Skip to contentSkip to conversation

kexec 2.0.4-rc1 won't work on an ARM kernel without device-tree

DISCUSSION Sven Neumann 1 participant 2 messages Nov 24, 2012 – Nov 28, 2012

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi,

I am trying to use kexec-2.0.4-rc1 on an ARM device running Linux 3.0.4. This old kernel does obviously not have device-tree support. Here's the error message I get:

# uname -a
Linux speaker 3.0.4 #1 PREEMPT Fri Nov 23 09:15:39 CET 2012 armv5tel GNU/Linux
# kexec --append="console=ttyS0,115200 root=ubi0:RootFS rootfstype=ubifs rw ubi.mtd=3" /boot/uImage --force
unrecoverable error: could not scan "/proc/device-tree/": No such file or directory

Looks like the release candidate depends on the very latest kernel features and won't work on older kernels? Is that intentional?

Regards,
Sven

Signature · 8 lines
Sven Neumann <[email protected]>
Raumfeld by Teufel


_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec
#1 lore raw

Hi,

Quoted 7 lines · Sven Neumann

I am trying to use kexec-2.0.4-rc1 on an ARM device running Linux 3.0.4. This old kernel does obviously not have device-tree support. Here's the error message I get:

# uname -a
Linux speaker 3.0.4 #1 PREEMPT Fri Nov 23 09:15:39 CET 2012 armv5tel GNU/Linux
Quoted 6 lines
# kexec --append="console=ttyS0,115200 root=ubi0:RootFS rootfstype=ubifs rw ubi.mtd=3" /boot/uImage --force
unrecoverable error: could not scan "/proc/device-tree/": No such file or directory

Looks like the release candidate depends on the very latest kernel features and won't work on older kernels? Is that intentional?

OK, in the meantime I figured out that I can use kexec-2.0.4-rc1 if I use the --atags command-line parameter. But unless I missed something, this change in behavior is nowhere documented and thus should be considered a severe regression. I am somewhat puzzled that no one here even cared about my email.

IMO the behavior should be changed so that kexec, if called on ARM without the --dtb or --atags parameters, checks if the running kernel provides /proc/device-tree. If it doesn't, then this should not be a fatal error, but it should assume --atags. That would restore the old behavior for kernels that don't use device-tree.

Regards,
Sven

Signature · 20 lines
Sven Neumann
Head of RAUMFELD Software Development

Lautsprecher Teufel GmbH
Gewerbehof Bülowbogen - Aufgang D1
Bülowstr. 66 - 10783 Berlin - Germany

Registernummer HRB 20271 beim Amtsgericht Berlin-Charlottenburg
Geschäftsführer: Thomas Stiegler, Edgar van Velzen

Tel. +49 (0)30 - 300 930 - 0, Direkt -153
[email protected]  

www.teufel.de  www.teufelaudio.com  http://blog.teufel.de


_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec