On Mon, Jul 28, 2014 at 7:42 PM, Stephen Warren [off-list ref] wrote:

On 07/28/2014 03:23 PM, Stephen Warren wrote:

quoted

On 07/28/2014 02:20 PM, Yufeng Shen wrote:

...

quoted

quoted

Where did you get the configuration file ? It is possible that we rely
too much on mxt_start to turn on the T9.CTRL bit and have neglected
its setting in the config file.
If you can tell me where you get the config file I can do a check.

It was already flashed into the touchpad when I received the board. I
did try to track down the firmware/config files a few months ago, but
didn't manage to; I was told since they were already flashed so I didn't
need them. The board is Venice2.

OK, I received the configuration and firmware file that's supposed to be in
the touchpad.

I can see that the config file I was given has the "83" byte in the T9
configuration, and in fact /almost/ exactly matches the configuration I
have. I don't know why my T9 configuration was wrong before, but I suspect
it's not worth trying to track that down.

Anyway, here's the diff between the two config files:

quoted

# diff -u mxt-save-after-t9-83-write.xml 224sl.raw

--- mxt-save-after-t9-83-write.xml      2014-07-25 19:41:45.000000000
+0000
+++ 224sl.raw   2014-07-28 23:25:49.000000000 +0000

@@ -1,8 +1,7 @@
 OBP_RAW V1
 82 01 10 AA 12 0C 16
 F5AF33
-000000
-0025 0000 0082 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00
+E21E65
 0026 0000 0008 00 00 00 00 00 00 00 00
 0007 0000 0004 20 10 32 00
 0008 0000 000A 1E 00 28 28 00 00 00 00 00 00

It seems that the T25(?) entry is missing in the new/expected configuration
file. I figured I'd try out the new/expected configuration file, so did:

T37 (0x25) is DEBUG_DIAGNOSTIC object which the host can read debugging info
from. It is not useful to have a initial config for it so usually CrOS
system would just
don't include configuration for this object.

Nick, I want to confirm with you that does T37 contribute to config
checksum computation ?

# ./obp-utils/mxt-app -d i2c-dev:1-004b --load 224sl.raw
# ./obp-utils/mxt-app -d i2c-dev:1-004b --save
mxt-save-after-loading-224sl.raw.xml

At this point, mxt-save-after-loading-224sl.raw.xml contains identical
content to mxt-save-after-t9-83-write.xml (my previous backup). It looks
like the new configuration isn't being loaded correctly, or perhaps
configuration loading doesn't delete entries that are simply not in the new
configuration file?

Yeah, I would guess since T37 is not in the config, so whatever in the NVRAM
stays the same and when you --save its original value gets dumped.

I subsequently did the following in case --save is reading from the NVRAM
rather than RAM:

# ./obp-utils/mxt-app -d i2c-dev:1-004b --backup
# ./obp-utils/mxt-app -d i2c-dev:1-004b --save
mxt-save-after-loading-224sl.raw.xml

... but that made no difference.

I haven't yet tried upgrading or otherwise using the new firmware image. I'd
like to make sure config load/save is fully working first, in case there's
any common problem between the two.