Re: [Xen-devel][PATCH] Fix the boot failure of xenU after destroying aHVM guest

[Keir Fraser <keir.fraser@xxxxxxxxxxxxx>]

It's only polite for a subsystem which trashes page_info to reinstate it
when it is done. We'll work out a patch for the p2m code.

 -- Keir

On 16/7/08 01:56, "Xu, Dongxiao" <dongxiao.xu@xxxxxxxxx> wrote:

> Hi, Keir, 
>     I saw in C/S 17899, you put the re-initialize code of the page lock in
> sh_set_allocation() instead of free_heap_pages(). In this condition, I think
> we should also re-initialize the page lock while freeing the shadow p2m page
> and maybe some other places. Because when alloc the shadow p2m page, it will
> use function sh_alloc_p2m_pages()->shadow_alloc() to initialize the page as
> shadow_page_info, so the re-initialize when freeing shadow p2m is necessary.
>     So I think if only put the page lock re-initialize code in
> free_heap_pages(), it is the most safe. And it could fix the bug 1287.
> 
> 1. Sometimes XenU guest cannot boot up on 32e host.
> http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=1287
> 
> Thanks
> -- Dongxiao
> 
> -----Original Message-----
> From: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
> [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of Xu, Dongxiao
> Sent: 2008年6月26日 14:12
> To: Keir Fraser; xen-devel@xxxxxxxxxxxxxxxxxxx
> Cc: jbeulich@xxxxxxxxxx
> Subject: [Xen-devel][PATCH] Fix the boot failure of xenU after destroying aHVM
> guest
> 
> Fix the boot failure of xenU after destroying a HVM guest.
> - Both page_info and shadow_page_info are stored in frame_table.
> At first, all spin locks in frame table are initialized as
> SPIN_LOCK_UNLOCKED.
> If one frame is used as shadow_page_info, we still should initialize the
> spin lock to be SPIN_LOCK_UNLOCKED when freeing that page.
> 
> Signed-off-by: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> 
> Best Regards,
> -- Dongxiao
> 



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