Hi, On Wed, Mar 19, 2008 at 10:45:15AM +0000, Samuel Thibault wrote: > [EMAIL PROTECTED], le Tue 18 Mar 2008 11:02:43 +0100, a écrit :
> > I don't know how the syncing works, so I can't really tell what the > > problem is. If there are blocking points before the superblock read, > > we need to change that somehow. If the superblock read is the first > > blocking point already, we need to change the scheduling, to make > > sure that the request is handled -- up to the first blocking point > > -- before returning to the requester. > > Yes, that's what I meant actually: the diskfs_sync_everything() > function is able to trigger a lot of thread creations. Err... Sorry, I don't understand this reply; it seems out of context... I guess we are talking past each other :-( > A way to have things work correctly would be by marking threads with a > "level", i.e. diskfs_sync_everything runs at level 0, threads that it > generates run at level 1, and threads that process their page faults > run at level 2, etc. Then we just need to limit the number of threads > for a given level. That should always permit termination of requests, > while still limiting the number of thread to a constant times the > maximum nesting of page faults. Well, I considered something like that, but I think it would stilll be more of a workaround than a real solution -- but complicated enough that the real solution might actually be easier... -antrik-
