== Issues ==
* None
== Progress ==
* Conditional compare (CCMP)
- Design tens of test cases to verify the codes.
- Handle CCMP in vrp pass. If the compare in CCMP can be simplified,
CCMP will be converted to a copy, compare or a bit operator.
- Investigating to handle CCMP in jump-thread related optimization.
- Investigating bootstrap issue when expanding conditional compare
to optimized RTL.
. It seams that there are illegal combinations of different
compares since bootstrap OK if the two compares of CCMP have the same
CODE.
. Current check bases on function "arm_select_dominance_cc_mode".
Need narrow down the root cause.
== Plan ==
* Continue on conditional compare.
== Panned leaves ==
* Sept. 3-6: internal meeting.
== Progress ==
* spec2k comparison between ARM and x86
- Initiated discussion and addressing the feedback
- Looking at register allocation (IRA + Reload) for reduced test
case and comparing with x86
- Analysed few more spec2k benchmarks
* Back-porting
- All the back ports are checked and reviewed. Thanks Yvan.
== Plan ==
* Continue with spec2k comparison between ARM and x86
* Wiki Cleanup
== Progress ==
* More work on Jenkins/Cbuildv2 integration.
* Finished reviewing all 81 wiki pages I had to review.
* Added a 'dryrun' option to Cbuildv2 that displays all configure
and make command lines without executing them as a learning aid.
* Added 'make check' to toolchain builds via Jenkins/Cbuildv2.
* Added a simple "Hello World" test case in C++ to toolchain builds
in Cbuildv2 to make sure the fresh build minimally works.
* Added several more sections to the Cbuildv2 Usage wiki page.
* Cbuildv2 can now downloads and builds several projects/ (QT, GTK,
etc...) as a toolchain test.
* Got the LAVA team to make many tweaks to Jenkins and other stuff
to support TCWG needs for Cbuildv2. Now Jenkins does a toolchain
build of all supported architectures for each commit to GCC
upstream.
* Two small wildland fires this week, plus 3 motorcycle accidents.
== Plan ==
* More work on binary releases via Cbuildv2.
* Import test results into Jenkins.
* Continued work on Jenkins/Cbuildv2 integration.
== Issues ==
* Libmudflap, needs -ldl, or make check fails when linking with
--as-needed (used on Ubuntu by default).
* On chromebook using a raring chroot, 'make check' for mpfr has
these three test cases: tasinh, tcot, and tcsc which consume all the
memory and hang when run with the other infrastructure libraries via
Cbuildv2. The ttan and tsinh tests segfault, and tgamma return an
incorrect result. These same test cases all pass if I use Cbuildv2
to only build and test mpfr. Weird...
== Progress ==
* GDB Record and Replay (TCWG-197)
Wrote instruction type decoder code for ARM Thumb32 instructions.
Wrote stub handlers for different type of Thumb32 instructions.
Wrote process record handlers for various type of load/store instruction.
Wrote a testing utility to test record and replay instruction decoding code.
* Visa Application
Received email notifcation for UK visa issuance for upcoming trip to Cambridge.
Filled out US visa application form DS-160, gathering information
required to complete my application.
== Plan ==
* GDB Record and Replay
Continue to write code for decoding and recording thumb32 instructions.
Try to find a way to generate test data of various instructions to
test record and replay code.
* Travel and Visas
Go to UK visa application center to collect my passport back.
Apply for US visa and make an appointment for the interview for the
week of 23rd September.
Prepare for UK travel for the week of 14th September.
== Progress ==
* Investigating Cross-Compilation with LLVM
- Gathering all options necessary to cross-compile
- Cross-compiling Clang+LLVM with Clang to ARM
- Writing down all issues found during the process
* Investigating the Clang Driver
- Refactoring some common decisions about targets
- Adding new CPUs, cleaning a few areas
== Plan ==
* Try to upstream some patches to make cross-compilation a bit easier
* Write a document on how to cross-compile with Clang on LLVM website
== Progress ==
* 4 day week (Bank Holiday Monday)
* Wiki cleanup pretty much done
* More malloc implementation work
* Further investigation of ptmalloc to see if it can be improved
* Fixed gdb build for non-Intel architectures
* Found an issue with clone in glibc 2.18 and committed a fix
* Committed strlen code to glibc
== Issues ==
* None
== Plan ==
* Get glibc patches applied upstream
* More malloc
* Look into stack guard and pointer guard stuff for ARM
--
Will Newton
Toolchain Working Group, Linaro
Hello,
Is there any LLVM cross-compiler built for ARM. If not, then could you
provide some pointers where I could start creating one?
--
Regards,
Pavan
Hi all,
There has been interest from LEG members to ensure that optimal library
routines are used on their platforms. My understanding is that the
"correct" way of doing this these days is to use ifuncs to select the
best implementation for a given system.
I see that glibc 2.18 contains an ifunc-ed version of memcpy. Does the
TCWG have a hit list of other functions that might get the same
treatment? If so, does it have a plan and the resources to implement
them? If it's a matter of resources, I think LEG might be able to help
there.
Cheers,
mwh
Hi,
Apologies in advance for any chinese whispers effects that happen, but
colleagues at Canonical are attempting to backport this change:
https://sourceware.org/git/?p=glibc.git;a=commit;h=ae65139d140ac85808c0666c…
to the (e)glibc in current versions of Ubuntu, 2.17, but are
encountering mysterious segfaults when building. These are the
additional patches being used:
arm/define-sfi_breg-and-PC_OFS.patch
arm/Add-Cortex-A15-optimized-NEON-and-VFP-memcpy-rou.patch
arm/Make-multiarch-memcpy-always-use-NEON-when-compi.patch
arm/Clean-up-__libc_ifunc_impl_list.patch
arm/Make-armv7-memcpy-implementations-SFI-friendly.patch
arm/Use-push-pop-mnemonics.patch
arm/Support-avoiding-pc-as-destination-register.patch
arm/BX_ALIGN_LOG2.patch
arm/Pass-dl_hwcap-to-IFUNC-resolver-functions.patch
-- all grabbed from glibc git. Are we missing something obvious? :)
Cheers,
mwh
