On Thursday 21 January 2010 19:56:15 Andriy Gapon wrote: > on 21/01/2010 19:52 David Naylor said the following: > > On Wednesday 20 January 2010 19:16:59 Andriy Gapon wrote: > >> Double-fault could indicate stack overflow. > > > > Thanks, I bumped KSTACK_PAGES to 32 (just to be on the safe side) and it > > is working fairly well now. Only crash I have had since was related to > > tmpfs. So far I have successfully build a port with 59 stacked unionfs (1 > > rw, 58 ro). > > Good that you found a workaround, bad that there is no better way to handle > this overflow. Or, is there? What would make it easier is having kern.kstack_pages tunable (and having the appropriate changes to use that variable). I do not know the internals of the kernel but is it possible for unionfs (assuming it runs in its own thread) to specify a stack size big enough?. It might even be possible to extend unionfs to mount multiple directories onto another, at once (thus eliminating the need for multiple mounts). And probably a measurable performance increase? P.S. Twice I have had processes freeze, these were restricted to directories affected by unionfs mounts.
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:00 UTC