On Fri, 14 Jan 2005 ragge@ludd.luth.se wrote:

>> On Thu, 13 Jan 2005 ragge@ludd.luth.se wrote:
>>
>>> There is a "backdoor" to get into DDB that may work:
>>>
>>> ^P
>>>>>> D/P 80000100 1
>>>>>> C
>>>
>>> If you're lucky you will enter DDB now.
>>
>> Okay, I managed to break into DDB that way.
>>
>> What I found out is that the init process is sleeping on "vmmapva".
>> Looking at the code, in uvm_map.c, it appears that it's waiting for other
>> processes to release resources, which is silly at this point.
>>
> Backtrace?

db> bt
panic: foo
Stack traceback :
0x849f7c18: 
uvm_map_prepare+0x138(0x827d3e00,0x83c2d000,0x40000,0x801af74c,0xfff
fffff,0xffffffff,0,0x1801727,0x849f7cb8)
0x849f7c64: 
uvm_map+0xad(0x827d3e00,0x849f7d20,0x40000,0x801af74c,0xffffffff,0xf
fffffff,0,0x1801727)
0x849f7cd4: 
uvm_km_valloc1+0x6c(0x827d3e00,0x40000,0,0xffffffff,0xffffffff,0)
0x849f7d24: sys_execve+0xe6(0x81ee2000,0x849f7f8c,0x849f7ef8)
0x849f7ea8: start_init+0x257(0x81ee2000)
db>

(Yes, I put in a panic right before the ltsleep)

 	Johnny

Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@update.uu.se           ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol