Subject: Re: kern/28869
To: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
From: Jesse Off <joff@embeddedARM.com>
List: netbsd-bugs
Date: 01/13/2005 08:05:29
I thought this was strange, printing each pool in DDB actually prints the
pool right before it:

panic: pool_put: kvakmem: page header missing
Stopped in pid 5.1 (pagedaemon) at      0xc0389230:     mov     r15, r14
db> show pool *pmap_pv_pool
POOL l2dtblpl: size 196, align 4, ioff 0, roflags 0x00000040
        alloc 0xc048314c
        minitems 40, minpages 2, maxpages 4294967295, npages 4
        itemsperpage 20, nitems 47, nout 33, hardlimit 4294967295

        nget 33, nfail 0, nput 0
        npagealloc 4, npagefree 0, hiwat 4, nidle 2
db> show pool *pmap_l2dtable_pool
POOL l2ptppl: size 1024, align 4, ioff 0, roflags 0x00000000
        alloc 0xc048314c
        minitems 16, minpages 4, maxpages 4294967295, npages 15
        itemsperpage 4, nitems 16, nout 44, hardlimit 4294967295

        nget 44, nfail 0, nput 0
        npagealloc 15, npagefree 0, hiwat 15, nidle 4
db> show pool *pmap_pmap_pool
POOL pvepl: size 16, align 4, ioff 0, roflags 0x00000040
        alloc 0xc0483f68
        minitems 512, minpages 3, maxpages 4294967295, npages 18
        itemsperpage 253, nitems 1353, nout 3201, hardlimit 4294967295

        nget 56673, nfail 0, nput 53472
        npagealloc 20, npagefree 2, hiwat 20, nidle 0


>> > I'm not sure which pool it is, I will attempt to find out.
>>
>> Seems like its draining l2dtblpl when it crashes
>>
>> db> show pool *drainpp
>> POOL l2dtblpl: size 196, align 4, ioff 0, roflags 0x00000040
>
> drainpp is a pool which will be drained the next time.
> i think you need to look at the previous pool in the pool_head list.
>
> YAMAMOTO Takashi
>
>