I will split the patch up into 6 pieces. I will submit them as a set of
sequential patches (must be applied in order) to perfmon_power5.c. Is
The pseries config does define HIGH_RES_TIMER by default.
I will try out the kernel patch you suggested today.. sorry that I was
not able to get to it earlier.
> Sorry for late reply. I think it would be better to split into several
> patches so that each commit
> will clearly identify the change.
> I ran the hrtimer modifications for > 12hours on my Core 2 and did not
> see the lockup. Now I
> discovered that the cb_mode only exists if you config for
> HIGH_RES_TIMER. Don't know if
> POWER has this enabled. On IA-64, they don't, but then they do not
> have the runqueue locked
> during ctxsw so they are fine.
IBM Linux Technology Center, Linux Toolchain