[Bug 234906] kernel panic: softdep_setup_inomapdep: dependency 0xfffff80107d15c00 for newinode already exists

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234906

--- Comment #13 from johan van Zanten <johanvz816 at giantfoo.org> ---
regarding the memory testing:

= memtest86+ -- my laptop can't boot the ISO image on a USB stick ; maybe a
UEFI compat problem.  I installed the memtest86+ package and tried to boot from
the FreeBSD menu (unload ; load /boot/opt/memtest86+).  As soon as i hit
return, i get a black screen, and then the machine immediately reboots without
doing any work.

= memtester -- i ran three instances concurrently to test 6.83 GB of the 8 GB
of RAM (any more RAM and the OS started paging).  After four loops, no errors
were found.

Given that i wasn't have any problems using this same config as a desktop
running FreeBSD 10 or 11, with the same sort of workload, it seems less likely
that RAM is the problem.

Lines 11214 to 11234 from /usr/src/sys/ufs/ffs/ffs_softdep.c:

        /*
         * If we have reached the end of the current list without
         * finding the just finished dependency, then it must be
         * on the future dependency list. Future dependencies cannot
         * be freed until they are moved to the current list.
         */
        if (listadp == NULL) {
#ifdef DEBUG
                if (adp->ad_state & EXTDATA)
                        listhead = &inodedep->id_newextupdt;
                else
                        listhead = &inodedep->id_newinoupdt;
                TAILQ_FOREACH(listadp, listhead, ad_next)
                        /* found our block */
                        if (listadp == adp)
                                break;
                if (listadp == NULL)
                        panic("handle_allocdirect_partdone: lost dep");
#endif /* DEBUG */
                return;
        }

---

That matches the panic sting from Wed Jan  9 23:41:29 CET 2019, but assuming
that's where the system is panicking, one still needs to find out how the
system is arriving in that state.

-- 
You are receiving this mail because:
You are the assignee for the bug.