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

Re: [Xen-devel] [RFC PATCH 4/9] xen/arm: p2m: Update IOMMU mapping whenever possible if page table is not shared



On Mon, Apr 24, 2017 at 2:41 PM, Julien Grall <julien.grall@xxxxxxx> wrote:
> Hi Oleksandr,
Hi, Julien

>
> On 21/04/17 19:44, Oleksandr Tyshchenko wrote:
>>
>> On Fri, Apr 21, 2017 at 7:27 PM, Julien Grall <julien.grall@xxxxxxx>
>> wrote:
>>>
>>> On 21/04/17 15:18, Oleksandr Tyshchenko wrote:
>>>>
>>>>
>>>> On Wed, Apr 19, 2017 at 8:46 PM, Julien Grall <julien.grall@xxxxxxx>
>>>> wrote:
>>
>>
>> Aha. Seems, I understand what you meant. I already have a check in IPMMU
>> driver:
>>
>> /*
>> * both the virtual address and the physical one, as well as
>> * the size of the mapping, must be aligned (at least) to the
>> * size of the smallest page supported by the hardware
>> */
>> if (!IS_ALIGNED(iova | paddr | size, min_pagesz)) {
>> printk("unaligned: iova 0x%lx pa 0x%"PRIx64" size 0x%zx min_pagesz
>> 0x%x\n",
>>       iova, paddr, size, min_pagesz);
>> return -EINVAL;
>> }
>>
>> where min_pagesz - is a minimum page size supported by hardware.
>> Hope, that this check can catch such case.
>
>
> I think will cover what I meant. Although, can't we do this check when the
> IOMMU is initialized at boot time rather than for every mapping?
I think, yes.

>
> For instance you know that if the mimimum page granularity supported  by the
> IOMMU is bigger than PAGE_SIZE, than you will likely get into trouble later
> on.
>
> So rather than randomly crashing at runtime, you can disable the IOMMU/panic
> at boot time.
Agree.

>
> Cheers,
>
> --
> Julien Grall



-- 
Regards,

Oleksandr Tyshchenko

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

 


Rackspace

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