Luis Henriques <lhenriques@xxxxxxx> wrote: > > Note that my fscache-iter branch[1] improves the situation where the disk > > I/O required to effect the destruction of a cache object delays the > > completion of relinquishment by inserting waits, but that oughtn't to help > > here. > > Right, I haven't looked at it yet (I'll try to) but that could make things > even worse, right? It forces fscache_acquire_cookie() to wait for cookie relinquishment to complete sufficiently; however, that doesn't seem to be the problem here since the cookie hasn't been relinquished yet. David
