xv + silkenmouse bug

[Keith Packard]

On Fri, 2005-09-30 at 16:39 +0200, Thomas Winischhofer wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> 
> It seems I am pushing the common code to its limits with my current
> work. This time it's about Xv:
> 
> 1) Xv in combination with silkenmouse:
> 
> Even if this is assumingly unneccessary, but please allow me to remind
> me and you that Silkenmouse means that the cursor update code is
> executed asynchronously, ie it can interrupt on-going other operations
> of the server.
> 
> The cursor code (pScrn->PointerMoved) is there to eventually pan the
> viewport. So it is likely that pScrn->AdjustFrame is executed, which is
> wrapped by Xv.

I think here's the place where we should fix things -- we should pend
the AdjustFrame call until 'normal' execution time. This means doing the
AdjustFrame from the input event processing code instead of the event
queueing code. The only thing that should be executed from the signal
handler is the hardware mouse positioning function and the event queuing
function.

(it looked like the rest of the message was dealing with AdjustFrame
called at signal handler time).

-keith

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://lists.x.org/archives/xorg/attachments/20050930/c53ea42a/attachment.pgp>