It's fixed in 00b0e039656a19ae9b6e013c90d49904f322301e, thanks to Matt. So we're good - as far as I know.
On Tue, Nov 5, 2013 at 11:17 AM, Venkatesh Srinivas <[email protected]>wrote: > On Tue, Nov 5, 2013 at 3:40 AM, John Marino <[email protected]> > wrote: > > On 11/5/2013 08:15, John Marino wrote: > >> On 11/5/2013 04:51, Justin Sherrill wrote: > >>> On Mon, Nov 4, 2013 at 3:19 PM, John Marino <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>> > >>> Are you tracking any must-fixes? > >>> Davshao posted this on http://bugs.dragonflybsd.org/issues/2597 : > >>> > >>> > >>> You wrote in that followup that you weren't able to duplicate it, and > >>> I'd think you'd encounter it a thousand times over with poudriere > >>> builds. Did it happen for you? > >> > >> The blade Poudriere doesn't use UFS, only tmpfs and hammer. > >> > >>> > >>> > >>> Are there others? > >>> Would the existence of others affect the branch decision? > >>> > >>> > >>> There's no pending problems that I'm aware of. > >>> > >> > >> Well, except this one. > >> John > > > > > > The issue is real. > > I just reproduced it two times out of two attempts on i386 + UFS. > > It tried to dump but the machine rebooted before the dump finished. > > This is a deal-breaker for Release in my opinion as well. > > John > > Can you post a copy of objdump -d of the kernel that paniced to the > bug tracker? vn_syncer_add() was either passed a null vp or a vnode w/ > a null mount in this bug. > > -- vs; >
