aa: too sensitive at the line end points for PNV platform

Submitted by marius.predut@intel.com on July 30, 2015, 2:51 p.m.

Details

Message ID 0E2B2CA4D522C14FBE8646D789EF72A01C3CB076@IRSMSX103.ger.corp.intel.com
State New
Headers show

Not browsing as part of any series.

Commit Message

marius.predut@intel.com July 30, 2015, 2:51 p.m.
Still not upstream.

-----Original Message-----
From: Predut, Marius 
Sent: Thursday, June 04, 2015 11:53 PM
To: piglit@lists.freedesktop.org
Cc: Predut, Marius
Subject: [PATCH] aa: too sensitive at the line end points for PNV platform

Adjust middles of the segments sample so that now the test pass also on PNV platform.

Bspec documentation G45: Volume 2: 3D/Media Last Pixel Enable  bit of  SF_STATE controls whether the last pixel of the last line in a LINESTRIP_xxx primitive or the last pixel of each line in a LINELIST_xxx primitive is rendered.

Signed-off-by: Marius Predut <marius.predut@intel.com>
---
 tests/general/line-aa-width.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

--
1.9.1

Patch hide | download patch | download mbox

diff --git a/tests/general/line-aa-width.c b/tests/general/line-aa-width.c index c0fd7c1..e7998c4 100644
--- a/tests/general/line-aa-width.c
+++ b/tests/general/line-aa-width.c
@@ -136,7 +136,7 @@  piglit_display(void)
 	 * coverage in each column
 	 */
 	for (x1 = 2; x1 < piglit_width; x1 += seg_width) {
-		int x2 = x1 + seg_width - 4;
+		int x2 = x1 + seg_width - 5;
 		int sample_x;
 		float y1, y2;
 		float avg = 0.0;
@@ -144,8 +144,8 @@  piglit_display(void)
 		float max = -100.0;
 		char *err = NULL;
 
-		if (x2 > piglit_width - 4)
-			x2 = piglit_width - 4;
+		if (x2 > piglit_width - 5)
+			x2 = piglit_width - 5;
 
 		/* If we don't have a couple of pixels to sample because we've
 		 * hit the edge of the window, we're done.

Comments

On 07/30/2015 04:51 PM, Predut, Marius wrote:
> Still not upstream.
> 
> -----Original Message-----
> From: Predut, Marius 
> Sent: Thursday, June 04, 2015 11:53 PM
> To: piglit@lists.freedesktop.org
> Cc: Predut, Marius
> Subject: [PATCH] aa: too sensitive at the line end points for PNV platform
> 
> Adjust middles of the segments sample so that now the test pass also on PNV platform.
> 
> Bspec documentation G45: Volume 2: 3D/Media Last Pixel Enable  bit of  SF_STATE controls whether the last pixel of the last line in a LINESTRIP_xxx primitive or the last pixel of each line in a LINELIST_xxx primitive is rendered.
> 
> Signed-off-by: Marius Predut <marius.predut@intel.com>
> ---
>  tests/general/line-aa-width.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/tests/general/line-aa-width.c b/tests/general/line-aa-width.c index c0fd7c1..e7998c4 100644
> --- a/tests/general/line-aa-width.c
> +++ b/tests/general/line-aa-width.c
> @@ -136,7 +136,7 @@ piglit_display(void)
>  	 * coverage in each column
>  	 */
>  	for (x1 = 2; x1 < piglit_width; x1 += seg_width) {
> -		int x2 = x1 + seg_width - 4;
> +		int x2 = x1 + seg_width - 5;
>  		int sample_x;
>  		float y1, y2;
>  		float avg = 0.0;
> @@ -144,8 +144,8 @@ piglit_display(void)
>  		float max = -100.0;
>  		char *err = NULL;
>  
> -		if (x2 > piglit_width - 4)
> -			x2 = piglit_width - 4;
> +		if (x2 > piglit_width - 5)
> +			x2 = piglit_width - 5;
>  
>  		/* If we don't have a couple of pixels to sample because we've
>  		 * hit the edge of the window, we're done.
> --
> 1.9.1
> 

Looks fine, though I don't have available HW to confirm the test passes
after applying corresponding Mesa patch. If somebody can confirm it,
this patch is:

Reviewed-By: Eduardo Lima Mitev <elima@igalia.com>
Ping...

> -----Original Message-----

> From: Piglit [mailto:piglit-bounces@lists.freedesktop.org] On Behalf Of

> Eduardo Lima Mitev

> Sent: Thursday, August 27, 2015 10:19 AM

> To: piglit@lists.freedesktop.org

> Subject: Re: [Piglit] [PATCH] aa: too sensitive at the line end points for PNV

> platform

> 

> On 07/30/2015 04:51 PM, Predut, Marius wrote:

> > Still not upstream.

