Bug #62

Linux kernel versions greater than 4.6 fail to boot when used as payload

Added by Timothy Pearson over 2 years ago. Updated over 1 year ago.

Status:ResolvedStart date:08/02/2016
Priority:NormalDue date:
Assignee:ron minnich% Done:

0%

Category:-
Target version:-

Description

Linux kernel versions greater than 4.6 fail to boot as a payload on both qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before decompression); gdb on a q35 qemu system indicates that at some point an lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first failing commit. The prior commit (6112a300c9e41993cc0dc56ac393743d28381284) works fine.

History

#1 Updated by Trammell Hudson over 2 years ago

The crash was traced to linux-4.7/arch/x86/boot/compressed/head_64.S use of the boot param init_size variable to determine where to decompress the kernel image: https://www.coreboot.org/pipermail/coreboot/2016-August/081891.html

#2 Updated by Trammell Hudson over 2 years ago

The just released Linux 4.9 kernel does not require any patches to boot as a payload under coreboot.

#3 Updated by ron minnich over 2 years ago

No, this would still not have worked save that I upstreamed a fix to
coreboot a few months ago and forgot to close this bug. :-)

How do we put a bug # in a commit and have it close the bug? is that
possible?

On Mon, Dec 12, 2016 at 7:53 AM coreboot@fe80.eu wrote:

Issue #62 has been updated by Trammell Hudson.

The just released Linux 4.9 kernel does not require any patches to boot as
a payload under coreboot.


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-215

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#4 Updated by ron minnich over 2 years ago

No, this would still not have worked save that I upstreamed a fix to
coreboot a few months ago and forgot to close this bug. :-)

How do we put a bug # in a commit and have it close the bug? is that
possible?

On Mon, Dec 12, 2016 at 7:53 AM coreboot@fe80.eu wrote:

Issue #62 has been updated by Trammell Hudson.

The just released Linux 4.9 kernel does not require any patches to boot as
a payload under coreboot.


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-215

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#5 Updated by ron minnich over 2 years ago

No, this would still not have worked save that I upstreamed a fix to
coreboot a few months ago and forgot to close this bug. :-)

How do we put a bug # in a commit and have it close the bug? is that
possible?

On Mon, Dec 12, 2016 at 7:53 AM coreboot@fe80.eu wrote:

Issue #62 has been updated by Trammell Hudson.

The just released Linux 4.9 kernel does not require any patches to boot as
a payload under coreboot.


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-215

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#6 Updated by ron minnich over 2 years ago

No, this would still not have worked save that I upstreamed a fix to
coreboot a few months ago and forgot to close this bug. :-)

How do we put a bug # in a commit and have it close the bug? is that
possible?

On Mon, Dec 12, 2016 at 7:53 AM coreboot@fe80.eu wrote:

Issue #62 has been updated by Trammell Hudson.

The just released Linux 4.9 kernel does not require any patches to boot as
a payload under coreboot.


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-215

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#7 Updated by ron minnich over 2 years ago

No, this would still not have worked save that I upstreamed a fix to
coreboot a few months ago and forgot to close this bug. :-)

How do we put a bug # in a commit and have it close the bug? is that
possible?

On Mon, Dec 12, 2016 at 7:53 AM coreboot@fe80.eu wrote:

Issue #62 has been updated by Trammell Hudson.

The just released Linux 4.9 kernel does not require any patches to boot as
a payload under coreboot.


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-215

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#8 Updated by Nico Huber almost 2 years ago

I suppose this is fixed? Please close the bug then.

#9 Updated by Timothy Pearson over 1 year ago

Would like to close, referencing the GIT hash of the commit that fixed the issue. Do you have the GIT hash available?

#10 Updated by ron minnich over 1 year ago

I'll check tomorrow
On Thu, Jul 13, 2017 at 8:17 PM coreboot@fe80.eu wrote:

Issue #62 has been updated by Timothy Pearson.

Would like to close, referencing the GIT hash of the commit that fixed the
issue. Do you have the GIT hash available?


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-347

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#11 Updated by Paul Menzel over 1 year ago

Ron, please strip the quote of, if you are top posting. Otherwise the HTML page for the issue it’s quite crowded(?)/confusing. No idea, if Redmine can be fixed.

#12 Updated by Paul Menzel over 1 year ago

Ron, do you mean commit fa74e4705d (cbfstool: set init_size for linux payloads.) [1]?

[1] https://review.coreboot.org/16781

#13 Updated by ron minnich over 1 year ago

yes, paul, thanks, that's it, fa74e4705dbb5ecf5ba466fd6a9bc309f7e3e8bf

On Fri, Jul 14, 2017 at 1:48 AM coreboot@fe80.eu wrote:

Issue #62 has been updated by Paul Menzel.

Ron, do you mean commit fa74e4705d (cbfstool: set init_size for linux
payloads.) [1]?

[1] https://review.coreboot.org/16781


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-350

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee:
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#14 Updated by Stefan Reinauer over 1 year ago

  • Assignee set to ron minnich

Didn't we fix this one, Ron?

#15 Updated by ron minnich over 1 year ago

Yes but I can't find the commit
On Mon, Aug 21, 2017 at 6:03 PM coreboot@fe80.eu wrote:

Issue #62 has been updated by Stefan Reinauer.

Assignee set to ron minnich

Didn't we fix this one, Ron?


Bug #62: Linux kernel versions greater than 4.6 fail to boot when used as
payload
https://ticket.coreboot.org/issues/62#change-396

  • Author: Timothy Pearson
  • Status: New
  • Priority: Normal
  • Assignee: ron minnich
  • Category:

* Target version:

Linux kernel versions greater than 4.6 fail to boot as a payload on both
qemu and real hardware (KGPE-D16).

The kernel locks up before early printk starts (likely in or before
decompression); gdb on a q35 qemu system indicates that at some point an
lret is attempted to uninitialized memory at 0x40000002.

Kernel commit ffd565e315aa8211365bb6864000f04e7b277f59 is the first
failing commit. The prior commit
(6112a300c9e41993cc0dc56ac393743d28381284) works fine.

--
You have received this notification because you have either subscribed to
it, or are involved in it.
To change your notification preferences, please click here:
https://ticket.coreboot.org/my/account

#16 Updated by Paul Menzel over 1 year ago

  • Status changed from New to Resolved

I noted the commit hash in comment #12 [1].

Ron, please trim your reply until Redmine is fixed. Otherwise the Web view is not usable.

[1] https://ticket.coreboot.org/issues/62#note-12

Also available in: Atom PDF