[Mesa-dev] [PATCH] mesa: add NV_read_{depth, stencil, depth_stencil} extensions
Marek Olšák
maraeo at gmail.com
Mon Aug 10 12:28:29 PDT 2015
On Mon, Aug 10, 2015 at 9:15 PM, Rob Clark <robdclark at gmail.com> wrote:
> On Mon, Aug 10, 2015 at 12:54 PM, Marek Olšák <maraeo at gmail.com> wrote:
>> On Mon, Aug 10, 2015 at 1:02 PM, Rob Clark <robdclark at gmail.com> wrote:
>>> From: Rob Clark <robclark at freedesktop.org>
>>>
>>> These extensions allow reading depth/stencil for GLES contexts, which is
>>> useful for tools like apitrace.
>>>
>>> Signed-off-by: Rob Clark <robclark at freedesktop.org>
>>> ---
>>> I have a patch, which I will send out after some cleanup, that makes
>>> apitrace able to dump depth/stencil buffers with GLES, thanks to this
>>> extension.
>>>
>>> src/mesa/main/extensions.c | 3 +++
>>> src/mesa/main/readpix.c | 25 +++++++++++++++++++------
>>> 2 files changed, 22 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c
>>> index 2dbfabd..d934d19 100644
>>> --- a/src/mesa/main/extensions.c
>>> +++ b/src/mesa/main/extensions.c
>>> @@ -385,6 +385,9 @@ static const struct extension extension_table[] = {
>>> { "GL_NV_point_sprite", o(NV_point_sprite), GL, 2001 },
>>> { "GL_NV_primitive_restart", o(NV_primitive_restart), GLL, 2002 },
>>> { "GL_NV_read_buffer", o(dummy_true), ES2, 2011 },
>>> + { "GL_NV_read_depth", o(dummy_true), ES2, 2011 },
>>> + { "GL_NV_read_depth_stencil", o(dummy_true), ES2, 2011 },
>>> + { "GL_NV_read_stencil", o(dummy_true), ES2, 2011 },
>>> { "GL_NV_texgen_reflection", o(dummy_true), GLL, 1999 },
>>> { "GL_NV_texture_barrier", o(NV_texture_barrier), GL, 2009 },
>>> { "GL_NV_texture_env_combine4", o(NV_texture_env_combine4), GLL, 1999 },
>>> diff --git a/src/mesa/main/readpix.c b/src/mesa/main/readpix.c
>>> index 2744232..65751aa 100644
>>> --- a/src/mesa/main/readpix.c
>>> +++ b/src/mesa/main/readpix.c
>>> @@ -917,7 +917,9 @@ read_pixels_es3_error_check(GLenum format, GLenum type,
>>> GLboolean is_unsigned_int = GL_FALSE;
>>> GLboolean is_signed_int = GL_FALSE;
>>>
>>> - if (!_mesa_is_color_format(internalFormat)) {
>>> + /* TODO just drop the check? Are there any formats to filter out? */
>>
>> Yes, please drop it.
>>
>>> + if (!(_mesa_is_color_format(internalFormat) ||
>>> + _mesa_is_depth_or_stencil_format(internalFormat))) {
>>> return GL_INVALID_OPERATION;
>>> }
>>>
>>> @@ -950,6 +952,22 @@ read_pixels_es3_error_check(GLenum format, GLenum type,
>>> (is_unsigned_int && type == GL_UNSIGNED_INT))
>>> return GL_NO_ERROR;
>>> break;
>>> + case GL_DEPTH_STENCIL:
>>> + if ((internalFormat == GL_DEPTH24_STENCIL8) &&
>>> + (type == GL_UNSIGNED_INT_24_8))
>>> + return GL_NO_ERROR;
>>> + if ((internalFormat == GL_DEPTH32F_STENCIL8) &&
>>> + (type == GL_FLOAT_32_UNSIGNED_INT_24_8_REV))
>>> + return GL_NO_ERROR;
>>> + break;
>>> + case GL_DEPTH_COMPONENT:
>>> + if ((internalFormat == GL_DEPTH_COMPONENT32F) &&
>>> + (type == GL_FLOAT))
>>> + return GL_NO_ERROR;
>>> + if ((internalFormat == GL_DEPTH_COMPONENT24) &&
>>> + (type == GL_UNSIGNED_INT))
>>> + return GL_NO_ERROR;
>>> + break;
>>
>> What about GL_STENCIL_INDEX? Mesa does support GL_OES_stencil8.
>
> hmm.. the gles3 glTexImage2D page (where I got the mapping of
> internalFormat and type) didn't include GL_STENCIL_INDEX. So I'm a
> bit unclear what (if any) the check for internalFormat should be? Or
> should we just check that type is GL_STENCIL_INDEX (which is all the
> NV_read_* spec says)?
This is ReadPixels, not TexImage.
GL_OES_stencil8 only allows GL_STENCIL_INDEX for renderbuffers.
Marek
More information about the mesa-dev
mailing list