HP ProBook x360 11 G1 EE Notebook PC succeeds but fails at exit

[sudburr]

FOG Server 1.4.4 with Kernel 4.13.4 on CentOS 7.4.
HP ProBook x360 11 G1 EE Notebook PC with UEFI BIOS 01.09 or 01.10 (latest)

UEFI network boot succeeds via internal ethernet.
Manually selected tasks from FOG menu succeed but then fail to exit properly. eg: Quick Registration, Deploy Image

See this video.

[Wayne Workman]

@sudburr said in HP ProBook x360 11 G1 EE Notebook PC succeeds but fails at exit:

Manually selected tasks from FOG menu succeed but then fail to exit properly

Have you tried every exit type individually in the UEFI exit modes?

[Sebastian Roth]

@sudburr Seems like this piece of hardware doesn’t like the Linux kernel.

Try out kernel parameter reboot=..., see here for the different options you can use: https://github.com/torvalds/linux/blob/master/Documentation/admin-guide/kernel-parameters.txt#L3656

Hope one of these will make it properly reboot. Other than that you might want to try booting Live Linux (like Ubuntu) to see if those hang on reboot as well. I remember a lengthy discussion with someone having reboot/shutdown issues on a different model. It turned out that the issue was not fully reproducible. E.g. when trying 50 times, half of the runs would hang and half were fine. So please try reboot a couple dozen times to see if it’s always the same. To test you don’t need to do a full deploy. Just schedule a debug deploy and key in reboot when you get to the shell. Should be exactly the same to what we do in the FOG scripts after a successful deploy.

[sudburr]

@wayne-workman Without doing a character by character comparison of the halting screen, all EFI exit types failed. Tasks complete, the FOG server is happy, but screen of bad remains.

[sudburr]

@sebastian-roth What precise wordage do you want me to try in [Host Kernel Arguments] ?

[sudburr]

On a debug deploy, entering “reboot” generates.

Below is a variable dump from FOG
osid=9
osname=Windows 10
mbrfile=
type=down
storage=10.12.40.14:/images/
img=Win_10v1709x64_C_20171114_UEFI
imgType=n
imgFormat=0
imgPartitionType=all
disks=/dev/sda
hd=/dev/sda

[Sebastian Roth]

@sudburr said in HP ProBook x360 11 G1 EE Notebook PC succeeds but fails at exit:

What precise wordage do you want me to try in [Host Kernel Arguments] ?

Try all these, one after the other, reboot=warm, reboot=cold, reboot=bios, reboot=smp, reboot=triple, reboot=kbd, reboot=acpi, reboot=efi, reboot=pci, reboot=force, reboot=gpio

See if any of these helps.

[sudburr]

Basically the same regardless of Host Kernel Argument; only a slight variation in the number displayed at the final line of output from the dump at completion.

/etc/int.d/S99fog: line 23: 41xx Segmentation fault reboot -f

[Sebastian Roth]

@sudburr The different numbers are just process ids. That’s not meaning much.

Can you take a steady slow motion video of the output on screen. Maybe there is more valuable information on the screen that we just can’t read in real time. I tried to skip through that last part of the video many times but just can’t read all of it.

As well, have you ever tried to boot some kind of Live Linux (CD or USB key) to see if there is the same happening on reboot?

[sudburr]

Here’s a 240fps video

[Sebastian Roth]

@sudburr Thanks for taking the time to record the video! Unfortunately I could not find anything useful in it. Have you tried booting other Linux live systems on that machine yet??

I compiled some fresh kernel images for you with debug symbols included. With those we should see the function names on screen instead of just the raw hex addresses. Find bzImage and bzImage32 here. Download those, move the current kernels in /var/www/fog/service/ipxe out of the way and put those in place. Try the same thing and take a picture of the kernel fault. Hopefully we’ll see function names this time.

[sudburr]

This may take a while longer to solve. I had to get the unit back to it’s user. Hopefully I’ll come across another one soon.

[Tom Elliott]

@sudburr What IPXE boot file are you using?

[Tom Elliott]

Oh I ask this question (for perspective here) is I had received some HP ProBok 645 G2’s who had a very similar type of problem.

Turns out it was from the firmware applied by bzImage and the ipxe file I was using (undionly.kpxe at the time). However, I found if I used ipxe.pxe it would work perfectly fine. In your case, I assume you’re using ipxe.efi?

[sudburr]

Yes, it was using ipxe.efi.

[Sebastian Roth]

@sudburr @Tom-Elliott Did anyone ever try the debug kernels I provided?

[sudburr]

I was in the throes of migrating all our images over to 10v1709 so I was only just barely able to accomplish what I’ve detailed above before I had to return the unit to its owner.