[ltp] T460s running Debian testing, latest kernels, HW keeps dying

[Thomas B. Rücker]

Hi,

On 11/21/18 2:12 PM, Thomas B. Rücker wrote:
> I'm turning to this list, as I have run out of ideas about the source of
> my issues two planar replacements ago.
>
> The problem is that in the last few months the device has died now 4 or
> 5 times, each time requiring a planar replacement. Each time it would
> become unbootable. The first occurence it had problems turning on and
> then just failed to do anything, despite disconnecting batteries, trying
> all reset buttons. The following times firmware was still alive. 

Of course I kinda figure it out /after/ I ask for help…

I suspect it was some sort of EFIVARfs problem. Turns out the firmware
they shipped on the latest board was really ancient.

After having that "hey, you didn't rule out that old EFIVAR problem"
thought, I went and tried to get it to boot in legacy mode and such.
That eventually succeeded. I also managed to switch back to UEFI and
boot the device.

I took the opportunity and deleted all files named "dump-type0*" from
/sys/firmware/efi/efivars - there were quite a few and they seem safe to
delete. I also updated firmware and ME firmware.

I still had that weird state once where it got stuck, but after more
reboots and changing boot mode it seems to have settled. Maybe the NVRAM
garbage collection ran, as AFAIU things don't get cleaned up
immediately, but depending on implementation every n boots.

I remain disappointed that Lenovo doesn't give you any way to do a
proper NVRAM/EFIvars purge. None of the factory default restores in
firmware seemed to accomplish anything in that regard.

I *hope* this was it…


Cheers,

TBR