Re: [PATCH v2 00/10] mfd: cros_ec: Add multi EC and proto v3 support

[Heiko Stuebner <heiko@sntech.de>]

Am Montag, 11. Mai 2015, 13:08:27 schrieb Alexandru Stan:
> On Mon, May 11, 2015 at 10:53 AM, Heiko Stuebner <heiko@sntech.de> wrote:
> > Doing an evtest /dev/input/event0 does not result in key events, now after
> > some keypresses I got a
> > [   48.352023] cros-ec-spi spi0.0: EC failed to respond in time
> > [   48.357695] cros-ec-spi spi0.0: Error transferring EC message -110
> > [   48.363884] cros-ec-spi spi0.0: failed to get keyboard state: -110
> > [  226.478588] cros-ec-spi spi0.0: Error transferring EC message -11
> > [  226.484690] cros-ec-spi spi0.0: failed to get keyboard state: -11 well
> > as the touchpad) the keyboard
> At first I tought it was because your EC firmware version is pretty old.
> A while ago we fixed a pretty nasty scheduler bug that would have
> caused similar symptoms:
> https://chromium-review.googlesource.com/#/c/248681/
> 
> I did my own testing with the trees you provided (v1 and v2) on
> veyron-minnie and
> I can definatelly confirm that v2 is broken.

for completenes sake, branches in use where:

clean apply of v2 on top of 4.1-rc3 + clk-next
(already emits the error when using evtest):
https://github.com/mmind/linux-rockchip/commits/tmp/crosec_v3_2

merged into working branch that produces something usable for veyron devices:
https://github.com/mmind/linux-rockchip/commits/devel/somewhat-stable

same branch rolled back to patch-series v1
(working keyboard on both pinky and minnie):
https://github.com/mmind/linux-rockchip/commits/tmp/cros_ec_to_v1


I'm at least somewhat sure, that there shouldn't be any big merge errors in 
there :-)


> 
> Here's a snippet of what a key press generates on the EC log:
> > [950.749097 KB poll]
> 
> [950.760357 HC 0x60]
> [950.760555 HC err 6]
> +[950.903657 HC 0x60]
> [950.905243 HC err 6]
> +[950.926239 KB wait]
> 
> Same key press on v1:
> > [37.107390 KB poll]
> 
> [37.118457 HC 0x60]
> [37.209218 HC 0x60]
> [37.278431 KB wait]
> 
> In the EC code, error 6 is defined as EC_RES_INVALID_VERSION.
> 
> > in my very first try it started similarly non-working but after some time
> > started magically working again, but now in subsequent tries I always get
> > the spi error + working keyboard after a reboot.
> 
> At first the keyboard doesn't work for me either (20 or so seconds).
> Then all keys show up and everything's fine (besides the errors on the EC
> log).
> > Strangely this really seems to be limited to the keyboard, touchpad works
> > flawlessly it seems.
> 
> Touchpad doesn't go through the EC, it's just a standard i2c device
> directly connected to the AP.

I should've looked at the dts before then to not overlook this ;-)


Heiko