Comment # 8 on bug 965505 from Egbert Eich

(In reply to Victor Victor from comment #6)
> Created attachment 664759 [details]
> X log

Thanks! PCI id of hw is 1002:683d.

> While having video open in SMplayer

Any special option to set? The stall happened in some text rendering function -
did you have subtitles turned on or anything like this?

(In reply to Victor Victor from comment #7)
> I don't really know, I'm not technical enough to understand what's going on.
> 
> 1. I wasn't doing anything GPU specific, and I didn't notice any
> disturbances other than not being sure if I finally had gotten rid of
> tearing in full screen videos.

The  message occurs when you move the mouse a lot while the Xserver is stalling
on some GPU rendering. Did you notice the screen becoming 'frozen' and moved
the mouse around trying to get it 'lose'?
Do you ever recall the mouse feeling 'frozen'? Ie the pointer on the screen not
folliwng your movements?

> 2. I can't remember, maybe I had SMplayer open. If I play a video with
> SMplayer, the error appears in the log.

Ok, so you are not sure if this was video related or happened at another time?

> 3. I have tried uploading the log. I get a billion other errors in the log
> when playing a video in SMplayer also. They are probably related to an X
> config file with Glamor I have added.

Not exactly billons but a few. This condition started 10 minutes after the
desktop started, occurred right after somethig queried/set a video mode and
lasted for roughly 6 seconds.
Does this ring a bell?

> So yes, not really useful, as I don't know what the error is about.
> 
> I don't know about module versions; I have tried modinfo but I get "modinfo:
> ERROR: Module evdev_drv.so not found."

No, you need to do an
rpm -qf /usr/lib64/dri/radeonsi_dri.so
on all files I've not listed in braces.
If the condition happens again, I'd be very interested to get the Xserver log
file and the output of dmesg.

>From looking at the log I doubt the GPU was wedged. The beginning of the stack
trace was identical but when the SIGIO happened the Xserver was in a different
function in radeonsi.