Re: port-sun2/45931: early kmem panic on sun2

[Lars Heidieker <lars%heidieker.de@localhost>]

On 08/11/2012 06:25 PM, Izumi Tsutsui wrote:
> The following reply was made to PR port-sun2/45931; it has been 
> noted by GNATS.
> 
> From: Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost> To: 
> para%NetBSD.org@localhost 
> Cc: gnats-bugs%NetBSD.org@localhost, tsutsui%ceres.dti.ne.jp@localhost 
> Subject: Re: 
> port-sun2/45931: early kmem panic on sun2 Date: Sun, 12 Aug 2012 
> 01:23:00 +0900
> 
>> NetBSD 5.99.64 (GENERIC) #0: Sun Feb  5 22:46:00 JST 2012 
>> tsutsui@mirage:/s/src/sys/arch/sun2/compile/obj.sun2/GENERIC 
>> Model: sun2 {120,170} fpu: no math support total memory = 7168 KB
>> avail memory = 4912 KB panic: pool_get: kmem-4096: page empty 
>> Stopped in pid 0.1 (system) at  netbsd:cpu_Debugger+0x6: unlk
>> a6
> 
> This still happens on netbsd-6, but HEAD sun2 kernels works.
> 
> Pulling sys/kern/subr_kmem.c rev 1.46: 
> http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/kern/subr_kmem.c#rev1.46
>
>
> 
> 
> Does it also include some implicit bug fixes, not only
> improvements for >4KB allocations? Or "decrease overall memory" is
> mandatory for poor tier II ports?
> 
> 

sun2 has a PAGE_SIZE of 2k which was not properly handled before this
change. A 4k pool was initialized before but did not work, this is
true for all archs with PAGE_SIZE < 4kb.
This is fixed with rev 1.46.
I think rev 1.46 should be pulled to netbsd-6.

Thanks for trying this out.

Lars