It is implemented combining the values returned by calls to the 32-bit
query _mesa_GetInternalformati32v.
The main reason is simplicity. The other option would be C&P how we
implemented the support of GL_MAX_{WIDTH/HEIGHT/DEPTH} and GL_SAMPLES.
Additionally, doing this way, we avoid adding checks on the code, as
are done by the call to the query itself.
MAX_COMBINED_DIMENSIONS is the only pname pointed on the spec of
needing a 64-bit query. We handle that possibility by packing the
returning value on the two first 32-bit integers of params. This
would work on the 32-bit query as far as the value is not greater
that INT_MAX. On the 64-bit query wrapper we unpack those values
in order to get the final value.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Implemented by calling GetIntegerv with the equivalent pname and
handling individually the exceptions related to dimensions.
All those pnames are used to get the maximum value for each dimension
of the given target. The only difference between this calls and
calling GetInteger with pnames like GL_MAX_TEXTURE_SIZE,
GL_MAX_3D_TEXTURE_SIZE, etc is that GetInternalformat allows to
specify a internalformat.
But at this moment, there is no reason to think that the values would
be different based on the internalformat. The spec already take that
into account, using these specific pnames as example on Issue 7 of
arb_internalformat_query2 spec.
So this seems like a hook to allow to return different values based on
the internalformat in the future.
It is worth to note that the piglit test associated to those pnames
are checking the returned values of GetInternalformat against the
values returned by GetInteger, and the test is passing with NVIDIA
proprietary drivers.
main/formatquery: support for MAX_{WIDTH/HEIGHT/DEPTH/LAYERS}
Implemented by calling GetIntegerv with the equivalent pname and
handling individually the exceptions related to dimensions.
All those pnames are used to get the maximum value for each dimension
of the given target. The only difference between this calls and
calling GetInteger with pnames like GL_MAX_TEXTURE_SIZE,
GL_MAX_3D_TEXTURE_SIZE, etc is that GetInternalformat allows to
specify a internalformat.
But at this moment, there is no reason to think that the values would
be different based on the internalformat. The spec already take that
into account, using these specific pnames as example on Issue 7 of
arb_internalformat_query2 spec.
So this seems like a hook to allow to return different values based on
the internalformat in the future.
It is worth to note that the piglit test associated to those pnames
are checking the returned values of GetInternalformat against the
values returned by GetInteger, and the test is passing with NVIDIA
proprietary drivers.
v2: use _mesa_has## instead of direct ctx->Extensions access (Nanley Chery)
Reviewed-by: Dave Airlie <airlied@redhat.com>
From arb_internalformat_query2 spec:
"IMAGE_FORMAT_COMPATIBILITY_TYPE: The matching criteria use for the
resource when used as an image textures is returned in
<params>. This is equivalent to calling GetTexParameter with <value>
set to IMAGE_FORMAT_COMPATIBILITY_TYPE."
Current implementation of GetTexParameter for this case returns a
field of a texture object, so the support of this pname was
implemented creating a temporal texture object and returning that
value.
It is worth to mention that right now that field is not reassigned
after initialization. So it is somehow hardcoded. An alternative
option would be return that value. That doesn't seems really scalable
though.
v2: use _mesa_has## instead of direct ctx->Extensions access (Nanley Chery)
Reviewed-by: Dave Airlie <airlied@redhat.com>
From arb_internalformat_query2 spec:
" If <internalformat> is not color-renderable, depth-renderable, or
stencil-renderable (as defined in section 4.4.4), or if <target>
does not support multiple samples (ie other than
TEXTURE_2D_MULTISAMPLE, TEXTURE_2D_MULTISAMPLE_ARRAY, or
RENDERBUFFER), <params> is not modified."
So there are cases where the buffer should not be modified. As the
64-bit query is a wrapper over the 32-bit query, we can't just copy
the values to the equivalent 32-bit buffer, as that would fail if the
original params contained values greater that INT_MAX. So we need to
copy-back only the values that got modified by the 32-bit query. We do
that by filling the temporal buffer by negatives, as the 32-bit query
should not return negative values ever.