> >

> > -----Original Message-----

> > From: Predut, Marius

> > Sent: Thursday, June 04, 2015 11:53 PM

> > To: piglit@lists.freedesktop.org

> > Cc: Predut, Marius

> > Subject: [PATCH] aa: too sensitive at the line end points for PNV platform

> >

> > Adjust middles of the segments sample so that now the test pass also on PNV

> platform.

> >

> > Bspec documentation G45: Volume 2: 3D/Media Last Pixel Enable  bit of

> SF_STATE controls whether the last pixel of the last line in a LINESTRIP_xxx

> primitive or the last pixel of each line in a LINELIST_xxx primitive is

> rendered.

> >

> > Signed-off-by: Marius Predut <marius.predut@intel.com>

> > ---

> >  tests/general/line-aa-width.c | 6 +++---

> >  1 file changed, 3 insertions(+), 3 deletions(-)

> >

> > diff --git a/tests/general/line-aa-width.c b/tests/general/line-aa-width.c

> index c0fd7c1..e7998c4 100644

> > --- a/tests/general/line-aa-width.c

> > +++ b/tests/general/line-aa-width.c

> > @@ -136,7 +136,7 @@ piglit_display(void)

> >  	 * coverage in each column

> >  	 */

> >  	for (x1 = 2; x1 < piglit_width; x1 += seg_width) {

> > -		int x2 = x1 + seg_width - 4;

> > +		int x2 = x1 + seg_width - 5;

> >  		int sample_x;

> >  		float y1, y2;

> >  		float avg = 0.0;

> > @@ -144,8 +144,8 @@ piglit_display(void)

> >  		float max = -100.0;

> >  		char *err = NULL;

> >

> > -		if (x2 > piglit_width - 4)

> > -			x2 = piglit_width - 4;

> > +		if (x2 > piglit_width - 5)

> > +			x2 = piglit_width - 5;

> >

> >  		/* If we don't have a couple of pixels to sample because we've

> >  		 * hit the edge of the window, we're done.

> > --

> > 1.9.1

> >

> 

> Looks fine, though I don't have available HW to confirm the test passes

> after applying corresponding Mesa patch. If somebody can confirm it,

> this patch is:

> 

> Reviewed-By: Eduardo Lima Mitev <elima@igalia.com>

> 

> _______________________________________________

> Piglit mailing list

> Piglit@lists.freedesktop.org

> http://lists.freedesktop.org/mailman/listinfo/piglit
On 09/09/2015 03:00 PM, Predut, Marius wrote:
> Ping...
> 

I have been waiting for the corresponding mesa patch to get clearance
before pushing this one. Otherwise we risk bringing a regression for an
undefined timespan.

Since I cannot test on PNV, I don't know if it the piglit test still
passes after applying this (and before having the fix in mesa). If you
can confirm it does, then it is no risk to push this already.

Eduardo

>> -----Original Message-----
>> From: Piglit [mailto:piglit-bounces@lists.freedesktop.org] On Behalf Of
>> Eduardo Lima Mitev
>> Sent: Thursday, August 27, 2015 10:19 AM
>> To: piglit@lists.freedesktop.org
>> Subject: Re: [Piglit] [PATCH] aa: too sensitive at the line end points for PNV
>> platform
>>
>> On 07/30/2015 04:51 PM, Predut, Marius wrote:
>>> Still not upstream.
>>>
>>> -----Original Message-----
>>> From: Predut, Marius
>>> Sent: Thursday, June 04, 2015 11:53 PM
>>> To: piglit@lists.freedesktop.org
>>> Cc: Predut, Marius
>>> Subject: [PATCH] aa: too sensitive at the line end points for PNV platform
>>>
>>> Adjust middles of the segments sample so that now the test pass also on PNV
>> platform.
>>>
>>> Bspec documentation G45: Volume 2: 3D/Media Last Pixel Enable  bit of
>> SF_STATE controls whether the last pixel of the last line in a LINESTRIP_xxx
>> primitive or the last pixel of each line in a LINELIST_xxx primitive is
>> rendered.
>>>
>>> Signed-off-by: Marius Predut <marius.predut@intel.com>
>>> ---
>>>  tests/general/line-aa-width.c | 6 +++---
>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/tests/general/line-aa-width.c b/tests/general/line-aa-width.c
>> index c0fd7c1..e7998c4 100644
>>> --- a/tests/general/line-aa-width.c
>>> +++ b/tests/general/line-aa-width.c
>>> @@ -136,7 +136,7 @@ piglit_display(void)
>>>  	 * coverage in each column
>>>  	 */
>>>  	for (x1 = 2; x1 < piglit_width; x1 += seg_width) {
>>> -		int x2 = x1 + seg_width - 4;
>>> +		int x2 = x1 + seg_width - 5;
>>>  		int sample_x;
>>>  		float y1, y2;
>>>  		float avg = 0.0;
>>> @@ -144,8 +144,8 @@ piglit_display(void)
>>>  		float max = -100.0;
>>>  		char *err = NULL;
>>>
>>> -		if (x2 > piglit_width - 4)
>>> -			x2 = piglit_width - 4;
>>> +		if (x2 > piglit_width - 5)
>>> +			x2 = piglit_width - 5;
>>>
>>>  		/* If we don't have a couple of pixels to sample because we've
>>>  		 * hit the edge of the window, we're done.
>>> --
>>> 1.9.1
>>>
>>
>> Looks fine, though I don't have available HW to confirm the test passes
>> after applying corresponding Mesa patch. If somebody can confirm it,
>> this patch is:
>>
>> Reviewed-By: Eduardo Lima Mitev <elima@igalia.com>
>>
>> _______________________________________________
>> Piglit mailing list
>> Piglit@lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/piglit
> _______________________________________________
> Piglit mailing list
> Piglit@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/piglit
>
I confirm that not piglit regression.


