On Tue, 2004-07-27 at 14:03, Gilles Chanteperdrix wrote: > Marc Kleine-Budde wrote: > > > If you want to avoid that, I suppose you should check that nkpod is not > > > NULL in several other places. > > > > Perhaps one should keep in mind that nkpod could be NULL during in the > > proc routined, this should not harm realtime performance. > > Yes, this is what I meant, add some other checks in the read_proc > function.
No. As far as I can tell reading fs/proc, you might be scheduled out between your test and the use of the pointer, letting rmmod do its nasty job, so you would refer to stale memory, which is even worse. -- Philippe.
