Do ARM Cortex-A9 PMU supoort oprofile
f-turgis at ti.com
Thu Mar 15 18:12:03 UTC 2012
> As far as I know this has not happened. Such a change would also have
> to include making the use of timer interrupts (as opposed to PMU counter
> interrupts) selectable at runtime. I believe the use of timer
> interrupts for oprofile data collection is still a configuration/boot
> option. Alternatively we could just decide to deconfigure PMU hardware
> counters by default to make timers accessible.
Missed that mail while in vacation.
OK, you confirm it did not happen. I read usage of PMU mostly to wake-up every X cycles so timer by default + tunable timer would be OK until we go to r3p0 revisions where PMU bug is solved. Or, if people are OK to lose some interrupts, please speak out, that would be Ok if losses are not too bursty ;-)
However, having the capability to read PMU counters everytime timer triggers would be important to track things like cache misses. I have seen this is adressed in parallel thread.
Texas Instruments France SA, 821 Avenue Jack Kilby, 06270 Villeneuve Loubet. 036 420 040 R.C.S Antibes. Capital de EUR 753.920
More information about the linaro-dev