Menu
▾
▴
Re: [Linuxwacom-devel] [PATCH 1/4 v2] Remove channel duplication code for generic devices
|
From: Chris B. <ch...@cn...> - 2012-12-17 17:39:25
|
On Sun, Dec 16, 2012 at 10:34 PM, Ping Cheng <pin...@gm...> wrote:
> Hi Chris,
>
> Thank you for your reply and your comments. Most importantly, thank you
> for your time.
>
> The patch does not have to be merged. Let's figure out if the duplicated
> channel is necessary or not.
>
>
> On Sun, Dec 16, 2012 at 2:56 PM, Chris Bagwell <ch...@cn...>wrote:
>
>>
>> On Thu, Dec 13, 2012 at 2:16 PM, Ping Cheng <pin...@gm...> wrote:
>>
>>> We use true MT protocol for MT devices in kernel now. This code
>>> was introduced to deal with ABS_TOOL_*TAP events loss issue. It
>>> is uncessary any more. And its existence makes it hard to support
>>> generic PAD cleanly.
>>>
>>>
>> I can't think of any negative effects removing these will do with
>> protocol 4 and generic devices with MT events using the latest input-wacom
>> drivers. There is the theoritical issue the code is solving so if its
>> possible to solve what ever the bug is without deleting this then that
>> would be my preference.
>>
>> Since I don't know exact issue your seeing, I can't really comment on if
>> its a good idea to delete this.
>>
>> If you do decide to delete this logic
>>
>
> We can keep this logic. But it does not provide the information required
> for generic device any more.
>
After discussing 3/4, I see now we need some form of this patch. We can
either delete this logic or keep it *but* save off device_type and
proximity and restore it after the structure copy. Maybe need serial # to.
I'm OK with either solution since I have no known issue it solves... except
for next comment.
> For P5 devices, we do not use it anyway.
>
> The theoritical issues only happen:
>
> if incoming tool lands on outgoing tool's axes, As you mentioned we deal
> with absolute values, this case is hard to happen, plus ignoring the first
> motion events would avoid this issue;
>
> if user pressing a button while bringing tool out of prox, then pressing
> the same button while bringing it back in. In this case, we want to send
> the last button up since we do not know if the tool is going to come back
> or not.
>
> In both cases, duplicated channel is unnecessary.
>
>
> then I'd say we official do not support the *old* 2 finger drivers and so
>> I'd delete the Protocol 4 DOUBLETAP and TRIPLETAP logic in wcmUSB.c and
>> wcmValidateDevice.c as part of this patch as well.
>>
>
> Those TAP code are used by ISDV4 touch devices, which I can not afford to
> delete.
>
>
>> Touchpad's using old drivers but without this code become close to
>> un-usable with all the cursor jumping that occurs.
>>
>
> Which old driver are we talking about, in input-wacom or in old kernel
> releases? Which versions?
>
> The current 2.6.30 kernel driver for bamboo in input-wacom is un-usable
> with the current xf86-input-wacom. We have to use xf86-input-synaptics. I
> thought that was on purpose.
>
Oh yeah, I got part of that wrong. We must keep DOUBLETAP logic. Its the
TRIPLETAP that I'm worried about. If any of the old drivers are used that
sends TRIPLETAP then it needs the logic that this patch is attempting to
delete to be somewhat usable (meaning your cursor doesn't constantly make
wild jumps).
I'd prefer to get rid of TRIPLETAP support in xf86-input-wacom *even* *if*
older kernels may be sending it since 1) it improves code readability, 2)
TRIPLETAP has known issues that xf86-input-wacom could never fully work
around and 3) removing code from xf86-input-wacom doesn't break working
with those older kernel drivers, it just removes second finger touch
support which you've mentioned is already broke anyways in current
xf86-input-wacom.
If you want to keep this version of patch (deleting logic), do you mind
also deleting lines 304-307 in wcmValidateDevice.c and 1343-1357 in
wcmUSB.c? That should be all thats needed to remove support for old style
TRIPLETAP.
