Jason and Chris, I am slow in reviewing the patches. But I do want to share my thoughts so I don't block your progress. See my comments inline.
Ping On Tue, Jan 19, 2010 at 8:21 PM, Jason Childs < obliv...@users.sourceforge.net> wrote: > Chris, > > Thanks for the feedback, I've got two more from tonight as well plus a > fix to patch 0011. In patch 0011 I messed up the logic on resetting > firstFinger to zero in the kernel code. The 0019 makes sure horizontal > scroll will work on TPC2FG's, and 0020 fixes a problem pointed out by > Favux with my change to wacomcpl-exec. > > More responses inline... > > On Tue, 2010-01-19 at 21:23 -0600, Chris Bagwell wrote: > > > Since there is quite a bit there, I suggest breaking the submits to > > CVS into batches. I suggest doing the earlier kernel patches first > > just because they can be verified on both versions of X drivers right > > now. Probably we should save the patch that disables buttons (0011) > > until its discussed a little more. > > > Agreed, I will need feedback on adding serial numbers and channels for > touch devices. I've got a mocked up patch that seems to work really > well (i.e. I've got buttons working again and gestures). It will > obviously need review to make sure I don't break existing functionality. > I am ok with adding serial numbers and channels for touch devices. I want to sleep on them tonight and take a fresh look at them tomorrow > before I send them out. Also, in wcmTouchFilter.c there are hard coded > values for the gesture min/max activation distances, obviously 300 and > 350 are a bit big for the 4x5 tablets, I'd like to get feedback on > making these configurable items via the common structure? > You can follow the existing way to make those two values configurable. > > Here is my feedback on your kernel driver patches; all tested with > > current git of xf86-input-wacom (you can add Tested-By: me if you > > like). The comments are more geared towards Ping or whom ever will > > submit the patches as a feel good measure then to Jason. > > > > 0006-Kernel-Module-Fix-shared-features-buffer-corruption.patch > > > > Seems an important fix. I saw no issues in review or testing. I > > really like appending Touch/Pen to name. Makes output of dmesg that > > much more useful. > > It may also add a way of having more than one of the same model tablet > connected since we can dynamically change the name in the kernel driver > now. Still thinking of what we can append that makes them unique > though. They would then show up as distinct devices to wacomcpl and X. > This is a very promising approach. If you can make the name unique for those same model tablets, we will no longer need to bug kernel/hal/udev/xorg to distinguish the devices for us. I like this independence. I guess we will need to see if this approach gives us a static name convention or not. By static I mean the name would not change for the same device after each logout/login and reboot as long as the user doesn't reconnect the device to a different physical port. > > > 0002-Kernel-Module-Fixes-for-Bamboo-P-T-issues.patch > > 0005-Kernel-Module-Change-Bamboo-P-T-features-length.patch > > > > Code definitely reads better using correct packet size everywhere. My > > tablet worked before and after patch though. I know Jason said it > > helps certain USB hubs and maybe more. Can't see a place it would > > hurt other tablets (wrapped in if()'s for Bamboo P&T specific > > functions). > > > > Yep, definitely required the correct size on my ATI OCHI chipset to get > the pen packets working. > These two were the leftover from last merge. I have no problem to commit them. > Patch 0002 contains some obvious fixes in x/y values that look good. > > > > 0001-Input-wacom-Debian-patch-for-device-ids.patch > > > > No real comment on this one. It worked for me and I saw no issues. > > I'm not sure I understand background of why patch was written. I > > think the reason may be in mailing list archive but I didn't look. > > This came from the devel list a few weeks ago. It was supposed to help > with the crashes I was seeing initially. I was fighting for spin locks > around all static shared memory access in the kernel driver. But I > promised Ping I'd back off on that since I haven't had any further > crashes with the changes from the 0001 patch :). > I want to be fair for both solutions. So, please "break" your promise if you ever crash your system with this patch applied. > 0003-X-Driver-Default-Bamboo-P-T-Touch-to-Relative.patch > > > > OK, the only xdrv patch I'll commit on for now. :-) Its already in > > xf86-input-wacom so might as well get it in linuxwacom as well. > > > > Chris > > > 0007, 0008, 0009 and 0013 are important for getting hotplug of a tablet > to work in X. They makes sure that the correct tablet initialization > values are being used when a tablet is plugged in after X is already up > and running. There is some other stuff mixed in (like gesture changes). I haven't had time to get into these patches yet. Talking about hot-plugging, I'd like you to test the attached .fdi file on XOrg 1.6 or older systems. From the driver point of view, this .fdi should work for Bamboo P&T. I need your testing result to prove if it is right or wrong. Ping
10-linuxwacom.fdi
Description: Binary data
------------------------------------------------------------------------------ Throughout its 18-year history, RSA Conference consistently attracts the world's best and brightest in the field, creating opportunities for Conference attendees to learn about information security's most important issues through interactions with peers, luminaries and emerging and established companies. http://p.sf.net/sfu/rsaconf-dev2dev
_______________________________________________ Linuxwacom-devel mailing list Linuxwacom-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxwacom-devel
