Re: [Mesa-dev] [Mesa-stable] [PATCH] Revert "glsl: don't reference shader prog data during cache fallback"

[Juan A. Suarez Romero]

On Thu, 2017-06-01 at 10:09 +1000, Timothy Arceri wrote:
> On 01/06/17 07:30, Juan A. Suarez Romero wrote:
> > On Tue, 2017-05-16 at 20:40 +0200, Nicolai Hähnle wrote:
> > > From: Nicolai Hähnle 
> > > 
> > > This reverts commit 0e9991f957e296f46cfff40a94ffba0adf2a58e1.
> > > 
> > > At least when we fallback because of TGSI, the gl_program objects are
> > > re-allocated, and we don't in fact already have a reference to the
> > > gl_shader_program_data.
> > > 
> > 
> > This patch didn't land in master yet. Any news on this?
> > 
> > Also, commit 80e64334 (st/mesa: don't mark the program as in
> > cache_fallback when there is cache miss) fixes the patch we are
> > reverting. Probably we will need to revert too.
> > 
> > 
> > J.A.
> 
> Hi,
> 
> First of all sorry for the confusion. Please ignore this patch, 80e64334 
> is the one we would like to land in stable.
> 
> Please let me know if there are any issue applying that one to stable.
> 

Thank you! I already queued that patch and found no issues so far.


J.A.

> Thanks,
> Tim
> 
> > 
> > 
> > 
> > > This fixes a crash that can be reproduced as follows:
> > > 
> > > 1. Run any GL program with MESA_GLSL=cache_info.
> > > 2. Note the SHA of some tgsi_tokens.
> > > 3. Remove the corresponding file from ~/.cache/mesa and re-run.
> > > 
> > > Cc: 17.1 
> > > ---
> > >   src/compiler/glsl/linker.cpp | 3 +--
> > >   src/mesa/main/shaderobj.c| 3 +--
> > >   2 files changed, 2 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/src/compiler/glsl/linker.cpp b/src/compiler/glsl/linker.cpp
> > > index 2e7dd2b..4776d09 100644
> > > --- a/src/compiler/glsl/linker.cpp
> > > +++ b/src/compiler/glsl/linker.cpp
> > > @@ -2243,22 +2243,21 @@ link_intrastage_shaders(void *mem_ctx,
> > >  struct gl_program *gl_prog =
> > > ctx->Driver.NewProgram(ctx,
> > >
> > > _mesa_shader_stage_to_program(shader_list[0]->Stage),
> > >prog->Name, false);
> > >  if (!gl_prog) {
> > > prog->data->LinkStatus = linking_failure;
> > > _mesa_delete_linked_shader(ctx, linked);
> > > return NULL;
> > >  }
> > >   
> > > -   if (!prog->data->cache_fallback)
> > > -  _mesa_reference_shader_program_data(ctx, &gl_prog->sh.data, 
> > > prog->data);
> > > +   _mesa_reference_shader_program_data(ctx, &gl_prog->sh.data, 
> > > prog->data);
> > >   
> > >  /* Don't use _mesa_reference_program() just take ownership */
> > >  linked->Program = gl_prog;
> > >   
> > >  linked->ir = new(linked) exec_list;
> > >  clone_ir_list(mem_ctx, linked->ir, main->ir);
> > >   
> > >  link_fs_inout_layout_qualifiers(prog, linked, shader_list, 
> > > num_shaders);
> > >  link_tcs_out_layout_qualifiers(prog, gl_prog, shader_list, 
> > > num_shaders);
> > >  link_tes_in_layout_qualifiers(prog, gl_prog, shader_list, 
> > > num_shaders);
> > > diff --git a/src/mesa/main/shaderobj.c b/src/mesa/main/shaderobj.c
> > > index 8a5fa5e..f6f7bf5 100644
> > > --- a/src/mesa/main/shaderobj.c
> > > +++ b/src/mesa/main/shaderobj.c
> > > @@ -427,22 +427,21 @@ _mesa_free_shader_program_data(struct gl_context 
> > > *ctx,
> > >   
> > >   
> > >   /**
> > >* Free/delete a shader program object.
> > >*/
> > >   void
> > >   _mesa_delete_shader_program(struct gl_context *ctx,
> > >   struct gl_shader_program *shProg)
> > >   {
> > >  _mesa_free_shader_program_data(ctx, shProg);
> > > -   if (!shProg->data->cache_fallback)
> > > -  _mesa_reference_shader_program_data(ctx, &shProg->data, NULL);
> > > +   _mesa_reference_shader_program_data(ctx, &shProg->data, NULL);
> > >  ralloc_free(shProg);
> > >   }
> > >   
> > >   
> > >   /**
> > >* Lookup a GLSL program object.
> > >*/
> > >   struct gl_shader_program *
> > >   _mesa_lookup_shader_program(struct gl_context *ctx, GLuint name)
> > >   {
> > 
> > ___
> > mesa-dev mailing list
> > mesa-dev@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
> > 
> 
> ___
> mesa-stable mailing list
> mesa-sta...@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-stable
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] i965: Reject unsupported formats in glEGLImageTargetTexture2D()

[Kenneth Graunke]

On Wednesday, May 31, 2017 4:54:59 PM PDT Chad Versace wrote:
> If the EGLImage's format is not a supported texture format according to
> brw_surface_formats.c, then refuse to create the miptree. This follows
> the precedent in glEGLImageRenderbufferStorage (implemented by
> intel_image_target_renderbuffer_storage), which rejects the EGLImage's
> format if is not renderable.
> 
> Tested on Broadwell with deqp@1ee59ff9 on X11/EGL:
> 
> > ./deqp-egl --deqp-case='dEQP-EGL.functional.image.*'
> Test run totals:
>   Passed:131/169 (77.5%)
>   Failed:0/169 (0.0%)
>   Not supported: 38/169 (22.5%)
>   Warnings:  0/169 (0.0%)
> ---
>  src/mesa/drivers/dri/i965/intel_tex_image.c | 4 
>  1 file changed, 4 insertions(+)
> 
> Tomasz, you're CC'd because this is related to R8G8B8X8 EGLConfigs on
> Skylake.
> 
> diff --git a/src/mesa/drivers/dri/i965/intel_tex_image.c 
> b/src/mesa/drivers/dri/i965/intel_tex_image.c
> index 7208d8e..649b390 100644
> --- a/src/mesa/drivers/dri/i965/intel_tex_image.c
> +++ b/src/mesa/drivers/dri/i965/intel_tex_image.c
> @@ -251,9 +251,13 @@ static struct intel_mipmap_tree *
>  create_mt_for_dri_image(struct brw_context *brw,
>  GLenum target, __DRIimage *image)
>  {
> +   struct gl_context *ctx = &brw->ctx;
> struct intel_mipmap_tree *mt;
> uint32_t draw_x, draw_y;
>  
> +   if (!ctx->TextureFormatSupported[image->format])
> +  return NULL;
> +
> /* Disable creation of the texture's aux buffers because the driver 
> exposes
>  * no EGL API to manage them. That is, there is no API for resolving the 
> aux
>  * buffer's content to the main buffer nor for invalidating the aux 
> buffer's
> 

Sounds reasonable to me.  Thanks Chad!

Reviewed-by: Kenneth Graunke 


signature.asc
Description: This is a digitally signed message part.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 0/1] r600g: work around shaders allocating too many superflous temporaries

[Gert Wollny]

Hello, 

I should probably add that I don't have write access to the mesa-git. 

Best, 
Gert 


Am Dienstag, den 30.05.2017, 14:14 +0200 schrieb Gert Wollny:
> Dear all,
> 
> like I pointed out yesterday I've got a workaround for shaders that
> allocate
> too many temporaries after the TGSI pass that can actually be
> discarded in the
> sb pass but would fail before that in the tgsi-to-bc stage.
> 
> The patch should not interfere with most shaders that worked before,
> the only
> problem could be with shaders that reserve a valid number of GPRs
> close to the
> limit of 124.
> 
> Since I'm only starting to learn how the tgsi to bytecode translation
> works
> I'm not sure about the implications when temporaries are reserved in
> the
> kcache bank range (even though they are later discarded). Maybe
> r600_shader_from_tgsi() should actually return with an error
> the moment ctx.temp_reg >= 124 orAm Dienstag, den 30.05.2017, 14:14
> +0200 schrieb Gert Wollny:
> > Dear all,
> > 
> > like I pointed out yesterday I've got a workaround for shaders that
> > allocate
> > too many temporaries after the TGSI pass that can actually be
> > discarded in the
> > sb pass but would fail before that in the tgsi-to-bc stage.
> > 
> > The patch should not interfere with most shaders that worked
> before,
> > the only
> > problem could be with shaders that reserve a valid number of GPRs
> > close to the
> > limit of 124.
> > 
> > Since I'm only starting to learn how the tgsi to bytecode
> translation
> > works
> > I'm not sure about the implications when temporaries are reserved
> in
> > the
> > kcache bank range (even though they are later discarded). Maybe
> > r600_shader_from_tgsi() should actually return with an error
> > the moment ctx.temp_reg >= 124 or r600_get_temp() > 124.
> > 
> > I'm also aware the it would actually be better to optimize the
> shader
> > in the
> > TGSI representation in order to avoid the problem altogether, but
> > before I
> > can look into that I'll probably have t learn a lot more about the
> > code.
> > 
> > In any case, this patch fixes the problem I had with the Unreal
> > Editor, the tests
> > are still passing, piglit didn't show me any regressions, and so
> far
> > I also didn't
> > see breakage in other program using mesa.
> > 
> > Many thanks for any comments,
> > Gert Wollny (1):
> >   r600g: work around shaders allocating too many superflous
> > temporaries
> > 
> >  src/gallium/drivers/r600/r600_shader.c | 52
> > +++---
> >  1 file changed, 42 insertions(+), 10 deletions(-) r600_get_temp()
> > 124.
> 
> I'm also aware the it would actually be better to optimize the shader
> in the TGSI representation in order to avoid the problem altogether,
> but before I can look into that I'll probably have t learn a lot more
> about the code.
> 
> In any case, this patch fixes the problem I had with the Unreal
> Editor, the tests are still passing, piglit didn't show me any
> regressions, and so far I also didn't see breakage in other program
> using mesa.
> 
> Many thanks for any comments,
> Gert Wollny (1):
>   r600g: work around shaders allocating too many superflous
> temporaries
> 
>  src/gallium/drivers/r600/r600_shader.c | 52
> +++---
>  1 file changed, 42 insertions(+), 10 deletions(-)
> 
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 00/30] i965: Overhaul resolves

[Pohjolainen, Topi]

On Wed, May 31, 2017 at 04:09:00PM +0300, Pohjolainen, Topi wrote:
> On Tue, May 30, 2017 at 11:48:15AM +0300, Pohjolainen, Topi wrote:
> > On Fri, May 26, 2017 at 04:30:04PM -0700, Jason Ekstrand wrote:
> > > This patch series does a complete overhaul of the current resolve handling
> > > framework inside the i916 OpenGL driver.  For HiZ and MCS, the current
> > > resolve code is ok but not optimal.  For CCS, however, it's pretty bad.
> > > I've been looking at the code for a week now and I still don't know how 
> > > Ben
> > > ever got it to do a partial resolve for his CCS modifiers series.  So far
> > > as I can tell, it's not capable of doing so.  The new resolve system is
> > > hopefully much easier to reason about.  For users of the system, there are
> > > fewer entry-points and depth and color are no longer separate.  The guts 
> > > of
> > > the system are much more explicit and, thanks to the new isl_aux_state
> > > enum, should be easier to understand.
> > > 
> > > As of my last Jenkins run, the series is still failing 2 piglit tests on
> > > Sandy Bridge and I have yet to do any benchmarking.  However, I wanted to
> > > send it out early so that I could get feedback on the structure of the
> > > system as quickly as possible.  Discussion of the structure can happen in
> > > parallel with final tweaking.  Personally, I'm fairly happy with it and I
> > > think this looks like a good way to go but I'd like more eyes.
> > > 
> > > The patch series itself is organized as follows
> > > 
> > >  * The first 13 patches are various cleanups which make later patches
> > >simpler.  They should be fairly benign.  These can easily land on their
> > >own as I think most of them are good clean-ups anyway.
> > > 
> > >  * Patch 14 adds the new isl_aux_state enum and the accompanying comment
> > > 
> > >  * Patch 15 adds the new interface for doing resolves.  All of the
> > >functions are just dummies which call the already existing functions.
> > > 
> > >  * Patches 16-26 convert everything over to using the new resolve
> > >interface.  I tried as hard as I could to not make any functional
> > >changes while doing so.  If you see any, they are probably bugs!
> > > 
> > >  * Patch 27 wholesale replaces the current color resolve scheme with a new
> > >one based on isl_aux_state.  It's a bit unfortunate that it all had to
> > >happen in one go but it's not easy to switch resolve schemes slowly.
> > > 
> > >  * Patch 28 replaces the HiZ resolve framework.  This one is not nearly as
> > >drastic as patch 27 because the current HiZ framework is already pretty
> > >good.
> > > 
> > >  * Patch 29 deletes the now unused intel_resolve_map struct
> > > 
> > >  * Patch 30 enables fast-clears for non-CCS_E capable surfaces.  In
> > >particular, this gives us fast-clears for sRGB.
> > > 
> > > I'd appreciate it if the initial review focussed on patches 14, 15, and 
> > > 27.
> > > Those are where you see the new resolve system in action.
> > > 
> > > This series is available here:
> > > 
> > > https://cgit.freedesktop.org/~jekstrand/mesa/log/?h=review/i965-resolve-rework
> > > 
> > > Happy Reviewing!
> > > 
> > > --Jason Ekstrand
> > > 
> > > Cc: Chad Versace 
> > > Cc: Kenneth Graunke 
> > > Cc: Nanley Chery 
> > > Cc: Topi Pohjolainen 
> > > 
> > > Jason Ekstrand (30):
> > >   i965: Mark depth surfaces as needing a HiZ resolve after blitting
> > >   i965/surface_state: Images can't handle CCS at all
> > >   intel/isl: Add a helper for determining if a color is 0/1
> > >   i965/miptree: Store fast clear colors in an isl_color_value
> > >   i965/miptree: Clean up the depth resolve helpers a little
> > >   i965/miptree: Refactor intel_miptree_resolve_color
> > >   i965: Get rid of intel_renderbuffer_resolve_*
> > >   i965: Inline renderbuffer_att_set_needs_depth_resolve
> > >   i965/miptree: Move color resolve on map to intel_miptree_map
> > >   i965/blorp: Take an explicit fast clear op in resolve_color
> > >   i965/blorp: Refactor do_single_blorp_clear
> > >   i965/blorp: Move MCS allocation earlier for clears
> > >   i965: Combine render target resolve code
> > 
> > Patches 3-13:
> > 
> > Reviewed-by: Topi Pohjolainen 
> > 
> > >   intel/isl: Add an enum for describing auxiliary compression state
> 
> This of course looks good, I'm just hoping we could somehow document CCS not
> having RESOLVED/AUX_INVALID.
> 
> > >   i965/miptree: Add new entrypoints for resolve management
> > >   i965: Use the new resolve function for several simple cases
> > >   i965: Finalize miptrees before prepare_texture
> > >   i965: Move texturing to the new resolve functions
> > >   i965: Move color rendering to the new resolve functions
> > >   i965: Remove an unneeded render_cache_set_check_flush
> > >   i965: Move framebuffer fetch to the new resolve functions
> > >   i965: Move images to the new resolve functions
> > >   i965: Move depth to the new resolve functions
> > >   i965: Move blorp to the new r

Re: [Mesa-dev] [PATCH] nvc0: Clean up unnecessary includes from gallium/auxiliary/vl/

[Samuel Pitoiset]

Pushed, thanks!

On 06/01/2017 12:48 AM, Rhys Kidd wrote:

Signed-off-by: Rhys Kidd 
---
  src/gallium/drivers/nouveau/nvc0/nvc0_screen.c | 3 ---
  1 file changed, 3 deletions(-)

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c 
b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
index f2ae4f3ae3..c636926994 100644
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
@@ -27,9 +27,6 @@
  #include "util/u_format_s3tc.h"
  #include "pipe/p_screen.h"
  
-#include "vl/vl_decoder.h"

-#include "vl/vl_video_buffer.h"
-
  #include "nouveau_vp3_video.h"
  
  #include "nvc0/nvc0_context.h"



___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 08/11] intel/isl: Add a new layout for HiZ and stencil on Sandy Bridge

[Pohjolainen, Topi]

On Tue, May 30, 2017 at 05:55:17PM -0700, Jason Ekstrand wrote:
> ---
>  src/intel/isl/isl.c | 162 
> ++--
>  src/intel/isl/isl.h |  40 +
>  2 files changed, 197 insertions(+), 5 deletions(-)
> 
> diff --git a/src/intel/isl/isl.c b/src/intel/isl/isl.c
> index 77b8a40..e4d7901 100644
> --- a/src/intel/isl/isl.c
> +++ b/src/intel/isl/isl.c
> @@ -479,6 +479,12 @@ isl_choose_array_pitch_span(const struct isl_device *dev,
> * compact QPitch possible in order to conserve memory.
> */
>return ISL_ARRAY_PITCH_SPAN_COMPACT;
> +
> +   case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
> +  /* Each array image in the gen6 stencil of HiZ surface is compact in 
> the
> +   * sense that every LOD is a compact array of the same size as LOD0.
> +   */
> +  return ISL_ARRAY_PITCH_SPAN_COMPACT;
> }
>  
> unreachable("bad isl_dim_layout");
> @@ -510,10 +516,15 @@ isl_choose_image_alignment_el(const struct isl_device 
> *dev,
>return;
> } else if (info->format == ISL_FORMAT_HIZ) {
>assert(ISL_DEV_GEN(dev) >= 6);
> -  /* HiZ surfaces are always aligned to 16x8 pixels in the primary 
> surface
> -   * which works out to 2x2 HiZ elments.
> -   */
> -  *image_align_el = isl_extent3d(2, 2, 1);
> +  if (ISL_DEV_GEN(dev) == 6) {
> + /* HiZ surfaces on Sandy Bridge are packed tightly. */
> + *image_align_el = isl_extent3d(1, 1, 1);
> +  } else {
> + /* On gen7+, HiZ surfaces are always aligned to 16x8 pixels in the
> +  * primary surface which works out to 2x2 HiZ elments.
> +  */
> + *image_align_el = isl_extent3d(2, 2, 1);
> +  }
>return;
> }
>  
> @@ -540,6 +551,11 @@ isl_surf_choose_dim_layout(const struct isl_device *dev,
> enum isl_surf_dim logical_dim,
> enum isl_tiling tiling)
>  {
> +   /* Sandy bridge needs a special layout for HiZ and stencil. */
> +   if (ISL_DEV_GEN(dev) == 6 &&
> +   (tiling == ISL_TILING_W || tiling == ISL_TILING_HIZ))
> +  return ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ;
> +
> if (ISL_DEV_GEN(dev) >= 9) {
>switch (logical_dim) {
>case ISL_SURF_DIM_1D:
> @@ -608,6 +624,7 @@ isl_calc_phys_level0_extent_sa(const struct isl_device 
> *dev,
>  
>case ISL_DIM_LAYOUT_GEN9_1D:
>case ISL_DIM_LAYOUT_GEN4_2D:
> +  case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
>   *phys_level0_sa = (struct isl_extent4d) {
>  .w = isl_align_npot(info->width, fmtl->bw),
>  .h = fmtl->bh,
> @@ -619,7 +636,8 @@ isl_calc_phys_level0_extent_sa(const struct isl_device 
> *dev,
>break;
>  
> case ISL_SURF_DIM_2D:
> -  assert(dim_layout == ISL_DIM_LAYOUT_GEN4_2D);
> +  assert(dim_layout == ISL_DIM_LAYOUT_GEN4_2D ||
> + dim_layout == ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ);
>  
>if (tiling == ISL_TILING_Ys && info->samples > 1)
>   isl_finishme("%s:%s: multisample TileYs layout", __FILE__, 
> __func__);
> @@ -684,6 +702,7 @@ isl_calc_phys_level0_extent_sa(const struct isl_device 
> *dev,
>  
>switch (dim_layout) {
>case ISL_DIM_LAYOUT_GEN9_1D:
> +  case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
>   unreachable("bad isl_dim_layout");
>  
>case ISL_DIM_LAYOUT_GEN4_2D:
> @@ -969,6 +988,67 @@ isl_calc_phys_total_extent_el_gen4_3d(
>  
>  /**
>   * A variant of isl_calc_phys_slice0_extent_sa() specific to
> + * ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ.
> + */
> +static void
> +isl_calc_phys_total_extent_el_gen6_stencil_hiz(
> +  const struct isl_device *dev,
> +  const struct isl_surf_init_info *restrict info,
> +  const struct isl_tile_info *tile_info,
> +  const struct isl_extent3d *image_align_sa,
> +  const struct isl_extent4d *phys_level0_sa,
> +  uint32_t *array_pitch_el_rows,
> +  struct isl_extent2d *phys_total_el)
> +{
> +   const struct isl_format_layout *fmtl = 
> isl_format_get_layout(info->format);
> +
> +   const struct isl_extent2d tile_extent_sa = {
> +  .w = tile_info->logical_extent_el.w * fmtl->bw,
> +  .h = tile_info->logical_extent_el.h * fmtl->bh,
> +   };
> +   /* Tile size is a multiple of image alignment */
> +   assert(tile_extent_sa.w % image_align_sa->w == 0);
> +   assert(tile_extent_sa.h % image_align_sa->h == 0);
> +
> +   const uint32_t W0 = phys_level0_sa->w;
> +   const uint32_t H0 = phys_level0_sa->h;
> +
> +   /* Each image has the same height as LOD0 because the hardware thinks
> +* everything is LOD0
> +*/
> +   const uint32_t H = isl_align(H0, image_align_sa->h) * phys_level0_sa->a;
> +
> +   uint32_t slice_top_w = 0;
> +   uint32_t slice_bottom_w = 0;
> +   uint32_t slice_h = 0;

I don't really understand why slice_bottom_w and slice_h are prefixed with
slice, I would rather call them total_bottom_w and total_h.

