Thursday 8 Dec 2011

Standup call.

This month's meetings

<< <  2011 / 12 >  >>
Mon Tue Wed Thu Fri Sat Sun
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30 31  



  • Release week -

Action Items from this Meeting

  • TBA

Action Items from Previous Meeting

  • TBA


  • Release Week
    • GCC
    • GCC testsuite looks fine - Looks like the second set of tests failed miserably - We need to see what happened with those tests. Benchmarking tests and ubutests are still running - Take status with benchmarks offline - Andrew will look this afternoon and see if any feedback from MichaelH about ubutests . Release notes to be written up and figure out how to sign tarballs -
      • GDB
      • Uli hasn't committed the remote core-file generation for this release because upstream has another set of discussions on the release. Single-stepping atomic sequences committed - Updated to 7.3.1 release level - has some fixes for other bits. Finished running on x86 and ARM but still in the queue for x86_64.
  • Round the group.
  • Andrew
    • 64 bit Neon stuff - one_cmpl patch approved and looking at the shifts using QImode shift amount being investigated. Release stuff as above.
  • Åsa
    • Not much Linaro work. Working on testing the Linaro toolchain on internal android builds. Running web-benchmarks there. Seeing some improvements with 4.6 toolchain against the standard Android 4.4.3 toolchain and regressions in some cases and trying to narrowing down. Some regressions in the js benchmark. Why are Google still hanging on to the old 4.4.x toolchain ? Worried about breakages ? bero is doing some work in that area -
  • Dave
    • Got a reply on memchr going into eglibc - Tests look ok - working on feedback recieved. Has a fix for one of the qemu issues with mapping of target_code_buffer. Testing still hard because of other issues.
  • Ira
    • Fixed a bug in the vectorizer. Most of the time spent on looking at the alignment issue.
  • Ken
    • Debugging the various errors and warnings that I
      • encountered with the binary CSL 2011.03 toolchain fleshed out the receipt for the external toolchain and now get a working core image mininal that boots and seems to be usable
    • started to build the core-image-sato (X and everything)
    • ran into more errors (currently cmake crosscompiling issue with
      • libproxy)
  • Richard
  • Uli
    • Release week status as above.
  • Ramana
    • Backported vcvt fixed / floating point patch to 4.6 only to quickly revert it. Backported A15 scheduling patches into Linaro 4.6 Looking at alignment hints in the backend along with Ira and what we are doing in these cases. It looks like more work in the areas of the neon move patterns and teaching them some kind of cost information about when to use what addressing modes, changing output_move_neon and a whole raft of issues. Backported 1 of the partial-partial PRE fixes to Linaro 4.6 - need to read it more and understand further.

WorkingGroups/ToolChain/Meetings/Archive/2011-12-08 (last modified 2013-08-30 11:48:31)