd also be interested if you think this class of optimization makes
little sense on ARM and then I'll stop and find something else to do :-)
Cheers,
mwh
The v8 Foundation Model User Guide has a bare metal hello world example that uses semi-hosting. The Makefile uses ARM tools, however. Is there equivalent support for this example using a bare metal version of the gnu tools, such as gcc-linaro-aarch64-none-elf-4.8-2013.04-20130422_linux.tar.xz? I took a look, but didn't see a way to do this.
Of course, running the Linaro linux port on the v8 Foundation Model allows one to run hello world and much more, but I'm currently only interested in a bare metal target using gnu tools.
Thanks, Don
== Progress ==
* Backed up laptop data and did new ubuntu installation which crashed for
some reason.
* Wrote python script with googledoc API to automate fill up of
googlespread sheet.
* Created and tested patch for arm assembler compatibility fixes for
gdb.dwarf test suite assembly files.
== Plan ==
* Identify arm bugs out of gdb7.6 test results and work towards fixing them.
* Update JIRA enteries with test suite failures on arm to track progress.
* More work on automating googledoc spreadsheet writing using python.
* 2 Day off on coming Friday and Monday.
== Progress ==
* AARCH64 - gprof support.
Completed gprof -c support for aarch64.
Got reviewed internally by Matt and Will.
Patch yet to be posted. Waiting for some feedback on copyright message.
*Testing GCC bootstrap and regression suite.
Created a large image with help of Bero.
Bootstrap fails with GCC trunk libgcc_eh.a (unwind-dw2-fde-dip.o)
hidden symbol __register_frame_info is referenced bu DSO
ld final link Bad value.
Drilling down
== Plan ==
* Post patches in gcc and binutils for gprof work
* Continue handling builtin_return_address when -fomit-frame-pointer
is enabled.
* Continue gcc bootstrap and regression test.
== Issues ==
* Cbuild down most of the week.
== Progress ==
* 4.6 and 4.7 releases
- Released after a painful week !
* LRA on ARM and AArch64:
- Enabled on AArch64, but it leads to an ICE too.
- Applied Brice's ARM patches didn't solved the issue.
- Looked at the documentation/comments to understand the process.
- Debug ongoing.
== Plan ==
* Continue on LRA
== Issues ==
* None
== Progress ==
* Continue on conditional compare.
- Mix fixes for bootstrap.
* Update shrink-wrap patches according to comments and retest them on
Pandaboard and Chromebook.
* Prebuild 2013.05 Linaro toolchain locally.
- gdb related local patches need rework.
== Plan ==
* Continue on conditional compare to bootstrap.
* Linaro toolchain 2013.05 binary release.
Best Regards!
-Zhenqiang
The Linaro Toolchain Working Group is pleased to announce the release
of Linaro GDB 7.6.
Linaro GDB 7.6 2013.05 is the first release in the 7.6 series.
***NOTE*** Linaro GDB 7.6 2013.05 is identical to the FSF GDB 7.6 release,
except for the change in version number and Linaro branding, since all
Linaro GDB features were already accepted upstream and are included in
the FSF release as-is. Future releases in the Linaro GDB 7.6 series may
include additional ARM-focused bug fixes and enhancements.
The source tarball is available at:
https://launchpad.net/gdb-linaro/+milestone/7.6-2013.05
More information on Linaro GDB is available at:
https://launchpad.net/gdb-linaro
--
Matthew Gretton-Dann
Linaro Toolchain Working Group
matthew.gretton-dann(a)linaro.org
== Progress ==
* 3.3 Release
- Bootstrapping, testing, fixing bugs, etc.
- RC1 released on Tuesday
http://people.linaro.org/~rengolin/llvm/
- Fix for C11 atomics on Linux
http://llvm.org/bugs/show_bug.cgi?id=15429
- Fix for zero extend vector bug in test-suite
http://llvm.org/bugs/show_bug.cgi?id=15970
* Test-Suite
- ClamAV fails on Calxeda, same as PPC64 (bad test)
* Self-Host Buildbot
- Re-purposing the Panda back to buildbot service
- Buildbot passing green, setting it up on Build Master
== Issues ==
Calxeda gets turned off quite regularly, which messes up any long term
commitment you might have for them.
== Plan ==
* Install Ubuntu on Chromebook, run benchmarks with 3.3 RC1
* Try to revive LLVM CBuild job if it's any different than our current
buildbot
* Try to setup benchmark jobs for LLVM, either buildbots, CBuild, or
whatever
* Stay alert for RC2 and re-run release process on them
== Progress ==
* Finished porting ld-ifunc tests to AArch64.
* Requested a relocation number for R_AARCH64_IRELATIVE.
* Tested AArch64 ifunc code natively to run generic ifunc tests.
* Investigated gdb bug #1175525 and submitted workaround patch to gdb-patches.
* Started DeveloperTools and LibraryPerformance wiki pages.
* Submitted initial version of AArch64 binutils patch with temporary
reloc number.
== Issues ==
* None.
== Plan ==
* Complete process of getting R_AARCH64_IRELATIVE allocated.
