Current state of the Linaro toolchain

Registered by Michael Hope on 2010-10-18

This is an open session to discuss the current state of the Linaro toolchain, our outputs, and what could be improved.

Blueprint information

Status:
Not started
Approver:
None
Priority:
Undefined
Drafter:
Yao Qi
Direction:
Needs approval
Assignee:
None
Definition:
New
Series goal:
None
Implementation:
Unknown
Milestone target:
None

Related branches

Sprints

Whiteboard

The next six months:
 * GCC
 * GDB
 * Library tuning
 * OpenOCD
 * Binutils related performance
 * Developer tools

Review of upcoming sessions:
 Future Linaro toolchain areas
Tuesday:
 Toolchain consumption models
 Cross-compilation environment
 Using QEMU for demonstration
 LTTng
 gdbserver
 Current and future GDB plans
Wednesday:
 Eclipse CDT support
 ARM cross-compiler packages
Thursday:
 NEON/vectoriser direction
 Performance inside GCC
 ARM specific library tuning
 Performance outside GCC
Friday:
 OpenOCD
 Hard float

People's main areas
Rotation into other areas
Priorities
Scheduling
Particular blueprints

- Current state of Linaro toolchain
  * 2010.10 release http://lists.linaro.org/pipermail/linaro-toolchain/2010-October/000363.html
  * Linaro GCC:
  * Linaro GDB: Better test suite pass rate on ARM and even on x86, compared with FSF GDB 7.2. (Can we claim on this?)
  * Valgrind: "developer preview" http://lists.linaro.org/pipermail/linaro-dev/2010-September/001001.html
  * ltrace:

- To be improved
  * Linaro GCC
     ** Performance is one of our focus, and can be better. How to keep performance advantage against FSF GCC? How to avoid speed regression?
     ** .....
  * Oprofile: (need kernel support)
  * QEMU: Performance?
  * Actively collect feedback from users. We've done a lot of things, such as bug fix, performance tuning, and new features development, but we don't know how much benefit users get from our work, for example,
       ** When code size optimization implementation is done, can we measure how much code size is reduced on linux kernel and other application?
       ** When neon auto-vect is done, how this enhancement benefit other applications?

- Near term areas:
  * LTO - get it up and running, get it robust
  * PGO - get it up and running

(?)

Work Items