UFS_DIRHASH stable?

[Simon Burge <simonb%NetBSD.org@localhost>]

Hi folks,

Is UFS_DIRHASH stable these days?

On an eight core amd64 I've been doing some stability tests with
the WAPBL journalling patches, right now to test if there's no
regressions in the non-logging cases.  With a non-UFS_DIRHASH kernel
and a non-logged filesystem, I was able to do three runs each of a
-j{2,4,8,16,32} build.sh without problems.  With a UFS_DIRHASH kernel
and a non-logged filesystem it completed the three -j{2,4} runs but
wedged on the first -j8 run.  Backtrace included below.

So back to the original question - is UFS_DIRHASH considered stable?
I'll re-run the same test, and also try a vanilla kernel (no WAPBL
patch) with UFS_DIRHASH too to gather more data.

Cheer,
Simon.

db{0}> bt  
breakpoint() at netbsd:breakpoint+0x5
comintr() at netbsd:comintr+0x51a  
Xintr_ioapic_edge4() at netbsd:Xintr_ioapic_edge4+0xef
--- interrupt ---
x86_pause() at netbsd:x86_pause+0x2
intr_biglock_wrapper() at netbsd:intr_biglock_wrapper+0x16
Xintr_ioapic_level10() at netbsd:Xintr_ioapic_level10+0xf7
--- interrupt ---
_atomic_add_ptr() at netbsd:_atomic_add_ptr+0x4
uvm_km_free() at netbsd:uvm_km_free+0xe4
execve1() at netbsd:execve1+0xc8b
syscall() at netbsd:syscall+0xa6
db{0}> mac cpu 1
using CPU 1
db{0}> bt  
mutex_vector_enter() at netbsd:mutex_vector_enter+0xd8
uvm_anfree() at netbsd:uvm_anfree+0x18c
amap_wipeout() at netbsd:amap_wipeout+0x116
uvm_unmap_detach() at netbsd:uvm_unmap_detach+0x4e
uvmspace_free() at netbsd:uvmspace_free+0x106 
exit1() at netbsd:exit1+0x243  
sys_exit() at netbsd:sys_exit+0x65
syscall() at netbsd:syscall+0xa6
db{0}> mac cpu 2
using CPU 2
db{0}> bt
_kernel_lock() at netbsd:_kernel_lock+0xb9 
do_sys_wait() at netbsd:do_sys_wait+0x40
sys_wait4() at netbsd:sys_wait4+0x50
syscall() at netbsd:syscall+0xa6
db{0}> mac cpu 3
using CPU 3 
db{0}> bt  
x86_pause() at netbsd:x86_pause+0x2  
uvm_fault_internal() at netbsd:uvm_fault_internal+0x18c9
trap() at netbsd:trap+0x7bb
uvm_fault(0xffff800068868470, 0x7f7ffda58000, 1) -> e
kernel: page fault trap, code=0
Faulted in DDB; continuing...
db{0}> mac cpu 4
using CPU 4
db{0}> bt
x86_mwait() at netbsd:x86_mwait+0xd
idle_loop() at netbsd:idle_loop+0x167
Bad frame pointer: 0xffff800064a14740
db{0}> mac cpu 5
using CPU 5
db{0}> bt  
mutex_vector_enter() at netbsd:mutex_vector_enter+0xe5
uvm_fault_internal() at netbsd:uvm_fault_internal+0x18c9
trap() at netbsd:trap+0x7bb
uvm_fault(0xffff800068868470, 0x7f7ff2f8a000, 1) -> e
kernel: page fault trap, code=0
Faulted in DDB; continuing...  
db{0}> mac cpu 6
using CPU 6
db{0}> bt
x86_pause() at netbsd:x86_pause+0x2
uvm_fault_internal() at netbsd:uvm_fault_internal+0x18c9
trap() at netbsd:trap+0x7bb
uvm_fault(0xffff800068868470, 0x7f7ffdff7000, 1) -> e
kernel: page fault trap, code=0
Faulted in DDB; continuing...
db{0}> mac cpu 7
using CPU 7 
db{0}> bt  
atomic_cas_ptr() at netbsd:atomic_cas_ptr+0x8
pmap_sync_pv() at netbsd:pmap_sync_pv+0x275
pmap_clear_attrs() at netbsd:pmap_clear_attrs+0xbd
genfs_getpages() at netbsd:genfs_getpages+0xa2b
VOP_GETPAGES() at netbsd:VOP_GETPAGES+0x7e
ufs_balloc_range() at netbsd:ufs_balloc_range+0xed
ffs_write() at netbsd:ffs_write+0x644
VOP_WRITE() at netbsd:VOP_WRITE+0x60
vn_write() at netbsd:vn_write+0xce
dofilewrite() at netbsd:dofilewrite+0x81
sys_write() at netbsd:sys_write+0x72 
syscall() at netbsd:syscall+0xa6