On Tue, Mar 31, 2015 at 01:41:04AM +0300, Gleb Smirnoff wrote: > On Tue, Mar 31, 2015 at 12:30:04AM +0200, Hans Petter Selasky wrote: > H> On 03/30/15 23:19, Gleb Smirnoff wrote: > H> > On Sat, Mar 28, 2015 at 03:02:28PM -0700, Manfred Antar wrote: > H> > M> I get the following panic on current svn ver r280793: > H> > M> > H> > M> Sat Mar 28 14:41:28 PDT 2015 > H> > M> > H> > M> FreeBSD/amd64 (pozo.com) (ttyu0) > H> > M> > H> > M> panic: Invalid CPU in callout 16 > H> > > H> > The same happened to me in the OFED code. After investigation > H> > it appeared that for some unknown reason, the OFED code used > H> > /usr/include/sys/callout.h instead of SYSDIR/sys/callout.h, > H> > that yield in wrong value passing as parameter. > H> > > H> > I failed to reproduce the problem. :( So, the fix is a rebuild > H> > of kernel. But the right fix is to understand what went wrong > H> > in the previous build. > H> > H> How did you compile the OFED stuff? Did you set the SYSDIR variable when > H> building? > > I just have this in my kernel config: > > options OFED > device mlxen Quick grep of the sys/ofed immediately shows the following: sys/ofed/include/linux/wait.h:#include <sys/param.h> sys/ofed/include/linux/wait.h:#include <sys/systm.h> sys/ofed/include/linux/wait.h:#include <sys/sleepqueue.h> sys/ofed/include/linux/wait.h:#include <sys/kernel.h> sys/ofed/include/linux/wait.h:#include <sys/proc.h> ys/ofed/include/linux/timer.h:#include <sys/callout.h> sys/ofed/include/linux/types.h:#include <sys/cdefs.h> sys/ofed/include/linux/types.h:#include <sys/types.h> sys/ofed/include/linux/types.h:#include <sys/param.h> sys/ofed/include/linux/types.h:#include <sys/systm.h> and so on.Received on Mon Mar 30 2015 - 21:51:53 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:56 UTC