Chris
> Ping
>
>
>
>>
>> Chris
>>
>>
>> Signed-off-by: Ping Cheng <pi...@wa...>
>>> Acked-by: Jason Gerecke <kil...@gm...>
>>> ---
>>> v2: only patch 2/4 has code change from Jason's comments. To ease the
>>> review
>>> and merge effort, I updated all to v2 with acked-by and reviewed-by tags.
>>> ---
>>> src/wcmUSB.c | 34 ++--------------------------------
>>> 1 file changed, 2 insertions(+), 32 deletions(-)
>>>
>>> diff --git a/src/wcmUSB.c b/src/wcmUSB.c
>>> index e192489..4b5f53b 100644
>>> --- a/src/wcmUSB.c
>>> +++ b/src/wcmUSB.c
>>> @@ -37,7 +37,6 @@ typedef struct {
>>> Bool wcmPenTouch;
>>> Bool wcmUseMT;
>>> int wcmMTChannel;
>>> - int wcmPrevChannel;
>>> int wcmEventCnt;
>>> struct input_event wcmEvents[MAX_USB_EVENTS];
>>> int nbuttons; /* total number of buttons */
>>> @@ -1601,11 +1600,8 @@ static void usbDispatchEvents(InputInfoPtr pInfo)
>>> return;
>>> }
>>>
>>> - /* Protocol 5 devices have some complications related to
>>> DUALINPUT
>>> - * support and can not use below logic to recover from input
>>> - * event filtering. Instead, just live with occasional dropped
>>> - * event. Since tools are dynamically assigned a channel #, the
>>> - * structure must be initialized to known starting values
>>> + /* Protocol 5 tools are dynamically assigned with channel
>>> numbers.
>>> + * The structure must be initialized to known starting values
>>> * when first entering proximity to discard invalid data.
>>> */
>>> if (common->wcmProtocolLevel == WCM_PROTOCOL_5)
>>> @@ -1614,32 +1610,6 @@ static void usbDispatchEvents(InputInfoPtr pInfo)
>>> memset(&common->wcmChannel[channel],0,
>>> sizeof(WacomChannel));
>>> }
>>> - else
>>> - {
>>> - /* Because of linux input filtering, each switch to a new
>>> - * tool is required to have its initial values match
>>> values
>>> - * of previous tool.
>>> - *
>>> - * For normal case, all tools are in channel 0 and so
>>> - * no issue. Protocol 4 2FGT devices split between
>>> - * two channels though and so need to copy data between
>>> - * channels to prevent loss of events; which could
>>> - * lead to cursor jumps.
>>> - *
>>> - * PAD device is special. It shares no events
>>> - * with other channels and is always in proximity.
>>> - * So it requires no copying of data from other
>>> - * channels.
>>> - */
>>> - if (private->wcmPrevChannel != channel &&
>>> - channel != PAD_CHANNEL &&
>>> - private->wcmPrevChannel != PAD_CHANNEL)
>>> - {
>>> - common->wcmChannel[channel].work =
>>> -
>>> common->wcmChannel[private->wcmPrevChannel].work;
>>> - private->wcmPrevChannel = channel;
>>> - }
>>> - }
>>>
>>> ds = &common->wcmChannel[channel].work;
>>> dslast = common->wcmChannel[channel].valid.state;
>>> --
>>> 1.7.10.4
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
>>> Remotely access PCs and mobile devices and provide instant support
>>> Improve your efficiency, and focus on delivering more value-add services
>>> Discover what IT Professionals Know. Rescue delivers
>>> http://p.sf.net/sfu/logmein_12329d2d
>>> _______________________________________________
>>> Linuxwacom-devel mailing list
>>> Lin...@li...
>>> https://lists.sourceforge.net/lists/listinfo/linuxwacom-devel
>>>
>>
>>
>
|