* Feedback on binutils AArch64 ifunc patch.
* Feedback on slightly hacky gdb patch.
* bionic memcpy.
--
Will Newton
Toolchain Working Group, Linaro
== Progress ==
* Disable-peeling: analyzed regression on one bench.
* Libsanitizer/aarch64:
- porting will require more effort than aarch32.
- some syscalls are not supported by aarch64
- libsanitizer expects frame to grow downwards (true on aarch32,
false on aarch64)
* PGO/LTO bug reported by Doko: updating board environment to reproduce it
* Internal support
== Next ==
* Disable-peeling: see how to disable too aggressive vectorization
* Revert-coalesce-vars: discuss how to handle it with the team
* Neon intrinsics: continue improving crc with vuzp
* Neon intrinsics tests: handle feedback if any
* PGO/LTO bug: reproduce it
Progress:
* VIRT-49, VIRT-50 [cp15 migration, reset]
** last bits of patch cleanup complete; realised I could test
KVM migration without any timer or vgic patches; did so and
sent first version of patches out to qemu-devel
* VIRT-55:
** started to draft basic notes on what we want to test:
https://wiki.linaro.org/PeterMaydell/MigrationTesting
* misc
** reviewed Huawei's aarch64 tcg target patches; there are some
issues they need to fix but overall looking good
** fixed a last-minute bug in the versatilepb PCI controller
model (some changes we made earlier in the release cycle
wouldn't work with newer Linux kernels)
Plans:
* deal with any code review feedback on patchset
* set up a raring/aarch64 cross build environment (got partway
through this, waiting for wookey to update some packages)
* more in-depth review/test of John Rigby's mach-virt, aarch64 patchsets
* VIRT-55: write up how to test migration (and talk to Andre
about what he's doing testing-wise)
-- PMM
Hi, All
The attachment file is analysed by the streamline tool.
How to do you think about the vsub.f32 hot spot there?
Is there any way we can improve that?
--
Thanks,
Yongqin Liu
---------------------------------------------------------------
#mailing list
linaro-android(a)lists.linaro.org <linaro-dev(a)lists.linaro.org>
http://lists.linaro.org/mailman/listinfo/linaro-android
linaro-validation(a)lists.linaro.org <linaro-dev(a)lists.linaro.org>
http://lists.linaro.org/pipermail/linaro-validation
== Progress ==
* Filled up the googledoc comparison sheets of gdb 7.6 test suite run in
different configuration on chromebook and x86.
* Filled up the googledoc comparison sheets of gdb 7.5.1 test suite run in
different configuration on chromebook and x86.
* Compared GDB 7.6 and 7.5.1 results which show good improvements in gdb
7.6.
== Plan ==
* Repair my laptop's ubuntu installation which crashed for some reason.
* Submit patch to fix dwarf test suite in gdb 7.6
* Investigate gdb 7.6 gdb.base failures on arm and try to figure out causes
+ possible solution.
* Try to figure out a way automate a comparison of two test suite results
and upload them to googledocs.
Hi,
I downloaded 13.04 version and tried to compile linaro-image-sdk and
linaro-image-lamp. Both these builds fail with the following error.
However, I could compile core-image-minimum. Can somebody help me with this
issue? Is there any quick workaround. Mostly, I do not need ltp package.
Thanks,
Aparna
aarch64-oe-linux-gcc -Wl,-O1 -Wl,--hash-style=gnu -Wl,--as-needed
--sysroot=/home/kvs/aparna/openembedded/build/tmp-eglibc/sysroots/genericarmv8
-O2 -pipe -g -feliminate-unused-debug-types -g -O2 -fno-strict-aliasing
-pipe -Wall -DTST_USE_NEWER64_SYSCALL=1
-I/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases/kernel/include
-I/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases/kernel/syscalls/getdents
-I/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases/kernel/syscalls/getdents/../utils
-I../../../../include -I../../../../include -D_FILE_OFFSET_BITS=64
-DOFF_T=__off64_t -c -o getdents04_64.o getdents04.c
| In file included from getdents01.c:57:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| In file included from getdents02.c:53:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| In file included from getdents03.c:56:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| In file included from getdents04.c:56:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| make[4]: *** [getdents01] Error 1
| make[4]: *** Waiting for unfinished jobs....
