FreeBSD 10-STABLE/sparc64 panic
Nathaniel W Filardo
nwf at cs.jhu.edu
Mon May 19 14:52:36 UTC 2014
On Sun, May 18, 2014 at 07:58:53PM -0400, Nathaniel W Filardo wrote:
> On Sun, May 18, 2014 at 01:06:41PM -0400, Chris Ross wrote:
> > Nathaniel, I assume you saw Eric's email saying he'd narrowed it down to
> > r263478? If that's accurate, it links it to four revisions from head. So, that
> > is a big head start on the bisection... I'd think r262763 is the obvious candidate
> > of the four mentioned, as it's the only big one. At that point we'll need someone
> > pretty familiar with the networking parts of the kernel and sparc64, and I'm
> > certain I can't fill that role.
> Indeed. After staring at the changes for a while and getting nowhere, on a
> whim, I removed everything of substance in that commit in a new build of the
> head of stable/10 branch (that is, I nuked all the counter_* calls and
> renamed the counter entry so that I was sure I didn't miss any) and it still
> panic()d in the same way. Nothing else looks like it could possibly
> manipulate spinlocks, so just to be sure I'm re-bisecting. If this
> bisection yields the same result, I'll probably try diffing the generated
> assembler, but I'm not sure what I expect to see from that.
I am having trouble bisecting, and it's slow going, but I should report that
a clean build of dbd9b17 (r262853), which is well before 03fdc293 (r263478)
-- also exhibits this problem. I believe 04e37d68 also problematic, just in
case anyone else is out there looking, but I may have flubbed testing that
one. Will report back soon.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 181 bytes
Desc: not available
More information about the freebsd-sparc64