Re: [Xen-devel] [PATCH v12] introduce XENMEM_reserved_device_memory_map

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 22.07.15 at 15:20, <tiejun.chen@xxxxxxxxx> wrote:
> On 2015/7/22 21:03, Jan Beulich wrote:
>>>>> On 22.07.15 at 14:55, <tiejun.chen@xxxxxxxxx> wrote:
>>>>   +#ifdef HAS_PASSTHROUGH
>>>> +        case XENMEM_reserved_device_memory_map:
>>>> +        {
>>>> +            struct get_reserved_device_memory grdm;
>>>> +
>>>> +            if ( unlikely(start_extent) )
>>>> +                return -ENOSYS;
>>>> +
>>>> +            if ( copy_from_guest(&grdm.map, compat, 1) ||
>>>> +                 !compat_handle_okay(grdm.map.buffer, grdm.map.nr_entries)
>>> )
>>>> +                return -EFAULT;
>>>> +
>>>> +            if ( grdm.map.flags & ~XENMEM_RDM_ALL )
>>>> +                return -EINVAL;
>>>> +
>>>> +            grdm.used_entries = 0;
>>>> +            rc = 
> iommu_get_reserved_device_memory(get_reserved_device_memory,
>>>> +                                                  &grdm);
>>>> +
>>>
>>> Just as you asked me previously,
>>>
>>> Here if RDM doesn't exist, so
>>>
>>> grdm.map.nr_entries = grdm.used_entries = 0, and rc = 0, right?
>>
>> No, grdm.map.nr_entries still holds whatever the caller passed.
> 
> What if the caller pass "0" like raising an inquiry? Indeed, this is 
> what we did in patch #11. I think this is reasonable since the caller 
> always doesn't know how much buffers should be allocated beforehand, so 
> instead, the caller prefer to make this sort of inquiry without any 
> buffers.

If the callers passes in 0, then we return back the actual count.
I don't think I see what problem you're seeing.

>>>> +            if ( !rc && grdm.map.nr_entries < grdm.used_entries )
>>>> +                rc = -ENOBUFS;
>>>> +            grdm.map.nr_entries = grdm.used_entries;
>>>> +            if ( __copy_to_guest(compat, &grdm.map, 1) )
>>>
>>> So can we still do this copy here?
>>
>> We not only can, we need to. The only case where we might skip it
>> is when the incoming grdm.map.nr_entries is unchanged.
> 
> If what I'm saying above is right, __copy_to_guest() would return a 
> error in this case, right?

Why would it? Are you mixing it up with copying to the actual
array (indirected through the handle inside the structure)?

Jan


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