> +
> +   for (uint32_t l = 0; l < info->levels; ++l) {
> +  const uint3

[Mesa-dev] [PATCH] util/rand_xor: add missing include statements

[Nicolas Dechesne]

Fixes for:

src/util/rand_xor.c:60:13: error: implicit declaration of function 'open' 
[-Werror=implicit-function-declaration]
int fd = open("/dev/urandom", O_RDONLY);
 ^~~~
src/util/rand_xor.c:60:34: error: 'O_RDONLY' undeclared (first use in this 
function)
int fd = open("/dev/urandom", O_RDONLY);
  ^~~~

Signed-off-by: Nicolas Dechesne 
---
 src/util/rand_xor.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/src/util/rand_xor.c b/src/util/rand_xor.c
index de05fa64b3..de04bbc284 100644
--- a/src/util/rand_xor.c
+++ b/src/util/rand_xor.c
@@ -25,6 +25,7 @@
 #if defined(__linux__)
 #include 
 #include 
+#include 
 #else
 #include 
 #endif
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 11/11] intel/blorp: Handle gen6 stencil/HiZ offsets in the back-end

[Pohjolainen, Topi]

On Tue, May 30, 2017 at 05:55:20PM -0700, Jason Ekstrand wrote:
> ---
>  src/intel/blorp/blorp_genX_exec.h| 32 -
>  src/intel/isl/isl_emit_depth_stencil.c   |  1 +
>  src/mesa/drivers/dri/i965/brw_blorp.c| 61 
> 
>  src/mesa/drivers/dri/i965/brw_wm_surface_state.c |  3 +-
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.c| 17 ---
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.h|  2 +-
>  6 files changed, 42 insertions(+), 74 deletions(-)
> 
> diff --git a/src/intel/blorp/blorp_genX_exec.h 
> b/src/intel/blorp/blorp_genX_exec.h
> index 8b9b8d2..59c1d36 100644
> --- a/src/intel/blorp/blorp_genX_exec.h
> +++ b/src/intel/blorp/blorp_genX_exec.h
> @@ -1384,9 +1384,23 @@ blorp_emit_depth_stencil_config(struct blorp_batch 
> *batch,
>if (info.hiz_usage == ISL_AUX_USAGE_HIZ) {
>   info.hiz_surf = ¶ms->depth.aux_surf;
>  
> + struct blorp_address hiz_address = params->depth.aux_addr;
> +#if GEN_GEN == 6
> + /* Sandy bridge hardware does not technically support mipmapped HiZ.
> +  * However, we have a special layout that allows us to make it work
> +  * anyway by manually offsetting to the specified miplevel.
> +  */
> + assert(info.hiz_surf->dim_layout == 
> ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ);
> + uint32_t offset_B;
> + isl_surf_get_image_offset_B_tile_sa(info.hiz_surf,
> + info.view->base_level, 0, 0,
> + &offset_B, NULL, NULL);
> + hiz_address.offset += offset_B;
> +#endif
> +
>   info.hiz_address =
>  blorp_emit_reloc(batch, dw + isl_dev->ds.hiz_offset / 4,
> - params->depth.aux_addr, 0);
> + hiz_address, 0);
>  
>   info.depth_clear_value = params->depth.clear_color.u32[0];
>}
> @@ -1395,9 +1409,23 @@ blorp_emit_depth_stencil_config(struct blorp_batch 
> *batch,
> if (params->stencil.enabled) {
>info.stencil_surf = ¶ms->stencil.surf;
>  
> +  struct blorp_address stencil_address = params->stencil.addr;
> +#if GEN_GEN == 6
> +  /* Sandy bridge hardware does not technically support mipmapped 
> stencil.
> +   * However, we have a special layout that allows us to make it work
> +   * anyway by manually offsetting to the specified miplevel.
> +   */
> +  assert(info.stencil_surf->dim_layout == 
> ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ);
> +  uint32_t offset_B;
> +  isl_surf_get_image_offset_B_tile_sa(info.stencil_surf,
> +  info.view->base_level, 0, 0,
> +  &offset_B, NULL, NULL);
> +  stencil_address.offset += offset_B;
> +#endif
> +
>info.stencil_address =
>   blorp_emit_reloc(batch, dw + isl_dev->ds.stencil_offset / 4,
> -  params->stencil.addr, 0);
> +  stencil_address, 0);
> }
>  
> isl_emit_depth_stencil_hiz_s(isl_dev, dw, &info);
> diff --git a/src/intel/isl/isl_emit_depth_stencil.c 
> b/src/intel/isl/isl_emit_depth_stencil.c
> index 41a01be..04bed07 100644
> --- a/src/intel/isl/isl_emit_depth_stencil.c
> +++ b/src/intel/isl/isl_emit_depth_stencil.c
> @@ -158,6 +158,7 @@ isl_genX(emit_depth_stencil_hiz_s)(const struct 
> isl_device *dev, void *batch,
>hiz.SurfaceBaseAddress = info->hiz_address;
>hiz.HierarchicalDepthBufferMOCS = info->mocs;
>hiz.SurfacePitch = info->hiz_surf->row_pitch - 1;
> +

Leftover?

>  #if GEN_GEN >= 8
>/* From the SKL PRM Vol2a:
> *
> diff --git a/src/mesa/drivers/dri/i965/brw_blorp.c 
> b/src/mesa/drivers/dri/i965/brw_blorp.c
> index 61c6bda..28be620 100644
> --- a/src/mesa/drivers/dri/i965/brw_blorp.c
> +++ b/src/mesa/drivers/dri/i965/brw_blorp.c
> @@ -118,36 +118,6 @@ brw_blorp_init(struct brw_context *brw)
>  }
>  
>  static void
> -apply_gen6_stencil_hiz_offset(struct isl_surf *surf,
> -  struct intel_mipmap_tree *mt,
> -  uint32_t lod,
> -  uint32_t *offset)
> -{
> -   assert(mt->array_layout == GEN6_HIZ_STENCIL);
> -
> -   if (mt->format == MESA_FORMAT_S_UINT8) {
> -  /* Note: we can't compute the stencil offset using
> -   * intel_miptree_get_aligned_offset(), because the miptree
> -   * claims that the region is untiled even though it's W tiled.
> -   */
> -  *offset = mt->level[lod].level_y * mt->pitch +
> -mt->level[lod].level_x * 64;
> -   } else {
> -  *offset = intel_miptree_get_aligned_offset(mt,
> - mt->level[lod].level_x,
> - mt->level[lod].level_y);
> -   }
> -
> -   surf->logical_level0_px.width = minify(surf->logical_level0_px.width, 
> lod);
> -   surf->logical_level0_px.height

Re: [Mesa-dev] [PATCH 00/11] intel/isl: Add support for HiZ and stencil on gen6

[Pohjolainen, Topi]

On Tue, May 30, 2017 at 05:55:09PM -0700, Jason Ekstrand wrote:
> This little series adds support to ISL for gen6 HiZ ans stencil via a new
> isl_dim_layout enum value.  Unfortunately, I couldn't easily add the new
> layout without doing a little refactoring.  The way ISL likes to calculate
> surface layouts today is fairly heavily based on the needs of 2D array
> textures.  The gen4 3D and gen9 1D layouts are bolted on but not really
> first-class citizens in some sense.  In order to better accommodate more
> exotic layouts, I re factored things so that the primary layout helper
> generates the total surface extent in elements instead of the slice0 extent
> that they were generating before.  I think the new code is somewhat easier
> to follow especially for 1D and 3D images.
> 
> The last patch moves some things around so that BLORP now handles gen6 HiZ
> and stencil properly in the back-end and there is no longer any need for
> brw_blorp to do surface offsetting.  I'm hoping that this will make Topi's
> life much easier as he tries to rework his "convert i965 to ISL" patches to
> accommodate the recent revelation about gen6 HiZ and stencil.

I had a few questions but looks good. Series:

Reviewed-by: Topi Pohjolainen 

I'll try to rebase my ISL work once I have figured out a few regressions I'm
seeing currently with depth surfaces.

> 
> Cc: Chad Versace 
> Cc: Nanley Chery 
> Cc: Topi Pohjolainen 
> 
> Jason Ekstrand (11):
>   intel/isl: Remove the device parameter from isl_tiling_get_info
>   intel/isl: Move isl_calc_array_pitch_el_rows higher up
>   intel/isl: Refactor isl_calc_array_pitch_el_rows
>   intel/isl: Add an isl_assert_div helper
>   intel/isl: Refactor to use a phys_total_el extent.
>   intel/isl: Don't check array pitch for gen4 3D textures
>   intel/isl: Generate phys_total_el from isl_calc_phys_extent
>   intel/isl: Add a new layout for HiZ and stencil on Sandy Bridge
>   intel/isl: Make get_intratile_offset_el take the element size in bits
>   intel/isl: Add a helper for getting the byte/tile offset of a subimage
>   intel/blorp: Handle gen6 stencil/HiZ offsets in the back-end
> 
>  src/intel/blorp/blorp_blit.c   |  15 +-
>  src/intel/blorp/blorp_genX_exec.h  |  32 +-
>  src/intel/isl/isl.c| 619 
> ++---
>  src/intel/isl/isl.h|  76 ++-
>  src/intel/isl/isl_emit_depth_stencil.c |   1 +
>  src/intel/isl/isl_priv.h   |   7 +
>  src/intel/isl/isl_surface_state.c  |   2 +-
>  .../isl/tests/isl_surf_get_image_offset_test.c |   1 -
>  src/mesa/drivers/dri/i965/brw_blorp.c  |  61 --
>  src/mesa/drivers/dri/i965/brw_wm_surface_state.c   |   3 +-
>  src/mesa/drivers/dri/i965/intel_blit.c |   3 +-
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.c  |  17 +-
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.h  |   2 +-
>  13 files changed, 535 insertions(+), 304 deletions(-)
> 
> -- 
> 2.5.0.400.gff86faf
> 
> ___
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH v2] egl/android: support for EGL_KHR_partial_update

[Harish Krupo]

This patch adds support for the EGL_KHR_partial_update extension for
android platform. It passes 36/37 tests in dEQP for EGL_KHR_partial_update.
1 test not supported.

v2: add fallback for eglSetDamageRegionKHR (Tapani)

Signed-off-by: Harish Krupo 
---
 src/egl/drivers/dri2/egl_dri2.c | 12 +
 src/egl/drivers/dri2/egl_dri2.h |  4 ++
 src/egl/drivers/dri2/platform_android.c | 33 +
 src/egl/main/eglapi.c   | 87 +
 src/egl/main/eglapi.h   |  2 +
 src/egl/main/egldisplay.h   |  1 +
 src/egl/main/eglentrypoint.h|  1 +
 src/egl/main/eglfallbacks.c |  1 +
 src/egl/main/eglsurface.c   |  8 +++
 src/egl/main/eglsurface.h   | 12 +
 10 files changed, 161 insertions(+)

diff --git a/src/egl/drivers/dri2/egl_dri2.c b/src/egl/drivers/dri2/egl_dri2.c
index 0be7132ac5..c58efbf4d6 100644
--- a/src/egl/drivers/dri2/egl_dri2.c
+++ b/src/egl/drivers/dri2/egl_dri2.c
@@ -1464,6 +1464,17 @@ dri2_swap_buffers_region(_EGLDriver *drv, _EGLDisplay 
*dpy, _EGLSurface *surf,
 }
 
 static EGLBoolean
+dri2_set_damage_region(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf,
+   EGLint *rects, EGLint n_rects)
+{
+   struct dri2_egl_display *dri2_dpy = dri2_egl_display(dpy);
+   if (dri2_dpy->vtbl->set_damage_region)
+  return dri2_dpy->vtbl->set_damage_region(drv, dpy, surf, rects, n_rects);
+
+   return EGL_FALSE;
+}
+
+static EGLBoolean
 dri2_post_sub_buffer(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf,
  EGLint x, EGLint y, EGLint width, EGLint height)
 {
@@ -2958,6 +2969,7 @@ _eglBuiltInDriverDRI2(const char *args)
dri2_drv->base.API.SwapBuffers = dri2_swap_buffers;
dri2_drv->base.API.SwapBuffersWithDamageEXT = dri2_swap_buffers_with_damage;
dri2_drv->base.API.SwapBuffersRegionNOK = dri2_swap_buffers_region;
+   dri2_drv->base.API.SetDamageRegion = dri2_set_damage_region;
dri2_drv->base.API.PostSubBufferNV = dri2_post_sub_buffer;
dri2_drv->base.API.CopyBuffers = dri2_copy_buffers,
dri2_drv->base.API.QueryBufferAge = dri2_query_buffer_age;
diff --git a/src/egl/drivers/dri2/egl_dri2.h b/src/egl/drivers/dri2/egl_dri2.h
index fdeffb3a85..d2aeda2c83 100644
--- a/src/egl/drivers/dri2/egl_dri2.h
+++ b/src/egl/drivers/dri2/egl_dri2.h
@@ -118,6 +118,10 @@ struct dri2_egl_display_vtbl {
   _EGLSurface *surface,
   const EGLint *rects, EGLint n_rects);
 
+   EGLBoolean (*set_damage_region)(_EGLDriver *drv, _EGLDisplay *dpy,
+   _EGLSurface *surface,
+   const EGLint *rects, EGLint n_rects);
+
EGLBoolean (*swap_buffers_region)(_EGLDriver *drv, _EGLDisplay *dpy,
  _EGLSurface *surf, EGLint numRects,
  const EGLint *rects);
diff --git a/src/egl/drivers/dri2/platform_android.c 
b/src/egl/drivers/dri2/platform_android.c
index 1ae779e59c..0752e330b4 100644
--- a/src/egl/drivers/dri2/platform_android.c
+++ b/src/egl/drivers/dri2/platform_android.c
@@ -651,6 +651,37 @@ droid_swap_buffers(_EGLDriver *drv, _EGLDisplay *disp, 
_EGLSurface *draw)
return EGL_TRUE;
 }
 
+static EGLBoolean
+droid_set_damage_region(_EGLDriver *drv,
+_EGLDisplay *disp,
+_EGLSurface *draw, const EGLint* rects, EGLint n_rects)
+{
+   struct dri2_egl_display *dri2_dpy = dri2_egl_display(disp);
+   struct dri2_egl_surface *dri2_surf = dri2_egl_surface(draw);
+   android_native_rect_t* droid_rects = NULL;
+   if (n_rects != 0)
+  droid_rects = (android_native_rect_t *)
+ calloc(n_rects, sizeof(android_native_rect_t));
+
+   EGLint surfWidth = dri2_surf->base.Width;
+   EGLint surfHeight = dri2_surf->base.Height;
+   EGLint dIndex;
+
+   for (dIndex = 0; dIndex < n_rects; dIndex++) {
+  EGLint i = dIndex * 4;
+  droid_rects[dIndex].left = rects[i]; // left == x
+  droid_rects[dIndex].bottom = rects[i + 1]; // bottom == y
+  droid_rects[dIndex].right = rects[i] + rects[i + 2]; // left + width
+  droid_rects[dIndex].top = rects[i + 1] + rects[i + 3]; // bottom + height
+   }
+
+   native_window_set_surface_damage(dri2_surf->window, droid_rects, n_rects);
+
+   free(droid_rects);
+
+   return EGL_TRUE;
+}
+
 static _EGLImage *
 droid_create_image_from_prime_fd_yuv(_EGLDisplay *disp, _EGLContext *ctx,
  struct ANativeWindowBuffer *buf, int fd)
@@ -1101,6 +1132,7 @@ static struct dri2_egl_display_vtbl droid_display_vtbl = {
.swap_buffers = droid_swap_buffers,
.swap_buffers_with_damage = dri2_fallback_swap_buffers_with_damage,
.swap_buffers_region = dri2_fallback_swap_buffers_region,
+   .set_damage_region = droid_set_damage_region,
.post_sub_buffer = dri2_fallback_post_sub_buffer,
.copy_buffers = dri2

[Mesa-dev] [PATCH] dri3/GLX: Fix drawable invalidation

[Thomas Hellstrom]

A number of internal VMware apitrace traces image comparisons fail with
dri3 because the viewport transformation becomes incorrect after an X
drawable resize. The incorrect viewport transformation sometimes persist
until the second draw-call after a swapBuffer.

Comparing with the dri2 glx code there are a couple of places where dri2
invalidates the drawable in the absence of server-triggered invalidation,
where dri3 doesn't do that. When these invalidation points are added to
dri3, the image comparisons become correct.

Cc: 
Signed-off-by: Thomas Hellstrom 
---
 src/glx/dri3_glx.c | 6 ++
 1 file changed, 6 insertions(+)

diff --git a/src/glx/dri3_glx.c b/src/glx/dri3_glx.c
index 4264fd8..7732279 100644
--- a/src/glx/dri3_glx.c
+++ b/src/glx/dri3_glx.c
@@ -235,6 +235,11 @@ dri3_bind_context(struct glx_context *context, struct 
glx_context *old,
if (!(*psc->core->bindContext) (pcp->driContext, dri_draw, dri_read))
   return GLXBadContext;
 
+   if (dri_draw)
+  (*psc->f->invalidate)(dri_draw);
+   if (dri_read)
+  (*psc->f->invalidate)(dri_read);
+
return Success;
 }
 
@@ -493,6 +498,7 @@ dri3_flush_front_buffer(__DRIdrawable *driDrawable, void 
*loaderPrivate)
 
loader_dri3_flush(draw, __DRI2_FLUSH_DRAWABLE, __DRI2_THROTTLE_FLUSHFRONT);
 
+   (*psc->f->invalidate)(driDrawable);
loader_dri3_wait_gl(draw);
 }
 
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [ANNOUNCE] mesa 17.0.7

[Emil Velikov]

Mesa 17.0.7 is now available.

Note: this is the final anticipated release in the 17.0 series. Users are
encouraged to migrate to the 17.1 series in order to obtain future fixes.

In this release we have:

The mesa GLVND GLX library now handles glXGetDriverConfig, as used by driconf.

For EGL the error checking within eglMakeCurrent has been relaxed.

On the driver side:
i965 and ANV have improved DXT1 and BC1 format handling. And the i965 vec4
backend has a few uniforms related fixes.

The nouveau, etnaviv and vc4 drivers have seen minor improvements.

Last but not least, we have several Wayland patches improving event queue
handling.


Andres Gomez (1):
  docs: add sha256 checksums for 17.0.6

Bartosz Tomczyk (1):
  mesa: Avoid leaking surface in st_renderbuffer_delete

Chad Versace (1):
  egl: Partially revert 23c86c74, fix eglMakeCurrent

Daniel Stone (7):
  vulkan: Fix Wayland uninitialised registry
  vulkan/wsi/wayland: Remove roundtrip when creating image
  vulkan/wsi/wayland: Use per-display event queue
  vulkan/wsi/wayland: Use proxy wrappers for swapchain
  egl/wayland: Don't open-code roundtrip
  egl/wayland: Use per-surface event queues
  egl/wayland: Ensure we get a back buffer

Emil Velikov (6):
  st/va: fix misplaced closing bracket
  anv: automake: list shared libraries after the static ones
  radv: automake: list shared libraries after the static ones
  egl/wayland: select the format based on the interface used
  Update version to 17.0.7
  docs: add release notes for 17.0.7

Eric Anholt (2):
  renderonly: Initialize fields of struct winsys_handle.
  vc4: Don't allocate new BOs to avoid synchronization when they're shared.

