keyboard keys reported stuck in self-test

[kop]

Running diagnostics (left game arrow on boot) sometimes some (a single one, which varies) keys are reported as stuck on by the keyboard test. The screen shows the key pressed when it is not.

Re-pressing the key makes it go away. The screen shows the "stuck" keys as still pressed when other keys are pressed and released. The other keys show as pressed/released on the screen.

I've gotten 2 keys to "stick" at the same time.

Another user on irc reported this behavior as well.

I have not used the keyboard much, at all, so it _could_ be hardware, but I've the feeling it's not. YMMV.

I cannot seem to get the keys to stick in the tam-tam activity (singing mode.)

I thought I saw it both while running my finger across the keyboard and by going down the row pressing individual keys.

G1G1 XO.

[walter]

It is not clear to me from the description that the problem occurs outside of the diagnostic program itself. Can you please clarify?

[kop]

I have not seen the problem occur outside the diagnostic program. Although, as I say, I've not used the machine much so it's possible that it does. The terminal activity does not appear to have keys that auto-repeat, so I don't know I'd notice a stuck key there.

The XO is pretty much "out of the box" new.

My guess is that it's a problem in the diagnostic program. (Which would make me suspect the problem might also occur elsewhere. It "feels" as if it were some some race condition the the debouncing or some other software/timing related issue, but that's founded on complete and utter ignorance.) I have not played with the problem much. I've only gotten it to 'stick' keys 5 or 6 times. I don't recall having noticed any tactile differences in the physical mechanism when either "sticking" or "unsticking" keys. I tried feeling for something different when "unsticking" once and noticed nothing.

[wad]

Closed as duplicate of #5658

Please add further comments to #5658.

[gnu]

According to later comments in #5658, this is *not* a duplicate of 5658. This bug is in the OFW diagnostic for the keyboard (or in something that it depends on, like the EC or the keyboard processor). It can be demonstrated in otherwise-working keyboards, on arbitrary keys (not just the Ctrl/Alt keys). It does not appear to happen while in Sugar and MiniTamTam.

Mitch: I wonder if adding a "view-lossage" type of display to the diagnostic would help figure out what's up. Do we get a key-down and then no key-up? Do we get two key-downs? Do we get a serial garble? Etc.

See: http://dev.laptop.org/ticket/5658#comment:61

http://dev.laptop.org/ticket/5658#comment:65 http://dev.laptop.org/ticket/5658#comment:50 http://dev.laptop.org/ticket/5658#comment:45 http://dev.laptop.org/ticket/5658#comment:46

We have with #5658 a high priority and pervasive flaky-keyboard problem (which seems to be hardware at this point). Since this seemingly unrelated bug that shows up in the OFW keyboard diagnostic tends to confuse and/or mask diagnosis of the pervasive problem, I raised the priority of this bug.

[kop]

More testing. No new results.

I have tried again to get keys to stick in tam-tam (the activity with the drum icon) by clicking on the singer and pressing and releasing keys rapidly. I can't get a sound to 'stick', but I do notice that really rapid key press/release sequences seem to produce no sound at all. These are the sorts of keypresses which would result in a 'stick' in the OFW diagnostic. I'm still able to get key sticks in the diagnostic.

FYI, I'm using the tamtam build 650 (IIRC). I've not upgraded to 653.