2008-10-20 05:04:35

by Ken'ichi Ohmichi

[permalink] [raw]
Subject: [PATCH] kdump: Fix invalid access on i386 sparsemem.


Hi,

Since linux-2.6.27, kdump has failed on i386 sparsemem kernel.
1st-kernel gets a panic just before switching to 2nd-kernel.

The cause is that a kernel accesses invalid mem_section by
page_to_pfn(image->swap_page) at machine_kexec().
image->swap_page is allocated if kexec for hibernation, but
it is not allocated if kdump. So if kdump, a kernel should
not access the mem_section corresponding to image->swap_page.
The attached patch fixes this invalid access.


Thanks
Ken'ichi Ohmichi

Signed-off-by: Ken'ichi Ohmichi <[email protected]>
---
diff -rpuN a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c
--- a/arch/x86/kernel/machine_kexec_32.c 2008-10-10 07:13:53.000000000 +0900
+++ b/arch/x86/kernel/machine_kexec_32.c 2008-10-17 20:38:37.000000000 +0900
@@ -162,7 +162,10 @@ void machine_kexec(struct kimage *image)
page_list[VA_PTE_0] = (unsigned long)kexec_pte0;
page_list[PA_PTE_1] = __pa(kexec_pte1);
page_list[VA_PTE_1] = (unsigned long)kexec_pte1;
- page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) << PAGE_SHIFT);
+
+ if (image->type == KEXEC_TYPE_DEFAULT)
+ page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page)
+ << PAGE_SHIFT);

/* The segment registers are funny things, they have both a
* visible and an invisible part. Whenever the visible part is
_


2008-10-20 07:15:11

by Ingo Molnar

[permalink] [raw]
Subject: Re: [PATCH] kdump: Fix invalid access on i386 sparsemem.


* Ken'ichi Ohmichi <[email protected]> wrote:

>
> Hi,
>
> Since linux-2.6.27, kdump has failed on i386 sparsemem kernel.
> 1st-kernel gets a panic just before switching to 2nd-kernel.
>
> The cause is that a kernel accesses invalid mem_section by
> page_to_pfn(image->swap_page) at machine_kexec().
> image->swap_page is allocated if kexec for hibernation, but
> it is not allocated if kdump. So if kdump, a kernel should
> not access the mem_section corresponding to image->swap_page.
> The attached patch fixes this invalid access.

applied to tip/x86/urgent, thanks!

Ingo