Monday 15th August 2011

This month's meetings

<< <  2011 / 8 >  >>
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        



  • Review action items from last meeting
  • Release week
    • lp:~ams-codesourcery/gcc-linaro/merge-from-fsf-20110811-4.5 (Todo)
    • lp:~michaelh1/gcc-linaro/merge-from-fsf-r177703-4.6 (Merged)
    • lp:~ramana/gcc-linaro/fix-lp-823548 (Approved, too late)
    • lp:~ams-codesourcery/gcc-linaro/widening-multiplies-4.6
    • Spin builds - Andrew to upload directly?
      • scp -P 7023 gcc-foo.tar.bz2

      • ~/cbuild/tools/ ~/snapshots/gcc-foo.tar.bz2

  • libav coordination
  • Graphite is now enabled
  • String routines next steps
  • QEMU release
  • Changing to monthly planning

Action Items from this Meeting

  • ACTION: Andrew to contact Martin Pool re: bzr issues
  • ACTION: Michael to do libav instrumentation by the end of the week

Action Items from Previous Meeting

  • DONE: Dave to contact server guys on our side re: getting it into Ubuntu
  • DONE: Andrew to contact Matthias re: kicking the import, then merge


* Dave, Asa, Andrew, Ulrich, Ken, Dave Rusling, Ramana * Matthias, Richard

Dave to contact server guys on our side re: getting it into Ubuntu:

  • Spoke with Mike (NCommander)
  • Will talk with others, will get back, need to chase
  • Matthias:
    • Can integrate if needed
    • membase requires it (reason people asked for this function)

Andrew to contact Matthias re: kicking the import, then merge:

  • GCC import is working now (again)
  • Working for 4.5, 4.6 branches
  • Andrew can't seem to merge
    • ACTION: Contact Martin Pool
  • Give it a run through, have plan B if needed

Release week:

  • Reviewed branches
  • Also mirrors to


  • Real bug, originally reported on ARM
  • Has impact on PowerPC. Fault is in the backend itself.
  • Fix is the same as on x86 and SPU
  • Fault is rare
  • Upstream fix state? Has
  • Presents as an ICE
  • Doesn't present in significant packages
  • How about the fix?
    • Focused, contained
    • Has history because it was the same fix as on others
    • Matthias will run the PowerPC build/test

Aligns labels work:

  • Ramana had suggested it for reducing variation between runs
  • Ran EEMBC with it, didn't compile


  • Who's doing what
    • Revital: microbenchmark from Richard, running with and without SMS
    • Richard: exercise more of the library
      • Was looking for standard performance tests
    • Michael's plan:
      • ACTION: Michael to do by the end of the week
  • Richard has currently picked based on the features such as interleaved load/stores
  • What features does he want to test?
    • SMS/IV
    • Want to run just C and share results with upstream

String routines next steps:

  • Update patch to send to EGLIBC based on feedback
  • Need to look at Newlib, see who uses it
  • Talked with Android guys re: 0xbench and library routines used
    • Are planning to profile, haven't done that yet
    • Don't have perf or much else
    • But they should be able to run it?
  • Why Newlib?
    • Seeds many deeply embedded projects
    • Good place for others to pick up string routines
  • A15 guys are back from holiday, Ramana will talk about running cortex-strings

QEMU release:

  • Who's doing it?
  • Has done testing, seems OK
  • Needs pressing of the button. Expect Peter to.
  • Ran Linaro Android Beagle image on QEMU
    • Not that flash, but this is the first time running
    • Green Android on boot is corrupt. Wrong idea on screen size?
    • Desktop isn't the same as what the Android guys expect
    • Should have first class support
  • Manual test takes about a day, easier this time
  • Adding Android? Perhaps an hour or two

Changing to monthly planning:

  • See the agenda
  • Will discuss next meeting

WorkingGroups/ToolChain/Meetings/Archive/2011-08-15 (last modified 2013-08-30 11:49:20)