On Sun, Nov 6, 2011 at 3:32 AM, Alcides Viamontes Esquivel <a.viamontes.esqui...@gmail.com> wrote: > Hi guys, > > I would like to say thanks for your wonderful job in supporting Wacom tablets. > > I got one CTH-470 yesterday and could get it cranking thanks to the > work of Chris Bagwell > (http://sourceforge.net/mailarchive/message.php?msg_id=28204872) , but > I found some sort of bug which I think is related to that thread in > GIMP discussion list but that also affects all other uses of the pen > in this tablet: the cursor jumps some times (all > coordinates/distance/pressures and a button touch=0) when dragging the > stylus for a while. This doesn't happen in Windows, so I assumed it > was some sort of software problem.
Thanks for the report and help in isolating it. I haven't seen the issues but I'm also not an extensive user of the tablet. > > Anyway, because I wanted it working today I went for a quick bug > hunting. I found that there is a weird event packet processed by > wacom_bpt_pen where data[1]==0, and that, of course, causes some fuzz. > I merged the latest changes that I could find to this driver (version > 1.52 in wacom.h, but I took the files from Linus' github repository) > with Chris' work and patched for this zero-packet. Now it works very > well for me (changed files attached, this driver works!). FYI: The driver is in a staging tree and scheduled to be in Linux 3.3. You can grab the latest files from Dmitri's "next" branch until 3.3-rc period. http://git.kernel.org/?p=linux/kernel/git/dtor/input.git;a=shortlog;h=refs/heads/next > > I suspect that the real bug is somewhere in the way the packets are > classified, and that somehow wacom_bpt_pen is processing a packet > which is not for her. But I'm really not a kernel developer, so, I > would appreciate any hints (in particular, how do you trace from > inside a kernel module?). > What code did you add to handle this? Do you ignore the whole packet (do a return) if data[1] == 0? I need to understand a little more whats going on in the big picture. If we did get this unwanted packets, the current driver should be telling xf86-input-wacom that the pen when out of proximity and then when next packet comes that is valid it will tell xf86-input-wacom that its back in proximity and with good values. Since we do not move the cursor while out of proximity and presumably have valid data for next packets, I'd think no issue should be seen beyond some pauses in drawing. So to really see whats happening, you may need to write some debug code that does a printk() for 1st packet when data[1] == 0 and then also printk() for the next 5 valid packets as well to see what their values are. Chris ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ Linuxwacom-devel mailing list Linuxwacom-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxwacom-devel