Reviewed-by: Dave Airlie <airlied@redhat.com>
It just does a wrapping on the existing 32-bit GetInternalformativ.
We will maintain the 32-bit query as default as it is likely that
it would be the one most used.
Reviewed-by: Dave Airlie <airlied@redhat.com>
From the ARB_internalformat_query2 spec:
"- INTERNALFORMAT_RED_SIZE
- INTERNALFORMAT_GREEN_SIZE
- INTERNALFORMAT_BLUE_SIZE
- INTERNALFORMAT_ALPHA_SIZE
- INTERNALFORMAT_DEPTH_SIZE
- INTERNALFORMAT_STENCIL_SIZE
- INTERNALFORMAT_SHARED_SIZE
For uncompressed internal formats, queries of these values return the
actual resolutions that would be used for storing image array components
for the resource.
For compressed internal formats, the resolutions returned specify the
component resolution of an uncompressed internal format that produces
an image of roughly the same quality as the compressed algorithm.
For textures this query will return the same information as querying
GetTexLevelParameter{if}v for TEXTURE_*_SIZE would return.
If the internal format is unsupported, or if a particular component is
not present in the format, 0 is written to <params>.
- INTERNALFORMAT_RED_TYPE
- INTERNALFORMAT_GREEN_TYPE
- INTERNALFORMAT_BLUE_TYPE
- INTERNALFORMAT_ALPHA_TYPE
- INTERNALFORMAT_DEPTH_TYPE
- INTERNALFORMAT_STENCIL_TYPE
For uncompressed internal formats, queries for these values return the
data type used to store the component.
For compressed internal formats the types returned specify how components
are interpreted after decompression.
For textures this query returns the same information as querying
GetTexLevelParameter{if}v for TEXTURE_*TYPE would return.
Possible values return include, NONE, SIGNED_NORMALIZED,
UNSIGNED_NORMALIZED, FLOAT, INT, UNSIGNED_INT, representing missing,
signed normalized fixed point, unsigned normalized fixed point,
floating-point, signed unnormalized integer and unsigned unnormalized
integer components. NONE is returned for all component types if the
format is unsupported."
Reviewed-by: Dave Airlie <airlied@redhat.com>
The new pnames accepted by the function are:
- INTERNALFORMAT_RED_SIZE
- INTERNALFORMAT_GREEN_SIZE
- INTERNALFORMAT_BLUE_SIZE
- INTERNALFORMAT_ALPHA_SIZE
- INTERNALFORMAT_DEPTH_SIZE
- INTERNALFORMAT_STENCIL_SIZE
It will be used by the ARB_internalformat_query2 implementation to
implement those pnames.
Reviewed-by: Dave Airlie <airlied@redhat.com>
The new pnames accepted by the function are:
- INTERNALFORMAT_RED_SIZE
- INTERNALFORMAT_GREEN_SIZE
- INTERNALFORMAT_BLUE_SIZE
- INTERNALFORMAT_ALPHA_SIZE
- INTERNALFORMAT_DEPTH_SIZE
- INTERNALFORMAT_STENCIL_SIZE
- INTERNALFORMAT_RED_TYPE
- INTERNALFORMAT_GREEN_TYPE
- INTERNALFORMAT_BLUE_TYPE
- INTERNALFORMAT_ALPHA_TYPE
- INTERNALFORMAT_DEPTH_TYPE
- INTERNALFORMAT_STENCIL_TYPE
It will be used by the ARB_internalformat_query2 implementation to
implement those pnames.