Hans de Goede (1):
  glxglvnddispatch: Add missing dispatch for GetDriverConfig

Ilia Mirkin (1):
  nvc0/ir: SHLADD's middle source must be an immediate

Jason Ekstrand (2):
  i965/blorp: Do and end-of-pipe sync on both sides of fast-clear ops
  i965: Round copy size to the nearest block in intel_miptree_copy

Lucas Stach (1):
  etnaviv: stop oversizing buffer resources

Nanley Chery (2):
  anv/formats: Update the three-channel BC1 mappings
  i965/formats: Update the three-channel DXT1 mappings

Pohjolainen, Topi (1):
  intel/isl/gen7: Use stencil vertical alignment of 8 instead of 4

Samuel Iglesias Gonsálvez (3):
  i965/vec4/gs: restore the uniform values which was overwritten
by failed vec4_gs_visitor execution
  i965/vec4: fix swizzle and writemask when loading an uniform
with constant offset
  i965/vec4: load dvec3/4 uniforms first in the push constant buffer

Tom Stellard (1):
  gallivm: Make sure module has the correct data layout when pass
manager runs

git tag: mesa-17.0.7

https://mesa.freedesktop.org/archive/mesa-17.0.7.tar.gz
MD5:  4b4e94d69f25161b3bc5267121469fac  mesa-17.0.7.tar.gz
SHA1: 20b0a3c8526426007fd7a8c14cdb7ec7e6dfd7f9  mesa-17.0.7.tar.gz
SHA256: bc68d13c6b1a053b855ac453ebf7e62bd89511adf44bad6c613e09f7fa13390a
 mesa-17.0.7.tar.gz
SHA512: 
09e3f3d6dcba798ad9cfa24b99312fc2776256531664dbe2e7061c6e84975503c39f7b7bbc9f1c47f1bed2ebce74987dcb73a40de83436034f69307954e08b23
 mesa-17.0.7.tar.gz
PGP:  https://mesa.freedesktop.org/archive/mesa-17.0.7.tar.gz.sig

https://mesa.freedesktop.org/archive/mesa-17.0.7.tar.xz
MD5:  8d088c8f7a099084ac745646ed1ad62b  mesa-17.0.7.tar.xz
SHA1: 33eecf0d137ad31bc91d2863447766aa42d2df43  mesa-17.0.7.tar.xz
SHA256: f6d75304a229c8d10443e219d6b6c0c342567dbab5a879ebe7cfa3c9139c4492
 mesa-17.0.7.tar.xz
SHA512: 
378d643cf63ec3ed5fd093c00079fb03375f6a87875d3583a5d4707d3d6b21934194b0f624f4ed261c9a05a2fe3e9f292dcd8466e82cfb95d919459e3eb635ea
 mesa-17.0.7.tar.xz
PGP:  https://mesa.freedesktop.org/archive/mesa-17.0.7.tar.xz.sig
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] Revert "egl/display: remove unnecessary code and make it easier to read"

[Eric Engestrom]

On Thursday, 2017-06-01 02:58:17 +0300, Grazvydas Ignotas wrote:
> This reverts commit 7adb9b094894a512c019b3378eb9e3c69d830edc.
> 
> The "No functional change" claim is wrong, after that commit, the
> _eglGetNativePlatformFromEnv() function is now called every time instead
> of just once it was called before.

Oops :/
You're entirely correct, it looks like I missed the fact native_platform
is static...
_eglLog() is also called every time now, which is spammy and might be
how you noticed it?

> What's worse, when
> _eglGetNativePlatform() is called from multiple threads, the static
> native_platform variable may now alternate between _EGL_INVALID_PLATFORM
> and some other value, which causes some of the threads to see and return
> unintended result.

I think this can be solved by a simple `if (detected_platform ==
_EGL_INVALID_PLATFORM)` before the first detection method. There's no
way the current detection code can return different results in different
threads, so no back-and-forth possible, and if this ever changes to make
that possible then the old code doesn't handle that either anyway.
Shout if I'm (again) missing something :)

As an alternative to your revert, how about operating on a local var,
reading the global one once at the start and writing it once at the end,
with a compare-and-swap?
With that, the worse that would happen is two threads doing the
detection and only one of them writing the result, but (for now at
least) they would always detect the same thing so it would just be a few
CPU cycles wasted.

Sending my variant in a minute.

Cheers,
  Eric

> 
> The old code is not exactly thread safe either, but because the code
> stops changing 'native_platform' as soon as it's not
> _EGL_INVALID_PLATFORM, it should not cause problems in practice.
> 
> Cc: Eric Engestrom 
> Fixes: 7adb9b0948 "egl/display: remove unnecessary code and make it easier to 
> read"
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101252
> Signed-off-by: Grazvydas Ignotas 
> ---
>  src/egl/main/egldisplay.c | 29 +++--
>  1 file changed, 15 insertions(+), 14 deletions(-)
> 
> diff --git a/src/egl/main/egldisplay.c b/src/egl/main/egldisplay.c
> index 2a1f027..6e007a1 100644
> --- a/src/egl/main/egldisplay.c
> +++ b/src/egl/main/egldisplay.c
> @@ -178,28 +178,29 @@ _eglNativePlatformDetectNativeDisplay(void 
> *nativeDisplay)
>   * Return the native platform.  It is the platform of the EGL native types.
>   */
>  _EGLPlatformType
>  _eglGetNativePlatform(void *nativeDisplay)
>  {
> -   static _EGLPlatformType native_platform;
> -   char *detection_method;
> -
> -   native_platform = _eglGetNativePlatformFromEnv();
> -   detection_method = "environment overwrite";
> -
> -   if (native_platform == _EGL_INVALID_PLATFORM) {
> -  native_platform = _eglNativePlatformDetectNativeDisplay(nativeDisplay);
> -  detection_method = "autodetected";
> -   }
> +   static _EGLPlatformType native_platform = _EGL_INVALID_PLATFORM;
> +   char *detection_method = NULL;
>  
> if (native_platform == _EGL_INVALID_PLATFORM) {
> -  native_platform = _EGL_NATIVE_PLATFORM;
> -  detection_method = "build-time configuration";
> +  native_platform = _eglGetNativePlatformFromEnv();
> +  detection_method = "environment overwrite";
> +  if (native_platform == _EGL_INVALID_PLATFORM) {
> + native_platform = 
> _eglNativePlatformDetectNativeDisplay(nativeDisplay);
> + detection_method = "autodetected";
> + if (native_platform == _EGL_INVALID_PLATFORM) {
> +native_platform = _EGL_NATIVE_PLATFORM;
> +detection_method = "build-time configuration";
> + }
> +  }
> }
>  
> -   _eglLog(_EGL_DEBUG, "Native platform type: %s (%s)",
> -   egl_platforms[native_platform].name, detection_method);
> +   if (detection_method != NULL)
> +  _eglLog(_EGL_DEBUG, "Native platform type: %s (%s)",
> +  egl_platforms[native_platform].name, detection_method);
>  
> return native_platform;
>  }
>  
>  
> -- 
> 2.7.4
> 
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH mesa 2/2] egl/display: make platform detection thread-safe

[Eric Engestrom]

If the detections methods ever become able to return different results
for different threads, the if chain might make threads go back and forth
between invalid and valid platforms.
Solve this by doing the detection in a local var and only overwriting
the global one at the end, if no other thread has updated it since.

This means the platform detected in the thread might not be the platform
returned by the function, but this is a different issue that will need
to be discussed when this becomes possible.

Reported-by: Grazvydas Ignotas 
Signed-off-by: Eric Engestrom 
---

This is unnecessary in my opinion, but doesn't hurt :)
Putting it out there as a possible fix, but my vote is to not merge it,
not until this per-thread-platform-detection becomes possible anyway.
---
 src/egl/main/egldisplay.c | 17 +++--
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/src/egl/main/egldisplay.c b/src/egl/main/egldisplay.c
index 3f7752f53d..92586def0c 100644
--- a/src/egl/main/egldisplay.c
+++ b/src/egl/main/egldisplay.c
@@ -36,6 +36,7 @@
 #include 
 #include 
 #include "c11/threads.h"
+#include "util/u_atomic.h"
 
 #include "eglcontext.h"
 #include "eglcurrent.h"
@@ -181,23 +182,27 @@ _EGLPlatformType
 _eglGetNativePlatform(void *nativeDisplay)
 {
static _EGLPlatformType native_platform = _EGL_INVALID_PLATFORM;
+   _EGLPlatformType detected_platform = native_platform;
char *detection_method = NULL;
 
-   if (native_platform == _EGL_INVALID_PLATFORM) {
-  native_platform = _eglGetNativePlatformFromEnv();
+   if (detected_platform == _EGL_INVALID_PLATFORM) {
+  detected_platform = _eglGetNativePlatformFromEnv();
   detection_method = "environment overwrite";
}
 
-   if (native_platform == _EGL_INVALID_PLATFORM) {
-  native_platform = _eglNativePlatformDetectNativeDisplay(nativeDisplay);
+   if (detected_platform == _EGL_INVALID_PLATFORM) {
+  detected_platform = _eglNativePlatformDetectNativeDisplay(nativeDisplay);
   detection_method = "autodetected";
}
 
-   if (native_platform == _EGL_INVALID_PLATFORM) {
-  native_platform = _EGL_NATIVE_PLATFORM;
+   if (detected_platform == _EGL_INVALID_PLATFORM) {
+  detected_platform = _EGL_NATIVE_PLATFORM;
   detection_method = "build-time configuration";
}
 
+   p_atomic_cmpxchg(&native_platform, _EGL_INVALID_PLATFORM,
+detected_platform);
+
if (detection_method != NULL) {
   _eglLog(_EGL_DEBUG, "Native platform type: %s (%s)",
   egl_platforms[native_platform].name, detection_method);
-- 
Cheers,
  Eric

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH mesa 1/2] egl/display: only detect the platform once

[Eric Engestrom]

My refactor missed the fact that `native_platform` is static.
Add the proper guard before the first detection method, as it might not
be necessary, and only print the debug message when a detection was
actually performed.

Reported-by: Grazvydas Ignotas 
Fixes: 7adb9b094894a512c019 ("egl/display: remove unnecessary code and
  make it easier to read")
Signed-off-by: Eric Engestrom 
---
 src/egl/main/egldisplay.c | 16 ++--
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/src/egl/main/egldisplay.c b/src/egl/main/egldisplay.c
index 2a1f02738d..3f7752f53d 100644
--- a/src/egl/main/egldisplay.c
+++ b/src/egl/main/egldisplay.c
@@ -180,11 +180,13 @@ _eglNativePlatformDetectNativeDisplay(void *nativeDisplay)
 _EGLPlatformType
 _eglGetNativePlatform(void *nativeDisplay)
 {
-   static _EGLPlatformType native_platform;
-   char *detection_method;
+   static _EGLPlatformType native_platform = _EGL_INVALID_PLATFORM;
+   char *detection_method = NULL;
 
-   native_platform = _eglGetNativePlatformFromEnv();
-   detection_method = "environment overwrite";
+   if (native_platform == _EGL_INVALID_PLATFORM) {
+  native_platform = _eglGetNativePlatformFromEnv();
+  detection_method = "environment overwrite";
+   }
 
if (native_platform == _EGL_INVALID_PLATFORM) {
   native_platform = _eglNativePlatformDetectNativeDisplay(nativeDisplay);
@@ -196,8 +198,10 @@ _eglGetNativePlatform(void *nativeDisplay)
   detection_method = "build-time configuration";
}
 
-   _eglLog(_EGL_DEBUG, "Native platform type: %s (%s)",
-   egl_platforms[native_platform].name, detection_method);
+   if (detection_method != NULL) {
+  _eglLog(_EGL_DEBUG, "Native platform type: %s (%s)",
+  egl_platforms[native_platform].name, detection_method);
+   }
 
return native_platform;
 }
-- 
Cheers,
  Eric

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH mesa 2/2] egl/display: make platform detection thread-safe

[Grazvydas Ignotas]

On Thu, Jun 1, 2017 at 2:15 PM, Eric Engestrom
 wrote:
> If the detections methods ever become able to return different results
> for different threads, the if chain might make threads go back and forth
> between invalid and valid platforms.
> Solve this by doing the detection in a local var and only overwriting
> the global one at the end, if no other thread has updated it since.
>
> This means the platform detected in the thread might not be the platform
> returned by the function, but this is a different issue that will need
> to be discussed when this becomes possible.
>
> Reported-by: Grazvydas Ignotas 

Not really, see https://bugs.freedesktop.org/show_bug.cgi?id=101252

> Signed-off-by: Eric Engestrom 
> ---
>
> This is unnecessary in my opinion, but doesn't hurt :)

It is necessary, without it things will work most of the time but
there will be rare failures.
Imagine there are 2 threads that both call _eglGetNativePlatform()
simultaneously:
- thread 1 completes the first "if (native_platform ==
_EGL_INVALID_PLATFORM)" check and is preemted to do something else
- thread 2 executes the whole function, does "native_platform =
_EGL_NATIVE_PLATFORM" and just before returning it's preemted
- thread 1 wakes up and calls _eglGetNativePlatformFromEnv() which
returns _EGL_INVALID_PLATFORM because no env vars are set, updates
native_platform and then gets preemted again
- thread 2 wakes up and returns wrong _EGL_INVALID_PLATFORM

> Putting it out there as a possible fix, but my vote is to not merge it,
> not until this per-thread-platform-detection becomes possible anyway.
> ---
>  src/egl/main/egldisplay.c | 17 +++--
>  1 file changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/src/egl/main/egldisplay.c b/src/egl/main/egldisplay.c
> index 3f7752f53d..92586def0c 100644
> --- a/src/egl/main/egldisplay.c
> +++ b/src/egl/main/egldisplay.c
> @@ -36,6 +36,7 @@
>  #include 
>  #include 
>  #include "c11/threads.h"
> +#include "util/u_atomic.h"
>
>  #include "eglcontext.h"
>  #include "eglcurrent.h"
> @@ -181,23 +182,27 @@ _EGLPlatformType
>  _eglGetNativePlatform(void *nativeDisplay)
>  {
> static _EGLPlatformType native_platform = _EGL_INVALID_PLATFORM;
> +   _EGLPlatformType detected_platform = native_platform;
> char *detection_method = NULL;
>
> -   if (native_platform == _EGL_INVALID_PLATFORM) {
> -  native_platform = _eglGetNativePlatformFromEnv();
> +   if (detected_platform == _EGL_INVALID_PLATFORM) {
> +  detected_platform = _eglGetNativePlatformFromEnv();
>detection_method = "environment overwrite";
> }
>
> -   if (native_platform == _EGL_INVALID_PLATFORM) {
> -  native_platform = _eglNativePlatformDetectNativeDisplay(nativeDisplay);
> +   if (detected_platform == _EGL_INVALID_PLATFORM) {
> +  detected_platform = 
> _eglNativePlatformDetectNativeDisplay(nativeDisplay);
>detection_method = "autodetected";
> }
>
> -   if (native_platform == _EGL_INVALID_PLATFORM) {
> -  native_platform = _EGL_NATIVE_PLATFORM;
> +   if (detected_platform == _EGL_INVALID_PLATFORM) {
> +  detected_platform = _EGL_NATIVE_PLATFORM;
>detection_method = "build-time configuration";
> }
>
> +   p_atomic_cmpxchg(&native_platform, _EGL_INVALID_PLATFORM,
> +detected_platform);

I'd make the first "if (detected_platform == _EGL_INVALID_PLATFORM) {"
block end here to avoid doing the slow cmpxchg every time, which makes
it similar to how it looked before your original  patch that I wanted
to revert.

Gražvydas
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [Bug 101252] eglGetDisplay() is not thread safe

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=101252

Emre Ucan  changed:

   What|Removed |Added

 CC||eu...@de.adit-jv.com

-- 
You are receiving this mail because:
You are the assignee for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 03/21] mesa: add KHR_no_error support for glBindVertexBuffers()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/ARB_multi_bind.xml |  2 +-
 src/mesa/main/varray.c| 13 +
 src/mesa/main/varray.h|  5 +
 3 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_multi_bind.xml 
b/src/mapi/glapi/gen/ARB_multi_bind.xml
index d58c2708cb..601680f30b 100644
--- a/src/mapi/glapi/gen/ARB_multi_bind.xml
+++ b/src/mapi/glapi/gen/ARB_multi_bind.xml
@@ -41,7 +41,7 @@
 
 
 
-
+
 
 
 
diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index 0cc8b56c64..a08b11fcf8 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -2249,6 +2249,19 @@ vertex_array_vertex_buffers_err(struct gl_context *ctx,
 
 
 void GLAPIENTRY
+_mesa_BindVertexBuffers_no_error(GLuint first, GLsizei count,
+ const GLuint *buffers, const GLintptr 
*offsets,
+ const GLsizei *strides)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   vertex_array_vertex_buffers(ctx, ctx->Array.VAO, first, count,
+   buffers, offsets, strides, true,
+   "glBindVertexBuffers");
+}
+
+
+void GLAPIENTRY
 _mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
 const GLintptr *offsets, const GLsizei *strides)
 {
diff --git a/src/mesa/main/varray.h b/src/mesa/main/varray.h
index 90ca8483ca..43b9fe905e 100644
--- a/src/mesa/main/varray.h
+++ b/src/mesa/main/varray.h
@@ -375,6 +375,11 @@ extern void GLAPIENTRY
 _mesa_VertexArrayVertexBuffer(GLuint vaobj, GLuint bindingIndex, GLuint buffer,
   GLintptr offset, GLsizei stride);
 
+void GLAPIENTRY
+_mesa_BindVertexBuffers_no_error(GLuint first, GLsizei count,
+ const GLuint *buffers, const GLintptr 
*offsets,
+ const GLsizei *strides);
+
 extern void GLAPIENTRY
 _mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
 const GLintptr *offsets, const GLsizei *strides);
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 02/21] mesa: add KHR_no_error support for glVertexArrayVertexBuffers()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/ARB_direct_state_access.xml |  2 +-
 src/mesa/main/varray.c | 15 +++
 src/mesa/main/varray.h |  6 ++
 3 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_direct_state_access.xml 
b/src/mapi/glapi/gen/ARB_direct_state_access.xml
index f13a1444a9..cb24d7981c 100644
--- a/src/mapi/glapi/gen/ARB_direct_state_access.xml
+++ b/src/mapi/glapi/gen/ARB_direct_state_access.xml
@@ -607,7 +607,7 @@
   

 
-   
+   
   
   
   
diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index c2c771c173..0cc8b56c64 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -2273,6 +2273,21 @@ _mesa_BindVertexBuffers(GLuint first, GLsizei count, 
const GLuint *buffers,
 
 
 void GLAPIENTRY
+_mesa_VertexArrayVertexBuffers_no_error(GLuint vaobj, GLuint first,
+GLsizei count, const GLuint *buffers,
+const GLintptr *offsets,
+const GLsizei *strides)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   struct gl_vertex_array_object *vao = _mesa_lookup_vao(ctx, vaobj);
+   vertex_array_vertex_buffers(ctx, vao, first, count,
+   buffers, offsets, strides, true,
+   "glVertexArrayVertexBuffers");
+}
+
+
+void GLAPIENTRY
 _mesa_VertexArrayVertexBuffers(GLuint vaobj, GLuint first, GLsizei count,
const GLuint *buffers,
const GLintptr *offsets, const GLsizei *strides)
diff --git a/src/mesa/main/varray.h b/src/mesa/main/varray.h
index 24e37a9bf7..90ca8483ca 100644
--- a/src/mesa/main/varray.h
+++ b/src/mesa/main/varray.h
@@ -379,6 +379,12 @@ extern void GLAPIENTRY
 _mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
 const GLintptr *offsets, const GLsizei *strides);
 
+void GLAPIENTRY
+_mesa_VertexArrayVertexBuffers_no_error(GLuint vaobj, GLuint first,
+GLsizei count, const GLuint *buffers,
+const GLintptr *offsets,
+const GLsizei *strides);
+
 extern void GLAPIENTRY
 _mesa_VertexArrayVertexBuffers(GLuint vaobj, GLuint first, GLsizei count,
const GLuint *buffers,
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 00/21] mesa: Some more KHR_no_error support patches

[Samuel Pitoiset]

Hi,

Found after profiling "Dawn Of War 3" with Sysprof. This series should
save few CPU cycles for this not yet released game.

Please review,
Thanks!

