IA64: a missing patch: fix-paddr-vmcore-ELF

[nanhai.zou at intel.com (Zou, Nanhai)]

> -----Original Message-----
> From: Jay Lan [mailto:jlan at sgi.com]
> Sent: 2007Äê2ÔÂ15ÈÕ 7:49
> To: Horms
> Cc: Magnus Damm; Zou, Nanhai; fastboot
> Subject: Re: IA64: a missing patch: fix-paddr-vmcore-ELF
> 
> Jay Lan wrote:
> > Jay Lan wrote:
> >> Jay Lan wrote:
> >>> Horms wrote:
> >>>> On Tue, Feb 13, 2007 at 07:07:14PM -0800, Jay Lan wrote:
> >>>>> I found this patch from Nan-hai was missing in your git tree.
> >>>>> I do not have the url of his original posting, but here is the
> >>>>> content. :)
> >>>> Thanks, I have applied this.
> >>>>
> >>>> This is actually the second time its been applied, unfortunately it got
> >>>> reverted as part of the crash_create_elf64_headers() patch,
> >>>> f038d30bff8510a1df1e72af08db1766581d1f2c.
> >>> I found problem in the crash_create_elf64_headers() patch more than
> >>> this one. For example, when calculating the size, the nr_cpus should not
> >>> be a long. It is an int. It will cause problem only when initializing
> >>> the crash utility.
> >>>
> >>> However, even after reverting f038d30bff8510a1df1e72af08db1766581d1f2c
> >>> i still had problem in booting up a crashdump kernel on SN platform.
> >>> I saw many many "Bad state" and "Bad pte" errors, probably when trying
> >>> to create the vmcore.
> >> BTW, i did my testing with 2.6.20 kernel and the kexec-tools-testing
> >> tot tree as of early yesterday.
> >
> > :(
> > I just tried 2.6.20 with a "working" kexec (the rhel5 kexec that
> > worked on rhel5 kernel), but it showed
> > put_kernel_page: page at 0xe000003018b44000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b44000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b3c000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b3c000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b3c000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b3c000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b3c000 not in reserved memory
> > put_kernel_page: page at 0xe000003018b3c000 not in reserved memory
> >
> > and followed by tons of "Bad page state" and "Bad pte" errors.
> 
> It took a long long time, but the kdump kernel eventually came up.
> 
> What essential IA64 patches do i need (besides the zero-size-vmcore)
> for kexec/kdump to work in 2.6.20?
> 
> Thanks,
>  - jay


I don't know..., I will try to see if I can reproduce it here.
My first guess is some pages are missed from initialization in memmap_init->virtual_memmap_init->memmap_init_zone.

Thanks
Zou Nan hai