| make[4]: *** [getdents02] Error 1
| In file included from getdents03.c:56:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| make[4]: *** [getdents03] Error 1
| In file included from getdents02.c:53:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| make[4]: *** [getdents04] Error 1
| make[4]: *** [getdents03_64.o] Error 1
| In file included from getdents04.c:56:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| In file included from getdents01.c:57:0:
| getdents.h: In function 'getdents':
| getdents.h:60:16: error: 'SYS_getdents' undeclared (first use in this
function)
| getdents.h:60:16: note: each undeclared identifier is reported only once
for each function it appears in
| make[4]: *** [getdents02_64.o] Error 1
| make[4]: *** [getdents04_64.o] Error 1
| make[4]: *** [getdents01_64.o] Error 1
| make[4]: Leaving directory
`/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases/kernel/syscalls/getdents'
| make[3]: *** [all] Error 2
| make[3]: Leaving directory
`/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases/kernel/syscalls'
| make[2]: *** [all] Error 2
| make[2]: Leaving directory
`/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases/kernel'
| make[1]: *** [all] Error 2
| make[1]: Leaving directory
`/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/ltp-20120903/testcases'
| make: *** [testcases-all] Error 2
| ERROR: oe_runmake failed
| ERROR: Function failed: do_compile (see
/home/kvs/aparna/openembedded/build/tmp-eglibc/work/aarch64-oe-linux/ltp/20120903-r2/temp/log.do_compile.6470
for further information)
ERROR: Task 349
(/home/kvs/aparna/openembedded/openembedded-core/meta/recipes-extended/ltp/
ltp_20120903.bb, do_compile) failed with exit code '1'
== Progress ==
* AARCH64 - gprof support.
Submitted GCC patches to generate profile call.
Also applied the review comments from Marcus and tested patches for
aarch64-none-elf.
http://gcc.gnu.org/ml/gcc-patches/2013-05/msg00597.html
* Setup a Ubuntu machine at Office.
Ran GCC testsuite with on openembeded/V8 model
ref: http://people.linaro.org/~hrw/oe/vekumar/.
This image has updated gcc and glibc and enables gprof.
During bootstrap GCC build failed due to space getting over.
Need to create a larger image.
* Looking at how to handle builtin_return_address when
-fomit-frame-pointer is enabled.
GCC saves frame pointer for the function alone where
builtin_return_address is used.
Callers of the function may not have set up the frame yet.
== Plan ==
* Continue gprof "-c" option support in binutils
* Continue handling builtin_return_address when -fomit-frame-pointer
is enabled.
* Continue gcc bootstrap and regression suite after creating a larger image.
== Progress ==
* 3.3 Release
- Running tests on Pandas and Calxeda, Ubuntu 12.04 and 12.10
- Chromebook arriving next week
- Calxedas died a couple of times this week for no reason (somebody else
turned it off?)
- Many bootstrap failures on 12.04: check-all, test-suite, etc on all
phases
- Fixing a bug that allows to bootstrap on Ubuntu 12.10
* EuroLLVM 2013
- Published the questionnaire results, wrote a post on the blog
- http://blog.llvm.org/2013/05/eurollvm-2013-paris-france.html
* Extra
- Studying C++11 atomics
== Issues ==
There's no way to tell when someone will turn the Calxeda server off, so
relying on its results or usability is folly. Not to mention that the
designers should have had a bit more care on the interface to power up and
configure nodes.
== Plan ==
* Spend the weekend testing on the Pandaboard with Ubuntu 12.10 and see if
the fix is good, commit
* Install Ubuntu on Chromebook arriving on Monday, run release tests on them
* Hope that Ubuntu 12.10 will be less hemorragic than 12.04 for the LLVM
release...
Short week (3 days)
== Progress ==
* Disable-peeling: got results vs reference, shared with team.
* Revert-coalesce vars: got results vs reference, showing regressions.
* Libsanitizer: committed upstream.
* Neon intrinsics: shared initial proposal of dejagnu-ization of my
existing tests.
* Branch reviews
* Internal support
== Next ==
* Disable-peeling: investigate regression on eon
* Revert-coalesce: discuss how to handle it with team
* Libsanitizer: enable on aarch64
* Neon intrinsics: continue improving crc with vuzp
* Neon intrinsics tests: continue dejagnu-ization
== Progress ==
* Four day week and traditional bank holiday cold slowed progress.
* Tested glibc memcpy patch on big endian and got it committed.
* Submitted some generic IFUNC patches for binutils upstream.
* Further work on AArch64 IFUNC support in binutils.
== Issues ==
* cbuild seems to not be doing much building at the moment.
== Plan ==
* Complete AArch64 IFUNC support and submit a patch.
* Look further at gdb bug if I get time.
--
Will Newton
Toolchain Working Group, Linaro
[three day week]
Progress:
* VIRT-49 [cp15 migration]
** lots of patch cleanup; nearly ready to submit but ideally
I'd like to test KVM migration with Andre's kernel changes first
** improved arndale setup by switching to USB hard disk; confirmed
I can build and boot my own kernel
* VIRT-50 [cp15 reset]
** it turns out that the VIRT-49 patches on their own break
reset handling for the KVM case, so we need to include VIRT-50
work in the same patchset. Fortunately it turns out to be a small
extension; patch done and tested, and will be submitted as part
of the VIRT-49 patchset
* misc
** discussions about early-printk in a virtual machine. I still think
it's important to be able to have a device tree binding to tell
the kernel where its early-printk uart is. Grant Likely said he
and Nico actually tossed this idea around in the past, it just
never got implemented.
Plans:
* VIRT-49/50: test against Andre's kernel timer save/load patches,
then submit
* more in-depth review and test of John Rigby's mach-virt and
aarch64 patchsets
-- PMM