Samuel Pitoiset (21):
  mesa: add vertex_array_vertex_buffers_err() helper
  mesa: add KHR_no_error support for glVertexArrayVertexBuffers()
  mesa: add KHR_no_error support for glBindVertexBuffers()
  mesa: add invalidate_buffer_subdata() helper
  mesa: add KHR_no_error support for glInvalidateBufferSubData()
  mesa: add KHR_no_error support for glInvalidateBufferData()
  mesa: add KHR_no_error support for glBindBuffer()
  mesa: make _mesa_blit_framebuffer() static
  mesa: add validate_stencil_buffer() helper
  mesa: add validate_depth_buffer() helper
  mesa: add validate_color_buffer() helper
  mesa: add 'no_error' parameter to blit_framebuffer()
  mesa: add KHR_no_error support for glBlitFramebuffer()
  mesa: add KHR_no_error support for glBlitNamedFramebuffer()
  mesa: rename prepare_target() to prepare_target_err()
  mesa: add prepare_target() helper
  mesa: add copy_image_subdata() helper
  mesa: add KHR_no_error support for glCopyImageSubData()
  mesa: rename texsubimage() to texsubimage_err()
  mesa: add texsubimage() helper
  mesa: add KHR_no_error support for glTexSubImage*D()

 src/mapi/glapi/gen/ARB_copy_image.xml  |   2 +-
 src/mapi/glapi/gen/ARB_direct_state_access.xml |   4 +-
 src/mapi/glapi/gen/ARB_framebuffer_object.xml  |   2 +-
 src/mapi/glapi/gen/ARB_invalidate_subdata.xml  |   4 +-
 src/mapi/glapi/gen/ARB_multi_bind.xml  |   2 +-
 src/mapi/glapi/gen/gl_API.xml  |   8 +-
 src/mesa/main/blit.c   | 585 ++---
 src/mesa/main/blit.h   |  21 +-
 src/mesa/main/bufferobj.c  |  44 +-
 src/mesa/main/bufferobj.h  |  11 +
 src/mesa/main/copyimage.c  | 159 +--
 src/mesa/main/copyimage.h  |   7 +
 src/mesa/main/teximage.c   |  95 +++-
 src/mesa/main/teximage.h   |  18 +
 src/mesa/main/varray.c | 136 --
 src/mesa/main/varray.h |  11 +
 16 files changed, 717 insertions(+), 392 deletions(-)

-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 04/21] mesa: add invalidate_buffer_subdata() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/bufferobj.c | 15 +++
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index 9e656a4c98..08f5a9e59b 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -4237,6 +4237,15 @@ _mesa_BindBuffersBase(GLenum target, GLuint first, 
GLsizei count,
}
 }
 
+static inline void
+invalidate_buffer_subdata(struct gl_context *ctx,
+  struct gl_buffer_object *bufObj, GLintptr offset,
+  GLsizeiptr length)
+{
+   if (ctx->Driver.InvalidateBufferSubData)
+  ctx->Driver.InvalidateBufferSubData(ctx, bufObj, offset, length);
+}
+
 void GLAPIENTRY
 _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
   GLsizeiptr length)
@@ -4286,8 +4295,7 @@ _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr 
offset,
   return;
}
 
-   if (ctx->Driver.InvalidateBufferSubData)
-  ctx->Driver.InvalidateBufferSubData(ctx, bufObj, offset, length);
+   invalidate_buffer_subdata(ctx, bufObj, offset, length);
 }
 
 void GLAPIENTRY
@@ -4324,8 +4332,7 @@ _mesa_InvalidateBufferData(GLuint buffer)
   return;
}
 
-   if (ctx->Driver.InvalidateBufferSubData)
-  ctx->Driver.InvalidateBufferSubData(ctx, bufObj, 0, bufObj->Size);
+   invalidate_buffer_subdata(ctx, bufObj, 0, bufObj->Size);
 }
 
 static void
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 01/21] mesa: add vertex_array_vertex_buffers_err() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/varray.c | 108 -
 1 file changed, 61 insertions(+), 47 deletions(-)

diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index 47528ba2a7..c2c771c173 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -2117,29 +2117,14 @@ _mesa_VertexArrayVertexBuffer(GLuint vaobj, GLuint 
bindingIndex, GLuint buffer,
 }
 
 
-static void
+static ALWAYS_INLINE void
 vertex_array_vertex_buffers(struct gl_context *ctx,
 struct gl_vertex_array_object *vao,
 GLuint first, GLsizei count, const GLuint *buffers,
 const GLintptr *offsets, const GLsizei *strides,
-const char *func)
+bool no_error, const char *func)
 {
-   GLuint i;
-
-   ASSERT_OUTSIDE_BEGIN_END(ctx);
-
-   /* The ARB_multi_bind spec says:
-*
-*"An INVALID_OPERATION error is generated if  + 
-* is greater than the value of MAX_VERTEX_ATTRIB_BINDINGS."
-*/
-   if (first + count > ctx->Const.MaxVertexAttribBindings) {
-  _mesa_error(ctx, GL_INVALID_OPERATION,
-  "%s(first=%u + count=%d > the value of "
-  "GL_MAX_VERTEX_ATTRIB_BINDINGS=%u)",
-  func, first, count, ctx->Const.MaxVertexAttribBindings);
-  return;
-   }
+   GLint i;
 
if (!buffers) {
   /**
@@ -2184,31 +2169,33 @@ vertex_array_vertex_buffers(struct gl_context *ctx,
for (i = 0; i < count; i++) {
   struct gl_buffer_object *vbo;
 
-  /* The ARB_multi_bind spec says:
-   *
-   *"An INVALID_VALUE error is generated if any value in
-   *  or  is negative (per binding)."
-   */
-  if (offsets[i] < 0) {
- _mesa_error(ctx, GL_INVALID_VALUE,
- "%s(offsets[%u]=%" PRId64 " < 0)",
- func, i, (int64_t) offsets[i]);
- continue;
-  }
+  if (!no_error) {
+ /* The ARB_multi_bind spec says:
+  *
+  *"An INVALID_VALUE error is generated if any value in
+  *  or  is negative (per binding)."
+  */
+ if (offsets[i] < 0) {
+_mesa_error(ctx, GL_INVALID_VALUE,
+"%s(offsets[%u]=%" PRId64 " < 0)",
+func, i, (int64_t) offsets[i]);
+continue;
+ }
 
-  if (strides[i] < 0) {
- _mesa_error(ctx, GL_INVALID_VALUE,
- "%s(strides[%u]=%d < 0)",
- func, i, strides[i]);
- continue;
-  }
+ if (strides[i] < 0) {
+_mesa_error(ctx, GL_INVALID_VALUE,
+"%s(strides[%u]=%d < 0)",
+func, i, strides[i]);
+continue;
+ }
 
-  if (ctx->API == API_OPENGL_CORE && ctx->Version >= 44 &&
-  strides[i] > ctx->Const.MaxVertexAttribStride) {
- _mesa_error(ctx, GL_INVALID_VALUE,
- "%s(strides[%u]=%d > "
- "GL_MAX_VERTEX_ATTRIB_STRIDE)", func, i, strides[i]);
- continue;
+ if (ctx->API == API_OPENGL_CORE && ctx->Version >= 44 &&
+ strides[i] > ctx->Const.MaxVertexAttribStride) {
+_mesa_error(ctx, GL_INVALID_VALUE,
+"%s(strides[%u]=%d > "
+"GL_MAX_VERTEX_ATTRIB_STRIDE)", func, i, strides[i]);
+continue;
+ }
   }
 
   if (buffers[i]) {
@@ -2234,6 +2221,33 @@ vertex_array_vertex_buffers(struct gl_context *ctx,
 }
 
 
+static void
+vertex_array_vertex_buffers_err(struct gl_context *ctx,
+struct gl_vertex_array_object *vao,
+GLuint first, GLsizei count,
+const GLuint *buffers, const GLintptr *offsets,
+const GLsizei *strides, const char *func)
+{
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   /* The ARB_multi_bind spec says:
+*
+*"An INVALID_OPERATION error is generated if  + 
+* is greater than the value of MAX_VERTEX_ATTRIB_BINDINGS."
+*/
+   if (first + count > ctx->Const.MaxVertexAttribBindings) {
+  _mesa_error(ctx, GL_INVALID_OPERATION,
+  "%s(first=%u + count=%d > the value of "
+  "GL_MAX_VERTEX_ATTRIB_BINDINGS=%u)",
+  func, first, count, ctx->Const.MaxVertexAttribBindings);
+  return;
+   }
+
+   vertex_array_vertex_buffers(ctx, vao, first, count, buffers, offsets,
+   strides, false, func);
+}
+
+
 void GLAPIENTRY
 _mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
 const GLintptr *offsets, const GLsizei *strides)
@@ -2252,9 +2266,9 @@ _mesa_BindVertexBuffers(GLuint first, GLsizei count, 
const GLuint *buffers,
   return;
}
 
-   vertex_array_vertex_buffers(ctx, ctx->Array.V

[Mesa-dev] [PATCH 05/21] mesa: add KHR_no_error support for glInvalidateBufferSubData()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/ARB_invalidate_subdata.xml |  2 +-
 src/mesa/main/bufferobj.c | 10 ++
 src/mesa/main/bufferobj.h |  5 +
 3 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_invalidate_subdata.xml 
b/src/mapi/glapi/gen/ARB_invalidate_subdata.xml
index 052816ad7e..79e8a917f5 100644
--- a/src/mapi/glapi/gen/ARB_invalidate_subdata.xml
+++ b/src/mapi/glapi/gen/ARB_invalidate_subdata.xml
@@ -19,7 +19,7 @@
 
   
 
-  
+  
 
 
 
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index 08f5a9e59b..171750e0c3 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -4247,6 +4247,16 @@ invalidate_buffer_subdata(struct gl_context *ctx,
 }
 
 void GLAPIENTRY
+_mesa_InvalidateBufferSubData_no_error(GLuint buffer, GLintptr offset,
+   GLsizeiptr length)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   struct gl_buffer_object *bufObj = _mesa_lookup_bufferobj(ctx, buffer);
+   invalidate_buffer_subdata(ctx, bufObj, offset, length);
+}
+
+void GLAPIENTRY
 _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
   GLsizeiptr length)
 {
diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h
index ff44fed0b8..2e49d33b8d 100644
--- a/src/mesa/main/bufferobj.h
+++ b/src/mesa/main/bufferobj.h
@@ -338,6 +338,11 @@ _mesa_BindBuffersRange(GLenum target, GLuint first, 
GLsizei count,
 void GLAPIENTRY
 _mesa_BindBuffersBase(GLenum target, GLuint first, GLsizei count,
   const GLuint *buffers);
+
+void GLAPIENTRY
+_mesa_InvalidateBufferSubData_no_error(GLuint buffer, GLintptr offset,
+   GLsizeiptr length);
+
 void GLAPIENTRY
 _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr offset,
   GLsizeiptr length);
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 06/21] mesa: add KHR_no_error support for glInvalidateBufferData()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/ARB_invalidate_subdata.xml | 2 +-
 src/mesa/main/bufferobj.c | 9 +
 src/mesa/main/bufferobj.h | 3 +++
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_invalidate_subdata.xml 
b/src/mapi/glapi/gen/ARB_invalidate_subdata.xml
index 79e8a917f5..2cbc4f63be 100644
--- a/src/mapi/glapi/gen/ARB_invalidate_subdata.xml
+++ b/src/mapi/glapi/gen/ARB_invalidate_subdata.xml
@@ -25,7 +25,7 @@
 
   
 
-  
+  
 
   
 
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index 171750e0c3..9323e8d9a9 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -4309,6 +4309,15 @@ _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr 
offset,
 }
 
 void GLAPIENTRY
+_mesa_InvalidateBufferData_no_error(GLuint buffer)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   struct gl_buffer_object *bufObj =_mesa_lookup_bufferobj(ctx, buffer);
+   invalidate_buffer_subdata(ctx, bufObj, 0, bufObj->Size);
+}
+
+void GLAPIENTRY
 _mesa_InvalidateBufferData(GLuint buffer)
 {
GET_CURRENT_CONTEXT(ctx);
diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h
index 2e49d33b8d..a8e174dc6b 100644
--- a/src/mesa/main/bufferobj.h
+++ b/src/mesa/main/bufferobj.h
@@ -348,6 +348,9 @@ _mesa_InvalidateBufferSubData(GLuint buffer, GLintptr 
offset,
   GLsizeiptr length);
 
 void GLAPIENTRY
+_mesa_InvalidateBufferData_no_error(GLuint buffer);
+
+void GLAPIENTRY
 _mesa_InvalidateBufferData(GLuint buffer);
 
 void GLAPIENTRY
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 07/21] mesa: add KHR_no_error support for glBindBuffer()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/gl_API.xml |  2 +-
 src/mesa/main/bufferobj.c | 10 ++
 src/mesa/main/bufferobj.h |  3 +++
 3 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/gl_API.xml b/src/mapi/glapi/gen/gl_API.xml
index 8f93318b95..5814d2ad92 100644
--- a/src/mapi/glapi/gen/gl_API.xml
+++ b/src/mapi/glapi/gen/gl_API.xml
@@ -5037,7 +5037,7 @@
 
 
 
-
+
 
 
 
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index 9323e8d9a9..e79d22ca86 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -1229,6 +1229,16 @@ _mesa_buffer_unmap_all_mappings(struct gl_context *ctx,
 /**/
 
 void GLAPIENTRY
+_mesa_BindBuffer_no_error(GLenum target, GLuint buffer)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   struct gl_buffer_object **bindTarget = get_buffer_target(ctx, target);
+   bind_buffer_object(ctx, bindTarget, buffer);
+}
+
+
+void GLAPIENTRY
 _mesa_BindBuffer(GLenum target, GLuint buffer)
 {
GET_CURRENT_CONTEXT(ctx);
diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h
index a8e174dc6b..44342840a8 100644
--- a/src/mesa/main/bufferobj.h
+++ b/src/mesa/main/bufferobj.h
@@ -156,6 +156,9 @@ _mesa_ClearBufferSubData_sw(struct gl_context *ctx,
  * API functions
  */
 void GLAPIENTRY
+_mesa_BindBuffer_no_error(GLenum target, GLuint buffer);
+
+void GLAPIENTRY
 _mesa_BindBuffer(GLenum target, GLuint buffer);
 
 void GLAPIENTRY
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 08/21] mesa: make _mesa_blit_framebuffer() static

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 29 ++---
 src/mesa/main/blit.h |  8 
 2 files changed, 14 insertions(+), 23 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index e739130f98..2c0300eab3 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -177,13 +177,12 @@ is_valid_blit_filter(const struct gl_context *ctx, GLenum 
filter)
 }
 
 
-void
-_mesa_blit_framebuffer(struct gl_context *ctx,
-   struct gl_framebuffer *readFb,
-   struct gl_framebuffer *drawFb,
-   GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
-   GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
-   GLbitfield mask, GLenum filter, const char *func)
+static void
+blit_framebuffer(struct gl_context *ctx,
+ struct gl_framebuffer *readFb, struct gl_framebuffer *drawFb,
+ GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
+ GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
+ GLbitfield mask, GLenum filter, const char *func)
 {
const GLbitfield legalMaskBits = (GL_COLOR_BUFFER_BIT |
  GL_DEPTH_BUFFER_BIT |
@@ -561,10 +560,10 @@ _mesa_BlitFramebuffer(GLint srcX0, GLint srcY0, GLint 
srcX1, GLint srcY1,
   dstX0, dstY0, dstX1, dstY1,
   mask, _mesa_enum_to_string(filter));
 
-   _mesa_blit_framebuffer(ctx, ctx->ReadBuffer, ctx->DrawBuffer,
-  srcX0, srcY0, srcX1, srcY1,
-  dstX0, dstY0, dstX1, dstY1,
-  mask, filter, "glBlitFramebuffer");
+   blit_framebuffer(ctx, ctx->ReadBuffer, ctx->DrawBuffer,
+srcX0, srcY0, srcX1, srcY1,
+dstX0, dstY0, dstX1, dstY1,
+mask, filter, "glBlitFramebuffer");
 }
 
 
@@ -612,8 +611,8 @@ _mesa_BlitNamedFramebuffer(GLuint readFramebuffer, GLuint 
drawFramebuffer,
else
   drawFb = ctx->WinSysDrawBuffer;
 
-   _mesa_blit_framebuffer(ctx, readFb, drawFb,
-  srcX0, srcY0, srcX1, srcY1,
-  dstX0, dstY0, dstX1, dstY1,
-  mask, filter, "glBlitNamedFramebuffer");
+   blit_framebuffer(ctx, readFb, drawFb,
+srcX0, srcY0, srcX1, srcY1,
+dstX0, dstY0, dstX1, dstY1,
+mask, filter, "glBlitNamedFramebuffer");
 }
diff --git a/src/mesa/main/blit.h b/src/mesa/main/blit.h
index 88dd4a9ec8..1ca4f83028 100644
--- a/src/mesa/main/blit.h
+++ b/src/mesa/main/blit.h
@@ -34,14 +34,6 @@ _mesa_regions_overlap(int srcX0, int srcY0,
   int dstX0, int dstY0,
   int dstX1, int dstY1);
 
-extern void
-_mesa_blit_framebuffer(struct gl_context *ctx,
-   struct gl_framebuffer *readFb,
-   struct gl_framebuffer *drawFb,
-   GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
-   GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
-   GLbitfield mask, GLenum filter, const char *func);
-
 extern void GLAPIENTRY
 _mesa_BlitFramebuffer(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
  GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 11/21] mesa: add validate_color_buffer() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 164 +++
 1 file changed, 86 insertions(+), 78 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 455a9a647f..95f97ed559 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -178,6 +178,90 @@ is_valid_blit_filter(const struct gl_context *ctx, GLenum 
filter)
 
 
 static void
+validate_color_buffer(struct gl_context *ctx, struct gl_framebuffer *readFb,
+  struct gl_framebuffer *drawFb, GLbitfield *mask,
+  GLenum filter, bool no_error, const char *func)
+{
+   const GLuint numColorDrawBuffers = drawFb->_NumColorDrawBuffers;
+   const struct gl_renderbuffer *colorReadRb = readFb->_ColorReadBuffer;
+   const struct gl_renderbuffer *colorDrawRb = NULL;
+   GLuint i;
+
+   /* From the EXT_framebuffer_object spec:
+*
+* "If a buffer is specified in  and does not exist in both
+* the read and draw framebuffers, the corresponding bit is silently
+* ignored."
+*/
+   if (!colorReadRb || numColorDrawBuffers == 0) {
+  *mask &= ~GL_COLOR_BUFFER_BIT;
+   } else if (!no_error) {
+  for (i = 0; i < numColorDrawBuffers; i++) {
+ colorDrawRb = drawFb->_ColorDrawBuffers[i];
+ if (!colorDrawRb)
+continue;
+
+ /* Page 193 (page 205 of the PDF) in section 4.3.2 of the OpenGL
+  * ES 3.0.1 spec says:
+  *
+  * "If the source and destination buffers are identical, an
+  * INVALID_OPERATION error is generated. Different mipmap levels
+  * of a texture, different layers of a three- dimensional texture
+  * or two-dimensional array texture, and different faces of a cube
+  * map texture do not constitute identical buffers."
+  */
+ if (_mesa_is_gles3(ctx) && (colorDrawRb == colorReadRb)) {
+_mesa_error(ctx, GL_INVALID_OPERATION,
+"%s(source and destination color buffer cannot be the "
+"same)", func);
+return;
+ }
+
+ if (!compatible_color_datatypes(colorReadRb->Format,
+ colorDrawRb->Format)) {
+_mesa_error(ctx, GL_INVALID_OPERATION,
+"%s(color buffer datatypes mismatch)", func);
+return;
+ }
+
+ /* extra checks for multisample copies... */
+ if (readFb->Visual.samples > 0 || drawFb->Visual.samples > 0) {
+/* color formats must match on GLES. This isn't checked on desktop
+ * GL because the GL 4.4 spec was changed to allow it.  In the
+ * section entitled “Changes in the released Specification of July
+ * 22, 2013” it says:
+ *
+ * “Relax BlitFramebuffer in section 18.3.1 so that format
+ * conversion can take place during multisample blits, since
+ * drivers already allow this and some apps depend on it.”
+ */
+if (_mesa_is_gles(ctx) &&
+!compatible_resolve_formats(colorReadRb, colorDrawRb)) {
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+   "%s(bad src/dst multisample pixel formats)", func);
+   return;
+}
+ }
+  }
+
+  if (filter != GL_NEAREST) {
+ /* From EXT_framebuffer_multisample_blit_scaled specification:
+  * "Calling BlitFramebuffer will result in an INVALID_OPERATION error
+  * if filter is not NEAREST and read buffer contains integer data."
+  */
+ GLenum type = _mesa_get_format_datatype(colorReadRb->Format);
+
+ if (type == GL_INT || type == GL_UNSIGNED_INT) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(integer color type)", func);
+return;
+ }
+  }
+   }
+}
+
+
+static void
 validate_stencil_buffer(struct gl_context *ctx, struct gl_framebuffer *readFb,
 struct gl_framebuffer *drawFb, GLbitfield *mask,
 bool no_error, const char *func)
@@ -348,84 +432,8 @@ blit_framebuffer(struct gl_context *ctx,
   return;
}
 
-   /* get color read/draw renderbuffers */
-   if (mask & GL_COLOR_BUFFER_BIT) {
-  const GLuint numColorDrawBuffers = drawFb->_NumColorDrawBuffers;
-  const struct gl_renderbuffer *colorReadRb = readFb->_ColorReadBuffer;
-  const struct gl_renderbuffer *colorDrawRb = NULL;
-  GLuint i;
-
-  /* From the EXT_framebuffer_object spec:
-   *
-   * "If a buffer is specified in  and does not exist in both
-   * the read and draw framebuffers, the corresponding bit is silently
-   * ignored."
-   */
-  if (!colorReadRb || numColorDrawBuffers == 0) {
- mask &= ~GL_COLOR_BUFFER_BIT;
-  }
-  else {
- for (i = 0; i < numColorDrawB

