Re: "per-process" limits (was: Showstopper list)

Paul Barton-Davis (pbd@Op.Net)
Fri, 16 Oct 1998 20:44:28 -0300

In message <19981016184116.A2784@castle.nmd.msu.ru>you write:
>It was me who hacked Alan's beancounter code.
>I've finished the generic interface for beancounter management
>and mostly implemented the accounting for virtual memory
>and unpageable kernel memory occupied by groups of tasks.
>As unpageable kernel memory the code currently
>accounts task+stack structures, mlock()ed memory and page
>directories. Socket buffer space and other accounting aren't
>implemented yet.
>
>I plan to finish the patch when I'll have more free time (probably
>at the beginning of November).

I hope to release a version of my task group patches before then. I
think it would make some sense to consider how these things might fit
together. Right now, my code is structured mostly to support
task-group-specific scheduling policies (and downloadable
schedulers). I think it could be used fairly easily to support
accounting groups as well. I think the biggest concern would be
running out space in the 2 pages allocated for the task_struct by
adding new pointers (sched_group, acct_group, etc.), but my sense at
the moment is that there is quite a bit of space left there.

--p

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