[PATCH] drm/i915: fix corruptions on i8xx due to relaxed fencing

[Indan Zupancic]

Hi,

On Wed, February 23, 2011 08:10, Daniel Vetter wrote:
> Am Mi, 23.02.2011, 07:59 schrieb Indan Zupancic:
>> On Tue, February 22, 2011 18:25, Daniel Vetter wrote:
>>> It looks like gen2 has a peculiar interleaved 2-row inter-tile
>>> layout. Probably inherited from i81x which had 2kb tiles (which
>>> naturally fit an even-number-of-tile-rows scheme to fit onto 4kb
>>> pages). There is no other mention of this in any docs (also not
>>> in the Intel internal documention according to Chris Wilson).
>>>
>>> Problem manifests itself in corruptions in the second half of the
>>> last tile row (if the bo has an odd number of tiles). Which can
>>> only happen with relaxed tiling (introduced in a00b10c360b35d6431a9).
>>>
>>> So reject set_tiling calls that don't satisfy this constrain to
>>> prevent broken userspace from causing havoc. While at it, also
>>> check the size for newer chipsets.
>>>
>>> LKML: https://lkml.org/lkml/2011/2/19/5
>>> Reported-by: Indan Zupancic <indan at nul.nu>
>>> Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>
>>> ---
>>>  drivers/gpu/drm/i915/i915_gem_tiling.c |   16 +++++++++++++++-
>>>  1 files changed, 15 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c
>>> b/drivers/gpu/drm/i915/i915_gem_tiling.c
>>> index 22a32b9..79a04fd 100644
>>> --- a/drivers/gpu/drm/i915/i915_gem_tiling.c
>>> +++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
>>> @@ -184,7 +184,7 @@ i915_gem_detect_bit_6_swizzle(struct drm_device
>>> *dev)
>>>  static bool
>>>  i915_tiling_ok(struct drm_device *dev, int stride, int size, int
>>> tiling_mode)
>>>  {
>>> -	int tile_width;
>>> +	int tile_width, tile_height;
>>>
>>>  	/* Linear is always fine */
>>>  	if (tiling_mode == I915_TILING_NONE)
>>> @@ -215,6 +215,20 @@ i915_tiling_ok(struct drm_device *dev, int stride,
>>> int size, int
>>> tiling_mode)
>>>  		}
>>>  	}
>>>
>>> +	if (IS_GEN2(dev) ||
>>> +	    (tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev)))
>>> +		tile_height = 32;
>>> +	else
>>> +		tile_height = 8;
>>> +	/* i8xx is strange: It has 2 interleaved rows of tiles, so needs an
>>> even
>>> +	 * number of tile rows. */
>>> +	if (IS_GEN2(dev))
>>> +		tile_height *= 2;
>>> +
>>> +	/* Size needs to be aligned to a full tile row */
>>> +	if (size & (tile_height * stride - 1))
>>> +		return false;
>>> +
>>>  	/* 965+ just needs multiples of tile width */
>>>  	if (INTEL_INFO(dev)->gen >= 4) {
>>>  		if (stride & (tile_width - 1))
>>
>> Tested-by: Indan Zupancic <indan at nul.nu>
>>
>> I tested with this patch and without the other ones you send and the
>> corruption is indeed gone.
>>
>> Not sure why you dropped lkml from CC, now people who stuble upon it
>> don't see the ending...
>
> Random incoherency in my brain. Re-added to cc.

This isn't in rc8, can someone make sure it gets into 2.6.38-rc9/2.6.38?

Thanks,

Indan