Bug 96308

Created attachment 124239 [details]
evemu-record output

evemu-record output in attachment

whoah, this really jumps all over the place. but it's a bit hard to debug because the recording contains almost 50000 lines, 2 minutes of events and it doesn't start from a neutral state. (also, clicking the link keeps crashing my browser). Please try to get a shorter recording, from a neutral state (i.e. finger off the touchpad), and only a single movement of a few seconds max. Thanks.

Created attachment 124267 [details]
evemu-record output few seconds

evemu-record output few seconds

Hello,
I have just sent you an attachement of a shorter event.

It's quite difficult to take a snapshot from a neutral state cause the passage from early state crazyness to deep crazyness is very fast.

If this log cannot help you I tought about a simple script that runs evemu-record, wait 2 seconds and then quit.

I call the script just "a" , and put it under /usr/local/bin so that when event happens , I can just do Alt+F2 --> "a" quickly to get more canches to record the events you need.


Thank you for your help,
Ciao
Daniele

wow, this is really messed up. Can you try a newer kernel please (grab one from F24). Looks familiar though, Benjamin, didn't we have a bug at some point where the ABS_MT_SLOT events where missing? this one looks exactly like that:

E: 0.030027 0000 0000 0000      # ------------ SYN_REPORT (0) ---------- +9ms
E: 0.038894 0003 0035 1088      # EV_ABS / ABS_MT_POSITION_X    1088
E: 0.038894 0003 0036 0162      # EV_ABS / ABS_MT_POSITION_Y    162
E: 0.038894 0003 0000 1088      # EV_ABS / ABS_X                1088
E: 0.038894 0003 0001 0162      # EV_ABS / ABS_Y                162
E: 0.038894 0000 0000 0000      # ------------ SYN_REPORT (0) ---------- +8ms
E: 0.047096 0003 0035 0384      # EV_ABS / ABS_MT_POSITION_X    384
E: 0.047096 0003 0036 0131      # EV_ABS / ABS_MT_POSITION_Y    131
E: 0.047096 0003 0000 0384      # EV_ABS / ABS_X                384
E: 0.047096 0003 0001 0131      # EV_ABS / ABS_Y                131
E: 0.047096 0000 0000 0000      # ------------ SYN_REPORT (0) ---------- +9ms
E: 0.054086 0003 0035 1223      # EV_ABS / ABS_MT_POSITION_X    1223
E: 0.054086 0003 0036 0405      # EV_ABS / ABS_MT_POSITION_Y    405
E: 0.054086 0003 0000 1223      # EV_ABS / ABS_X                1223
E: 0.054086 0003 0001 0405      # EV_ABS / ABS_Y                405
E: 0.054086 0000 0000 0000      # ------------ SYN_REPORT (0) ---------- +7ms
E: 0.062658 0003 0036 0449      # EV_ABS / ABS_MT_POSITION_Y    449
E: 0.062658 0003 0001 0449      # EV_ABS / ABS_Y                449

(In reply to Peter Hutterer from comment #5)
> wow, this is really messed up. Can you try a newer kernel please (grab one
> from F24). Looks familiar though, Benjamin, didn't we have a bug at some
> point where the ABS_MT_SLOT events where missing? this one looks exactly
> like that:

Hmm, not sure of which bug you are talking about, but this has been common in the early days of hid-multitouch when the protocol was still uncertain.

Daniele, please attach a dmesg and a hid-recorder output (package hid-replay in Fedora, run with "sudo hid-recorder"). In the hid recorder output, I'd like you to you to follow this procedure (whether the touchpad is working or not under X/wayland):
- single finger touch/release
- wait few secs
- one finger down, second finger down, first finger release, second finger release
- wait few secs
- then as put and move many fingers possible on the sensor (usually 10, but those small touchpads might be limited to 5 fingers, no need to call all your friends to stick too many fingers on the touchpad :-P )

Created attachment 124338 [details]
dmesg

dmesg

Created attachment 124339 [details]
hid-record output

hid-record output

Created attachment 124340 [details]
hid-record output from crazy state

hid-record output from crazy state

Peter,
let's try Benjamin suggestions before updating the kernel.

I have uploaded the output from dmesg and hid-recorer.

Firs hid-record output did not cause a crazy state, the second did.
Do not know if it is a mine suggestion but it seams that putting 5 finger on the touchpad calm the crazyness.


The cursor then is blocked for some seconds and after few seconds comes back to live again.



Ciao and thank you for your help Peter and Benjamin.



Daniele

Thanks for the logs. From the look of it, the first hid-recorder seems OK (though confidence is not set how I wanted it to be). On the second one, it really goes crazy:

  0.000000 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1175 | Y:    185 | Scan Time:  49461 | Contact Count:    1 | B1: 0 | # 
  0.009717 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1178 | Y:      0 | Scan Time:  49560 | Contact Count:    1 | B1: 0 | # 
  0.019703 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1223 | Y:    232 | Scan Time:  49655 | Contact Count:    1 | B1: 0 | # 
  0.028705 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1186 | Y:    145 | Scan Time:  49754 | Contact Count:    1 | B1: 0 | # 
  0.038611 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1178 | Y:    109 | Scan Time:  49844 | Contact Count:    1 | B1: 0 | # 
  0.048116 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1176 | Y:    215 | Scan Time:  49941 | Contact Count:    1 | B1: 0 | # 
  0.057634 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1223 | Y:      0 | Scan Time:  50036 | Contact Count:    1 | B1: 0 | # 
  0.066883 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1177 | Y:      0 | Scan Time:  50130 | Contact Count:    1 | B1: 0 | # 
  0.076872 ReportID: 3 / Confidence: 0 | Tip Switch: 1 | Contact Id:  1 | # | X:   1223 | Y:    268 | Scan Time:  50222 | Contact Count:    1 | B1: 0 | # 

Y gets reset to 0 quite a lot of times, while the Scan Time increments properly. I'd say you also have the same HW issue than the one in the link you shared (http://www.zachpoff.com/diy-resources/acer-laptop-touchpad-fix/) but I'll reach out to Synaptics to check with them if there is a software solution for this.

Synaptics' answer is that there is not much we can do on the software side. We "could" reset the touchpad when we detect this but this has a lot of issues (stuck fingers, takes time).

This has been fixed in the late production series (as seen in the link on the one you provided http://www.iq-tm.de/TP%20freeze.htm).

So basically, there is not much we can do unless you can crack open the laptop and add a proper ground to the touchpad :(

Thank you Benjamin,

I bought this netbook after my notebook has been stolen, just to continue working while deciding what to buy next. I will not risk to break it before I buy a new one.

As workaround I installed "touchpad indicator" gnome plugin instead. 
With a mouse plugged in , the touchpad is automatically disabled and I can continue to work.

Thank you again for the help.

I will let you know if I will do some hacks with the welder.


Ciao
Daniele