[Xen-devel] Re: [PATCH][dom0] Add a module exit function to BLKBACK

[Keir Fraser <Keir.Fraser@xxxxxxxxxxxx>]

Looks to completely ignore that you may have blkfronts connected and I/Os still in flight. This will only work for a totally quiescent blkback.

 -- Keir

On 22/1/08 05:46, "Shan, Haitao" <haitao.shan@xxxxxxxxx> wrote:

> Hi, Keir, 
>
> This little patch will add a module exit function for block backend. Can you have a look? 
>
> Signed-off-by: Dugger Donald D <donald.d.dugger@xxxxxxxxx> 
> Signed-off-by: Haitao Shan        <haitao.shan@xxxxxxxxx> 
> diff -r 854690f06ed9 drivers/xen/blkback/blkback.c 
> --- a/drivers/xen/blkback/blkback.c     Tue Jan 15 15:59:31 2008 +0000 
> +++ b/drivers/xen/blkback/blkback.c     Mon Jan 21 22:37:54 2008 +0800 
> @@ -165,6 +165,7 @@ static void fast_flush_area(pending_req_ 
>  { 
>         struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST]; 
>         unsigned int i, invcount = 0; 
> +       unsigned long pfn; 
>         grant_handle_t handle; 
>         int ret; 
>  
> @@ -176,6 +177,8 @@ static void fast_flush_area(pending_req_ 
>                                     handle); 
>                 pending_handle(req, i) = BLKBACK_INVALID_HANDLE; 
>                 invcount++; 
> +               pfn = page_to_pfn(pending_pages[vaddr_pagenr(req, i)]); 
> +               set_phys_to_machine(pfn, INVALID_P2M_ENTRY); 
>         } 
>  
>         ret = HYPERVISOR_grant_table_op( 
> @@ -612,6 +615,20 @@ static int __init blkif_init(void) 
>         return -ENOMEM; 
>  } 
>  
> +static void __exit blkif_exit(void) 
> +{ 
> +       int mmap_pages; 
> + 
> +       mmap_pages = blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST; 
> + 
> +       kfree(pending_reqs); 
> +       kfree(pending_grant_handles); 
> +       free_empty_pages_and_pagevec(pending_pages, mmap_pages); 
> + 
> +       return; 
> +} 
> + 
>  module_init(blkif_init); 
> +module_exit(blkif_exit); 
>  
>  MODULE_LICENSE("Dual BSD/GPL"); 
>  <<exit.patch>> 
>
> Best Regards 
> Haitao Shan 




_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel