Re: [PATCH 04/13] vmcore: introduce types for objects copied in 2nd kernel

[Vivek Goyal <vgoyal@redhat.com>]

On Thu, Feb 14, 2013 at 07:12:05PM +0900, HATAYAMA Daisuke wrote:
> Some parts of old memory need to be copied in buffers on 2nd kernel to
> be remapped to user-space. To distinguish objects in the buffer on 2nd
> kernel and the ones on old memory, enum vmcore_type is introduced: the
> object in the buffer on 2nd kernel has VMCORE_2ND_KERNEL type, and the
> one on old memory has VMCORE_OLD_MEMORY type.
> 
> Signed-off-by: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>
> ---
> 
>  include/linux/proc_fs.h |   11 ++++++++++-
>  1 files changed, 10 insertions(+), 1 deletions(-)
> 
> diff --git a/include/linux/proc_fs.h b/include/linux/proc_fs.h
> index 32676b3..4b153ed 100644
> --- a/include/linux/proc_fs.h
> +++ b/include/linux/proc_fs.h
> @@ -97,11 +97,20 @@ struct kcore_list {
>  	int type;
>  };
>  
> +enum vmcore_type {
> +	VMCORE_OLD_MEMORY,
> +	VMCORE_2ND_KERNEL,

This VMCORE_2ND_KERNEL tag looks bad.

How about introducing a "unsigned int flag" element in "struct vmcore"
and set the flag MEM_TYPE_OLDMEM for any contents which come from oldmem.

If MEM_TYPE_OLDMEM is not set, it is assumed that contents are to be
fetched from current kernel using pointer vmcore->buf.

BTW, which elements you need to copy from first kernel?

> +};
> +
>  struct vmcore {
>  	struct list_head list;
> -	unsigned long long paddr;
> +	union {
> +		unsigned long long paddr;
> +		char *buf;
> +	};
>  	unsigned long long size;
>  	loff_t offset;
> +	enum vmcore_type type;
>  };
>  

Thanks
Vivek

>  #ifdef CONFIG_PROC_FS

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec