The Linaro Toolchain Working Group (TCWG) is pleased to announce the
2017.02 snapshot of both Linaro GCC 5 and Linaro GCC 6 source
packages.
Linaro GCC 6 monthly snapshot[1] is based on FSF GCC 6.3+svn245201 and
includes performance improvements and bug fixes backported from
mainline GCC. This snapshot contents will be part of the 2017.05
stable[2] quarterly release.
This snapshot tarball is available on:
http://snapshots.linaro.org/components/toolchain/gcc-linaro/6.3-2017.02/
Interesting changes in this GCC source package snapshot include:
* Updates to GCC 6.3+svn245201
* Linaro BZ #2562: [PR target/78253] Call weak function instead of
strong when called through pointer
* Backport of [Bugfix] [AArch32] PR target/71270 Fix
neon_valid_immediate for big-endian
* Backport of [Bugfix] [AArch32] PR target/77439 Wrong code for
sibcall with longcall, APCS frame and VFP
* Backport of [Bugfix] [AArch32] PR target/78364 Add proper
restrictions to zero and sign_extract patterns operands
* Backport of [Bugfix] [AArch32] PR target/78694 Avoid invalid RTL
sharing in minipool code
* Backport of [Bugfix] [AArch32] PR target/79145 Fix xordi3 expander
for immediate operands in iWMMXt
* Backport of [Bugfix] [AArch64] PR target/78362 Make sure to only
take REGNO of a register
* Backport of [Bugfix] PR rtl-optimization/79121 Incorrect expansion
of extend plus left shift
* Backport of [AArch32] 1/2 Use generic_extra_costs in all remaining
tuning structs
* Backport of [AArch32] 2/2 Remove old rtx costs
* Backport of [AArch32] arm_neon.h: Add artificial and gnu_inline
* Backport of [AArch32] Improve Cortex-a53 integer scheduler
* Backport of [AArch32] Improve Thumb allocation order
* Backport of [AArch32] Merge negdi2 patterns
* Backport of [AArch64] 1/2 Add bfx attribute
* Backport of [AArch64] 2/2 Add bfx attribute
* Backport of [AArch64] Fix bootstrap on aarch64-*-freebsd
* Backport of [AArch64] Improve SHA1 scheduling
* Backport of [AArch64] Purge leftover occurrences of
aarch64_nopcrelative_literal_loads
* Backport of [AArch64] Split X-reg UBFIZ into W-reg LSL when possible
* Backport of [AArch64] Split X-reg UBFX into W-reg LSR when possible
* Backport of [AArch64] Tweak Cortex-A57 vector cost
* Backport of [Testsuite] [AArch64] PR target/77634 some vectorized
testcases fail with -mcpu=thunderx
* Backport of [Testsuite] [AArch64] PR target/77635 load/store pair
testcases need to use -mcpu=generic
* Backport of [Testsuite] Fix format string in AdvSIMD tests
* Backport of [Testsuite] Require shared effective target for some lto.exp tests
Linaro GCC 5 monthly snapshot[1] is based on FSF GCC 5.4+svn245200 and
includes performance improvements and bug fixes backported from
mainline GCC. This snapshot contents will be part of the next
maintenance release.
This snapshot tarball is available on:
http://snapshots.linaro.org/components/toolchain/gcc-linaro/5.4-2017.02/
Interesting changes in this GCC source package snapshot include:
* Updates to GCC 5.4+svn245200
* Linaro BZ #2785: PR target/66785 Internal compiler error in record_operand_use
* Linaro BZ #2562: [PR target/78253] Call weak function instead of
strong when called through pointer
** 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 bugzilla against GCC product:
http://bugs.linaro.org/enter_bug.cgi?product=GCC
* Interested in commercial support? inquire at "Linaro support":
mailto:support@linaro.org
~ Progress ~
* AArch64 OpenOCD. TCWG-947, [2/10]
After asking Peter G. about OpenOCD + Hikey, he sent me his working
Hikey+Cable to me.
* SVE GDB patches review. TCWG-1040, [4/10]
Review some patches, and suggest Alan how to split the patch to easily
make some progress.
Think about the needed change to GDB target description.
* Kernel awareness debugging patch review. [2/10]
Manage to write a small test case so that we can somehow test the new
feature in user space in gdb testsuite.
Think about how does Linaro move on about this project.
