kdump, ia64: always reserve elfcore header memory in crash kernel

kdump, ia64: always reserve elfcore header memory in crash kernel

[Simon Horman]

elfcore header memory needs to be reserved in a crash kernel.
This means that the relevant code should be protected
by CONFIG_CRASH_DUMP rather than CONFIG_PROC_VMCORE.

Signed-off-by: Simon Horman <horms@verge.net.au>

--- 
Andrew, this patch fixes bug in the (unlikely) case where
an ia64 crashdump kernel does not have CONFIG_PROC_FS set.
I think it is worth including in 2.6.27. But breakage cases are
likely to be minimal to non-existent, so I am comfortable
with post 2.6.27 too.

This patch should be appended to the series,
"is_kdump_kernel() cleanup and related patches".

http://lists.infradead.org/pipermail/kexec/2008-July/002270.html

Index: linux-2.6/arch/ia64/kernel/efi.c
===================================================================
--- linux-2.6.orig/arch/ia64/kernel/efi.c	2008-08-05 18:31:13.000000000 +1000
+++ linux-2.6/arch/ia64/kernel/efi.c	2008-08-05 18:31:21.000000000 +1000
@@ -1334,7 +1334,7 @@ kdump_find_rsvd_region (unsigned long si
 }
 #endif
 
-#ifdef CONFIG_PROC_VMCORE
+#ifdef CONFIG_CRASH_DUMP
 /* locate the size find a the descriptor at a certain address */
 unsigned long __init
 vmcore_find_descriptor_size (unsigned long address)
Index: linux-2.6/arch/ia64/kernel/setup.c
===================================================================
--- linux-2.6.orig/arch/ia64/kernel/setup.c	2008-08-05 18:31:20.000000000 +1000
+++ linux-2.6/arch/ia64/kernel/setup.c	2008-08-05 18:31:21.000000000 +1000
@@ -352,7 +352,7 @@ reserve_memory (void)
 	}
 #endif
 
-#ifdef CONFIG_PROC_VMCORE
+#ifdef CONFIG_CRASH_KERNEL
 	if (reserve_elfcorehdr(&rsvd_region[n].start,
 			       &rsvd_region[n].end) == 0)
 		n++;
@@ -496,9 +496,7 @@ static int __init parse_elfcorehdr(char 
 	return 0;
 }
 early_param("elfcorehdr", parse_elfcorehdr);
-#endif
 
-#ifdef CONFIG_PROC_VMCORE
 int __init reserve_elfcorehdr(unsigned long *start, unsigned long *end)
 {
 	unsigned long length;

Re: kdump, ia64: always reserve elfcore header memory in crash kernel

[Andrew Morton]

On Tue, 5 Aug 2008 19:48:27 +1000 Simon Horman <horms@verge.net.au> wrote:

> elfcore header memory needs to be reserved in a crash kernel.
> This means that the relevant code should be protected
> by CONFIG_CRASH_DUMP rather than CONFIG_PROC_VMCORE.
> 
> Signed-off-by: Simon Horman <horms@verge.net.au>
> 
> --- 
> Andrew, this patch fixes bug in the (unlikely) case where
> an ia64 crashdump kernel does not have CONFIG_PROC_FS set.
> I think it is worth including in 2.6.27. But breakage cases are
> likely to be minimal to non-existent, so I am comfortable
> with post 2.6.27 too.
> 
> This patch should be appended to the series,
> "is_kdump_kernel() cleanup and related patches".

hm, that means that we would need to apply a moderate-size seven patch
series to fix one little bug.

I think that if we want to fix this in 2.6.27 then the basic single
patch would be preferable.  Or we leave it until 2.6.28/

Re: kdump, ia64: always reserve elfcore header memory in crash kernel

[Simon Horman]

On Tue, Aug 05, 2008 at 02:54:42AM -0700, Andrew Morton wrote:
> On Tue, 5 Aug 2008 19:48:27 +1000 Simon Horman <horms@verge.net.au> wrote:
> 
> > elfcore header memory needs to be reserved in a crash kernel.
> > This means that the relevant code should be protected
> > by CONFIG_CRASH_DUMP rather than CONFIG_PROC_VMCORE.
> > 
> > Signed-off-by: Simon Horman <horms@verge.net.au>
> > 
> > --- 
> > Andrew, this patch fixes bug in the (unlikely) case where
> > an ia64 crashdump kernel does not have CONFIG_PROC_FS set.
> > I think it is worth including in 2.6.27. But breakage cases are
> > likely to be minimal to non-existent, so I am comfortable
> > with post 2.6.27 too.
> > 
> > This patch should be appended to the series,
> > "is_kdump_kernel() cleanup and related patches".
> 
> hm, that means that we would need to apply a moderate-size seven patch
> series to fix one little bug.
> 
> I think that if we want to fix this in 2.6.27 then the basic single
> patch would be preferable.  Or we leave it until 2.6.28/

I think it would be better to wait until 2.6.28 than mess around
with a minimal and then a proper fix. I doubt that anyone is hitting
this.

