Tails

#1 Updated by intrigeri 2019-11-10 15:36:18

Hi,

> given that tails has been running for a few hours, starting big applications ( like LibreOffice, GNU image manipulation etc..) make GNOME crash

> two error lines show up on the screen for a few seconds before going to a static non responding black screen then tails has to be restarted manually by the user:

> amnesia kernel: nouveau 0000:01:00.0: bus: MMIO write of 0000001f FAULT at xxxxxx [ IBUS ]
> amnesia kernel: nouveau 0000:01:00.0: bus: MMIO write of badf1001 FAULT at xxxxxx [ IBUS ]

> note that these errors also appear during boot but doesnt stop tails from starting normally

I suspect that these 2 lines are unrelated to GNOME crashing, and were actually already present on the console before GNOME crashed.
If my hunch is correct, then we have 2 problems here:

• this nouveau driver error messages → as long as this does not break anything, I think we can safely ignore them
• GNOME crashing → that’s a problem, indeed; let’s focus on this one for now

> maybe it has something to do with running out of RAM, in which case I suggest warning users they dont have enough memory left to run certain applications

Right, it could be the case. But it could be something else too.

To debug this, we would need someone who’s affected by this problem to send us the output of the sudo tails-debugging-info command, run from a text console after the GNOME crash. Do you have the skills and motivation to do that?

#2 Updated by Cozain 2019-11-10 18:05:58

intrigeri wrote:

> I suspect that these 2 lines are unrelated to GNOME crashing, and were actually already present on the console before GNOME crashed.

those exact two lines show on the screen each time this happens with a time stamp showing it is actually after crashing

further, if console shows any last 2 lines then it should change each time which doesnt happen

> Right, it could be the case. But it could be something else too.

running out of RAM should show a warning to prevent a crash regardless of this ticket

it is an important function which is missing in tails

>send us the output of the sudo tails-debugging-info command

I rather get direction to investigate this after the crash

how can I get to command line after this? it is not possible, the computer just hangs

#3 Updated by Cozain 2019-11-10 18:15:56

@intrigeri

if you like to try replicating this then fill memory till almost reaching the end then start an application that requires more memory and see what happens

check if you get the same result described here or that you get different errors

#4 Updated by Cozain 2019-11-10 18:20:05

@intrigeri

I already mentioned earlier in the ticket that the outcome is a total freeze and that the system has to be restarted, maybe you missed this part

#5 Updated by intrigeri 2019-11-11 11:43:12

Hi,

> I already mentioned earlier in the ticket that the outcome is a total freeze and that the system has to be restarted, maybe you missed this part

Indeed, this was not 100% clear to me so far. I don’t know where you’re coming from so when you write “static non responding black screen”, it’s hard for me to guess what you mean. For example:

• for some users, the lack of a mouse cursor is enough to qualify as “non responding”
• for some users, no feedback when typing qualifies as “non responding”
• other users would try various CTRL+ALT+Fn to try to reach another tty, and only if that fails, they would say that Tails is not responding; did you try this?

Please send a WhisperBack report before the crash, referencing Bug #17218 in it, so I can check what exact hardware is affected.

Also, we have a bunch of workarounds of workarounds documented for various NVIDIA graphics cards. One of them (nomodeset, modprobe.blacklist=nouveau) are mentioned in the links you pointed us to. Please try these workarounds and tell me what they give you.

Finally, a newer kernel might help, so please try an image from https://nightly.tails.boum.org/build_Tails_ISO_bugfix-17124-17161-linux-5.3-from-sid-force-all-tests/lastSuccessful/archive/build-artifacts/ and see if it is affected as well.

Thanks in advance!

#6 Updated by Cozain1 2019-11-20 18:37:35

intrigeri wrote:
> Hi,
>
> > I already mentioned earlier in the ticket that the outcome is a total freeze and that the system has to be restarted, maybe you missed this part
>
> Indeed, this was not 100% clear to me so far. I don’t know where you’re coming from so when you write “static non responding black screen”, it’s hard for me to guess what you mean. For example:
>
> * for some users, the lack of a mouse cursor is enough to qualify as “non responding”
> * for some users, no feedback when typing qualifies as “non responding”
> * other users would try various CTRL+ALT+Fn to try to reach another tty, and only if that fails, they would say that Tails is not responding; did you try this?
>
> Please send a WhisperBack report before the crash, referencing Bug #17218 in it, so I can check what exact hardware is affected.
>
> Also, we have a bunch of workarounds of workarounds documented for various NVIDIA graphics cards. One of them (nomodeset, modprobe.blacklist=nouveau) are mentioned in the links you pointed us to. Please try these workarounds and tell me what they give you.
>
> Finally, a newer kernel might help, so please try an image from https://nightly.tails.boum.org/build_Tails_ISO_bugfix-17124-17161-linux-5.3-from-sid-force-all-tests/lastSuccessful/archive/build-artifacts/ and see if it is affected as well.
>
> Thanks in advance!

@intrigeri

I dug down much further on this and narrowed this down to this error:

ACPI: \SB.PCI0.PEG0.PEGP: failed to evaluate _DSM

This suggests an issue with NVIDIA card I am using (GF117M) and the current Kernel on tails 4.0 (I might be wrong about this conclusion though, because I am not that familiar with GPU issues)

This issue could also affect other kinds of hardware as I also dont know if any other users are facing this (I need a reference from you)

I did not test the workarounds which you said is documented but didnt provide a link, could you please do?

I still didnt test nomodeset, modprobe.blacklist=nouveau, can this be a startup option or should I just blacklist the driver after login to tails?

Note that this was all based on tails 4.0 not the nightly builds yet

Finally, the question of why this only happens if tails has been running for a long time (day or two) while running the same applications on fresh tails boot works fine, is still unclear

#7 Updated by intrigeri 2019-11-22 11:07:00

Hi,

redmine@redmine.tails.boum.org:
> This suggests an issue with NVIDIA card I am using (GF117M) and the current Kernel on tails 4.0 (I might be wrong about this conclusion though, because I am not that familiar with GPU issues)

Yes, it could be. That’s why I need the debugging info I’ve asked you previously :)

> Note that this was all based on tails 4.0 not the nightly builds yet

It would be great if you could first test the nightly builds: if the newer kernel that will be in Tails 4.1 fixes the problem, then it may not be worth spending more time on this.

> I did not test the workarounds which you said is documented but didnt provide a link, could you please do?
> I still didnt test nomodeset, modprobe.blacklist=nouveau, can this be a startup option or should I just blacklist the driver after login to tails?

Oops, sorry, indeed I messed up a copy’n’paste. Please test all the NVIDIA-related workarounds mentioned on https://tails.boum.org/support/known_issues/graphics/.
That page will point you to https://tails.boum.org/doc/first_steps/startup_options/index.en.html#boot_loader_menu, which should answer your second question.