[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] x86/setup: don't relocate the VGA hole



On 07/02/2013 21:29, "Ian Campbell" <ian.campbell@xxxxxxxxxx> wrote:

> On Thu, 2013-02-07 at 20:38 +0000, Keir Fraser wrote:
>> On 07/02/2013 20:20, "Ian Campbell" <ian.campbell@xxxxxxxxxx> wrote:
>> 
>>>> @@ -840,10 +839,8 @@ void __init __start_xen(unsigned long mb
>>>>               * data until after we have switched to the relocated
>>>> pagetables!
>>>>               */
>>>>              barrier();
>>>> -            dst = move_memory(e, 0, (unsigned long)&_end - XEN_VIRT_START,
>>>> 1);
>>>> -
>>>> -            /* Poison low 1MB to detect stray pointers to physical 0-1MB.
>>>> */
>>>> -            memset(dst, 0x55, 1U << 20);
>>> 
>>> Is this poisoning not still desirable?
>> 
>> If we don't copy that low 1MB I don't see any reason to poison it at the
>> destination.
> 
> The poisoning was to protect against stray pointers out of the low 1MB
> region? I assumed it was against pointers into it...

I mean any stray pointers into that region are only going to see garbage at
the relocated highmem address anyway, as contents of 0-1MB are no longer
being copied. So deliberate poisoning seems even more paranoid than it
already was (it was already really quite paranoid).

 -- Keir

> Ian.
> 



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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.