* Dmitry Torokhov wrote:

On Thu, Aug 25, 2011 at 12:56:25AM -0700, Dmitry Torokhov wrote:

quoted

On Thu, Aug 25, 2011 at 09:52:54AM +0200, Thierry Reding wrote:

quoted

* Dmitry Torokhov wrote:

quoted

On Thu, Aug 25, 2011 at 09:06:57AM +0200, Thierry Reding wrote:

quoted

* Dmitry Torokhov wrote:

quoted

On Thu, Aug 25, 2011 at 08:21:59AM +0200, Thierry Reding wrote:

quoted

* Dmitry Torokhov wrote:

quoted

This is completely untested but I believe it should work better than original
workqueue solution, especially when we do not have get_pendown_state method
available. I wonder if it will help Richard's issue with IRQ getting re-raised
after pen up event.

Anyone who has hardware - please give it a spin.

I can see no noticeable changes with this patch.

That's the result I was aiming for ;) Thanks for testing.

Oh, crap... Does it still work if you change IRQ flags in
request_threaded_irq() from 0 to IRQF_ONESHOT? 

Also, while we are at it would you ming trying out another patch
(below)?

With both your patch applied and the change from 0 to IRQF_ONESHOT the driver
oopses when loaded. I didn't have time to investigate further, but I think I
could perhaps test a couple more patches if you provide them.

Hmm, can't spot the problem... Where does it die?

I'll see if I can track it down, but I have very limited time.

Don't worry, there is no rush whatsoever... I'll try to ponder the code
some more.

I think we fail in tsc2007_open(), instead of "if (err) {" it should say
"if (err < 0) {"... but I still wonder why this causes oops...

It was indeed the open() call that was failing and changing "if (err) {" to
"if (err < 0) {" does indeed fix the issue.

Again, no negative nor positive changes in the touchscreen behaviour. So
again,

Tested-by: Thierry Reding <redacted>