Hi Eric,
On 2020/3/12 17:59, Auger Eric wrote:
Hi Zenghui,
On 3/12/20 10:19 AM, Zenghui Yu wrote:
On 2020/3/11 22:00, Marc Zyngier wrote:
That is still a problem with the ITS. There is no architectural way
to report an error, even if the error numbers are architected...
One thing we could do though is to implement the stall model (as
described
in 5.3.2). It still doesn't give us the error, but at least the command
queue would stop on detecting an error.
It would be interesting to see the buggy guest's behavior under the
stall mode. I've used the following diff (absolutely *not* a formal
patch, don't handle CREADR.Stalled and CWRITER.Retry at all) to have
a try, and caught another command error in the 'its-trigger' test.
logs/its-trigger.log:
" INT dev_id=2 event_id=20
lib/arm64/gic-v3-its-cmd.c:194: assert failed: false: INT timeout! "
dmesg:
[13297.711958] ------------[ cut here ]------------
[13297.711964] ITS command error encoding 0x10307
It's the last INT test in test_its_trigger() who has triggered this
error, Eric?
Yes it may be the culprit. Anyway I removed the collection unmap in v6.
I forgot to mention that this is based on your v6. I'll reply to it.
By the way are you OK now with v6? I think Drew plans to send a pull
request by the end of this week.
Sorry I haven't looked at it yet (v5 already looks good except for
some minor issues).
Thanks,
Zenghui
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
