On Fri, Mar 5, 2010 at 2:12 PM, Chris Bagwell <ch...@cnpbagwell.com> wrote:
> Because your still seeing "bad magic" messages, I've got a concern
> that we are throwing out packets that are related to going out of
> proximity. If this occurs, we will never switch tools because for
> case of pen to eraser, it will only make this decision on transition
> from out of proximity to in proximity. That work around logic related
> to side button seems to be related to why eraser to pen always works.
>
> It seems that in error cases, the current code is throwing away to
> much data. It would be hard to ever get back in sync this way. Can
> you search in wcmISDV4.c and look for this code:
Upon further review, I think I was wrong about the wcmSerialValidate()
consuming to much data.. but I think I see other issues.
I think it would be safer to move this validation logic to earlier to
prevent a case were a corrupt packet could be treated as a valid touch
packet and consume wcmPktLength data; further corrupting next packet.
/* Coordinate data bit check */
if (data[0] & 0x40) /* control data */
return common->wcmPktLength;
else if ((n = wcmSerialValidate(common,data)) > 0)
return n;
Move right after:
if (len < common->wcmPktLength)
return 0;
Also, it might be safer to rephase validation logic like this so we
can verify the overall packet looks valid before we consume a full
wcmPktLength worth of "control data". I'm assuming this "control
data" follows same rules and has a 0x80 magic header with nothing else
following it having 0x80.
if ((n = wcmSerialValidate(common,data)) > 0)
return n;
/* Coordinate data bit check */
if (data[0] & 0x40) /* control data */
return common->wcmPktLength;
Chris
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Linuxwacom-devel mailing list
Linuxwacom-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linuxwacom-devel
