Howto for fglrx / catalyst (Ati driver)

[Hlingler]

result is that you get to the login screen, but then it locks up after moving mouse to click on login

Sounds like the same exact results as Ravnos above ?

Looks like some problem between FGLRX/Catalyst-9.8 and mouse/drivers.

V

[Jeff72]

Originally Posted by Hlingler

Sounds like the same exact results as Ravnos above ?

Looks like some problem between FGLRX/Catalyst-9.8 and mouse/drivers.

V

Moving the mouse around itself is fine. Using the mouse to move any window around causes a quick lockup of the display.

I also tried disabling selinux but I still have the same issue.

[HMSS013]

anyone have luck with the ATI 9.8 Binary in F11?!? ...or am i gonna pretty much nuke my box if i give it a whirl?!

[Ravnos]

Just an update, I had no problems without the mouse for hours, even manipulating windows and switching between them, using 3D effects, using firefox, thunderbird, rhythmbox, skype ... Tried moving the mouse for a while and crashes again ...

Tried not loading evdev and loading mouse driver instead but it crashed the same way. No doubt the problem is an interaction between fglrx and the cursor movement.

[TrentZ]

Adding

	Option	    "SWCursor"	"true"

to the "Device" section in xorg.conf fixes the freeze problem, but as a side effect corruption appears around the cursor if compiz is enabled.

[Hlingler]

Try:

Option	    "SWCursor"	"true"
Option	    "HWCursor"	"off"

Re-start of X11 required to apply.

V

And note:

Originally Posted by Hlingler

There are of course a number of xorg.conf options that you might try - check the ATI FGLRX/Catalyst docs (if available).

[TrentZ]

Option	    "HWCursor"	"off"

That is completely redundant if SWCursor is present.

[Hlingler]

Not true, so far as I have always understood. In any case: is it not worth a try?

V

[TrentZ]

Originally Posted by Hlingler

Not true, so far as I have always understood.

Based on what?

Originally Posted by Hlingler

In any case: is it not worth a try?

What is the point of adding something that is unnecessary? I think that most people would agree is not worth adding a redundant line to a config file.

Anyway. to check if SWCursor is enabled, see your Xorg log file like this:

cat /var/log/Xorg.0.log | grep cursor

And look for this:

(**) fglrx(0): Option "SWcursor" "true"
(**) fglrx(0): Using software cursor

[iamroot]

Originally Posted by Ravnos

It does work if you do not use the mouse too much, I will spend all the day long using my computer without a mouse, no problems so far. Maybe it just need some tweaks in xorg.conf to be stable, and people say it works on other distros using kernel 2.6.29 with no problems. And it might be some odd problem here too.

So the drivers have been released? RPMFusion still doesnt have the flgrx driver.

[Hlingler]

OK, since you seem to know so much, and also have the target hardware, I'll stop making suggestions here, and leave the rest of the members in your capable hands. I'm sure that you'll be able to help them all fix whatever problems they have.

Good Luck,
V

[mck182]

Originally Posted by Hlingler

OK, since you seem to know so much, and also have the target hardware, I'll stop making suggestions here, and leave the rest of the members in your capable hands. I'm sure that you'll be able to help them all fix whatever problems they have.

Good Luck,
V

Forget that guy, I'd rather try even redundant options, if it's supposed to help. It definitely won't hurt.

Thanks for all your help so far...and please, stay with us

[Ravnos]

Originally Posted by iamroot

So the drivers have been released? RPMFusion still doesnt have the flgrx driver.

There are no RPMFusion packages yet, but ATI released a new driver that supports the current F11 kernel, but unless it becomes stable do not expect to see them soon in RPMFusion.

TrentZ: Thanks ! It works here, I guess I can live with this mouse corruption, and the problem became more clear to be fixed.

Hlingler: Thanks, but unfortunately adding HWCursor off did not help.

About the redundant option, I did not found anything in the xorg.conf manual, and sometimes X does not think things are redundant and load them both, resulting in two drivers reading inputs from the same device, so I guess it was a valid try ...

[Hlingler]

Originally Posted by mck182

Forget that guy... Thanks for all your help so far...and please, stay with us

Leigh is on holiday, and the other <ahem> "expert" appears to be either unable or unwilling to field questions, so you're stuck with me for now.

Originally Posted by mck182

I'd rather try even redundant options, if it's supposed to help. It definitely won't hurt.

Originally Posted by Ravnos

Hlingler: Thanks, but unfortunately adding HWCursor off did not help.

About the redundant option, I did not found anything in the xorg.conf manual, and sometimes X does not think things are redundant and load them both, resulting in two drivers reading inputs from the same device, so I guess it was a valid try ...

Exactly: it didn't help (as it turns out), but it was extremely unlikely to hurt (probability >= 99.99%). Also: I concede that 'Option "HWCursor" "off" ' is redundant and should not be needed, BUT as we often see, things don't always work as they're supposed to (case in point: why are we having the current discussion... ?), and small things like this are very easy to try, and might help. Unfortunately, this idea did not.

Additional things to try (from man xorg.conf):

INPUTDEVICE SECTION

       Option "AlwaysCore"  "boolean"

       Option "SendCoreEvents"  "boolean"
              Both of these options are equivalent, and when enabled cause the
              input device to always report core events.  This  can  be  used,
              for  example,  to allow an additional pointer device to generate
              core pointer events (like moving the cursor, etc).

       Option "HistorySize"  "number"
           Sets the motion history size.  Default: 0.

       Option "SendDragEvents"  "boolean"
              ???

Not sure what value to suggest for 'Option "HistorySize" ' - this manual page is not very helpful....

Also try the ati-config utility included with the binary package.

V

[DteK]

Adding:
Option "SWCursor" "true"

as TrentZ mentioned seems to have resolved the issue for me so far. I can certainly live without compiz for the moment. just glad to finally get this running.

Thanks alot Hlingler and TrentZ