Re: further touchscreen / keyboard issues in the newest cvs version

Hello Daniel,

Sunday, February 4, 2007, 7:48:11 AM, you wrote:

> Hi all,

> So it seems that the new touchscreen stuff interferes with the keyboard
> module for the h4350. A few days ago the h4000_defconfig file in CVS
> did not include support for the touchscreen. When I compiled from CVS
> the keyboard worked as usual. However when I then went back and
> included the 2 lines for touchscreen support - CONFIG_INPUT_TOUCHSCREEN
> and CONFIG_TOUCHSCREEN_ADS7846_SSP the keyboard no longer worked.

  Well, that's strange. These modules are not directly related to
keyboard in any way. Can you make clean-room test: use CVS HEAD and
h4000_defconfig as the base, but disable those 2 options, and check
keyboard works, then use pristine h4000_defconfig and check that it
doesn't?

> I
> figured I had done something wrong, but Paul has since released a new
> version of the defconfig file, and it has the same problem. I have
> included the output from dmesg at the end of this email. I have noticed
> four interesting things. The first is that when I load the keyboard
> module the keys light up on button press, but no input is recieved (this
> is true reguardless of whether I am in opie or the terminal). The
> second is that after I unload the module the keys still light up on
> keypress even though the output of lsmod does not show the module
> present (though they don't before the module is loaded for the first
> time). The third is that each time the module is loaded it increments
> the inputX number that is used (it starts at /dev/input/input3, and goes
> up from there). And the fourth is that the output of lsmod when the
> module is loaded does not show that it is in use.

> Thanks,
> Daniel Steen

> P.S. I have included the entire output of dmesg. The kbd stuff is at
> the bottom.

[]

> IRQ handler type mismatch for IRQ 158
> current handler: Record button
> [<c0026dac>] (dump_stack+0x0/0x14) from [<c005ba40>] (setup_irq+0x1bc/0x1e0)
> [<c005b884>] (setup_irq+0x0/0x1e0) from [<c005bc3c>] (request_irq+0xac/0xc8)
> r8 = BF0585A8 r7 = 00000040 r6 = BF0582B8 r5 = 0000009E
> r4 = C3601FA0

   This means that h4300_kbd.c is grown somewhat old and needs some
rewrite. But that's not news, of course. It still depends on filling
in content of h4300_detect.c, which I still hope some of h4300 guys
will go for, because I've still got too much other stuff to work on.

[]

-- 
Best regards,
 Paul                            mailto:pmiscml_at_gmail.com