Wednesday 9th Feb 2011

This month's meetings

<<MonthCalendar: execution failed [Argument "anniversary" must be a boolean value, not "workinggroups/toolchain/meetingtemplate"] (see also the log)>>

Attendees

Agenda

  • Stand up call

Minutes

Mounir:

  • Looking at toolchain general status re: technical requirements
  • Chart looks at ~40% or less, seems wrong
  • Asks group to have a look through the blueprints
  • Ulrich: some of this is due to tickets not counting as work items
  • binutils-linaro exists, and there are tickets on it, but no-one is tracking
  • Worked with Robbie from Ubuntu to tidy up some of the pulled in blueprints

Dave:

  • Reviewed libffi patch to see if it is OK, cleaned up
  • Added copyright headers as appropriate
  • Re-worked test code to load test routines as a shared library as it was skewing numbers
  • Built Qt to get some RTL out of it
  • Still wiating for sign-off

Richard:

  • Asked for new IFUNC relocation to be added, have provisional number
  • Have enough useful stuff that he can submit IFUNC upstream
  • Start submitting other parts now, wait for final EABI approval
  • Talking with Ira about vectorisation, how to represent optimisations

Marchin:

  • Backport PPA for Lucid and Mavrick now, need to announce them
  • Has ugly source package to use cross toolchain packages on Debian
  • Binary tarballs are next. Waiting for EGLIBC to build before next patch.
  • Two weeks for now at emdebian sprint, will talk about how packages work
  • Michael's tarballs http://builds.linaro.org

Ulrich:

  • Hardware breakpoint and watchpoint support
  • Got a more recent kernel on his IGEPv2
  • Seemed to be working, but has a few problems
  • ARM architecture has two methods for accessing the debug features - CP14 and memory mapped
  • Kernel supports only CP14, IGEP only has memory-mapped
  • Doesn't report failure correctly

Ken:

  • Looked into the Redhat crash utility and extensions
  • SIAL isn't working, but looks easy
  • SNAP extension creates snapshots of a running system, not working
  • Backtrace is pretty rough, is missing arguments, etc
  • LTTng, looked at user space tracing
  • Userspace depends on liburcu, but rest of userspace hasn't been ported
  • latrace, but relies on the audit API from eglibc
  • Looks pretty neat
  • Michael: get DS5 license for Ken

Revital:

  • Do-loop patch was submitted two weeks ago, has been approved
  • Can't commit at the moment due to license issues
  • Ulrich: lawyers have been re-reading contract and seems the assumption is wrong
  • No one from Linaro can commit at the moment as there's no FSF copyright assignment
  • Will pick up on Monday's call
  • Looking into DENBench more closely
  • Seeing a good improvement on the mp3 player benchmark

Andrew:

  • Working on getting the recently reverted patches properly in
  • Starting moving 4.5 patches over to 4.6

Peter:

  • qemu-linaro release on Tuesday
  • Fixed the swap bug in the MMC model
  • Doing reviews on the upstream mailing list
  • Found many issues with float16/float32 conversions
  • Done a basic estimate of the vexpress model
    • About 2 weeks to have a model without sound or USB
    • Mostly plumbing as it's mostly ARM primecells
  • Looked at ARMARM, it is possible to probe where the debug memory map lives

Michael:

  • Monday was the 2011.02 releases
  • Past that, scripting benchmarks to get historical results

WorkingGroups/ToolChain/Meetings/Archive/2011-02-09 (last modified 2013-08-30 11:48:14)