[Cbe-oss-dev] [PATCH] fix reboot fail on kexec for CBE

[André Detsch]

Akira Tsukamoto wrote:
>> Thanks for the Information! Is this patch actually sufficient in that
>> it enables kexec to work as expected, or did you need additional changes
>> beyond it?
> 
> In our usage of our cell blades does not require additional changes to 
> boot second kernel after kexec.

I've tried your patch (applying it on 2.6.20-arnd3), but second kernel 
hangs after the following message:
NET: Registered protocol family 16

Command was:
# kexec -p ./vmlinux_kdump \
          --initrd=/boot/vmlinux-s --args-linux \
          --append="init 1 noirqdistrib console=hvc0 xmon=off"
(as suggested at the other thread)

Previously, using other kexec arguments:
# kexec -p vmlinux_kdump --initrd=/boot/vmlinux-s \
    --append="console=hvc0 root=/dev/hda3"

results varied: sometimes the kernel got a panic, sometimes it hung at 
the end of the second kernel load, with an "unable to open initial 
console" message. At one try, however, the second boot went fine, and I 
got to use the shell. Trying to copy the /proc/vmcore file, however, 
lead to a "cp: reading `vmcore': Invalid argument" message. About 500MB 
of data were copied, though.

You are using it on a QS20, right? What args for the first kernel? What 
FW version (e.g. /proc/device-tree/openprom/ibm,fw-temp-bank)? Are you 
able to copy the /proc/vmcore file?

Thanks!

--
Andre Detsch