Messages in this thread | | | Date | Sat, 31 Jul 1999 13:10:11 -0700 (PDT) | From | Linus Torvalds <> | Subject | Re: [patch] 'multimedia latency patch', lowlatency-2.2.10-M6 |
| |
On Sat, 31 Jul 1999, Ingo Molnar wrote: > On Sat, 31 Jul 1999, Benno Senoner wrote: > > > as Roger Larrson suspected, there are some parts in the kernel > > which have a too long execution path: > > On his PPro with 512MB RAM, d_lookup takes up to 80ms to execute ! > > yes. I've attached a patch that is supposed to make things much better.
No. Check the numbers first, d_lookup() doesn't do anything heavy (it just follows a linked list of entries - even assuming a lot of cache misses that would probably imply more dentries than you can fit in 512MB).
Basically, if d_lookup() takes 80ms, there's a bug somewhere. Don't try to paper it over by adding "reschedule" calls.
The bug may be a completely buggered hash function or similar.
Or the bug may be the latency measurement.
In either case, trying to add band-aid over it is the wrong solution.
Linus
- 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/
| |