[SOLVED] Kernel not booting, could use help with debugging

steabert · 2014-08-13 19:46:07

Hi,

for a while now I could not boot several linux versions (since 3.14, boot consistently fails for every version), and I thought it had to do with an outstanding UEFI bug. That bug seems to be resolved now in 3.16, but I still cannot boot, so I decided to investigate this issue closer. The kernel I _can_ currently boot is 3.10.46-1 (I'm preventing linux-lts from upgrading to 3.14.16, as that one does not boot).

I've followed the recommended ways to activate debug print levels of the kernel from the ArchWiki page Boot debugging, with the 'extreme' and 'insane' options. In both cases, the logging ends with the last line flickering (so I can't read it) and nothing further happening (I have a video of that but I guess it's too large to post anywhere).

So, currently, I'm a bit stuck. I did think of filing a kernel bug report, but they require you to select a subsystem first, and I have no clue whatsoever as to where I should report this.

thanks for any help/suggestions,
Steven

Last edited by steabert (2014-08-16 16:00:12)

Lone_Wolf · 2014-08-14 10:24:25

Booting an archlinux system goes through several stages, and you don't give clues at which stage your boot fails.

Please read https://wiki.archlinux.org/index.php/Boot_process and tell us where it goes wrong.

steabert · 2014-08-14 12:52:19

I boot under UEFI, gummiboot loads, I select the kernel, then I get a blank screen.
Only with the kernel parameters 'debug' and 'earlyprintk=efi' do I get any lines printed to the screen.
So, I'm guessing it hangs during the 'Kernel' stage of the page you referred me to?

The exact place where it hangs is difficult to tell, as at some point the lines that get printed to the screen
are not printed any more on a new line, but over the existing last line. The last part of the screen output reads:

...
[0.000000]  [mem 0x100000000-0x1fbffffff] page 2M
[0.000000] init_memory_mapping: [mem 0x200000000-0x21bffffff]
[0.000000]  [mem 0x200000000-0x21bffffff] page 2M
[0.000000] log_buf_len: 16777216
[0.000000] early log buf free: 479196(91%)
[0.000000] RAMDISK: [mem 0x7fcf7000-0x7fffffff]
[0.000000] ACPI: RSDP 00000000db67ef98 00024 (v02 DELL  )
[0.000000] ACPI: XSDT 00000000db67de18 00064 (v01 DELL    E2      06222004 MSFT 00010013)
...(illegible overlapping lines)...

Pse · 2014-08-14 13:47:30

Have you tried other bootloaders?

steabert · 2014-08-14 13:49:54

Pse wrote:

Have you tried other bootloaders?

I tried grub as well, with the same results: boot loaders loads fine, kernel stuck at the same place.

alphaniner · 2014-08-14 14:03:59

Some basic h/w info would be useful. Maybe something in here could be relevant, but who would know?

steabert · 2014-08-14 14:07:57

I'm getting this problem on a laptop, a Dell Latitude e4310.

Sorry for having been so terse, I already searched several specific places (Arch forum, Arch bug tracker, kernel bug tracker) for the problem and did not find anything related to this specific hardware and the problem I'm encountering.

Lone_Wolf · 2014-08-16 10:12:41

Steabert,

Have you tried booting into a live environment like that of the arch install isos ?

If not, please try multiple releases of the arch install isos.

While they do use kernels that don't work on your installed system, they use other methods to boot the system that tend to be very reliable
(though difficult to setup yourself i think).

steabert · 2014-08-16 11:13:37

Lone_Wolf wrote:

Have you tried booting into a live environment like that of the arch install isos ?
If not, please try multiple releases of the arch install isos.

Thanks for this great suggestion, didn't think of it myself.

I tried the most recent arch iso (from first of august) and that one fails to boot too (linux 3.15.7).
However, if I choose 'Legacy' boot for the USB device from the boot menu (I guess this is BIOS) instead of UEFI, the kernel boots fine.
So, I guess this must still be some kind of issue with UEFI, either with the firmware or the kernel?

Lone_Wolf · 2014-08-16 11:48:12

Yes, it looks more and more like an UEFI related issue.

Most UEFI firmwares have an option to boot with a Compatibility Support Module .
The CSM tries to make OSes see standard BIOS devices instead of UEFI devices.

Is your system running latest uefi firmware ?

steabert · 2014-08-16 15:58:51

Lone_Wolf wrote:

Is your system running latest uefi firmware ?

Turns out that was the problem, I was way behind with firmware updates. Thanks for the swift analysis which pointed me to the correct solution! I feel a bit stupid that I didn't check for more recent firmware updates earlier.

In case anyone else runs into this for a Dell Latitude E4310 and want to update the firmware, here is a short summary of what I did:

I used FreeDOS to update the firmware based on instructions here, for images that are too large for a floppy.
Downloaded the A07, A09, and A14 updated from here.
Since the updates are about 5M in size, I could only copy one of them onto the image, so I proceded in steps with one update at a time.

Arch Linux

#1 2014-08-13 19:46:07

[SOLVED] Kernel not booting, could use help with debugging

#2 2014-08-14 10:24:25

Re: [SOLVED] Kernel not booting, could use help with debugging

#3 2014-08-14 12:52:19

Re: [SOLVED] Kernel not booting, could use help with debugging

#4 2014-08-14 13:47:30

Re: [SOLVED] Kernel not booting, could use help with debugging

#5 2014-08-14 13:49:54

Re: [SOLVED] Kernel not booting, could use help with debugging

#6 2014-08-14 14:03:59

Re: [SOLVED] Kernel not booting, could use help with debugging

#7 2014-08-14 14:07:57

Re: [SOLVED] Kernel not booting, could use help with debugging

#8 2014-08-16 10:12:41

Re: [SOLVED] Kernel not booting, could use help with debugging

#9 2014-08-16 11:13:37

Re: [SOLVED] Kernel not booting, could use help with debugging

#10 2014-08-16 11:48:12

Re: [SOLVED] Kernel not booting, could use help with debugging

#11 2014-08-16 15:58:51

Re: [SOLVED] Kernel not booting, could use help with debugging

Board footer