[Mesa-dev] [PATCH 09/21] mesa: add validate_stencil_buffer() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 111 +++
 1 file changed, 58 insertions(+), 53 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 2c0300eab3..207ce7d501 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -178,6 +178,62 @@ is_valid_blit_filter(const struct gl_context *ctx, GLenum 
filter)
 
 
 static void
+validate_stencil_buffer(struct gl_context *ctx, struct gl_framebuffer *readFb,
+struct gl_framebuffer *drawFb, GLbitfield *mask,
+bool no_error, const char *func)
+{
+   struct gl_renderbuffer *readRb =
+  readFb->Attachment[BUFFER_STENCIL].Renderbuffer;
+   struct gl_renderbuffer *drawRb =
+  drawFb->Attachment[BUFFER_STENCIL].Renderbuffer;
+
+   /* From the EXT_framebuffer_object spec:
+*
+* "If a buffer is specified in  and does not exist in both
+* the read and draw framebuffers, the corresponding bit is silently
+* ignored."
+*/
+   if (readRb == NULL || drawRb == NULL) {
+  *mask &= ~GL_STENCIL_BUFFER_BIT;
+   } else if (!no_error) {
+  int read_z_bits, draw_z_bits;
+
+  if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(source and destination stencil buffer cannot be the "
+ "same)", func);
+ return;
+  }
+
+  if (_mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS) !=
+  _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS)) {
+ /* There is no need to check the stencil datatype here, because
+  * there is only one: GL_UNSIGNED_INT.
+  */
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(stencil attachment format mismatch)", func);
+ return;
+  }
+
+  read_z_bits = _mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS);
+  draw_z_bits = _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS);
+
+  /* If both buffers also have depth data, the depth formats must match
+   * as well.  If one doesn't have depth, it's not blitted, so we should
+   * ignore the depth format check.
+   */
+  if (read_z_bits > 0 && draw_z_bits > 0 &&
+  (read_z_bits != draw_z_bits ||
+   _mesa_get_format_datatype(readRb->Format) !=
+   _mesa_get_format_datatype(drawRb->Format))) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(stencil attachment depth format mismatch)", func);
+ return;
+  }
+   }
+}
+
+static void
 blit_framebuffer(struct gl_context *ctx,
  struct gl_framebuffer *readFb, struct gl_framebuffer *drawFb,
  GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
@@ -317,59 +373,8 @@ blit_framebuffer(struct gl_context *ctx,
   }
}
 
-   if (mask & GL_STENCIL_BUFFER_BIT) {
-  struct gl_renderbuffer *readRb =
- readFb->Attachment[BUFFER_STENCIL].Renderbuffer;
-  struct gl_renderbuffer *drawRb =
- drawFb->Attachment[BUFFER_STENCIL].Renderbuffer;
-
-  /* From the EXT_framebuffer_object spec:
-   *
-   * "If a buffer is specified in  and does not exist in both
-   * the read and draw framebuffers, the corresponding bit is silently
-   * ignored."
-   */
-  if ((readRb == NULL) || (drawRb == NULL)) {
- mask &= ~GL_STENCIL_BUFFER_BIT;
-  }
-  else {
- int read_z_bits, draw_z_bits;
-
- if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(source and destination stencil "
-"buffer cannot be the same)", func);
-return;
- }
-
- if (_mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS) !=
- _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS)) {
-/* There is no need to check the stencil datatype here, because
- * there is only one: GL_UNSIGNED_INT.
- */
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(stencil attachment format mismatch)", func);
-return;
- }
-
- read_z_bits = _mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS);
- draw_z_bits = _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS);
-
- /* If both buffers also have depth data, the depth formats must match
-  * as well.  If one doesn't have depth, it's not blitted, so we should
-  * ignore the depth format check.
-  */
- if (read_z_bits > 0 && draw_z_bits > 0 &&
- (read_z_bits != draw_z_bits ||
-  _mesa_get_format_datatype(readRb->Format) !=
-  _mesa_get_format_datatype(drawRb->Format))) {
-
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(stencil attachment depth format mismatch)", func);
-retur

[Mesa-dev] [PATCH 12/21] mesa: add 'no_error' parameter to blit_framebuffer()

[Samuel Pitoiset]

The whole GLES3 block has been moved before the buffer validation
checks.

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 177 ++-
 1 file changed, 90 insertions(+), 87 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 95f97ed559..aa95e9c96c 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -376,12 +376,9 @@ blit_framebuffer(struct gl_context *ctx,
  struct gl_framebuffer *readFb, struct gl_framebuffer *drawFb,
  GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
  GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
- GLbitfield mask, GLenum filter, const char *func)
+ GLbitfield mask, GLenum filter, bool no_error,
+ const char *func)
 {
-   const GLbitfield legalMaskBits = (GL_COLOR_BUFFER_BIT |
- GL_DEPTH_BUFFER_BIT |
- GL_STENCIL_BUFFER_BIT);
-
FLUSH_VERTICES(ctx, 0);
 
/* Update completeness status of readFb and drawFb. */
@@ -397,104 +394,110 @@ blit_framebuffer(struct gl_context *ctx,
   return;
}
 
-   /* check for complete framebuffers */
-   if (drawFb->_Status != GL_FRAMEBUFFER_COMPLETE_EXT ||
-   readFb->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) {
-  _mesa_error(ctx, GL_INVALID_FRAMEBUFFER_OPERATION_EXT,
-  "%s(incomplete draw/read buffers)", func);
-  return;
-   }
-
-   if (!is_valid_blit_filter(ctx, filter)) {
-  _mesa_error(ctx, GL_INVALID_ENUM, "%s(invalid filter %s)", func,
-  _mesa_enum_to_string(filter));
-  return;
-   }
-
-   if ((filter == GL_SCALED_RESOLVE_FASTEST_EXT ||
-filter == GL_SCALED_RESOLVE_NICEST_EXT) &&
-(readFb->Visual.samples == 0 || drawFb->Visual.samples > 0)) {
-  _mesa_error(ctx, GL_INVALID_OPERATION, "%s(%s: invalid samples)", func,
-  _mesa_enum_to_string(filter));
-  return;
-   }
-
-   if (mask & ~legalMaskBits) {
-  _mesa_error(ctx, GL_INVALID_VALUE, "%s(invalid mask bits set)", func);
-  return;
-   }
-
-   /* depth/stencil must be blitted with nearest filtering */
-   if ((mask & (GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT))
-&& filter != GL_NEAREST) {
-  _mesa_error(ctx, GL_INVALID_OPERATION,
- "%s(depth/stencil requires GL_NEAREST filter)", func);
-  return;
-   }
+   if (!no_error) {
+  const GLbitfield legalMaskBits = (GL_COLOR_BUFFER_BIT |
+GL_DEPTH_BUFFER_BIT |
+GL_STENCIL_BUFFER_BIT);
 
-   if (mask & GL_COLOR_BUFFER_BIT)
-  validate_color_buffer(ctx, readFb, drawFb, &mask, filter, false, func);
+  /* check for complete framebuffers */
+  if (drawFb->_Status != GL_FRAMEBUFFER_COMPLETE_EXT ||
+  readFb->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) {
+ _mesa_error(ctx, GL_INVALID_FRAMEBUFFER_OPERATION_EXT,
+ "%s(incomplete draw/read buffers)", func);
+ return;
+  }
 
-   if (mask & GL_STENCIL_BUFFER_BIT)
-  validate_stencil_buffer(ctx, readFb, drawFb, &mask, false, func);
+  if (!is_valid_blit_filter(ctx, filter)) {
+ _mesa_error(ctx, GL_INVALID_ENUM, "%s(invalid filter %s)", func,
+ _mesa_enum_to_string(filter));
+ return;
+  }
 
-   if (mask & GL_DEPTH_BUFFER_BIT)
-  validate_depth_buffer(ctx, readFb, drawFb, &mask, false, func);
-
-   if (_mesa_is_gles3(ctx)) {
-  /* Page 194 (page 206 of the PDF) in section 4.3.2 of the OpenGL ES
-   * 3.0.1 spec says:
-   *
-   * "If SAMPLE_BUFFERS for the draw framebuffer is greater than zero,
-   * an INVALID_OPERATION error is generated."
-   */
-  if (drawFb->Visual.samples > 0) {
- _mesa_error(ctx, GL_INVALID_OPERATION,
- "%s(destination samples must be 0)", func);
+  if ((filter == GL_SCALED_RESOLVE_FASTEST_EXT ||
+   filter == GL_SCALED_RESOLVE_NICEST_EXT) &&
+   (readFb->Visual.samples == 0 || drawFb->Visual.samples > 0)) {
+ _mesa_error(ctx, GL_INVALID_OPERATION, "%s(%s: invalid samples)", 
func,
+ _mesa_enum_to_string(filter));
  return;
   }
 
-  /* Page 194 (page 206 of the PDF) in section 4.3.2 of the OpenGL ES
-   * 3.0.1 spec says:
-   *
-   * "If SAMPLE_BUFFERS for the read framebuffer is greater than zero,
-   * no copy is performed and an INVALID_OPERATION error is generated
-   * if the formats of the read and draw framebuffers are not
-   * identical or if the source and destination rectangles are not
-   * defined with the same (X0, Y0) and (X1, Y1) bounds."
-   *
-   * The format check was made above because desktop OpenGL has the same
-   * requirement.
-   */
-  if (readFb->Visual.samples > 0
-  && (s

[Mesa-dev] [PATCH 10/21] mesa: add validate_depth_buffer() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 105 +++
 1 file changed, 56 insertions(+), 49 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 207ce7d501..455a9a647f 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -233,6 +233,60 @@ validate_stencil_buffer(struct gl_context *ctx, struct 
gl_framebuffer *readFb,
}
 }
 
+
+static void
+validate_depth_buffer(struct gl_context *ctx, struct gl_framebuffer *readFb,
+  struct gl_framebuffer *drawFb, GLbitfield *mask,
+  bool no_error, const char *func)
+{
+   struct gl_renderbuffer *readRb =
+  readFb->Attachment[BUFFER_DEPTH].Renderbuffer;
+   struct gl_renderbuffer *drawRb =
+  drawFb->Attachment[BUFFER_DEPTH].Renderbuffer;
+
+   /* From the EXT_framebuffer_object spec:
+*
+* "If a buffer is specified in  and does not exist in both
+* the read and draw framebuffers, the corresponding bit is silently
+* ignored."
+*/
+   if (readRb == NULL || drawRb == NULL) {
+  *mask &= ~GL_DEPTH_BUFFER_BIT;
+   } else if (!no_error) {
+  int read_s_bit, draw_s_bit;
+
+  if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(source and destination depth buffer cannot be the "
+ "same)", func);
+ return;
+  }
+
+  if ((_mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS) !=
+   _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS)) ||
+  (_mesa_get_format_datatype(readRb->Format) !=
+   _mesa_get_format_datatype(drawRb->Format))) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(depth attachment format mismatch)", func);
+ return;
+  }
+
+  read_s_bit = _mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS);
+  draw_s_bit = _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS);
+
+  /* If both buffers also have stencil data, the stencil formats must
+   * match as well.  If one doesn't have stencil, it's not blitted, so
+   * we should ignore the stencil format check.
+   */
+  if (read_s_bit > 0 && draw_s_bit > 0 && read_s_bit != draw_s_bit) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(depth attachment stencil bits mismatch)", func);
+ return;
+  }
+   }
+}
+
+
 static void
 blit_framebuffer(struct gl_context *ctx,
  struct gl_framebuffer *readFb, struct gl_framebuffer *drawFb,
@@ -376,55 +430,8 @@ blit_framebuffer(struct gl_context *ctx,
if (mask & GL_STENCIL_BUFFER_BIT)
   validate_stencil_buffer(ctx, readFb, drawFb, &mask, false, func);
 
-   if (mask & GL_DEPTH_BUFFER_BIT) {
-  struct gl_renderbuffer *readRb =
- readFb->Attachment[BUFFER_DEPTH].Renderbuffer;
-  struct gl_renderbuffer *drawRb =
- drawFb->Attachment[BUFFER_DEPTH].Renderbuffer;
-
-  /* From the EXT_framebuffer_object spec:
-   *
-   * "If a buffer is specified in  and does not exist in both
-   * the read and draw framebuffers, the corresponding bit is silently
-   * ignored."
-   */
-  if ((readRb == NULL) || (drawRb == NULL)) {
- mask &= ~GL_DEPTH_BUFFER_BIT;
-  }
-  else {
- int read_s_bit, draw_s_bit;
-
- if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(source and destination depth "
-"buffer cannot be the same)", func);
-return;
- }
-
- if ((_mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS) !=
-  _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS)) ||
- (_mesa_get_format_datatype(readRb->Format) !=
-  _mesa_get_format_datatype(drawRb->Format))) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(depth attachment format mismatch)", func);
-return;
- }
-
- read_s_bit = _mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS);
- draw_s_bit = _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS);
-
- /* If both buffers also have stencil data, the stencil formats must
-  * match as well.  If one doesn't have stencil, it's not blitted, so
-  * we should ignore the stencil format check.
-  */
- if (read_s_bit > 0 && draw_s_bit > 0 && read_s_bit != draw_s_bit) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(depth attachment stencil bits mismatch)", func);
-return;
- }
-  }
-   }
-
+   if (mask & GL_DEPTH_BUFFER_BIT)
+  validate_depth_buffer(ctx, readFb, drawFb, &mask, false, func);
 
if (_mesa_is_gles3(ctx)) {
   /* Page 194 (page 206 of the PDF) in section 4.3.2 of the OpenGL ES
-- 
2.13.0

_

[Mesa-dev] [PATCH 13/21] mesa: add KHR_no_error support for glBlitFramebuffer()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/ARB_framebuffer_object.xml |  2 +-
 src/mesa/main/blit.c  | 15 +++
 src/mesa/main/blit.h  |  6 ++
 3 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_framebuffer_object.xml 
b/src/mapi/glapi/gen/ARB_framebuffer_object.xml
index 76114eb32a..26f495f8bb 100644
--- a/src/mapi/glapi/gen/ARB_framebuffer_object.xml
+++ b/src/mapi/glapi/gen/ARB_framebuffer_object.xml
@@ -271,7 +271,7 @@

 
 
-
+
 
 
 
diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index aa95e9c96c..4e5f7b27bf 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -569,6 +569,21 @@ blit_framebuffer(struct gl_context *ctx,
  * when the samples must be resolved to a single color.
  */
 void GLAPIENTRY
+_mesa_BlitFramebuffer_no_error(GLint srcX0, GLint srcY0, GLint srcX1,
+   GLint srcY1, GLint dstX0, GLint dstY0,
+   GLint dstX1, GLint dstY1,
+   GLbitfield mask, GLenum filter)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   blit_framebuffer(ctx, ctx->ReadBuffer, ctx->DrawBuffer,
+srcX0, srcY0, srcX1, srcY1,
+dstX0, dstY0, dstX1, dstY1,
+mask, filter, true, "glBlitFramebuffer");
+}
+
+
+void GLAPIENTRY
 _mesa_BlitFramebuffer(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
   GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
   GLbitfield mask, GLenum filter)
diff --git a/src/mesa/main/blit.h b/src/mesa/main/blit.h
index 1ca4f83028..6397518dbd 100644
--- a/src/mesa/main/blit.h
+++ b/src/mesa/main/blit.h
@@ -34,6 +34,12 @@ _mesa_regions_overlap(int srcX0, int srcY0,
   int dstX0, int dstY0,
   int dstX1, int dstY1);
 
+void GLAPIENTRY
+_mesa_BlitFramebuffer_no_error(GLint srcX0, GLint srcY0, GLint srcX1,
+   GLint srcY1, GLint dstX0, GLint dstY0,
+   GLint dstX1, GLint dstY1,
+   GLbitfield mask, GLenum filter);
+
 extern void GLAPIENTRY
 _mesa_BlitFramebuffer(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
  GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 14/21] mesa: add KHR_no_error support for glBlitNamedFramebuffer()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/ARB_direct_state_access.xml |  2 +-
 src/mesa/main/blit.c   | 30 ++
 src/mesa/main/blit.h   |  9 
 3 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_direct_state_access.xml 
b/src/mapi/glapi/gen/ARB_direct_state_access.xml
index cb24d7981c..54992e5304 100644
--- a/src/mapi/glapi/gen/ARB_direct_state_access.xml
+++ b/src/mapi/glapi/gen/ARB_direct_state_access.xml
@@ -247,7 +247,7 @@
   

 
-   
+   
   
   
   
diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 4e5f7b27bf..95a71fca60 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -606,6 +606,36 @@ _mesa_BlitFramebuffer(GLint srcX0, GLint srcY0, GLint 
srcX1, GLint srcY1,
 
 
 void GLAPIENTRY
+_mesa_BlitNamedFramebuffer_no_error(GLuint readFramebuffer,
+GLuint drawFramebuffer,
+GLint srcX0, GLint srcY0,
+GLint srcX1, GLint srcY1,
+GLint dstX0, GLint dstY0,
+GLint dstX1, GLint dstY1,
+GLbitfield mask, GLenum filter)
+{
+   struct gl_framebuffer *readFb, *drawFb;
+
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (readFramebuffer)
+  readFb = _mesa_lookup_framebuffer(ctx, readFramebuffer);
+   else
+  readFb = ctx->WinSysReadBuffer;
+
+   if (drawFramebuffer)
+  drawFb = _mesa_lookup_framebuffer(ctx, drawFramebuffer);
+   else
+  drawFb = ctx->WinSysDrawBuffer;
+
+   blit_framebuffer(ctx, readFb, drawFb,
+srcX0, srcY0, srcX1, srcY1,
+dstX0, dstY0, dstX1, dstY1,
+mask, filter, true, "glBlitNamedFramebuffer");
+}
+
+
+void GLAPIENTRY
 _mesa_BlitNamedFramebuffer(GLuint readFramebuffer, GLuint drawFramebuffer,
GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
diff --git a/src/mesa/main/blit.h b/src/mesa/main/blit.h
index 6397518dbd..39021e7be6 100644
--- a/src/mesa/main/blit.h
+++ b/src/mesa/main/blit.h
@@ -45,6 +45,15 @@ _mesa_BlitFramebuffer(GLint srcX0, GLint srcY0, GLint srcX1, 
GLint srcY1,
  GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
  GLbitfield mask, GLenum filter);
 
+void GLAPIENTRY
+_mesa_BlitNamedFramebuffer_no_error(GLuint readFramebuffer,
+GLuint drawFramebuffer,
+GLint srcX0, GLint srcY0,
+GLint srcX1, GLint srcY1,
+GLint dstX0, GLint dstY0,
+GLint dstX1, GLint dstY1,
+GLbitfield mask, GLenum filter);
+
 extern void GLAPIENTRY
 _mesa_BlitNamedFramebuffer(GLuint readFramebuffer, GLuint drawFramebuffer,
GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 16/21] mesa: add prepare_target() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/copyimage.c | 24 
 1 file changed, 24 insertions(+)

diff --git a/src/mesa/main/copyimage.c b/src/mesa/main/copyimage.c
index 07c2359dda..6a74f7a1a8 100644
--- a/src/mesa/main/copyimage.c
+++ b/src/mesa/main/copyimage.c
@@ -214,6 +214,30 @@ prepare_target_err(struct gl_context *ctx, GLuint name, 
GLenum target,
return true;
 }
 
+static void
+prepare_target(struct gl_context *ctx, GLuint name, GLenum target,
+   int level, int z,
+   struct gl_texture_image **texImage,
+   struct gl_renderbuffer **renderbuffer)
+{
+   if (target == GL_RENDERBUFFER) {
+  struct gl_renderbuffer *rb = _mesa_lookup_renderbuffer(ctx, name);
+
+  *renderbuffer = rb;
+  *texImage = NULL;
+   } else {
+  struct gl_texture_object *texObj = _mesa_lookup_texture(ctx, name);
+
+  if (target == GL_TEXTURE_CUBE_MAP) {
+ *texImage = texObj->Image[z][level];
+  }
+  else {
+ *texImage = _mesa_select_tex_image(texObj, target, level);
+  }
+
+  *renderbuffer = NULL;
+   }
+}
 
