Thursday 29th March 2012

This month's meetings

WorkingGroups/ToolChain/Meetings
<< <  2012 / 3 >  >>
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  

Attendees

  • Michael Hope
  • Peter Maydell
  • Ramana Radhakrishnan
  • Rusty Russell
  • Ulrich Weigand
  • Zhenqiang Chen

Agenda

  • Standup call

Minutes

Rusty:

  • Wrapping up before the time away
  • Tidy up patches
  • Rejecting the idea that the guest should check for presence of the PMU
  • Is a Cortex-A15 so supply a PMU
  • After back is Thumb-2 work

Zhenqiang:

  • Binary toolchain release done this week
  • Code size regression investigation
  • Checking against 4.7, 4.6, and CSL
  • 4.7 is better but not as good

Peter

  • Working on the CP15 rework
  • Made some progress
  • Rebasing on top patches that move QEMU CPU into QEMU objects
  • Hope to have a set ready for the end of the week but depend on the other patch set
  • Handful of registers past that

Ulrich:

  • Fixed the crash in SLP that were exposed by patch that does pattern detection using SLP
  • Bug was routine that detects smallest time used in a gimple statement
  • Needs to know widening instructions, didn't know of widening shifts
  • Upstream, backported
  • Preparing a presentation for the Linux Foundation Collaboration Summit
  • On improvements done in GDB on ARM
  • Next will do the vld1 patches
  • Looking into big endian modes

Ramana:

  • Catching up on email etc after last week
  • ivopts changes are approved

Michael:

  • Backported rearnshaw's NEON intrinsic and NEON lex patches
  • Finished KVM blueprints
  • Looked into guest filesystems
  • Packaged boot-wrapper
  • Finished the binaries release

Initialisation of registers:

  • Kernel should init registers
  • Some registers depend on the machine you're emulating
  • and QEMU initialises many of them
  • Peter: QEMU is device side, host kernel is CPU emulation
  • Should initialise the CPU things in the kernel?
  • PowerPC uses set feature set on the kernel
  • x86 is half and half

WorkingGroups/ToolChain/Meetings/Archive/2012-03-29 (last modified 2013-08-30 11:50:22)