On Wed, Nov 05, 2008 at 08:02:38PM -0500, Bill Sommerfeld wrote:
> I'd report this to scm-migration-dev at opensolaris.org for initial triage.
Okay!
Hello scm-migration-dev folks!
> > Here's the deal:
> >
> > kebe(common/inet)[0]% hg outgoing -v
> > comparing with /home/danmcd/zclones/onnv
> > searching for changes
> >
> > changeset: 8018:8d3d37257d10
> > tag: tip
> > user: Dan McDonald <danmcd at sun.com>
> > date: Wed Nov 05 19:47:25 2008 -0500
> >
> > description:
> > 6762791 race condition found in ipsecah during ipsec-persock test
> >
> > modified:
> > usr/src/uts/common/inet/ip/sadb.c
> >
> > kebe(common/inet)[0]% diff -r . ~/zclones/onnv/usr/src/uts/common/inet |
> > egrep -v "Common subdir|Only in" | grep diff
> > diff -r ./sadb.h /home/danmcd/zclones/onnv/usr/src/uts/common/inet/sadb.h
> > diff -r ./ip/ip_sadb.c
> > /home/danmcd/zclones/onnv/usr/src/uts/common/inet/ip/ip_sadb.c
> > diff -r ./ip/sadb.c
> > /home/danmcd/zclones/onnv/usr/src/uts/common/inet/ip/sadb.c
> > kebe(common/inet)[1]%
> >
> > So definitely hg isn't picking up my changes in sadb.h and ip_sadb.c. As
> > the
> > kids say... WTF? :)
> I have a theory for something you can try.
>
> Before you do any modifications to your workspace, try a "zfs snapshot"
> to preserve the crime scene.
I've done that. It's not NFS-exported, but it's on
kebe.east:/export/home/danmcd/crimescene/.
> once you've done that, try "touch sadb.h" and "touch ip_sadb.c" to
> change the mod times on the file to see if it picks it up.
kebe(common/inet)[0]% touch sadb.h ip/ip_sadb.c
kebe(common/inet)[0]% hg list
modified:
usr/src/uts/common/inet/ip/sadb.c
kebe(common/inet)[0]%
> i believe that hg caches mod times somehow or other in a "dirstate"
> file, but (because of python limitations) only has 1-second granularity
> of timestamps.
Could be, but just touching the files didn't work.
Help?!?
Dan
