Re: [PATCH] crash_dump: export is_kdump_kernel to modules, consolidate elfcorehdr_addr, setup_elfcorehdr and saved_max_pfn

[Olaf Hering <olaf@aepfle.de>]

On Wed, Jan 26, Vivek Goyal wrote:

> Olaf, 
> 
> I think your HVM guest has booted into kdump kernel after the crash and
> now you want Xen PV driver to reset the connection with Dom0.

Yes, thats the case.

> If that's the case, I think instead of exporting is_kdump_kernel(), better
> make use of symbol "reset_devices". Kexec-tools passes this kernel command
> line option to kdump kernel which tells the kernel that it is booting
> in an unreliable environment and device drivers should reset the
> underlying devices and not rely on BIOS to have put devices in a known
> good state. So if reset_devices=1, then Xen PV driver should reset the
> underlying device hence the connection with Dom0.

I will use this variable, its passed unconditionally from the boot.kdump
script.


> Apart from that, moving the code from arch specific section to generic
> code helps. So that bit you can retain and get rid of export of
> is_kdump_kernel() and elfcorehdr_addr.


Andrew, do you want a new patch with the export of elfcorehdr_addr
removed, or can you edit my patch and remove that one line?

Olaf