Re: kern/48372: sunblade 2500 hangs under memory pressure -- loop between uvm/pool/vmem

[Mindaugas Rasiukevicius <rmind%netbsd.org@localhost>]

Lars Heidieker <lars%heidieker.de@localhost> wrote:
>  Hi,
>  
>  one thing puzzles me about the stack trace, how can uvm_km_kmem_alloc
>  call pool_cache_put_* without going via vmem_free?
>  Is it missing in the stack trace?
>  Anyway I think I found it, there is a bug fixed by the change of
>  revision: 1.125 src/sys/uvm/uvm_km.c
>  It should be pulled up to the netbsd-6 branch.

That is a good spot, indeed!

>  
>  Lars

-- 
Mindaugas