 /**
  * Check that the x,y,z,width,height,region is within the texture image
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 15/21] mesa: rename prepare_target() to prepare_target_err()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/copyimage.c | 34 ++
 1 file changed, 18 insertions(+), 16 deletions(-)

diff --git a/src/mesa/main/copyimage.c b/src/mesa/main/copyimage.c
index cf25159e88..07c2359dda 100644
--- a/src/mesa/main/copyimage.c
+++ b/src/mesa/main/copyimage.c
@@ -57,16 +57,16 @@ enum mesa_block_class {
  * \return true if success, false if error
  */
 static bool
-prepare_target(struct gl_context *ctx, GLuint name, GLenum target,
-   int level, int z, int depth,
-   struct gl_texture_image **tex_image,
-   struct gl_renderbuffer **renderbuffer,
-   mesa_format *format,
-   GLenum *internalFormat,
-   GLuint *width,
-   GLuint *height,
-   GLuint *num_samples,
-   const char *dbg_prefix)
+prepare_target_err(struct gl_context *ctx, GLuint name, GLenum target,
+   int level, int z, int depth,
+   struct gl_texture_image **tex_image,
+   struct gl_renderbuffer **renderbuffer,
+   mesa_format *format,
+   GLenum *internalFormat,
+   GLuint *width,
+   GLuint *height,
+   GLuint *num_samples,
+   const char *dbg_prefix)
 {
if (name == 0) {
   _mesa_error(ctx, GL_INVALID_VALUE,
@@ -484,14 +484,16 @@ _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, 
GLint srcLevel,
   return;
}
 
-   if (!prepare_target(ctx, srcName, srcTarget, srcLevel, srcZ, srcDepth,
-   &srcTexImage, &srcRenderbuffer, &srcFormat,
-   &srcIntFormat, &src_w, &src_h, &src_num_samples, "src"))
+   if (!prepare_target_err(ctx, srcName, srcTarget, srcLevel, srcZ, srcDepth,
+   &srcTexImage, &srcRenderbuffer, &srcFormat,
+   &srcIntFormat, &src_w, &src_h, &src_num_samples,
+   "src"))
   return;
 
-   if (!prepare_target(ctx, dstName, dstTarget, dstLevel, dstZ, srcDepth,
-   &dstTexImage, &dstRenderbuffer, &dstFormat,
-   &dstIntFormat, &dst_w, &dst_h, &dst_num_samples, "dst"))
+   if (!prepare_target_err(ctx, dstName, dstTarget, dstLevel, dstZ, srcDepth,
+   &dstTexImage, &dstRenderbuffer, &dstFormat,
+   &dstIntFormat, &dst_w, &dst_h, &dst_num_samples,
+   "dst"))
   return;
 
_mesa_get_format_block_size(srcFormat, &src_bw, &src_bh);
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 17/21] mesa: add copy_image_subdata() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/copyimage.c | 78 ---
 1 file changed, 47 insertions(+), 31 deletions(-)

diff --git a/src/mesa/main/copyimage.c b/src/mesa/main/copyimage.c
index 6a74f7a1a8..16a8132544 100644
--- a/src/mesa/main/copyimage.c
+++ b/src/mesa/main/copyimage.c
@@ -474,6 +474,50 @@ copy_format_compatible(const struct gl_context *ctx,
return false;
 }
 
+static void
+copy_image_subdata(struct gl_context *ctx,
+   struct gl_texture_image *srcTexImage,
+   struct gl_renderbuffer *srcRenderbuffer,
+   int srcX, int srcY, int srcZ, int srcLevel,
+   struct gl_texture_image *dstTexImage,
+   struct gl_renderbuffer *dstRenderbuffer,
+   int dstX, int dstY, int dstZ, int dstLevel,
+   int srcWidth, int srcHeight, int srcDepth)
+{
+   int i;
+
+   /* loop over 2D slices/faces/layers */
+   for (i = 0; i < srcDepth; ++i) {
+  int newSrcZ = srcZ + i;
+  int newDstZ = dstZ + i;
+
+  if (srcTexImage &&
+  srcTexImage->TexObject->Target == GL_TEXTURE_CUBE_MAP) {
+ /* need to update srcTexImage pointer for the cube face */
+ assert(srcZ + i < MAX_FACES);
+ srcTexImage = srcTexImage->TexObject->Image[srcZ + i][srcLevel];
+ assert(srcTexImage);
+ newSrcZ = 0;
+  }
+
+  if (dstTexImage &&
+  dstTexImage->TexObject->Target == GL_TEXTURE_CUBE_MAP) {
+ /* need to update dstTexImage pointer for the cube face */
+ assert(dstZ + i < MAX_FACES);
+ dstTexImage = dstTexImage->TexObject->Image[dstZ + i][dstLevel];
+ assert(dstTexImage);
+ newDstZ = 0;
+  }
+
+  ctx->Driver.CopyImageSubData(ctx,
+   srcTexImage, srcRenderbuffer,
+   srcX, srcY, newSrcZ,
+   dstTexImage, dstRenderbuffer,
+   dstX, dstY, newDstZ,
+   srcWidth, srcHeight);
+   }
+}
+
 void GLAPIENTRY
 _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, GLint srcLevel,
GLint srcX, GLint srcY, GLint srcZ,
@@ -490,7 +534,6 @@ _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, 
GLint srcLevel,
GLuint src_bw, src_bh, dst_bw, dst_bh;
GLuint src_num_samples, dst_num_samples;
int dstWidth, dstHeight, dstDepth;
-   int i;
 
if (MESA_VERBOSE & VERBOSE_API)
   _mesa_debug(ctx, "glCopyImageSubData(%u, %s, %d, %d, %d, %d, "
@@ -606,34 +649,7 @@ _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, 
GLint srcLevel,
   return;
}
 
-   /* loop over 2D slices/faces/layers */
-   for (i = 0; i < srcDepth; ++i) {
-  int newSrcZ = srcZ + i;
-  int newDstZ = dstZ + i;
-
-  if (srcTexImage &&
-  srcTexImage->TexObject->Target == GL_TEXTURE_CUBE_MAP) {
- /* need to update srcTexImage pointer for the cube face */
- assert(srcZ + i < MAX_FACES);
- srcTexImage = srcTexImage->TexObject->Image[srcZ + i][srcLevel];
- assert(srcTexImage);
- newSrcZ = 0;
-  }
-
-  if (dstTexImage &&
-  dstTexImage->TexObject->Target == GL_TEXTURE_CUBE_MAP) {
- /* need to update dstTexImage pointer for the cube face */
- assert(dstZ + i < MAX_FACES);
- dstTexImage = dstTexImage->TexObject->Image[dstZ + i][dstLevel];
- assert(dstTexImage);
- newDstZ = 0;
-  }
-
-  ctx->Driver.CopyImageSubData(ctx,
-   srcTexImage, srcRenderbuffer,
-   srcX, srcY, newSrcZ,
-   dstTexImage, dstRenderbuffer,
-   dstX, dstY, newDstZ,
-   srcWidth, srcHeight);
-   }
+   copy_image_subdata(ctx, srcTexImage, srcRenderbuffer, srcX, srcY, srcZ,
+  srcLevel, dstTexImage, dstRenderbuffer, dstX, dstY, dstZ,
+  dstLevel, srcWidth, srcHeight, srcDepth);
 }
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 18/21] mesa: add KHR_no_error support for glCopyImageSubData()

[Samuel Pitoiset]

---
 src/mapi/glapi/gen/ARB_copy_image.xml |  2 +-
 src/mesa/main/copyimage.c | 23 +++
 src/mesa/main/copyimage.h |  7 +++
 3 files changed, 31 insertions(+), 1 deletion(-)

diff --git a/src/mapi/glapi/gen/ARB_copy_image.xml 
b/src/mapi/glapi/gen/ARB_copy_image.xml
index 9ee2ba304f..fb4c9b1c24 100644
--- a/src/mapi/glapi/gen/ARB_copy_image.xml
+++ b/src/mapi/glapi/gen/ARB_copy_image.xml
@@ -5,7 +5,7 @@
 
 
 
-
+
 
 
 
diff --git a/src/mesa/main/copyimage.c b/src/mesa/main/copyimage.c
index 16a8132544..4a67f420de 100644
--- a/src/mesa/main/copyimage.c
+++ b/src/mesa/main/copyimage.c
@@ -519,6 +519,29 @@ copy_image_subdata(struct gl_context *ctx,
 }
 
 void GLAPIENTRY
+_mesa_CopyImageSubData_no_error(GLuint srcName, GLenum srcTarget, GLint 
srcLevel,
+GLint srcX, GLint srcY, GLint srcZ,
+GLuint dstName, GLenum dstTarget, GLint 
dstLevel,
+GLint dstX, GLint dstY, GLint dstZ,
+GLsizei srcWidth, GLsizei srcHeight, GLsizei 
srcDepth)
+{
+   struct gl_texture_image *srcTexImage, *dstTexImage;
+   struct gl_renderbuffer *srcRenderbuffer, *dstRenderbuffer;
+
+   GET_CURRENT_CONTEXT(ctx);
+
+   prepare_target(ctx, srcName, srcTarget, srcLevel, srcZ, &srcTexImage,
+  &srcRenderbuffer);
+
+   prepare_target(ctx, dstName, dstTarget, dstLevel, dstZ, &dstTexImage,
+  &dstRenderbuffer);
+
+   copy_image_subdata(ctx, srcTexImage, srcRenderbuffer, srcX, srcY, srcZ,
+  srcLevel, dstTexImage, dstRenderbuffer, dstX, dstY, dstZ,
+  dstLevel, srcWidth, srcHeight, srcDepth);
+}
+
+void GLAPIENTRY
 _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, GLint srcLevel,
GLint srcX, GLint srcY, GLint srcZ,
GLuint dstName, GLenum dstTarget, GLint dstLevel,
diff --git a/src/mesa/main/copyimage.h b/src/mesa/main/copyimage.h
index 40e95b6631..ea2f15b435 100644
--- a/src/mesa/main/copyimage.h
+++ b/src/mesa/main/copyimage.h
@@ -35,6 +35,13 @@
 extern "C" {
 #endif
 
+void GLAPIENTRY
+_mesa_CopyImageSubData_no_error(GLuint srcName, GLenum srcTarget, GLint 
srcLevel,
+GLint srcX, GLint srcY, GLint srcZ,
+GLuint destName, GLenum destTarget, GLint 
destLevel,
+GLint destX, GLint destY, GLint destZ,
+GLsizei srcWidth, GLsizei srcHeight, GLsizei 
srcDepth);
+
 extern void GLAPIENTRY
 _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, GLint srcLevel,
GLint srcX, GLint srcY, GLint srcZ,
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 19/21] mesa: rename texsubimage() to texsubimage_err()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/teximage.c | 34 +-
 1 file changed, 17 insertions(+), 17 deletions(-)

diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index fed1dad262..9285a96b68 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -3238,11 +3238,11 @@ _mesa_texture_sub_image(struct gl_context *ctx, GLuint 
dims,
  * Must split this out this way because of GL_TEXTURE_CUBE_MAP.
  */
 static void
-texsubimage(struct gl_context *ctx, GLuint dims, GLenum target, GLint level,
-GLint xoffset, GLint yoffset, GLint zoffset,
-GLsizei width, GLsizei height, GLsizei depth,
-GLenum format, GLenum type, const GLvoid *pixels,
-const char *callerName)
+texsubimage_err(struct gl_context *ctx, GLuint dims, GLenum target, GLint 
level,
+GLint xoffset, GLint yoffset, GLint zoffset,
+GLsizei width, GLsizei height, GLsizei depth,
+GLenum format, GLenum type, const GLvoid *pixels,
+const char *callerName)
 {
struct gl_texture_object *texObj;
struct gl_texture_image *texImage;
@@ -3402,10 +3402,10 @@ _mesa_TexSubImage1D( GLenum target, GLint level,
  const GLvoid *pixels )
 {
GET_CURRENT_CONTEXT(ctx);
-   texsubimage(ctx, 1, target, level,
-   xoffset, 0, 0,
-   width, 1, 1,
-   format, type, pixels, "glTexSubImage1D");
+   texsubimage_err(ctx, 1, target, level,
+   xoffset, 0, 0,
+   width, 1, 1,
+   format, type, pixels, "glTexSubImage1D");
 }
 
 
@@ -3417,10 +3417,10 @@ _mesa_TexSubImage2D( GLenum target, GLint level,
  const GLvoid *pixels )
 {
GET_CURRENT_CONTEXT(ctx);
-   texsubimage(ctx, 2, target, level,
-   xoffset, yoffset, 0,
-   width, height, 1,
-   format, type, pixels, "glTexSubImage2D");
+   texsubimage_err(ctx, 2, target, level,
+   xoffset, yoffset, 0,
+   width, height, 1,
+   format, type, pixels, "glTexSubImage2D");
 }
 
 
@@ -3433,10 +3433,10 @@ _mesa_TexSubImage3D( GLenum target, GLint level,
  const GLvoid *pixels )
 {
GET_CURRENT_CONTEXT(ctx);
-   texsubimage(ctx, 3, target, level,
-   xoffset, yoffset, zoffset,
-   width, height, depth,
-   format, type, pixels, "glTexSubImage3D");
+   texsubimage_err(ctx, 3, target, level,
+   xoffset, yoffset, zoffset,
+   width, height, depth,
+   format, type, pixels, "glTexSubImage3D");
 }
 
 void GLAPIENTRY
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 20/21] mesa: add texsubimage() helper

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/teximage.c | 18 ++
 1 file changed, 18 insertions(+)

diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index 9285a96b68..ca8090f653 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -3282,6 +3282,24 @@ texsubimage_err(struct gl_context *ctx, GLuint dims, 
GLenum target, GLint level,
 }
 
 
+static void
+texsubimage(struct gl_context *ctx, GLuint dims, GLenum target, GLint level,
+GLint xoffset, GLint yoffset, GLint zoffset,
+GLsizei width, GLsizei height, GLsizei depth,
+ GLenum format, GLenum type, const GLvoid *pixels)
+{
+   struct gl_texture_object *texObj;
+   struct gl_texture_image *texImage;
+
+   texObj = _mesa_get_current_tex_object(ctx, target);
+   texImage = _mesa_select_tex_image(texObj, target, level);
+
+   _mesa_texture_sub_image(ctx, dims, texObj, texImage, target, level,
+   xoffset, yoffset, zoffset, width, height, depth,
+   format, type, pixels, false);
+}
+
+
 /**
  * Implement all the glTextureSubImage1/2/3D() functions.
  * Must split this out this way because of GL_TEXTURE_CUBE_MAP.
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 21/21] mesa: add KHR_no_error support for glTexSubImage*D()

[Samuel Pitoiset]

Signed-off-by: Samuel Pitoiset 
---
 src/mapi/glapi/gen/gl_API.xml |  6 +++---
 src/mesa/main/teximage.c  | 43 +++
 src/mesa/main/teximage.h  | 18 ++
 3 files changed, 64 insertions(+), 3 deletions(-)

diff --git a/src/mapi/glapi/gen/gl_API.xml b/src/mapi/glapi/gen/gl_API.xml
index 5814d2ad92..d7c6ef8ca7 100644
--- a/src/mapi/glapi/gen/gl_API.xml
+++ b/src/mapi/glapi/gen/gl_API.xml
@@ -3299,7 +3299,7 @@
 
 
 
-
+
 
 
 
@@ -3311,7 +3311,7 @@
 
 
 
-
+
 
 
 
@@ -4025,7 +4025,7 @@
 
 
 
-
+
 
 
 
diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index ca8090f653..613bc13a5f 100644
--- a/src/mesa/main/teximage.c
+++ b/src/mesa/main/teximage.c
@@ -3414,6 +3414,20 @@ texturesubimage(struct gl_context *ctx, GLuint dims,
 
 
 void GLAPIENTRY
+_mesa_TexSubImage1D_no_error(GLenum target, GLint level,
+ GLint xoffset, GLsizei width,
+ GLenum format, GLenum type,
+ const GLvoid *pixels)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   texsubimage(ctx, 1, target, level,
+   xoffset, 0, 0,
+   width, 1, 1,
+   format, type, pixels);
+}
+
+
+void GLAPIENTRY
 _mesa_TexSubImage1D( GLenum target, GLint level,
  GLint xoffset, GLsizei width,
  GLenum format, GLenum type,
@@ -3428,6 +3442,21 @@ _mesa_TexSubImage1D( GLenum target, GLint level,
 
 
 void GLAPIENTRY
+_mesa_TexSubImage2D_no_error(GLenum target, GLint level,
+ GLint xoffset, GLint yoffset,
+ GLsizei width, GLsizei height,
+ GLenum format, GLenum type,
+ const GLvoid *pixels)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   texsubimage(ctx, 2, target, level,
+   xoffset, yoffset, 0,
+   width, height, 1,
+   format, type, pixels);
+}
+
+
+void GLAPIENTRY
 _mesa_TexSubImage2D( GLenum target, GLint level,
  GLint xoffset, GLint yoffset,
  GLsizei width, GLsizei height,
@@ -3442,6 +3471,20 @@ _mesa_TexSubImage2D( GLenum target, GLint level,
 }
 
 
+void GLAPIENTRY
+_mesa_TexSubImage3D_no_error(GLenum target, GLint level,
+ GLint xoffset, GLint yoffset, GLint zoffset,
+ GLsizei width, GLsizei height, GLsizei depth,
+ GLenum format, GLenum type,
+ const GLvoid *pixels)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   texsubimage(ctx, 3, target, level,
+   xoffset, yoffset, zoffset,
+   width, height, depth,
+   format, type, pixels);
+}
+
 
 void GLAPIENTRY
 _mesa_TexSubImage3D( GLenum target, GLint level,
diff --git a/src/mesa/main/teximage.h b/src/mesa/main/teximage.h
index bc4e235d38..295e35abb1 100644
--- a/src/mesa/main/teximage.h
+++ b/src/mesa/main/teximage.h
@@ -271,12 +271,24 @@ _mesa_TexImage3DEXT( GLenum target, GLint level, GLenum 
internalformat,
 extern void GLAPIENTRY
 _mesa_EGLImageTargetTexture2DOES( GLenum target, GLeglImageOES image );
 
+void GLAPIENTRY
+_mesa_TexSubImage1D_no_error(GLenum target, GLint level, GLint xoffset,
+ GLsizei width,
+ GLenum format, GLenum type,
+ const GLvoid *pixels);
+
 extern void GLAPIENTRY
 _mesa_TexSubImage1D( GLenum target, GLint level, GLint xoffset,
  GLsizei width,
  GLenum format, GLenum type,
  const GLvoid *pixels );
 
+void GLAPIENTRY
+_mesa_TexSubImage2D_no_error(GLenum target, GLint level,
+ GLint xoffset, GLint yoffset,
+ GLsizei width, GLsizei height,
+ GLenum format, GLenum type,
+ const GLvoid *pixels);
 
 extern void GLAPIENTRY
 _mesa_TexSubImage2D( GLenum target, GLint level,
@@ -285,6 +297,12 @@ _mesa_TexSubImage2D( GLenum target, GLint level,
  GLenum format, GLenum type,
  const GLvoid *pixels );
 
+void GLAPIENTRY
+_mesa_TexSubImage3D_no_error(GLenum target, GLint level,
+ GLint xoffset, GLint yoffset, GLint zoffset,
+ GLsizei width, GLsizei height, GLsizei depth,
+ GLenum format, GLenum type,
+ const GLvoid *pixels);
 
 extern void GLAPIENTRY
 _mesa_TexSubImage3D( GLenum target, GLint level,
-- 
2.13.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH mesa 2/2] egl/display: make platform detection thread-safe

[Emil Velikov]

Hi guys,

On 1 June 2017 at 12:56, Grazvydas Ignotas  wrote:
> On Thu, Jun 1, 2017 at 2:15 PM, Eric Engestrom
>  wrote:
>> If the detections methods ever become able to return different results
>> for different threads, the if chain might make threads go back and forth
>> between invalid and valid platforms.
>> Solve this by doing the detection in a local var and only overwriting
>> the global one at the end, if no other thread has updated it since.
>>
>> This means the platform detected in the thread might not be the platform
>> returned by the function, but this is a different issue that will need
>> to be discussed when this becomes possible.
>>
>> Reported-by: Grazvydas Ignotas 
>
> Not really, see https://bugs.freedesktop.org/show_bug.cgi?id=101252
>
>> Signed-off-by: Eric Engestrom 
>> ---
>>
>> This is unnecessary in my opinion, but doesn't hurt :)
>
> It is necessary, without it things will work most of the time but
> there will be rare failures.
> Imagine there are 2 threads that both call _eglGetNativePlatform()
> simultaneously:
> - thread 1 completes the first "if (native_platform ==
> _EGL_INVALID_PLATFORM)" check and is preemted to do something else
> - thread 2 executes the whole function, does "native_platform =
> _EGL_NATIVE_PLATFORM" and just before returning it's preemted
> - thread 1 wakes up and calls _eglGetNativePlatformFromEnv() which
> returns _EGL_INVALID_PLATFORM because no env vars are set, updates
> native_platform and then gets preemted again
> - thread 2 wakes up and returns wrong _EGL_INVALID_PLATFORM
>
Afaict this/similar fix is necessary, yet let me put an alternative
idea - add locking via _eglGlobal.Mutex.
May be an bit of overkill, but it's what we consistently use
throughout the code base.