-----Original Message-----
From: Piglit [mailto:piglit-bounces@lists.freedesktop.org] On Behalf Of Eduardo Lima Mitev

Sent: Thursday, September 10, 2015 11:44 AM
To: piglit@lists.freedesktop.org
Subject: Re: [Piglit] [PATCH] aa: too sensitive at the line end points for PNV platform

On 09/09/2015 03:00 PM, Predut, Marius wrote:
> Ping...

> 


I have been waiting for the corresponding mesa patch to get clearance before pushing this one. Otherwise we risk bringing a regression for an undefined timespan.

Since I cannot test on PNV, I don't know if it the piglit test still passes after applying this (and before having the fix in mesa). If you can confirm it does, then it is no risk to push this already.

Eduardo

>> -----Original Message-----

>> From: Piglit [mailto:piglit-bounces@lists.freedesktop.org] On Behalf 

>> Of Eduardo Lima Mitev

>> Sent: Thursday, August 27, 2015 10:19 AM

>> To: piglit@lists.freedesktop.org

>> Subject: Re: [Piglit] [PATCH] aa: too sensitive at the line end 

>> points for PNV platform

>>

>> On 07/30/2015 04:51 PM, Predut, Marius wrote:

>>> Still not upstream.

>>>

>>> -----Original Message-----

>>> From: Predut, Marius

>>> Sent: Thursday, June 04, 2015 11:53 PM

>>> To: piglit@lists.freedesktop.org

>>> Cc: Predut, Marius

>>> Subject: [PATCH] aa: too sensitive at the line end points for PNV 

>>> platform

>>>

>>> Adjust middles of the segments sample so that now the test pass also 

>>> on PNV

>> platform.

>>>

>>> Bspec documentation G45: Volume 2: 3D/Media Last Pixel Enable  bit 

>>> of

>> SF_STATE controls whether the last pixel of the last line in a 

>> LINESTRIP_xxx primitive or the last pixel of each line in a 

>> LINELIST_xxx primitive is rendered.

>>>

>>> Signed-off-by: Marius Predut <marius.predut@intel.com>

>>> ---

>>>  tests/general/line-aa-width.c | 6 +++---

>>>  1 file changed, 3 insertions(+), 3 deletions(-)

>>>

>>> diff --git a/tests/general/line-aa-width.c 

>>> b/tests/general/line-aa-width.c

>> index c0fd7c1..e7998c4 100644

>>> --- a/tests/general/line-aa-width.c

>>> +++ b/tests/general/line-aa-width.c

>>> @@ -136,7 +136,7 @@ piglit_display(void)

>>>  	 * coverage in each column

>>>  	 */

>>>  	for (x1 = 2; x1 < piglit_width; x1 += seg_width) {

>>> -		int x2 = x1 + seg_width - 4;

>>> +		int x2 = x1 + seg_width - 5;

>>>  		int sample_x;

>>>  		float y1, y2;

>>>  		float avg = 0.0;

>>> @@ -144,8 +144,8 @@ piglit_display(void)

>>>  		float max = -100.0;

>>>  		char *err = NULL;

>>>

>>> -		if (x2 > piglit_width - 4)

>>> -			x2 = piglit_width - 4;

>>> +		if (x2 > piglit_width - 5)

>>> +			x2 = piglit_width - 5;

>>>

>>>  		/* If we don't have a couple of pixels to sample because we've

>>>  		 * hit the edge of the window, we're done.

