Re: LOR #155

On Fri, 16 Dec 2005, Anish Mistry wrote:

> I got LOR #155 *I think* when unload acpi_video while testing some of
> jhb's patches.

The backtrace is a bit different from #155 (maybe some code changed,
maybe it's another code path).

That had been private patches not in HEAD?
If it's not in HEAD I'd rather not add it to the LOR page. It looks
like the LOR is because of the trap anyway.

> acpi_video0: detached
> lock order reversal: (sleepable after non-sleepable)
> 1st 0xc104e8c8 mt_zone (UMA zone) _at_ /usr/src/sys/vm/uma_core.c:2452
> 2nd 0xc2424d34 user map (user map) _at_ /usr/src/sys/vm/vm_map.c:2993
> KDB: stack backtrace:
> witness_checkorder(c2424d34,9,c062abe6,bb1,cc9dbaac) at
> witness_checkorder+0x5aa
> _sx_xlock(c2424d34,c062abe6,bb1,1000001,cc9dbaac) at _sx_xlock+0x3c
> vm_map_lookup(cc9dbaac,0,1,cc9dbab0,cc9dbaa0,cc9dbaa4,cc9dba87,cc9dba88)
> at vm_map_lookup+0x24
> vm_fault(c2424cf0,0,1,0,c25ed1a0) at vm_fault+0x63
> trap_pfault(15) at trap_pfault+0x12c
> trap(8,28,28,0,c104e8c0) at trap+0x37a
> calltrap() at calltrap+0x5
> --- trap 0xc, eip = 0xc05a8ec8, esp = 0xcc9dbc08, ebp = 0xcc9dbc1c ---
> uma_zfree_internal(0,1,3) at uma_zfree_internal+0xb8
> malloc_uninit(c078f020,c078ee1c,0,cc9dbc70,c04aebc2) at
> malloc_uninit+0xb3
> linker_file_unload(c2049b00,0,0,c25ed1a0,c25ebadc) at
> linker_file_unload+0x2f8
> kern_kldunload(0,cc9dbd04,2,0,3) at kern_kldunload+0x68
> syscall(3b,3b,3b,5,bfbfe9de) at syscall+0x166
> Xint0x80_syscall() at Xint0x80_syscall+0x1f
> --- syscall (444, FreeBSD ELF32, kldunloadf), eip = 0x280ab48b, esp =
> 0xbfbfe464, ebp = 0xbfbfe8a8 ---

-- 
Bjoern A. Zeeb				bzeeb at Zabbadoz dot NeT