[pvrusb2] Looking for kernel oops evidence...
isely at isely.net
Fri Sep 9 19:57:14 CDT 2005
I've had one new report of kernel oops with the pvrusb2 driver. Actually,
it isn't "new"; the person only reported it to me a few days ago though he
has been seeing this problem for a long time apparently.
I'm looking into this now, but just from inspecting the code this looks to
be an impossible condition. The null pointer he's hit can only happen if
the structure in question has been scribbled on or deleted. But (1) the
only place that pointer gets set to null is when the buffer is freed and
that only happens when streaming is being torn down (but his report
implies that this is happening for him in the middle of streaming), and
(2) if it got scribbled on I'd expect to see a garbage value not zero.
I'm also a little puzzled as to why this is the *only* report I've gotten
of a problem like this. It seems like if there is a race or dangling
pointer problem of this nature then by now somebody else should have
gotten burned by it. So I was just wondering - has ANYONE here
experienced a problem similar to the quoted text below?
(The person who reported this is not on the mailing list so I've deleted
identifying info from the message.)
I have been using your driver for two months now and i am very happy with it.
However, i have an issue with it that happens from times to times: after
keeping mplayer running for a long time (basically, all the night), i get a
kernel oops :( Sometimes, mplayer justs exits without crashing the driver.
I have captured the kernel logs, which you can find here:
It happens in pvr2_buffer_set_idle, in the following line:
it seems that sp is null...
Do you have an idea of what can be happening there?
If you need more information, you can ask me!
| Mike Isely | PGP fingerprint
Spammers Die!! | | 03 54 43 4D 75 E5 CC 92
| isely @ pobox (dot) com | 71 16 01 E2 B5 F5 C1 E8
More information about the pvrusb2