Re: kern/54908: kernel semi-lock-up

[Nick Hudson <nick.hudson%gmx.co.uk@localhost>]

On 01/02/2020 07:59, Lars Reichardt wrote:
[snip]

> Most pools allocate from the kmem arena the non exhaustive list of
> exceptions I remember are:
> - the buffer caches do have a special pool_allocator backend
> - execargs
> - uarea

I added pmap_l1tt_cache recently which has it's own pool_allocator

http://src.illumos.org/source/xref/netbsd-src/sys/arch/arm/arm32/pmap.c#6580

Along with this new pool I dropped the page size from 8KB to 4KB.

[snip]
> Martin gave it try with the kmem arena limit raised
> to 256mb but the system didn't boot which puzzels me as it seems to have
> 2g kvm.

I just tried with qemu and it booted with 256MB. I'll try real hardware.

While working on the pmap_l1tt_cache/4KB page size change I noticed bugs
in pmap_growkernel and in the initial kernel VM setup.  Maybe there are more
bugs here.


> The reason for this is 128M for kmem isn't much for 2g physical memory
> eg i386 raises this to about 320mb some others to 256mb.

I'm happy to bump this for _ARM_ARCH_6, i.e. modern arm after fixing the
problem Martin saw with it.

Nick