FreeBSD 10-STABLE/sparc64 panic

[Nathaniel W Filardo]

I am experiencing this, too, after a recent fresh install of stable/10 on my
V240 (2x1.53GHz).  When /etc/rc.d/netif kicks in, bge0 announces its
transition to DOWN, but all I get is the "b" of the next line, which is, I
think, bge0 announcing its transition to UP.

Curiously, setting net.link.log_link_state_change=0 seems to fix that
particular bit of sadness sometimes, but not always.  When it fails
despite that, I am told:

spin lock 0xc0742eb0 (smp rendezvous) held by 0xfffff8000827b240 (tid
100329) too long
timeout stopping cpus
panic: spin lock held too long
cpuid = 1
KDB: stack backtrace:
panic() at panic+0x1d4
_mtx_lock_spin_failed() at _mtx_lock_spin_failed+0x50
_mtx_lock_spin_cookie() at _mtx_lock_spin_cookie+0xb8
tick_get_timecount_mp() at tick_get_timecount_mp+0xdc
binuptime() at binuptime+0x3c
timercb() at timercb+0x6c
tick_intr() at tick_intr+0x220
-- interrupt level=0xe pil=0x4 %o7=0xc02e19d4 --
smp_rendezvous_action() at smp_rendezvous_action+0x260
-- interrupt level=0x4 pil=0 %o7=0xc02bdbe0 --
sched_idletd() at sched_idletd+0x3a8
fork_exit() at fork_exit+0x80
fork_trampoline() at fork_trampoline+0x8
Uptime: 1m12s

Once, the system, with net.link.log_link_state_change=0 in sysctl.conf
managed to boot as far as bringing up pf and pflog before failing with a
suspiciously similar panic:

spin lock 0xc0742eb0 (smp rendezvous) held by 0xfffff8001904c000 (tid
100608) too long
timeout stopping cpus
panic: spin lock held too long
cpuid = 1
KDB: stack backtrace:
panic() at panic+0x1d4
_mtx_lock_spin_failed() at _mtx_lock_spin_failed+0x50
_mtx_lock_spin_cookie() at _mtx_lock_spin_cookie+0xb8
tick_get_timecount_mp() at tick_get_timecount_mp+0xdc
binuptime() at binuptime+0x3c
timercb() at timercb+0x6c
tick_intr() at tick_intr+0x220
-- interrupt level=0xe pil=0x4 %o7=0xc02e19d4 --
smp_rendezvous_action() at smp_rendezvous_action+0xec
-- interrupt level=0x4 pil=0 %o7=0xc02bdbe0 --
sched_idletd() at sched_idletd+0x3b8
fork_exit() at fork_exit+0x80
fork_trampoline() at fork_trampoline+0x8
Uptime: 2m40s

Manually bringing up bge0 (with net.link.log_link_state_change=0) at least
got me a system where I could fetch the 10-RELEASE kernel.txz file.  This
bisection is going to be very painful, but here goes nothing.

I don't know if this is informative or useless data.  Hopefully the former.

--nwf;
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-sparc64/attachments/20140518/d6d1ce4f/attachment.sig>