Re: [Xen-devel] [PATCH for-xen-4-5 v3] xen/arm: dump guest stack even if not the current VCPU

[Frediano Ziglio <freddy77@xxxxxxxxx>]

Agreed, but this is a performance improvement and not a fix.

diff --git a/xen/arch/arm/p2m.c b/xen/arch/arm/p2m.c
index 2345199..fab0f57 100644
--- a/xen/arch/arm/p2m.c
+++ b/xen/arch/arm/p2m.c
@@ -1198,6 +1198,8 @@ struct page_info *get_page_from_gva(struct
domain *d, vaddr_t va,
         rc = gvirt_to_maddr(va, &maddr, flags);
     }

+    spin_unlock(&p2m->lock);
+
     if ( rc )
         goto err;

@@ -1211,7 +1213,6 @@ struct page_info *get_page_from_gva(struct
domain *d, vaddr_t va,
         page = NULL;

 err:
-    spin_unlock(&p2m->lock);
     return page;
 }


Frediano

Frediano


2014-10-23 12:56 GMT+01:00 Julien Grall <julien.grall@xxxxxxxxxx>:
> On 10/23/2014 12:31 PM, Ian Campbell wrote:
>> On Thu, 2014-10-23 at 09:46 +0100, Frediano Ziglio wrote:
>>> From: Frediano Ziglio <frediano.ziglio@xxxxxxxxxx>
>>>
>>> If show_guest_stack was called from Xen context (for instance hitting
>>> '0' key on Xen console) get_page_from_gva was not able to get the
>>> page returning NULL.
>>> Detecting different domain and changing VTTBR register make
>>> get_page_from_gva works for different domains.
>>>
>>> Signed-off-by: Frediano Ziglio <frediano.ziglio@xxxxxxxxxx>
>>
>> Acked + applied (since Konrad said ok in v2).
>>
>>> - collapse change in a single if to improve performances.
>>
>> Not 100% keen on the duplicated call to gvirt_to_maddr, but it'll do.
>
> I think having two if with gvirt_to_maddr in the middle would have been
> fine. My main concern was disabling the IRQ before taking the lock.
>
> Anyway, I plan to send a patch for 4.6 to improve this function. The p2m
> is taken for a long time for nothing.
>
> Regards,
>
> --
> Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel