[Mesa-dev] [PATCH] mesa: Fix copy-and-paste bug in do_row_3D

[Ian Romanick]

On 02/01/2012 02:28 PM, Brian Paul wrote:
> On 02/01/2012 02:21 PM, Ian Romanick wrote:
>> From: Ian Romanick<ian.d.romanick at intel.com>
>>
>> Several of the half-float cases used 4 as the texel size when it
>> should have been some smaller value.
>>
>> Signed-off-by: Ian Romanick<ian.d.romanick at intel.com>
>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43324
>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43325
>> ---
>> src/mesa/main/mipmap.c | 6 +++---
>> 1 files changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c
>> index 03ce536..756316a 100644
>> --- a/src/mesa/main/mipmap.c
>> +++ b/src/mesa/main/mipmap.c
>> @@ -991,7 +991,7 @@ do_row_3D(GLenum datatype, GLuint comps, GLint
>> srcWidth,
>> }
>> }
>> else if ((datatype == GL_HALF_FLOAT_ARB)&& (comps == 3)) {
>> - DECLARE_ROW_POINTERS(GLhalfARB, 4);
>> + DECLARE_ROW_POINTERS(GLhalfARB, 3);
>>
>> for (i = j = 0, k = k0; i< (GLuint) dstWidth;
>> i++, j += colStride, k += colStride) {
>> @@ -1001,7 +1001,7 @@ do_row_3D(GLenum datatype, GLuint comps, GLint
>> srcWidth,
>> }
>> }
>> else if ((datatype == GL_HALF_FLOAT_ARB)&& (comps == 2)) {
>> - DECLARE_ROW_POINTERS(GLhalfARB, 4);
>> + DECLARE_ROW_POINTERS(GLhalfARB, 2);
>>
>> for (i = j = 0, k = k0; i< (GLuint) dstWidth;
>> i++, j += colStride, k += colStride) {
>> @@ -1010,7 +1010,7 @@ do_row_3D(GLenum datatype, GLuint comps, GLint
>> srcWidth,
>> }
>> }
>> else if ((datatype == GL_HALF_FLOAT_ARB)&& (comps == 1)) {
>> - DECLARE_ROW_POINTERS(GLhalfARB, 4);
>> + DECLARE_ROW_POINTERS(GLhalfARB, 1);
>>
>> for (i = j = 0, k = k0; i< (GLuint) dstWidth;
>> i++, j += colStride, k += colStride) {
>
> Looks good. Reviewed-by: Brian Paul <brianp at vmware.com>
>
> Do we actually have any piglit tests that exercise 3D mipmap generation?
> I'm not sure I've ever seen an app use it, but anyway.

I'm sure it's roughly proportional to the number of apps using 3D 
textures in general... and that number is small.

We don't have any piglit tests yet.  We hit this in Intel's internal, 
closed-source test suite.  It's on our (very, very long) list of new 
tests and test infrastructure that need to be written over the next 
couple months.