LOR g_xdown vs user map
Brian Fundakowski Feldman
green at freebsd.org
Sun Oct 10 00:55:54 PDT 2004
On Sat, Oct 09, 2004 at 04:51:43AM -0400, Ketrien I. Saihr-Kesenchedra wrote:
> Bjoern A. Zeeb wrote:
>
> ><>On Fri, 8 Oct 2004, Ketrien I. Saihr-Kesenchedra wrote:
> >
> >Hi,
> >thanks for reporting. I have added it with LOR ID 041:
> >http://sources.zabbadoz.net/freebsd/lor.html#041
> >
> >I am Cc:ing current@ to let the others know about it
>
> <supersnip>
> Yeah, I'm lazy, so I didn't do that. ;)
> There are other issues with amr(4) on the system currently, and scottl
> is working on them AFAIK. This one is new as of today. System producing
> it is a 4-way Opteron 848 with 8GB, running -CUR sources from about
> 1800 EDT Friday, with Brian Feldman's uma patch applied additionally.
> ADAPTIVE_GIANT was enabled at the time, additionally. (Yes, you can
> shoot me for forgetting that,) but I've been unable to reproduce reliably.
> And what the heck is triggering alltraps_with_regs_pushed()? I've never
> seen that before on any systems.
Is it okay that amr(4) calls bus_dmamap_load(), doesn't check the return
value, then immediately calls bus_dmamap_sync() -- when the tags in use
were not allocated with BUS_DMAMEM_ALLOCNOW?
--
Brian Fundakowski Feldman \'[ FreeBSD ]''''''''''\
<> green at FreeBSD.org \ The Power to Serve! \
Opinions expressed are my own. \,,,,,,,,,,,,,,,,,,,,,,\
More information about the freebsd-current
mailing list