Reverting the patch (as suggested by Grazvydas) does not fully address
the problem, but only makes it less likely to hit.

-Emil
P.S. Even with this resolved we're likely to hit another issue as
mentioned here https://bugs.freedesktop.org/show_bug.cgi?id=54971
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] util/rand_xor: add missing include statements

[Eric Engestrom]

On Thursday, 2017-06-01 12:13:18 +0200, Nicolas Dechesne wrote:
> Fixes for:
> 
> src/util/rand_xor.c:60:13: error: implicit declaration of function 'open' 
> [-Werror=implicit-function-declaration]
> int fd = open("/dev/urandom", O_RDONLY);
>  ^~~~
> src/util/rand_xor.c:60:34: error: 'O_RDONLY' undeclared (first use in this 
> function)
> int fd = open("/dev/urandom", O_RDONLY);
>   ^~~~
> 
> Signed-off-by: Nicolas Dechesne 

Thanks, r-b and pushed, and welcome to Mesa :)

> ---
>  src/util/rand_xor.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/src/util/rand_xor.c b/src/util/rand_xor.c
> index de05fa64b3..de04bbc284 100644
> --- a/src/util/rand_xor.c
> +++ b/src/util/rand_xor.c
> @@ -25,6 +25,7 @@
>  #if defined(__linux__)
>  #include 
>  #include 
> +#include 
>  #else
>  #include 
>  #endif
> -- 
> 2.13.0
> 
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [RFC] some glsl->tgsi optimisations

[Samuel Pitoiset]

Hi Dave,

Can you rebase your series on top of master because it doesn't apply.

I'm fine with the general idea. Note that I have a shader which spends 
5s in copy_propagate()... That's huge! Looks like this series might 
improve the situation here, but needs more testing.


Thanks!

On 05/30/2017 07:52 AM, Dave Airlie wrote:

While looking at the fp64 emulation code, we were spending minutes
in the glsl->tgsi passes as some tests were producing shaders with

32000 temporaries. Now it might be possible to reduce these earlier

with some GLSL passes, but this code is pretty bad as-is.

This reduces one test execution time from 4m30 -> 5s.

I've not regression tested these much, but someone interested
might want to throw them at some slow compile situations to see
if they help any.

I'll give them a lot more testing before I consider them non RFC.

Dave.

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 4/4] st_glsl_to_tgsi: replace variables tracking list with a hash table

[Samuel Pitoiset]

On 05/30/2017 07:52 AM, Dave Airlie wrote:

From: Dave Airlie 

This removes the linear search which is fail when number of variables
goes up to 3 or so.
---
  src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 44 +-
  1 file changed, 31 insertions(+), 13 deletions(-)

diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp 
b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index 0e59aca..87c4b10 100644
--- a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -56,6 +56,7 @@
  #include "st_nir.h"
  #include "st_shader_cache.h"
  
+#include "util/hash_table.h"

  #include 
  
  #define PROGRAM_ANY_CONST ((1 << PROGRAM_STATE_VAR) |\

@@ -310,7 +311,9 @@ public:
 const struct tgsi_opcode_info *info;
  };
  
-class variable_storage : public exec_node {

+class variable_storage {
+   DECLARE_RZALLOC_CXX_OPERATORS(variable_storage)
+
  public:
 variable_storage(ir_variable *var, gl_register_file file, int index,
  unsigned array_id = 0)
@@ -488,7 +491,7 @@ public:
 st_src_reg result;
  
 /** List of variable_storage */

-   exec_list variables;
+   struct hash_table *variables;
  
 /** List of immediate_storage */

 exec_list immediates;
@@ -1306,13 +1309,15 @@ glsl_to_tgsi_visitor::get_temp(const glsl_type *type)
  variable_storage *
  glsl_to_tgsi_visitor::find_variable_storage(ir_variable *var)
  {
+   struct hash_entry *entry = _mesa_hash_table_search(this->variables,
+  var);
+   variable_storage *storage;
+   if (!entry)
+  return NULL;
  
-   foreach_in_list(variable_storage, entry, &this->variables) {

-  if (entry->var == var)
- return entry;
-   }
+   storage = (variable_storage *)entry->data;
  
-   return NULL;

+   return storage;
  }


I would suggest to write to:

{
  struct hash_entry *entry;

  entry = _mesa_hash_table_search(this->variables, var);
  if (!entry)
return NULL;
  return (variable_storage *)entry->data;
}

I'm just wondering if this might affect very small shaders. I think a 
full shaderdb run can answer the question. :)


  
  void

@@ -1345,7 +1350,8 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
if (i == ir->get_num_state_slots()) {
   /* We'll set the index later. */
   storage = new(mem_ctx) variable_storage(ir, PROGRAM_STATE_VAR, -1);
- this->variables.push_tail(storage);
+
+ _mesa_hash_table_insert(this->variables, ir, storage);
  
   dst = undef_dst;

} else {
@@ -1360,7 +1366,7 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
   storage = new(mem_ctx) variable_storage(ir, dst.file, dst.index,
   dst.array_id);
  
- this->variables.push_tail(storage);

+ _mesa_hash_table_insert(this->variables, ir, storage);
}
  
  
@@ -2595,7 +2601,7 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)

case ir_var_uniform:
   entry = new(mem_ctx) variable_storage(var, PROGRAM_UNIFORM,
 var->data.param_index);
- this->variables.push_tail(entry);
+ _mesa_hash_table_insert(this->variables, var, entry);
   break;
case ir_var_shader_in: {
   /* The linker assigns locations for varyings and attributes,
@@ -2642,7 +2648,7 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
 decl->array_id);
   entry->component = component;
  
- this->variables.push_tail(entry);

+ _mesa_hash_table_insert(this->variables, var, entry);
   break;
}
case ir_var_shader_out: {
@@ -2700,7 +2706,8 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
   }
   entry->component = component;
  
- this->variables.push_tail(entry);

+ _mesa_hash_table_insert(this->variables, var, entry);
+
   break;
}
case ir_var_system_value:
@@ -2714,7 +2721,7 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
  
   entry = new(mem_ctx) variable_storage(var, src.file, src.index,

 src.array_id);
- this->variables.push_tail(entry);
+ _mesa_hash_table_insert(this->variables, var, entry);
  
   break;

}
@@ -4569,10 +4576,19 @@ glsl_to_tgsi_visitor::glsl_to_tgsi_visitor()
 have_fma = false;
 use_shared_memory = false;
 has_tex_txf_lz = false;
+   variables = NULL;
+}
+
+static void var_destroy(struct hash_entry *entry)
+{
+   variable_storage *storage = (variable_storage *)entry->data;
+
+   delete storage;
  }
  
  glsl_to_tgsi_visitor::~glsl_to_tgsi_visitor()

  {
+   _mesa_hash_table_destroy(variables, var_destroy);
 free(array_sizes);
 ralloc_free(mem_ctx);
  }
@@ -6772,6 +6788,8 @@ get_mesa_program_tgsi(struct gl_context *ctx,
 v->h

Re: [Mesa-dev] [PATCH 1/4] st_glsl_to_tgsi: bump index back up to 32-bit

[Samuel Pitoiset]

Reviewed-by: Samuel Pitoiset 

On 05/30/2017 07:52 AM, Dave Airlie wrote:

From: Dave Airlie 

with some of the fp64 emulation, we are seeing shaders coming in with

32K temps, they go out with 40 or so used, but while doing register

renumber we need to store a lot of them.

So bump this fields back up to 32-bit.
---
  src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp 
b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index 1eacbb1..e87c241 100644
--- a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -165,7 +165,7 @@ public:
  
 explicit st_src_reg(st_dst_reg reg);
  
-   int16_t index; /**< temporary index, VERT_ATTRIB_*, VARYING_SLOT_*, etc. */

+   int32_t index; /**< temporary index, VERT_ATTRIB_*, VARYING_SLOT_*, etc. */
 int16_t index2D;
 uint16_t swizzle; /**< SWIZZLE_XYZWONEZERO swizzles from Mesa. */
 int negate:4; /**< NEGATE_XYZW mask from mesa */
@@ -239,7 +239,7 @@ public:
  
 explicit st_dst_reg(st_src_reg reg);
  
-   int16_t index; /**< temporary index, VERT_ATTRIB_*, VARYING_SLOT_*, etc. */

+   int32_t index; /**< temporary index, VERT_ATTRIB_*, VARYING_SLOT_*, etc. */
 int16_t index2D;
 gl_register_file file:5; /**< PROGRAM_* from Mesa */
 unsigned writemask:4; /**< Bitfield of WRITEMASK_[XYZW] */


___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 09/21] mesa: add validate_stencil_buffer() helper

[Gustaw Smolarczyk]

This time, send it to the list too.
Gustaw


1 cze 2017 15:45 "Gustaw Smolarczyk"  napisał(a):

1 cze 2017 15:07 "Samuel Pitoiset"  napisał(a):

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 111 +++---
-
 1 file changed, 58 insertions(+), 53 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 2c0300eab3..207ce7d501 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -178,6 +178,62 @@ is_valid_blit_filter(const struct gl_context *ctx,
GLenum filter)


 static void


Shouldn't it return a bool so that the caller will know whether the
validation passed or not?

+validate_stencil_buffer(struct gl_context *ctx, struct gl_framebuffer
*readFb,
+struct gl_framebuffer *drawFb, GLbitfield *mask,
+bool no_error, const char *func)
+{
+   struct gl_renderbuffer *readRb =
+  readFb->Attachment[BUFFER_STENCIL].Renderbuffer;
+   struct gl_renderbuffer *drawRb =
+  drawFb->Attachment[BUFFER_STENCIL].Renderbuffer;
+
+   /* From the EXT_framebuffer_object spec:
+*
+* "If a buffer is specified in  and does not exist in both
+* the read and draw framebuffers, the corresponding bit is silently
+* ignored."
+*/
+   if (readRb == NULL || drawRb == NULL) {
+  *mask &= ~GL_STENCIL_BUFFER_BIT;
+   } else if (!no_error) {
+  int read_z_bits, draw_z_bits;
+
+  if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(source and destination stencil buffer cannot be
the "
+ "same)", func);
+ return;
+  }
+
+  if (_mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS) !=
+  _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS)) {
+ /* There is no need to check the stencil datatype here, because
+  * there is only one: GL_UNSIGNED_INT.
+  */
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(stencil attachment format mismatch)", func);
+ return;
+  }
+
+  read_z_bits = _mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS);
+  draw_z_bits = _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS);
+
+  /* If both buffers also have depth data, the depth formats must match
+   * as well.  If one doesn't have depth, it's not blitted, so we
should
+   * ignore the depth format check.
+   */
+  if (read_z_bits > 0 && draw_z_bits > 0 &&
+  (read_z_bits != draw_z_bits ||
+   _mesa_get_format_datatype(readRb->Format) !=
+   _mesa_get_format_datatype(drawRb->Format))) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(stencil attachment depth format mismatch)", func);
+ return;
+  }
+   }
+}
+
+static void
 blit_framebuffer(struct gl_context *ctx,
  struct gl_framebuffer *readFb, struct gl_framebuffer
*drawFb,
  GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
@@ -317,59 +373,8 @@ blit_framebuffer(struct gl_context *ctx,
   }
}

-   if (mask & GL_STENCIL_BUFFER_BIT) {
-  struct gl_renderbuffer *readRb =
- readFb->Attachment[BUFFER_STENCIL].Renderbuffer;
-  struct gl_renderbuffer *drawRb =
- drawFb->Attachment[BUFFER_STENCIL].Renderbuffer;
-
-  /* From the EXT_framebuffer_object spec:
-   *
-   * "If a buffer is specified in  and does not exist in both
-   * the read and draw framebuffers, the corresponding bit is
silently
-   * ignored."
-   */
-  if ((readRb == NULL) || (drawRb == NULL)) {
- mask &= ~GL_STENCIL_BUFFER_BIT;
-  }
-  else {
- int read_z_bits, draw_z_bits;
-
- if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(source and destination stencil "
-"buffer cannot be the same)", func);
-return;
- }
-
- if (_mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS) !=
- _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS)) {
-/* There is no need to check the stencil datatype here, because
- * there is only one: GL_UNSIGNED_INT.
- */
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(stencil attachment format mismatch)", func);
-return;
- }
-
- read_z_bits = _mesa_get_format_bits(readRb->Format,
GL_DEPTH_BITS);
- draw_z_bits = _mesa_get_format_bits(drawRb->Format,
GL_DEPTH_BITS);
-
- /* If both buffers also have depth data, the depth formats must
match
-  * as well.  If one doesn't have depth, it's not blitted, so we
should
-  * ignore the depth format check.
-  */
- if (read_z_bits > 0 && draw_z_bits > 0 &&
- (read_z_bits != draw_z_bits ||
-  _mesa_get_format_da

Re: [Mesa-dev] [PATCH 09/21] mesa: add validate_stencil_buffer() helper

[Samuel Pitoiset]

Yeah, will fix locally, thanks!

On 06/01/2017 03:47 PM, Gustaw Smolarczyk wrote:

This time, send it to the list too.
Gustaw


1 cze 2017 15:45 "Gustaw Smolarczyk" > napisał(a):


1 cze 2017 15:07 "Samuel Pitoiset" mailto:samuel.pitoi...@gmail.com>> napisał(a):

Signed-off-by: Samuel Pitoiset mailto:samuel.pitoi...@gmail.com>>
---
  src/mesa/main/blit.c | 111
+++
  1 file changed, 58 insertions(+), 53 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 2c0300eab3..207ce7d501 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -178,6 +178,62 @@ is_valid_blit_filter(const struct
gl_context *ctx, GLenum filter)


  static void


Shouldn't it return a bool so that the caller will know whether the
validation passed or not?

+validate_stencil_buffer(struct gl_context *ctx, struct
gl_framebuffer *readFb,
+struct gl_framebuffer *drawFb,
GLbitfield *mask,
+bool no_error, const char *func)
+{
+   struct gl_renderbuffer *readRb =
+  readFb->Attachment[BUFFER_STENCIL].Renderbuffer;
+   struct gl_renderbuffer *drawRb =
+  drawFb->Attachment[BUFFER_STENCIL].Renderbuffer;
+
+   /* From the EXT_framebuffer_object spec:
+*
+* "If a buffer is specified in  and does not
exist in both
+* the read and draw framebuffers, the corresponding bit
is silently
+* ignored."
+*/
+   if (readRb == NULL || drawRb == NULL) {
+  *mask &= ~GL_STENCIL_BUFFER_BIT;
+   } else if (!no_error) {
+  int read_z_bits, draw_z_bits;
+
+  if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(source and destination stencil buffer
cannot be the "
+ "same)", func);
+ return;
+  }
+
+  if (_mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS) !=
+  _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS)) {
+ /* There is no need to check the stencil datatype
here, because
+  * there is only one: GL_UNSIGNED_INT.
+  */
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(stencil attachment format mismatch)",
func);
+ return;
+  }
+
+  read_z_bits = _mesa_get_format_bits(readRb->Format,
GL_DEPTH_BITS);
+  draw_z_bits = _mesa_get_format_bits(drawRb->Format,
GL_DEPTH_BITS);
+
+  /* If both buffers also have depth data, the depth
formats must match
+   * as well.  If one doesn't have depth, it's not blitted,
so we should
+   * ignore the depth format check.
+   */
+  if (read_z_bits > 0 && draw_z_bits > 0 &&
+  (read_z_bits != draw_z_bits ||
+   _mesa_get_format_datatype(readRb->Format) !=
+   _mesa_get_format_datatype(drawRb->Format))) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(stencil attachment depth format
mismatch)", func);
+ return;
+  }
+   }
+}
+
+static void
  blit_framebuffer(struct gl_context *ctx,
   struct gl_framebuffer *readFb, struct
gl_framebuffer *drawFb,
   GLint srcX0, GLint srcY0, GLint srcX1, GLint
srcY1,
@@ -317,59 +373,8 @@ blit_framebuffer(struct gl_context *ctx,
}
 }

-   if (mask & GL_STENCIL_BUFFER_BIT) {
-  struct gl_renderbuffer *readRb =
- readFb->Attachment[BUFFER_STENCIL].Renderbuffer;
-  struct gl_renderbuffer *drawRb =
- drawFb->Attachment[BUFFER_STENCIL].Renderbuffer;
-
-  /* From the EXT_framebuffer_object spec:
-   *
-   * "If a buffer is specified in  and does not
exist in both
-   * the read and draw framebuffers, the corresponding
bit is silently
-   * ignored."
-   */
-  if ((readRb == NULL) || (drawRb == NULL)) {
- mask &= ~GL_STENCIL_BUFFER_BIT;
-  }
-  else {
- int read_z_bits, draw_z_bits;
-
- if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s

Re: [Mesa-dev] [PATCH 10/21] mesa: add validate_depth_buffer() helper

[Gustaw Smolarczyk]

Same problem as in the previous patch.

Regards,
Gustaw


1 cze 2017 15:07 "Samuel Pitoiset"  napisał(a):

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/blit.c | 105 +++---
-
 1 file changed, 56 insertions(+), 49 deletions(-)

diff --git a/src/mesa/main/blit.c b/src/mesa/main/blit.c
index 207ce7d501..455a9a647f 100644
--- a/src/mesa/main/blit.c
+++ b/src/mesa/main/blit.c
@@ -233,6 +233,60 @@ validate_stencil_buffer(struct gl_context *ctx, struct
gl_framebuffer *readFb,
}
 }

+
+static void
+validate_depth_buffer(struct gl_context *ctx, struct gl_framebuffer
*readFb,
+  struct gl_framebuffer *drawFb, GLbitfield *mask,
+  bool no_error, const char *func)
+{
+   struct gl_renderbuffer *readRb =
+  readFb->Attachment[BUFFER_DEPTH].Renderbuffer;
+   struct gl_renderbuffer *drawRb =
+  drawFb->Attachment[BUFFER_DEPTH].Renderbuffer;
+
+   /* From the EXT_framebuffer_object spec:
+*
+* "If a buffer is specified in  and does not exist in both
+* the read and draw framebuffers, the corresponding bit is silently
+* ignored."
+*/
+   if (readRb == NULL || drawRb == NULL) {
+  *mask &= ~GL_DEPTH_BUFFER_BIT;
+   } else if (!no_error) {
+  int read_s_bit, draw_s_bit;
+
+  if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(source and destination depth buffer cannot be the
"
+ "same)", func);
+ return;
+  }
+
+  if ((_mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS) !=
+   _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS)) ||
+  (_mesa_get_format_datatype(readRb->Format) !=
+   _mesa_get_format_datatype(drawRb->Format))) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(depth attachment format mismatch)", func);
+ return;
+  }
+
+  read_s_bit = _mesa_get_format_bits(readRb->Format, GL_STENCIL_BITS);
+  draw_s_bit = _mesa_get_format_bits(drawRb->Format, GL_STENCIL_BITS);
+
+  /* If both buffers also have stencil data, the stencil formats must
+   * match as well.  If one doesn't have stencil, it's not blitted, so
+   * we should ignore the stencil format check.
+   */
+  if (read_s_bit > 0 && draw_s_bit > 0 && read_s_bit != draw_s_bit) {
+ _mesa_error(ctx, GL_INVALID_OPERATION,
+ "%s(depth attachment stencil bits mismatch)", func);
+ return;
+  }
+   }
+}
+
+
 static void
 blit_framebuffer(struct gl_context *ctx,
  struct gl_framebuffer *readFb, struct gl_framebuffer
*drawFb,
@@ -376,55 +430,8 @@ blit_framebuffer(struct gl_context *ctx,
if (mask & GL_STENCIL_BUFFER_BIT)
   validate_stencil_buffer(ctx, readFb, drawFb, &mask, false, func);

-   if (mask & GL_DEPTH_BUFFER_BIT) {
-  struct gl_renderbuffer *readRb =
- readFb->Attachment[BUFFER_DEPTH].Renderbuffer;
-  struct gl_renderbuffer *drawRb =
- drawFb->Attachment[BUFFER_DEPTH].Renderbuffer;
-
-  /* From the EXT_framebuffer_object spec:
-   *
-   * "If a buffer is specified in  and does not exist in both
-   * the read and draw framebuffers, the corresponding bit is
silently
-   * ignored."
-   */
-  if ((readRb == NULL) || (drawRb == NULL)) {
- mask &= ~GL_DEPTH_BUFFER_BIT;
-  }
-  else {
- int read_s_bit, draw_s_bit;
-
- if (_mesa_is_gles3(ctx) && (drawRb == readRb)) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(source and destination depth "
-"buffer cannot be the same)", func);
-return;
- }
-
- if ((_mesa_get_format_bits(readRb->Format, GL_DEPTH_BITS) !=
-  _mesa_get_format_bits(drawRb->Format, GL_DEPTH_BITS)) ||
- (_mesa_get_format_datatype(readRb->Format) !=
-  _mesa_get_format_datatype(drawRb->Format))) {
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(depth attachment format mismatch)", func);
-return;
- }
-
- read_s_bit = _mesa_get_format_bits(readRb->Format,
GL_STENCIL_BITS);
- draw_s_bit = _mesa_get_format_bits(drawRb->Format,
GL_STENCIL_BITS);
-
- /* If both buffers also have stencil data, the stencil formats
must
-  * match as well.  If one doesn't have stencil, it's not blitted,
so
-  * we should ignore the stencil format check.
-  */
- if (read_s_bit > 0 && draw_s_bit > 0 && read_s_bit != draw_s_bit)
{
-_mesa_error(ctx, GL_INVALID_OPERATION,
-"%s(depth attachment stencil bits mismatch)",
func);
-return;
- }
-  }
-   }
-
+   if (mask & GL_DEPTH_BUFFER_BIT)
+  validate_depth_buffer(ctx, readFb, drawFb, &mask, false, func);

if (_mesa_is_gles3(ctx)) {
   /* Pa

[Mesa-dev] [PATCH mesa] tree-wide: remove trailing backslash

[Eric Engestrom]

Simple search for a backslash followed by two newlines.
If one of the newlines were to be removed, this would cause issues, so
let's just remove these trailing backslashes.

