Messages in this thread | | | Subject | Re: combinatorial explosion in lockdep | From | Peter Zijlstra <> | Date | Wed, 30 Jul 2008 09:21:07 +0200 |
| |
On Tue, 2008-07-29 at 23:56 -0700, David Miller wrote: > From: David Miller <davem@davemloft.net> > Date: Tue, 29 Jul 2008 21:45:03 -0700 (PDT) > > > If we can agree on this as a fix, it should definitely be backported > > and submitted for -stable :-) > > > > ---------------------------------------- > > > > lockdep: Fix combinatorial explosion in lock subgraph traversal. > > So the next problem I run into is that 8192 is not a large > enough MAX_LOCKDEP_ENTRIES for this larger cpu count boxes. > 8192 isn't even big enough for my 64 cpu box. > > It needs to be scaled by NR_CPUS, or something like that.
Or something like that indeed - if distro kernels are going to be build using NR_CPUS=4096, then these tables might explode - then again, lockdep is only enabled on -debug kernels anyway...
So yeah.
| |