Reviewed-by: Dave Airlie <airlied@redhat.com>
It will be used by the ARB_internalformat_query2 implementation
to check if the target is valid for those <pnames> that are said
in the spec that should return the same values than the
'glGetTexLevelParameter{if}v' function:
- INTERNALFORMAT_RED_SIZE
- INTERNALFORMAT_GREEN_SIZE
- INTERNALFORMAT_BLUE_SIZE
- INTERNALFORMAT_ALPHA_SIZE
- INTERNALFORMAT_DEPTH_SIZE
- INTERNALFORMAT_STENCIL_SIZE
- INTERNALFORMAT_SHARED_SIZE
- INTERNALFORMAT_RED_TYPE
- INTERNALFORMAT_GREEN_TYPE
- INTERNALFORMAT_BLUE_TYPE
- INTERNALFORMAT_ALPHA_TYPE
- INTERNALFORMAT_DEPTH_TYPE
- INTERNALFORMAT_STENCIL_TYPE
- IMAGE_FORMAT_COMPATIBILITY_TYPE
Reviewed-by: Dave Airlie <airlied@redhat.com>
From the ARB_internalformat_query2 specification:
"The INTERNALFORMAT_SUPPORTED <pname> can be used to determine if
the internal format is supported, and the other <pnames> are defined
in terms of whether or not the format is supported."
v2: Consider also FBO base formats when checking if the internalformat is
supported.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Checks that the 'resource', as defined by the ARB_internalformat_query2
specification, is supported by the implementation for those 'pnames'
that require this check.
Reviewed-by: Dave Airlie <airlied@redhat.com>
This would allow to use this method if you are just querying if it is
allowed, like for arb_internalformat_query2.
Reviewed-by: Dave Airlie <airlied@redhat.com>
It will be used by the ARB_internalformat_query2 implementation
to check if a certain compressed 'internalformat' is supported
by texture 'targets'.
Reviewed-by: Dave Airlie <airlied@redhat.com>
It will be used by the ARB_internalformat_query2 implementation
to check if the 'internalformat' passed is supported by texture
MULTISAMPLE 'targets'.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Equivalent to commit bda540 (that added GL_ARB_internalformat_query)
v2: include the new xml to to API_XML list at Makefile.am (Emil Velikov)
Reviewed-by: Dave Airlie <airlied@redhat.com>
The goal is to extend the GetInternalformativ query to implement the
ARB_internalformat_query2 specification, keeping the behaviour defined
by the ARB_internalformat_query if ARB_internalformat_query2 is not
supported.
v2: Don't require ARB_internalformat_query when profile is GLES3.
Reviewed-by: Dave Airlie <airlied@redhat.com>
From the ARB_internalformat_query2 spec:
"If the particular <target> and <internalformat> combination do not make
sense, or if a particular type of <target> is not supported by the
implementation the "unsupported" answer should be given. This is not an
error."
This function checks if the <target> is supported by the implementation.
v2: Allow RENDERBUFFER targets also on GLES 3 profiles.
Reviewed-by: Dave Airlie <airlied@redhat.com>
The ARB_internalformat_query2 specification defines which is the
reponse best representing "not supported" or "not applicable" for
each <pname>.
Queries for unsupported features, targets, internalformats, combinations
of: target and internalformat, target and pname, pname and internalformat,
do not return an error but the corresponding 'unsupported' response.
We will use that response as the default answer.
For SAMPLES the 'unsupported' response is to not modify the 'params' buffer.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Handles the cases where an error should be returned according
to the ARB_internalformat_query and ARB_internalformat_query2
specifications.
Reviewed-by: Dave Airlie <airlied@redhat.com>
At this point, all uses have been replaced by the more general hook
QueryInternalFormat, introduced by ARB_internalformat_query2.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Implements SAMPLES and NUM_SAMPLE_COUNTS queries using the new generic
driver call QueryInternalFormat, which is being introduced as replacement
of QuerySamplesForFormat to support ARB_internalformat_query2.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Currently, the number of integers returned in the response to
GetInternalFormativ is being tracked by a 'count' variable.
This is so only the modified elements from the temporary buffer are copied into
the original user buffer.