Re: kdump, ia64: always reserve elfcore header memory in crash kernel

[Vivek Goyal]

On Tue, Aug 05, 2008 at 07:48:27PM +1000, Simon Horman wrote:
> elfcore header memory needs to be reserved in a crash kernel.
> This means that the relevant code should be protected
> by CONFIG_CRASH_DUMP rather than CONFIG_PROC_VMCORE.
> 
> Signed-off-by: Simon Horman <horms@verge.net.au>
> 
> --- 
> Andrew, this patch fixes bug in the (unlikely) case where
> an ia64 crashdump kernel does not have CONFIG_PROC_FS set.
> I think it is worth including in 2.6.27. But breakage cases are
> likely to be minimal to non-existent, so I am comfortable
> with post 2.6.27 too.
> 
> This patch should be appended to the series,
> "is_kdump_kernel() cleanup and related patches".
> 
> http://lists.infradead.org/pipermail/kexec/2008-July/002270.html
> 
> Index: linux-2.6/arch/ia64/kernel/efi.c
> ===================================================================
> --- linux-2.6.orig/arch/ia64/kernel/efi.c	2008-08-05 18:31:13.000000000 +1000
> +++ linux-2.6/arch/ia64/kernel/efi.c	2008-08-05 18:31:21.000000000 +1000
> @@ -1334,7 +1334,7 @@ kdump_find_rsvd_region (unsigned long si
>  }
>  #endif
>  
> -#ifdef CONFIG_PROC_VMCORE
> +#ifdef CONFIG_CRASH_DUMP
>  /* locate the size find a the descriptor at a certain address */
>  unsigned long __init
>  vmcore_find_descriptor_size (unsigned long address)
> Index: linux-2.6/arch/ia64/kernel/setup.c
> ===================================================================
> --- linux-2.6.orig/arch/ia64/kernel/setup.c	2008-08-05 18:31:20.000000000 +1000
> +++ linux-2.6/arch/ia64/kernel/setup.c	2008-08-05 18:31:21.000000000 +1000
> @@ -352,7 +352,7 @@ reserve_memory (void)
>  	}
>  #endif
>  
> -#ifdef CONFIG_PROC_VMCORE
> +#ifdef CONFIG_CRASH_KERNEL
>  	if (reserve_elfcorehdr(&rsvd_region[n].start,
>  			       &rsvd_region[n].end) == 0)
>  		n++;

Hi Simon,

Thanks for the patch. One minor nit. I think compilation will fail
in the case of CONFIG_PROC_VMCORE=n. Because reserve_elfcorehdr will
try to use vmcore_unusabe() which is not available now?

Thanks
Vivek

Re: kdump, ia64: always reserve elfcore header memory in crash kernel

[Simon Horman]

On Tue, Aug 05, 2008 at 08:56:34AM -0400, Vivek Goyal wrote:
> On Tue, Aug 05, 2008 at 07:48:27PM +1000, Simon Horman wrote:
> > elfcore header memory needs to be reserved in a crash kernel.
> > This means that the relevant code should be protected
> > by CONFIG_CRASH_DUMP rather than CONFIG_PROC_VMCORE.
> > 
> > Signed-off-by: Simon Horman <horms@verge.net.au>
> > 
> > --- 
> > Andrew, this patch fixes bug in the (unlikely) case where
> > an ia64 crashdump kernel does not have CONFIG_PROC_FS set.
> > I think it is worth including in 2.6.27. But breakage cases are
> > likely to be minimal to non-existent, so I am comfortable
> > with post 2.6.27 too.
> > 
> > This patch should be appended to the series,
> > "is_kdump_kernel() cleanup and related patches".
> > 
> > http://lists.infradead.org/pipermail/kexec/2008-July/002270.html
> > 
> > Index: linux-2.6/arch/ia64/kernel/efi.c
> > ===================================================================
> > --- linux-2.6.orig/arch/ia64/kernel/efi.c	2008-08-05 18:31:13.000000000 +1000
> > +++ linux-2.6/arch/ia64/kernel/efi.c	2008-08-05 18:31:21.000000000 +1000
> > @@ -1334,7 +1334,7 @@ kdump_find_rsvd_region (unsigned long si
> >  }
> >  #endif
> >  
> > -#ifdef CONFIG_PROC_VMCORE
> > +#ifdef CONFIG_CRASH_DUMP
> >  /* locate the size find a the descriptor at a certain address */
> >  unsigned long __init
> >  vmcore_find_descriptor_size (unsigned long address)
> > Index: linux-2.6/arch/ia64/kernel/setup.c
> > ===================================================================
> > --- linux-2.6.orig/arch/ia64/kernel/setup.c	2008-08-05 18:31:20.000000000 +1000
> > +++ linux-2.6/arch/ia64/kernel/setup.c	2008-08-05 18:31:21.000000000 +1000
> > @@ -352,7 +352,7 @@ reserve_memory (void)
> >  	}
> >  #endif
> >  
> > -#ifdef CONFIG_PROC_VMCORE
> > +#ifdef CONFIG_CRASH_KERNEL
> >  	if (reserve_elfcorehdr(&rsvd_region[n].start,
> >  			       &rsvd_region[n].end) == 0)
> >  		n++;
> 
> Hi Simon,
> 
> Thanks for the patch. One minor nit. I think compilation will fail
> in the case of CONFIG_PROC_VMCORE=n. Because reserve_elfcorehdr will
> try to use vmcore_unusabe() which is not available now?

Yes, sorry. I forgot to send a revised version of the patch that
creates vmcore_unusabe() so that it is present if CONFIG_PROC_VMCORE=n.
I will do so.