Ticket #4635 (new defect)

Opened 6 years ago

Last modified 6 years ago

When hitting the power button to resume, the laptop reboots instead

Reported by: tomeu Owned by: dilinger
Priority: blocker Milestone: 8.2.0 (was Update.2)
Component: kernel Version:
Keywords: Cc: pascal, wad
Action Needed: Verified: no
Deployments affected: Blocked By:
Blocking:

Description

What the title says, in joyride 224.

Change History

  Changed 6 years ago by pascal

  • cc pascal added

  Changed 6 years ago by cjb

I've just seen this too. We'd presumed it was just old hardware, but it happens on my C-Test, which is usually more reliable than this. It happens almost every time for me.

I'd like to upgrade this to blocker until we try the machines Wad is bringing back and find out whether it happens there too.

  Changed 6 years ago by jg

  • cc wad added
  • owner changed from jg to wad
  • component changed from distro to hardware
  • priority changed from normal to blocker

Ouch. Might not be hardware, though.... Wad, what's your guess?

  Changed 6 years ago by wad

  • status changed from new to assigned

The ticket doesn't mention it explicitly, but these are B4 and C1 units, right ?

Chris, are you seeing this on your golden C2 ?

There is a similar hardware problem (#4683, son of #1835(b)), but it hasn't been seen on button presses and the frequency is so rare that it shouldn't show up as "almost every time", unless...

Is the DCON actually being power cycled now ? On older machines this REALLY whacked the +3.3V line --- we will probably need a software workaround to only do this on C1 or better (and the C1 should be modified or it will still see it.)

  Changed 6 years ago by tomeu

This happens in my B4 with joyride 232, but doesn't happen in a B3 with 210.

  Changed 6 years ago by pascal

Mine is a B4. And for me the problem is more accurately that resume doesn't work when I suspend using the power button. Joyride-217 was fine.

It will crash/reboot on resume from, so far, all kinds of events. Closing the lid after pressing the powerbutton, opening the lid, pressing a button. Anything, really.

follow-up: ↓ 13   Changed 6 years ago by pascal

I did some more research. 217 is affected too. suspend/resume doesn't work. It seems to be modulated by network activity: when connected to my AP, it is easier to trigger than when not associated. When not associated, the B4 just hangs.

I went 'back' to 623 to see how that works. No problems what so ever. I can suspend/resume dozens of times without issue.

  Changed 6 years ago by pascal

joyride 204, the oldest I have available has the problem too.

  Changed 6 years ago by wad

This should be reassigned. Smells like software....

  Changed 6 years ago by kimquirk

  • owner changed from wad to dilinger
  • status changed from assigned to new
  • component changed from hardware to kernel

  Changed 6 years ago by kimquirk

  • milestone changed from Never Assigned to Update.1

  Changed 6 years ago by pascal

after talking in #olpc, it seems this is caused by B4 hardware not coping with the really deep suspend of more recent kernels. So software, yes, expected too.

in reply to: ↑ 7   Changed 6 years ago by dilinger

Replying to pascal:

I did some more research. 217 is affected too. suspend/resume doesn't work. It seems to be modulated by network activity: when connected to my AP, it is easier to trigger than when not associated. When not associated, the B4 just hangs. I went 'back' to 623 to see how that works. No problems what so ever. I can suspend/resume dozens of times without issue.

I'm going to ignore this unless you can reproduce it with a newer kernel. Joyride builds have been pulling in the wrong kernels, and there is all kinds of problems with them.

follow-up: ↓ 15   Changed 6 years ago by dilinger

Ok, joyride build 256 includes a proper kernel; please try and reproduce with that build.

in reply to: ↑ 14   Changed 6 years ago by pascal

Replying to dilinger:

Ok, joyride build 256 includes a proper kernel; please try and reproduce with that build.

Same issue, sorry. I don't know, but smithbone was quite adamant in #olpc this was to be expected for B4's, I'm not too worried about it. After update.1 I'll build myself a kernel without suspend support or something.

follow-up: ↓ 17   Changed 6 years ago by kimquirk

I saw this on a B4 joyride 258; 11/12/07. This build of joyride has suspend and resume running.

You have to hit a key or touch the pad to resume and twice (already) I have had one of my 5 machines reboot instead of resume.

in reply to: ↑ 16   Changed 6 years ago by dilinger

Replying to kimquirk:

I saw this on a B4 joyride 258; 11/12/07. This build of joyride has suspend and resume running. You have to hit a key or touch the pad to resume and twice (already) I have had one of my 5 machines reboot instead of resume.

Is it the same machine doing it every time, or does it happen randomly amongst all 5 of those machines? Also, can you reproduce it on C1+?

Note: See TracTickets for help on using tickets.