However, with the introduction of ARB_internalformat_query2, keeping track
of 'count' would complicate the code a lot, considering the high number of
queries.
So, we propose to forget about tracking count, and move all the 16 elements
in the temporary buffer, back to the user buffer (clamped to user buffer size
of course). This is basically a trade-off between performance and code clarity.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Use QueryInternalFormat instead of QuerySamplesForFormat to obtain the
highest supported sample. QuerySamplesForFormat is to be removed.
Reviewed-by: Dave Airlie <airlied@redhat.com>
This effectively disables old QuerySamplesForFormat driver hook, since it is
never called by Mesa anymore.
v2: Call brw_query_samples_for_format() with a dummy buffer to calculate num
samples, to avoid modifying the original buffer.
Reviewed-by: Dave Airlie <airlied@redhat.com>
By default, we call back the driver's hook fallback function that has generic
implementations for the all the queries.
Reviewed-by: Dave Airlie <airlied@redhat.com>
This new function queries different driver parameters for a particular target
and texture format. It is basically a driver hook to support
ARB_internalformat_query2.
Since ARB_internalformat_query2 introduced several new query parameters
over ARB_internalformat_query, having one driver hook for each parameter
is no longer feasible. So this is the generic entry-point for calls
to glGetInternalFormativ and glGetInternalFormati64v.
Reviewed-by: Dave Airlie <airlied@redhat.com>
When we find indirect indexing into an array, the current implementation
of the array spliiting optimization pass does not look further into the
expression tree. However, if the variable expression involves variable
indexing into other arrays, we can miss that these other arrays also have
variable indexing. If that happens, the pass will crash later on after
hitting an assertion put there to ensure that split arrays are in fact
always indexed via constants:
shader_runner: opt_array_splitting.cpp:296:
void ir_array_splitting_visitor::split_deref(ir_dereference**): Assertion `constant' failed.
This patch fixes the problem by letting the pass step into the variable
index expression to identify these cases properly.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89607
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
There is an old if statement (dated to 2011) that prevented doing
endian swap for colorformat, in case the buffer is marked as
PIPE_USAGE_STAGING.
This is now wrong because st_ReadPixels() reads into a destination
texture that is marked with PIPE_USAGE_STAGING. Therefore, even if
the texture is rendered correctly to the monitor, when reading it
back we get unswapped/wrong values.
This patch makes the check_rgba() function in gl-1.0-readpixsanity
piglit test pass in big-endian.
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
There is an old if statement (dated to 2011) that prevented doing
endian swap for colorformat, in case the buffer is marked
as PIPE_USAGE_STAGING.
This is now wrong because st_ReadPixels() reads into a destination
texture that is marked with PIPE_USAGE_STAGING. Therefore, even if
the texture is rendered correctly to the monitor, when reading it
back we get unswapped/wrong values.
This patch makes the check_rgba() function in gl-1.0-readpixsanity
piglit test pass in big-endian.
v2: removed duplicate call to r600_colorformat_endian_swap() inside
evergreen_init_color_surface_rat()
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested on Linux (centos, ubuntu, and suse variants)
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Acked-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Acked-by: Jose Fonseca <jfonseca@vmware.com>
swr driver which is written in C++ needs access to some more
gallium utility functions than are currently exposed.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Acked-by: Jose Fonseca <jfonseca@vmware.com>
OpenSWR is a new software rasterizer for x86 processors designed
for high performance and high scalablility on visualization workloads.
Acked-by: Roland Scheidegger <sroland@vmware.com>
Acked-by: Jose Fonseca <jfonseca@vmware.com>
Texture is already allocated before calling this meta function. So,
the value of 'allocate_storage' passed to the function is always false.
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Antia Puentes
Alejandro Piñeiro
Eduardo Lima Mitev
Iago Toral Quiroga
Oded Gabbay
Tim Rowley
Timothy Arceri
Anuj Phogat