On 2019-04-27 15:56:33, Stefan Sperling <[email protected]> wrote:
> On Sat, Apr 27, 2019 at 12:04:51PM +0200, Krystian Lewandowski wrote:
> > I observed similar issues:
> >
> > iwm0: hw rev 0x230, fw ver 22.361476.0, address e4:0e:ee:81:3d:a0
> > iwm0: unhandled firmware response 0x58/0xc10f rx ring 0[0]
> > iwm0: unhandled firmware response 0xc4/0xc10e rx ring 0[0]
> > iwm0: unhandled firmware response 0x5c/0xc10e rx ring 0[0]
> > iwm0: unhandled firmware response 0xce/0xa502028d rx ring 12[85]
> > iwm0: unhandled firmware response 0x42/0xa5020301 rx ring 12[85]
> > iwm0: unhandled firmware response 0x4f/0xa502040e rx ring 12[85]
> > iwm0: unhandled firmware response 0xe4/0x304288 rx ring 1[238]
> > [...]
> >
> > With kern.bufcachepercent=20, so far so good. But didn't test much.
>
> Thanks for confirming that the problem seems to be related to bufcachepercent.
> I will need help from elder wizards to learn what could be done about this.
>
Hello,
Just in case it might be useful, I thought mentioning a similar
issue I had encountered in the past might help.
In my case, I had issues with the em(4) driver disconnecting from
the system while mbuf usage increased, so this may ultimately
prove be a separate issue.
https://marc.info/?l=openbsd-bugs&m=152413838210478&w=2
Nevertheless, I figured mentioning it might possibly be of help,
since I ultimately figured out that the issues were caused by
me setting kern.bufcachepercent=90.
Setting it back to kern.bufcachepercent=20 fixed all my issues.
Through trial and error, I found that I could crank it up to 60%
without issues, which is where it's been since.
Again, my issue may be completely separate from the OPs issue, but
since both of them involved networking and the bufcache, I thought
I might at least mention it just in case it helps to reveal some
more information somehow.
--
Bryan
