#12551 HIGH 13.1.0: XO-4 resume hang on interactive wakeup with 13.1.0 os29 Q7B15/Q7B15pg

[Zarro Boogs per Child]

#12551: XO-4 resume hang on interactive wakeup with 13.1.0 os29 Q7B15/Q7B15pg
---------------------------------+------------------------------------------
           Reporter:  greenfeld  |       Owner:                                   
               Type:  defect     |      Status:  new                              
           Priority:  high       |   Milestone:  13.1.0                           
          Component:  kernel     |     Version:  Development build as of this date
         Resolution:             |    Keywords:                                   
        Next_action:  diagnose   |    Verified:  0                                
Deployment_affected:             |   Blockedby:                                   
           Blocking:             |  
---------------------------------+------------------------------------------
Description changed by greenfeld:

Old description:

> I keep running into a resume hang with 13.1.0 os29 and multiple C1 XOs.
>
> When this happens no kernel serial output is seen upon resume (although
> the power LED becomes solid green again) and SysRq fails to work.  Adding
> "debug" and/or "no_console_suspend" does not result in any additional
> resume output.
>
> Reproduction (?):
>  1. Start a XO and get to Sugar's main circle of activities.
>  2. {Optionally} Switch to a Linux text console.
>  3. Edit /etc/powerd/powerd.conf and configure powerd to idle suspend if
> the idle level is above 80% and to suspend after 10 seconds (not 100%
> necessary, but makes reproduction easier given we have trouble reaching
> 85% idle at the moment).
>  4. Wait until the system goes into suspend, and the power LED starts
> blinking.
>  5. Press the spacebar or another key to wake the system up without
> touching the touchpad first.  The power LED will turn on again but the
> main system may not wake up.
>  6. If the above is not to cause this to happen, having a SD card
> inserted while doing the above step seems to be a good jinx (be sure to
> remove it afterward).
>
> Once the system gets in this hung state once, it really seems to easily
> get in this hung state, although that just may be my interpretation of
> things.  A hard power-off followed by restarting the system may result in
> both the keyboard and touchpad being unusable while the system still
> responds to touchscreen and power button input.  A subsequent power cycle
> will result in the keyboard and touchpad being usable again.  The
> touchpad may hang the system if used to wake up from suspend, even if it
> did not before.
>
> Disabling wifi in Sugar, changing the video card to fbdev, pulling out
> the battery, etc., does not seem to be enough to get out of this repeated
> hanging state.  This is not #12541 in that I wait until the system has
> fully suspended and the EC starts flashing the power LED before trying to
> wake up the system again.
>
> I went all morning today without seeing this; but once I saw it once
> while using SD cards, I could not seem to get rid of the issue for
> several hours on the same XO without it almost always repeating the rest
> of the day.
>
> Of note is that I have had perfectly good S/R cycles with external SD
> cards in the XO-4, although #11647 can wreck havoc with this and open
> files during a S/R cycle on a XO-1.75.

New description:

 I keep running into a resume hang with 13.1.0 os29 and multiple C1 XOs.

 When this happens no kernel serial output is seen upon resume (although
 the power LED becomes solid green again) and SysRq fails to work.  Adding
 "debug" and/or "no_console_suspend" does not result in any additional
 resume output.

 Reproduction (?):
  1. Start a XO and get to Sugar's main circle of activities.
  2. {Optionally} Switch to a Linux text console.
  3. Edit /etc/powerd/powerd.conf and configure powerd to idle suspend if
 the idle level is above 80% and to suspend after 10 seconds (not 100%
 necessary, but makes reproduction easier given we have trouble reaching
 85% idle at the moment).
  4. Wait until the system goes into suspend, and the power LED starts
 blinking.
  5. Press the spacebar or another key to wake the system up without
 touching the touchpad first.  The power LED will turn on again but the
 main system may not wake up.
  6. If the above is not enough to cause this to happen, having a SD card
 inserted while doing the above step seems to be a good jinx (be sure to
 remove it afterward).

 Once the system gets in this hung state once, it really seems to easily
 get in this hung state, although that just may be my interpretation of
 things.  A hard power-off followed by restarting the system may result in
 both the keyboard and touchpad being unusable while the system still
 responds to touchscreen and power button input.  A subsequent power cycle
 will result in the keyboard and touchpad being usable again.  The touchpad
 may hang the system if used to wake up from suspend, even if it did not
 before.

 Disabling wifi in Sugar, changing the video card to fbdev, pulling out the
 battery, etc., does not seem to be enough to get out of this repeated
 hanging state.  This is not #12541 in that I wait until the system has
 fully suspended and the EC starts flashing the power LED before trying to
 wake up the system again.

 I went all morning today without seeing this; but once I saw it once while
 using SD cards, I could not seem to get rid of the issue for several hours
 on the same XO without it almost always repeating the rest of the day.

 Of note is that I have had perfectly good S/R cycles with external SD
 cards in the XO-4, although #11647 can wreck havoc with this and open
 files during a S/R cycle on a XO-1.75.

--

-- 
Ticket URL: <http://dev.laptop.org/ticket/12551#comment:2>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system