LOCKDEBUG panic: Mutex error: mutex_vector_enter: spinout

[Nicolas Joly <njoly%pasteur.fr@localhost>]

Hi,

I do have a big testcase that systematically triggers a LOCKDEBUG
panic on my amd64 machine. It uses 8 instances of NCBI blast search
tool (running under compat linux) at the same time ... When the 1st
process ends the machine panic with the following trace :

Mutex error: mutex_vector_enter: spinout

lock address : 0xffffffff80e85c08 type     :               spin
initialized  : 0xffffffff80724266
shared holds :                  0 exclusive:                  1
shares wanted:                  0 exclusive:                  2
current cpu  :                  1 last held:                  7
current lwp  : 0xfffffe81b4dc1800 last held: 0xfffffe81eabd0800
last locked* : 0xffffffff80723430 unlocked : 0xffffffff80722dd7
owner field  : 0x0000000000010600 wait/spin:                0/1

panic: LOCKDEBUG
cpu1: Begin traceback...
printf_nolog() at netbsd:printf_nolog
lockdebug_more() at netbsd:lockdebug_more
mutex_enter() at netbsd:mutex_enter+0x414
vmem_xalloc() at netbsd:vmem_xalloc+0x14e
vmem_alloc() at netbsd:vmem_alloc+0x105
bt_refill() at netbsd:bt_refill+0x4f
vmem_xalloc() at netbsd:vmem_xalloc+0x177
vmem_alloc() at netbsd:vmem_alloc+0x105
pool_page_alloc_meta() at netbsd:pool_page_alloc_meta+0x2c
pool_grow() at netbsd:pool_grow+0x34
pool_get() at netbsd:pool_get+0x82
pool_cache_put_slow() at netbsd:pool_cache_put_slow+0x1c7
pool_cache_put_paddr() at netbsd:pool_cache_put_paddr+0x10f
pmap_free_pvs() at netbsd:pmap_free_pvs+0x2d
pmap_remove() at netbsd:pmap_remove+0x219
uvm_unmap_remove() at netbsd:uvm_unmap_remove+0x256
sys_munmap() at netbsd:sys_munmap+0x82
linux_syscall() at netbsd:linux_syscall+0x85
cpu1: End traceback...

I don't have much time right now to dig into it ...

Thanks.

-- 
Nicolas Joly

Projects and Developments in Bioinformatics
Institut Pasteur, Paris.