* Schengen visa application. [2/10]
Visa application centre doesn't accept my documents because my documents
copy are tow-side printed. Has to book another appointment and go to
London again .
~ Plan ~
* TCWG-947, Try opencod with Peter G.'s board.
* TCWG-1040, prototype of new GDB/GDBserver target description.
--
Yao Qi
== Progress ==
* Validation
- more work on use of containers
* GCC
- reported a few failures upstream
- added a few missing poly64_t intrinsics (arm, aarch64)
* misc (conf-calls, meetings, emails, ....)
== Next ==
* ABE & Jenkins jobs patches reviews and bug fixes
== This Week ==
* gimple-fe (4/10)
a) Add support for abs_expr
b) Add support for fma_expr
* TCWG-1010 (4/10)
- Continue working on prototype, dealing with unexpected ICE's :(
* Misc (2/10)
- Reading up on gcc's inliner
- Meetings
== Next Week ==
- TCWG-1010, TCWG-1005, gimple-fe
- Benchmarking LTO using spec2xxx-utils on tx1/tk1
=== This Week ===
LLDB ARM/AArch64 hardware breakpoints [TCWG-717] [9/10]
-- Submitted AArch64 hardware breakpoint patch for review.
-- Updates after upstream comments.
-- Added Arm hardware breakpoint support code.
-- Testing and debugging of Arm code on android arm device.
Miscellaneous Activities [1/10]
-- Meetings, Emails etc.
=== Next Week ===
LLDB ARM/AArch64 hardware breakpoints [TCWG-717]
-- Write breakpoint command based test-cases.
-- Write RSP packets based test-cases.
-- Submit patch for review after edits and upgrades.
GDB Kernel Awareness
-- Background study and understand existing patches
== Progress ==
[TCWG-617] Upstreamed a patch to make copy relocations use synthetic
sections. This allows us to remove the output section relative dynamic
relocation which is important for range extension Thunks
Did some design work for range extension Thunks.
Did some refactoring to merge PLT and IFunc PLT at maintainers request
[BUD17]
Researched my forthcoming Connect presentation and wrote about 3/5 of the slides
== Plan ==
Finish Connect presentation early next week as I'll be on holiday the
week before Connect.
Start implementing range extension Thunks.
== Progress ==
* [ARM GlobalISel] Add support for fp arguments [TCWG-1029] [5/10]
- Committed support for single precision hard float and soft-fp ABIs
- Work in progress on supporting double precision
* Misc [5/10]
- Meetings, mailing lists, code reviews
- Reverted / fixed a few patches that broke the buildbots
- Tested and uploaded binaries for release candidate 2 on AArch64
== Plan ==
* More GlobalISel
Hi All,
I am working on log10/qsort benchmarks on ARM64 (ARMv8) processor,
I want to check if we have experience with these benchmarks.
Actually i am looking for a compiler version which gives best results with
these benchmarks and specific compiler optimization (in my case is see O3
gives best numbers) ?
I have tried GCC-4.9 and GCC-6.2 with log10 benchmark and my observations
are:
1) With gcc 4.9 - 140 us
2) With GCC 6.2 - 150 us
My compilation flags are "-O3 -ftree-vectorize -funroll-all-loops --param
max-inline-insns-auto=550 --param case-values-threshold=30
-falign-functions=32 -ftracer"
So it seems like gcc-6.2 is better, am i missing something, should i use
some better compiler flags?
Thanks
-Bharat
~ Progress ~
* AArch64 OpenOCD. [6/10]
Solder and desolder the JTAG joint on HiKey. Tried different JTAG
options, but still can't get OpenOCD working.
* Patches review. [3/10]
** ILP32 GDB. Help cavium to reduce fails from 500 to 70.
GDB patch is OK, but there are still some regressions on ilp32
vs. lp64 on aarch64. Steve is still investigating on them.
** SVE GDB. Remove MAX_REGISTER_SIZE. Look people tend to agree
the way removing MAX_REGISTER_SIZE. Looking forward to Alan's
patches :)
** Kernel awareness debugging. Reviewed the cover letter of IBM's
patches. Want to share the common kernel debugging part between
IBM and Linaro.
* Schengen visa. [1/10] Fill in more in application form. Document
preparation.
~ Plan ~
* Kernel awareness debugging. Continue reviewing IBM's patches, and
figure out how to share the code.
* Go to London for visa application on Friday.
--
Yao Qi
