Re: [PATCH V3] Add the pagefault count into memcg stats

On Tue, Mar 29, 2011 at 6:16 PM, KOSAKI Motohiro
[off-list ref] wrote:

Hi

sorry, I didn't see past discussion of this thread. then, I may be missing
something.

quoted

Two new stats in per-memcg memory.stat which tracks the number of
page faults and number of major page faults.

"pgfault"
"pgmajfault"

They are different from "pgpgin"/"pgpgout" stat which count number of
pages charged/discharged to the cgroup and have no meaning of reading/
writing page to disk.

It is valuable to track the two stats for both measuring application's
performance as well as the efficiency of the kernel page reclaim path.
Counting pagefaults per process is useful, but we also need the aggregated
value since processes are monitored and controlled in cgroup basis in memcg.

Currently, memory cgroup don't restrict number of page fault. And we already have
this feature by CONFIG_CGROUP_PERF if my understanding is correct. Why don't you
use perf cgroup?

In the other words, after your patch, we have four pagefault counter. Do we
really need *four*? Can't we consolidate them?

1. tsk->maj_flt
2. perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ)
3. count_vm_event(PGMAJFAULT);
4. mem_cgroup_count_vm_event(PGMAJFAULT);

The first three are per-process and per-system level counters. What I
did in this patch is to add per-memcg counters for pgfault and
pgmajfault. This purpose is not to do any limiting but monitoring. I
am not sure about the CONFIG_CGROUP_PERF, does it require
CONFIG_PERF_EVENTS?

Thanks

--Ying

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>