>>> --

>>> 1.9.1

>>>

>>

>> Looks fine, though I don't have available HW to confirm the test 

>> passes after applying corresponding Mesa patch. If somebody can 

>> confirm it, this patch is:

>>

>> Reviewed-By: Eduardo Lima Mitev <elima@igalia.com>

>>

>> _______________________________________________

>> Piglit mailing list

>> Piglit@lists.freedesktop.org

>> http://lists.freedesktop.org/mailman/listinfo/piglit

> _______________________________________________

> Piglit mailing list

> Piglit@lists.freedesktop.org

> http://lists.freedesktop.org/mailman/listinfo/piglit

> 


_______________________________________________
Piglit mailing list
Piglit@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/piglit
On 11/09/2015 07:28 PM, Predut, Marius wrote:
> I confirm that not piglit regression.
> 

Pushed to piglit master.
Thanks!

Eduardo

> 
> -----Original Message-----
> From: Piglit [mailto:piglit-bounces@lists.freedesktop.org] On Behalf Of Eduardo Lima Mitev
> Sent: Thursday, September 10, 2015 11:44 AM
> To: piglit@lists.freedesktop.org
> Subject: Re: [Piglit] [PATCH] aa: too sensitive at the line end points for PNV platform
> 
> On 09/09/2015 03:00 PM, Predut, Marius wrote:
>> Ping...
>>
> 
> I have been waiting for the corresponding mesa patch to get clearance before pushing this one. Otherwise we risk bringing a regression for an undefined timespan.
> 
> Since I cannot test on PNV, I don't know if it the piglit test still passes after applying this (and before having the fix in mesa). If you can confirm it does, then it is no risk to push this already.
> 
> Eduardo
> 
>>> -----Original Message-----
>>> From: Piglit [mailto:piglit-bounces@lists.freedesktop.org] On Behalf 
>>> Of Eduardo Lima Mitev
>>> Sent: Thursday, August 27, 2015 10:19 AM
>>> To: piglit@lists.freedesktop.org
>>> Subject: Re: [Piglit] [PATCH] aa: too sensitive at the line end 
>>> points for PNV platform
>>>
>>> On 07/30/2015 04:51 PM, Predut, Marius wrote:
>>>> Still not upstream.
>>>>
>>>> -----Original Message-----
>>>> From: Predut, Marius
>>>> Sent: Thursday, June 04, 2015 11:53 PM
>>>> To: piglit@lists.freedesktop.org
>>>> Cc: Predut, Marius
>>>> Subject: [PATCH] aa: too sensitive at the line end points for PNV 
>>>> platform
>>>>
>>>> Adjust middles of the segments sample so that now the test pass also 
>>>> on PNV
>>> platform.
>>>>
>>>> Bspec documentation G45: Volume 2: 3D/Media Last Pixel Enable  bit 
>>>> of
>>> SF_STATE controls whether the last pixel of the last line in a 
>>> LINESTRIP_xxx primitive or the last pixel of each line in a 
>>> LINELIST_xxx primitive is rendered.
>>>>
>>>> Signed-off-by: Marius Predut <marius.predut@intel.com>
>>>> ---
>>>>  tests/general/line-aa-width.c | 6 +++---
>>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/tests/general/line-aa-width.c 
>>>> b/tests/general/line-aa-width.c
>>> index c0fd7c1..e7998c4 100644
>>>> --- a/tests/general/line-aa-width.c
>>>> +++ b/tests/general/line-aa-width.c
>>>> @@ -136,7 +136,7 @@ piglit_display(void)
>>>>  	 * coverage in each column
>>>>  	 */
>>>>  	for (x1 = 2; x1 < piglit_width; x1 += seg_width) {
>>>> -		int x2 = x1 + seg_width - 4;
>>>> +		int x2 = x1 + seg_width - 5;
>>>>  		int sample_x;
>>>>  		float y1, y2;
>>>>  		float avg = 0.0;
>>>> @@ -144,8 +144,8 @@ piglit_display(void)
>>>>  		float max = -100.0;
>>>>  		char *err = NULL;
>>>>
>>>> -		if (x2 > piglit_width - 4)
>>>> -			x2 = piglit_width - 4;
>>>> +		if (x2 > piglit_width - 5)
>>>> +			x2 = piglit_width - 5;
>>>>
>>>>  		/* If we don't have a couple of pixels to sample because we've
>>>>  		 * hit the edge of the window, we're done.
>>>> --
>>>> 1.9.1
>>>>
>>>
>>> Looks fine, though I don't have available HW to confirm the test 
>>> passes after applying corresponding Mesa patch. If somebody can 
>>> confirm it, this patch is:
>>>
>>> Reviewed-By: Eduardo Lima Mitev <elima@igalia.com>
>>>
>>> _______________________________________________
>>> Piglit mailing list
>>> Piglit@lists.freedesktop.org
>>> http://lists.freedesktop.org/mailman/listinfo/piglit
>> _______________________________________________
>> Piglit mailing list
>> Piglit@lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/piglit
>>
> 
> _______________________________________________
> Piglit mailing list
> Piglit@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/piglit
>