Michael Tokarev <[email protected]> writes:
> 13.07.2010 15:58, Bjørn Mork wrote:
>> tags 588899 - patch
>> thanks
>>
>> Sorry, please disregard the patch. It merely avoids the warning
>> message, but leaves the real problem: The busy loop.
>>
>> strace shows (very quickly repeating):
>>
>> [pid 8454] select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) = 1 (in [17],
>> left {0, 999992})
>> [pid 8454] poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1, 0) = 1
>> ([{fd=17, revents=POLLIN}])
>> [pid 8454] select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) = 1 (in [17],
>> left {0, 999992})
>> [pid 8454] poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1, 0) = 1
>> ([{fd=17, revents=POLLIN}])
> ...
>
> Bjørn, can you please tell us what's the state of this
> on your machine?
>
> We uploaded qemu-kvm 0.12.5+dfsg-1 to -unstable about 10
> days ago, and that version includes a bugfix you were
> referred to above (that patch), it's now included upstream,
> but there's nothing else relevant to this issue.
Thanks, didn't notice as I am preferring testing at the moment.
I just installed and tested 0.12.5+dfsg-1, and the result was as
expected based on the previous patch testing: The warning is gone, but
the busy loop persists.
A snippet of "strace -ttfp <kvmpid>" running 0.12.5+dfsg-1:
[pid 18808] 09:56:51.005215 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.005285 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.005344 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.005413 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.005472 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.005541 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.005600 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.005670 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.005729 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.005798 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.005857 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.005927 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.005985 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.006062 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.006121 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.006190 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.006249 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
[pid 18808] 09:56:51.006318 poll([{fd=17, events=POLLIN|POLLERR|POLLNVAL}], 1,
0) = 1 ([{fd=17, revents=POLLIN}])
[pid 18808] 09:56:51.006377 select(25, [0 5 10 12 17 22 24], [], [], {1, 0}) =
1 (in [17], left {0, 999992})
> And/or, maybe you can provide some more details about
> your setup - what do you do to trigger the bug? Maybe
> it depends on the audio card in use, maybe it's some
> specific version of alsa...
I have the latest squeeze version (unless there is some other package I
should check?):
alsa-base:
Installed: 1.0.23+dfsg-1
Candidate: 1.0.23+dfsg-1
Version table:
*** 1.0.23+dfsg-1 0
990 http://ftp.no.debian.org squeeze/main Packages
600 http://ftp.no.debian.org sid/main Packages
100 /var/lib/dpkg/status
1.0.17.dfsg-4 0
700 http://ftp.no.debian.org lenny/main Packages
alsa-utils:
Installed: 1.0.23-2
Candidate: 1.0.23-2
Version table:
*** 1.0.23-2 0
990 http://ftp.no.debian.org squeeze/main Packages
600 http://ftp.no.debian.org sid/main Packages
100 /var/lib/dpkg/status
1.0.16-2 0
700 http://ftp.no.debian.org lenny/main Packages
libasound2:
Installed: 1.0.23-1
Candidate: 1.0.23-1
Version table:
*** 1.0.23-1 0
990 http://ftp.no.debian.org squeeze/main Packages
600 http://ftp.no.debian.org sid/main Packages
100 /var/lib/dpkg/status
1.0.16-2 0
700 http://ftp.no.debian.org lenny/main Packages
> The thing is, I can't reproduce this on any of my machines...
That's weird. But it is also the same malc told me...
It makes me suspect that this might be related to the specific Windows
version and/or driver I am running as guest. But I did find some
indications that I'm not alone, e.g. this bug which I see you have found
as well: https://bugs.launchpad.net/qemu/+bug/510612
My KVM command line is
kvm -m 512 -monitor stdio -hda image -vga std \
-net nic,model=e1000 -net vde,sock=/var/run/vde2/tap0.ctl,port=7 \
-usb -localtime \
-soundhw ac97
The guest image is "Microsoft Windows XP Professional Version 2002
Service Pack 3" and the Device Manager reports an "Intel(r) 8201AA AC'97
Audio Controller" using a Microsoft provided driver version 5.1.2535.0
But I do note that the Ubuntu bug reporter has verified the bug with the
es1370 driver as well, which undermines this theory.
> Also, I think I remember you posted this question to one
> of upstream mailinglists (either qemu-devel@ or kvm@),
> but I can't find that thread anymore for some reason --
> was there any resolution?
No, there wasn't any resolution. But I got a lot of help trying to
debug it and a number of suggested solutions in a private discussion
with malc (who is the main upstream alsa/audio developer, I believe).
Unfortunately, none of the proposed solutions did solve the problem. And
I was not able to create an image based on free software which would
trigger the problem either. Which of course makes it difficult to debug
for anyone else, since I am unable to provide the image which triggers
the bug.
> Thank you for your patience!
Well, this is not the kind of bug where I expect someone else to come
running and fix it for me. I have already gotten far more help from
malc than can be expected.
I do not expect anyone to work on a bug without a test case.
Bjørn
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
