Hi, > On Tue, 2004-04-20 at 18:57, Jan Kara wrote: > > > For 2.4 I think your patch is the best solution. Please push it. For > > 2.6 I've written a patch which implements per-sb lists of dirty dquots > > That's only part of the problem: a million dquots on a single sb will > still show this performance degradation currently. Rotating the list > head after each progress will make sure that even if we are processing > unnecessary dquots, we at least do so only once per sync (which is way > better than 0.5*N^2 times!) If I have dirty dquot list for sb+type (which is what I actually have) then I just always get first entry from the list, sync it, delete it and continue.. This way I will have the same asymptotic performance as with moving on inuse_list (ie. O(N)), no? And I will get some constant bonus for smaller list... Or do I miss something? Honza _______________________________________________ Ext3-users@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/ext3-users
