Hmm, there's a stack of patches for bugs related to the p2m / PoD /
ept code fixed in XenServer product that I neglected to send upstream
when I found. Unfortunately the p2m code has changed sufficiently
since 3.4 that they don't apply cleanly, and I've had a lot of trouble
actually getting a usable OSS system that I can test the p2m code on.
If you like, I can post the raw patches to the list, and you can see
if any of your bugs might be fixed among them...
On Mon, Nov 22, 2010 at 12:40 PM, Olaf Hering <olaf@xxxxxxxxx> wrote:
> On Mon, Nov 22, George Dunlap wrote:
>> Dang, 8 megs per cpu -- but I guess that's really not so much overhead
>> on a big machine; and it's definitely worth getting around the lost
>> records issue. Send the T_INFO_PAGES patch to the list, and see what
>> Keir thinks.
> Yes, will do.
> Sometimes there is a boost of records, like set_p2m_entry and the like.
> They all happen on the same cpu, so the per_cpu buffer fills up quickly.
> And thats what I'm debugging right now. Too bad, tracing adds overhead
> so the bug disappears. So, back to reading source code and find the race
> that way.
Xen-devel mailing list