Re: [Fastboot] Fw: Re: kexec?

[Vivek Goyal <vgoyal@in.ibm.com>]

On Tue, May 10, 2005 at 07:32:25PM -0700, Andrew Morton wrote:
> 
> We have some kexec/kdump testers over on
> linux-kernel.  Results: not good.
> 
> 
> 
> Begin forwarded message:
> 
> Date: Tue, 10 May 2005 21:11:31 +0800
> From: Coywolf Qi Hunt <coywolf@gmail.com>
> To: Borislav Petkov <petkov@uni-muenster.de>
> Cc: "Randy.Dunlap" <rddunlap@osdl.org>, Ralf Hildebrandt <Ralf.Hildebrandt@charite.de>, linux-kernel@vger.kernel.org
> Subject: Re: kexec?
> 
> 
> On 5/10/05, Borislav Petkov <petkov@uni-muenster.de> wrote:
> > I've been doing some kexec tests (as described in Documentation/kdump.txt) too
> > but can't get to load the image and get similar error messages. Let me know
> > if you need more info about the hardware. The first_kernel was booted with
> > "crashkernel=64M@16M" and the 16M value was configured into the second during
> > kconfig in "Physical address where the kernel is loaded" as 0x1000000.
> > 
> > [root@zmei]: kexec -p vmlinux --args-linux --append="root=/dev/hda1 maxcpus=1
> > init 1"
> 
>  kexec-tools-1.101 loads for me, but if cmdline is used, it hangs up
> after "Starting new kernel"

maxcpus=1 is the culprit. Even through bios/grub kernel does not boot with 
option maxcpus=1. It is a known issue with got notieced 2.6.12-rc2-mm1 onwards. 
So build second kernel as uniprocessor kernel and don't specify maxcpus=1 and 
test it out. It should work.

Thanks
Vivek