[Intel-gfx] [PATCH 2/2] drm/i915: Limit the scattergather coalescing to 32bits

[Tvrtko Ursulin]

On 14/10/2016 16:43, Chris Wilson wrote:
> On Fri, Oct 14, 2016 at 04:38:35PM +0100, Tvrtko Ursulin wrote:
>> On 14/10/2016 16:18, Chris Wilson wrote:
>>> The scattergather list uses a 32bit size counter, we should avoid
>>> exceeding it.
>>>
>>> Fixes: 871dfbd67d4e ("drm/i915: Allow compaction upto SWIOTLB max segment size")
>>> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
>>> ---
>>>   drivers/gpu/drm/i915/i915_gem.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
>>> index 0d1dc04302ec..858569419134 100644
>>> --- a/drivers/gpu/drm/i915/i915_gem.c
>>> +++ b/drivers/gpu/drm/i915/i915_gem.c
>>> @@ -2238,7 +2238,7 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
>>>   	max_segment = swiotlb_max_size();
>>>   	if (!max_segment)
>>> -		max_segment = obj->base.size;
>>> +		max_segment = rounddown(UINT_MAX, PAGE_SIZE);
>>>   	st = kmalloc(sizeof(*st), GFP_KERNEL);
>>>   	if (st == NULL)
>> I suppose it could be possible to get more that 4GiB of of contiguous pages.
>>
>> sg_alloc_table_from_pages seems broken as well, for this, and more
>> so, for using longs and stuffing them into ints.
> Noticed I was using unsigned long max_segment. Any complaints if I
> change that to unsigned int as well?

Not from me, should be changed!

Regards,

Tvrtko