Mailing List Archive: swap map

swap map

Jan 30, 2007, 3:12 AM

Post #1 of 3 (272 views)

In the it-would-be-nice department... While I was twiddling with swappiness
and benchmarking some code, I thought it would be pretty cool if there was a
node in /proc/<pid> that would show which pages of a process are resident or
nonresident. I'm not sure that it would be useful very often, but I was
thinking one could get a snapshot of that, correlated with traces from a
malloc profiler, to show what portions of a program's memory usage was in
active use vs idle.
--
-- Howard Chu
Chief Architect, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc
OpenLDAP Core Team http://www.openldap.org/project/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Re: swap map [ In reply to ]

hch at infradead

Jan 30, 2007, 3:20 AM

Post #2 of 3 (242 views)

Permalink

On Tue, Jan 30, 2007 at 03:12:35AM -0800, Howard Chu wrote:
> In the it-would-be-nice department... While I was twiddling with swappiness
> and benchmarking some code, I thought it would be pretty cool if there was
> a node in /proc/<pid> that would show which pages of a process are resident
> or nonresident. I'm not sure that it would be useful very often, but I was
> thinking one could get a snapshot of that, correlated with traces from a
> malloc profiler, to show what portions of a program's memory usage was in
> active use vs idle.

That's be a remote mincore(). This should be more or less trivial,
do_mincore would need a mm_struct argument instead of always using
current->mm and we'd need a simple seq_file interface iterating over it.

Any volunteers?
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Re: swap map [ In reply to ]

nickpiggin at yahoo

Jan 30, 2007, 3:33 AM

Post #3 of 3 (242 views)

Permalink

Christoph Hellwig wrote:
> On Tue, Jan 30, 2007 at 03:12:35AM -0800, Howard Chu wrote:
>
>>In the it-would-be-nice department... While I was twiddling with swappiness
>>and benchmarking some code, I thought it would be pretty cool if there was
>>a node in /proc/<pid> that would show which pages of a process are resident
>>or nonresident. I'm not sure that it would be useful very often, but I was
>>thinking one could get a snapshot of that, correlated with traces from a
>>malloc profiler, to show what portions of a program's memory usage was in
>>active use vs idle.
>
>
> That's be a remote mincore(). This should be more or less trivial,
> do_mincore would need a mm_struct argument instead of always using
> current->mm and we'd need a simple seq_file interface iterating over it.
>
> Any volunteers?

Well the first thing needed is an mincore that actually works for anonymous
memory!

I've implemented some fixups in that department, which should get merged
into 2.6.21. That reminds me: I'll send the patch to linux-mm for review...

We cannot get real atomic snapshots in general, but I don't think that would
be a problem if you are just doing some profiling.

--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/