Signed-off-by: Eric Engestrom 
---

I can split the patch by module if you want, but this seems trivial
enough?
A simple ack from a couple people would be enough IMO.
---
 src/amd/common/amd_kernel_code_t.h  | 2 +-
 src/compiler/nir/nir_builder.h  | 2 +-
 src/gallium/auxiliary/draw/draw_gs_tmp.h| 2 +-
 src/gallium/auxiliary/draw/draw_prim_assembler_tmp.h| 2 +-
 src/gallium/auxiliary/draw/draw_so_emit_tmp.h   | 2 +-
 src/gallium/drivers/r600/sb/sb_bc.h | 2 +-
 src/gallium/drivers/svga/svga_winsys.h  | 2 +-
 src/gallium/drivers/swr/rasterizer/memory/ClearTile.cpp | 4 ++--
 src/intel/compiler/brw_inst.h   | 2 +-
 src/mesa/drivers/dri/r200/r200_vertprog.c   | 2 +-
 src/mesa/drivers/x11/xmesaP.h   | 2 +-
 src/mesa/math/m_debug_util.h| 2 +-
 src/mesa/sparc/sparc_matrix.h   | 2 +-
 src/mesa/x86/mmx_blend.S| 2 +-
 src/mesa/x86/read_rgba_span_x86.S   | 2 +-
 15 files changed, 16 insertions(+), 16 deletions(-)

diff --git a/src/amd/common/amd_kernel_code_t.h 
b/src/amd/common/amd_kernel_code_t.h
index d0d7809da1..f8e9508518 100644
--- a/src/amd/common/amd_kernel_code_t.h
+++ b/src/amd/common/amd_kernel_code_t.h
@@ -36,7 +36,7 @@
 
 // Gets bits for specified mask from specified src packed instance.
 #define AMD_HSA_BITS_GET(src, mask)
\
-  ((src & mask) >> mask ## _SHIFT) 
\
+  ((src & mask) >> mask ## _SHIFT)
 
 /* Every amd_*_code_t has the following properties, which are composed of
  * a number of bit fields. Every bit field has a mask (AMD_CODE_PROPERTY_*),
diff --git a/src/compiler/nir/nir_builder.h b/src/compiler/nir/nir_builder.h
index 7dbf8efbb3..7c65886356 100644
--- a/src/compiler/nir/nir_builder.h
+++ b/src/compiler/nir/nir_builder.h
@@ -621,7 +621,7 @@ nir_load_system_value(nir_builder *build, nir_intrinsic_op 
op, int index)
nir_load_##name(nir_builder *build)   \
{ \
   return nir_load_system_value(build, nir_intrinsic_load_##name, 0); \
-   } \
+   }
 
 #include "nir_intrinsics.h"
 
diff --git a/src/gallium/auxiliary/draw/draw_gs_tmp.h 
b/src/gallium/auxiliary/draw/draw_gs_tmp.h
index b10bbc413d..bf276d3822 100644
--- a/src/gallium/auxiliary/draw/draw_gs_tmp.h
+++ b/src/gallium/auxiliary/draw/draw_gs_tmp.h
@@ -22,7 +22,7 @@
   default:\
  break;   \
   }   \
-   } while (0)\
+   } while (0)
 
 #define POINT(i0) gs_point(gs,i0)
 #define LINE(flags,i0,i1) gs_line(gs,i0,i1)
diff --git a/src/gallium/auxiliary/draw/draw_prim_assembler_tmp.h 
b/src/gallium/auxiliary/draw/draw_prim_assembler_tmp.h
index bff6d556ed..145a8ca74e 100644
--- a/src/gallium/auxiliary/draw/draw_prim_assembler_tmp.h
+++ b/src/gallium/auxiliary/draw/draw_prim_assembler_tmp.h
@@ -19,7 +19,7 @@
   return;   \
default: \
   break;\
-   }\
+   }
 
 
 #define POINT(i0) prim_point(asmblr, i0)
diff --git a/src/gallium/auxiliary/draw/draw_so_emit_tmp.h 
b/src/gallium/auxiliary/draw/draw_so_emit_tmp.h
index 282a52d1c0..c3a4695c1f 100644
--- a/src/gallium/auxiliary/draw/draw_so_emit_tmp.h
+++ b/src/gallium/auxiliary/draw/draw_so_emit_tmp.h
@@ -22,7 +22,7 @@
   default:\
  break;   \
   }   \
-   } while (0)\
+   } while (0)
 
 #define POINT(i0)so_point(so,i0)
 #define LINE(flags,i0,i1)so_line(so,i0,i1)
diff --git a/src/gallium/drivers/r600/sb/sb_bc.h 
b/src/gallium/drivers/r600/sb/sb_bc.h
index 2c662acf91..fed041cf50 100644
--- a/src/gallium/drivers/r600/sb/sb_bc.h
+++ b/src/gallium/drivers/r600/sb/sb_bc.h
@@ -787,7 +787,7 @@ public: \
} \
unsigned get_##name() const { \
return 
(value>>(first_bit))&((1ull<<((last_bit)-(first_bit)+1))-1); 

Re: [Mesa-dev] [PATCH] glsl: fix bounds check in blob_overwrite_bytes

[Iago Toral]

Reviewed-by: Iago Toral Quiroga 

On Wed, 2017-05-31 at 19:14 -0500, Rob Herring wrote:
> clang gives a warning in blob_overwrite_bytes because offset type is
> size_t which is unsigned:
> 
> src/compiler/glsl/blob.c:110:15: warning: comparison of unsigned
> expression < 0 is always false [-Wtautological-compare]
>    if (offset < 0 || blob->size - offset < to_write)
>    ~~ ^ ~
> 
> Remove the less than 0 check to fix this.
> 
> Additionally, if offset is greater than blob->size, the 2nd check
> would
> be false due to unsigned math. Rewrite the check to avoid
> subtraction.
> 
> Signed-off-by: Rob Herring 
> ---
>  src/compiler/glsl/blob.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/src/compiler/glsl/blob.c b/src/compiler/glsl/blob.c
> index 769ebf1a0232..db3625206508 100644
> --- a/src/compiler/glsl/blob.c
> +++ b/src/compiler/glsl/blob.c
> @@ -107,7 +107,7 @@ blob_overwrite_bytes(struct blob *blob,
>   size_t to_write)
>  {
> /* Detect an attempt to overwrite data out of bounds. */
> -   if (offset < 0 || blob->size - offset < to_write)
> +   if (blob->size < offset + to_write)
>    return false;
>  
> memcpy(blob->data + offset, bytes, to_write);
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH mesa] configure.ac: simplify --enable-libunwind=auto check

[Eric Engestrom]

Signed-off-by: Eric Engestrom 
---
 configure.ac | 8 ++--
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/configure.ac b/configure.ac
index b57be07e24..5caf316089 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1066,16 +1066,12 @@ AC_SUBST([LLVM_INCLUDEDIR])
 dnl
 dnl libunwind
 dnl
+PKG_CHECK_EXISTS(libunwind, [HAVE_LIBUNWIND=yes], [HAVE_LIBUNWIND=no])
 AC_ARG_ENABLE([libunwind],
 [AS_HELP_STRING([--enable-libunwind],
 [Use libunwind for backtracing (default: auto)])],
 [LIBUNWIND="$enableval"],
-[LIBUNWIND="auto"])
-
-PKG_CHECK_EXISTS(libunwind, [HAVE_LIBUNWIND=yes], [HAVE_LIBUNWIND=no])
-if test "x$LIBUNWIND" = "xauto"; then
-LIBUNWIND="$HAVE_LIBUNWIND"
-fi
+[LIBUNWIND="$HAVE_LIBUNWIND"])
 
 if test "x$LIBUNWIND" = "xyes"; then
 PKG_CHECK_MODULES(LIBUNWIND, libunwind)
-- 
Cheers,
  Eric

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] util/rand_xor: add missing include statements

[Nicolas Dechesne]

On Thu, Jun 1, 2017 at 3:27 PM, Eric Engestrom
 wrote:
> On Thursday, 2017-06-01 12:13:18 +0200, Nicolas Dechesne wrote:
>> Fixes for:
>>
>> src/util/rand_xor.c:60:13: error: implicit declaration of function 'open' 
>> [-Werror=implicit-function-declaration]
>> int fd = open("/dev/urandom", O_RDONLY);
>>  ^~~~
>> src/util/rand_xor.c:60:34: error: 'O_RDONLY' undeclared (first use in this 
>> function)
>> int fd = open("/dev/urandom", O_RDONLY);
>>   ^~~~
>>
>> Signed-off-by: Nicolas Dechesne 
>
> Thanks, r-b and pushed, and welcome to Mesa :)

thanks , starting small ;)

I would like this patch on the 17.1 branch as well. Is that possible?
I actually found the issue on 17.1.1 release.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] Status of Clover (AMDGPU)

[Luke A. Guest]

> I'd suggest that you check out the recently-open-sourced OpenCL
> conformance test suite [0] or my slightly-more-linux-build-friendly
> fork [1] and give it a run on top of mesa/clover. It'll give you a
> good list of things to work on.
Thanks, I took a quick look. Yeah, shame they didn't put in a proper
build system or provide a doc stating what the tests are or a main app
that runs them all in a correct order and outputs a report.

>
> I ran the integer_ops sub-tests last night, and besides the missing CL
> 1.2 popcount() function, that entire sub-suite succeeds. Feel free to
> attempt to implement that built-in function in the libclc[2] library
> to get your feet wet.
>
> There are also issues with buffer copies (using host-pointers,
> copySubRect), and there are a subset of the floating-point math
> built-ins which are either missing or fail some test scenarios. Some
> atomic operations seem to be incorrect, and that's about as far as
> I've managed to get so far in looking at the results. I'm sure plenty
> of other things need work/polish. Note that many of these things are
> gpu-independent and can be implemented either in the clover
> state-tracker, or in the generic (non-GPU-specific) portion of libclc.
>
> I've been looking into implementing the
> clCreateCommandQueueWithProperties API function as well, since there
> seem to be a bunch of CL programs that just blindly assume the API
> function exists for all platforms when using the ocl-icd loader, and
> segfault as a result.
>
> One big missing piece is image support. It's an optional feature of
> OpenCL, but there are a good number of programs used by a certain
> benchmarking site that depend on it.
>
Phoronix?

Yeah, it's on the todo list to see if I can get anywhere with it. My
immediate aim is to get Blender cycles with opensubdiv working, so
that's gonna need a lot of work with images.

Thanks,
Luke.

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 01/21] mesa: add vertex_array_vertex_buffers_err() helper

[Erik Faye-Lund]

On Jun 1, 2017 15:06, "Samuel Pitoiset"  wrote:

Signed-off-by: Samuel Pitoiset 
---
 src/mesa/main/varray.c | 108 --
---
 1 file changed, 61 insertions(+), 47 deletions(-)

diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index 47528ba2a7..c2c771c173 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -2117,29 +2117,14 @@ _mesa_VertexArrayVertexBuffer(GLuint vaobj, GLuint
bindingIndex, GLuint buffer,
 }


-static void
+static ALWAYS_INLINE void


This seems unrelated to the rest of the patch... It probably needs a
motivation also.

 vertex_array_vertex_buffers(struct gl_context *ctx,
 struct gl_vertex_array_object *vao,
 GLuint first, GLsizei count, const GLuint
*buffers,
 const GLintptr *offsets, const GLsizei
*strides,
-const char *func)
+bool no_error, const char *func)
 {
-   GLuint i;
-
-   ASSERT_OUTSIDE_BEGIN_END(ctx);
-
-   /* The ARB_multi_bind spec says:
-*
-*"An INVALID_OPERATION error is generated if  + 
-* is greater than the value of MAX_VERTEX_ATTRIB_BINDINGS."
-*/
-   if (first + count > ctx->Const.MaxVertexAttribBindings) {
-  _mesa_error(ctx, GL_INVALID_OPERATION,
-  "%s(first=%u + count=%d > the value of "
-  "GL_MAX_VERTEX_ATTRIB_BINDINGS=%u)",
-  func, first, count, ctx->Const.MaxVertexAttribBindings);
-  return;
-   }
+   GLint i;

if (!buffers) {
   /**
@@ -2184,31 +2169,33 @@ vertex_array_vertex_buffers(struct gl_context *ctx,
for (i = 0; i < count; i++) {
   struct gl_buffer_object *vbo;

-  /* The ARB_multi_bind spec says:
-   *
-   *"An INVALID_VALUE error is generated if any value in
-   *  or  is negative (per binding)."
-   */
-  if (offsets[i] < 0) {
- _mesa_error(ctx, GL_INVALID_VALUE,
- "%s(offsets[%u]=%" PRId64 " < 0)",
- func, i, (int64_t) offsets[i]);
- continue;
-  }
+  if (!no_error) {
+ /* The ARB_multi_bind spec says:
+  *
+  *"An INVALID_VALUE error is generated if any value in
+  *  or  is negative (per binding)."
+  */
+ if (offsets[i] < 0) {
+_mesa_error(ctx, GL_INVALID_VALUE,
+"%s(offsets[%u]=%" PRId64 " < 0)",
+func, i, (int64_t) offsets[i]);
+continue;
+ }

-  if (strides[i] < 0) {
- _mesa_error(ctx, GL_INVALID_VALUE,
- "%s(strides[%u]=%d < 0)",
- func, i, strides[i]);
- continue;
-  }
+ if (strides[i] < 0) {
+_mesa_error(ctx, GL_INVALID_VALUE,
+"%s(strides[%u]=%d < 0)",
+func, i, strides[i]);
+continue;
+ }

-  if (ctx->API == API_OPENGL_CORE && ctx->Version >= 44 &&
-  strides[i] > ctx->Const.MaxVertexAttribStride) {
- _mesa_error(ctx, GL_INVALID_VALUE,
- "%s(strides[%u]=%d > "
- "GL_MAX_VERTEX_ATTRIB_STRIDE)", func, i, strides[i]);
- continue;
+ if (ctx->API == API_OPENGL_CORE && ctx->Version >= 44 &&
+ strides[i] > ctx->Const.MaxVertexAttribStride) {
+_mesa_error(ctx, GL_INVALID_VALUE,
+"%s(strides[%u]=%d > "
+"GL_MAX_VERTEX_ATTRIB_STRIDE)", func, i,
strides[i]);
+continue;
+ }
   }

   if (buffers[i]) {
@@ -2234,6 +2221,33 @@ vertex_array_vertex_buffers(struct gl_context *ctx,
 }


+static void
+vertex_array_vertex_buffers_err(struct gl_context *ctx,
+struct gl_vertex_array_object *vao,
+GLuint first, GLsizei count,
+const GLuint *buffers, const GLintptr
*offsets,
+const GLsizei *strides, const char *func)
+{
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   /* The ARB_multi_bind spec says:
+*
+*"An INVALID_OPERATION error is generated if  + 
+* is greater than the value of MAX_VERTEX_ATTRIB_BINDINGS."
+*/
+   if (first + count > ctx->Const.MaxVertexAttribBindings) {
+  _mesa_error(ctx, GL_INVALID_OPERATION,
+  "%s(first=%u + count=%d > the value of "
+  "GL_MAX_VERTEX_ATTRIB_BINDINGS=%u)",
+  func, first, count, ctx->Const.MaxVertexAttribBindings);
+  return;
+   }
+
+   vertex_array_vertex_buffers(ctx, vao, first, count, buffers, offsets,
+   strides, false, func);
+}
+
+
 void GLAPIENTRY
 _mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers,
 const GLintptr *offsets, const GLsizei *strides)
@@ -2252,9 +2266,9 @@ _mesa_BindVertexBuff

Re: [Mesa-dev] [PATCH v2] egl/android: support for EGL_KHR_partial_update

[Emil Velikov]

Hi Harish

On 1 June 2017 at 11:27, Harish Krupo  wrote:
> This patch adds support for the EGL_KHR_partial_update extension for
> android platform. It passes 36/37 tests in dEQP for EGL_KHR_partial_update.
> 1 test not supported.
>
> v2: add fallback for eglSetDamageRegionKHR (Tapani)
>
AFAICT the native_window_set_surface_damage API was introduced in 2015
[1]. Is it available all the way back to Android L - earliest
supported version in Mesa?
Alternatively can you please reintroduce the ANDROID_VERSION define
[2] and guard the new code accordingly.

As a follow-up you can use
native_window_set_surface_damage()/droid_set_damage_region to provide
a proper .swap_buffers_with_damage implementation and enable the
EXT/KHR extensions.

Thanks
Emil

[1] 
https://android.googlesource.com/platform/system/core/+/04274a29ff534000ca38cdb627daab98cfa2878d%5E1..04274a29ff534000ca38cdb627daab98cfa2878d/
[2] 
https://cgit.freedesktop.org/mesa/mesa/commit/?id=0e1ff22d55816c9a3710257c2e705a98ad3282bc
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH mesa] configure.ac: simplify --enable-libunwind=auto check

[Emil Velikov]

On 1 June 2017 at 15:10, Eric Engestrom  wrote:
> Signed-off-by: Eric Engestrom 
> ---
>  configure.ac | 8 ++--
>  1 file changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/configure.ac b/configure.ac
> index b57be07e24..5caf316089 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -1066,16 +1066,12 @@ AC_SUBST([LLVM_INCLUDEDIR])
>  dnl
>  dnl libunwind
>  dnl
> +PKG_CHECK_EXISTS(libunwind, [HAVE_LIBUNWIND=yes], [HAVE_LIBUNWIND=no])
>  AC_ARG_ENABLE([libunwind],
>  [AS_HELP_STRING([--enable-libunwind],
>  [Use libunwind for backtracing (default: auto)])],
>  [LIBUNWIND="$enableval"],
> -[LIBUNWIND="auto"])
> -
> -PKG_CHECK_EXISTS(libunwind, [HAVE_LIBUNWIND=yes], [HAVE_LIBUNWIND=no])
> -if test "x$LIBUNWIND" = "xauto"; then
> -LIBUNWIND="$HAVE_LIBUNWIND"
> -fi
> +[LIBUNWIND="$HAVE_LIBUNWIND"])
>
Hmm we could have kept PKG_CHECK_EXISTS only as needed by moving it down a line.

Either way - the patch as-is or implementing the above idea is
Reviewed-by: Emil Velikov 

-Emil
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH mesa 2/2] egl/display: make platform detection thread-safe

[Grazvydas Ignotas]

On Thu, Jun 1, 2017 at 4:23 PM, Emil Velikov  wrote:
> Hi guys,
>
> On 1 June 2017 at 12:56, Grazvydas Ignotas  wrote:
>> On Thu, Jun 1, 2017 at 2:15 PM, Eric Engestrom
>>  wrote:
>>> If the detections methods ever become able to return different results
>>> for different threads, the if chain might make threads go back and forth
>>> between invalid and valid platforms.
>>> Solve this by doing the detection in a local var and only overwriting
>>> the global one at the end, if no other thread has updated it since.
>>>
>>> This means the platform detected in the thread might not be the platform
>>> returned by the function, but this is a different issue that will need
>>> to be discussed when this becomes possible.
>>>
>>> Reported-by: Grazvydas Ignotas 
>>
>> Not really, see https://bugs.freedesktop.org/show_bug.cgi?id=101252
>>
>>> Signed-off-by: Eric Engestrom 
>>> ---
>>>
>>> This is unnecessary in my opinion, but doesn't hurt :)
>>
>> It is necessary, without it things will work most of the time but
>> there will be rare failures.
>> Imagine there are 2 threads that both call _eglGetNativePlatform()
>> simultaneously:
>> - thread 1 completes the first "if (native_platform ==
>> _EGL_INVALID_PLATFORM)" check and is preemted to do something else
>> - thread 2 executes the whole function, does "native_platform =
>> _EGL_NATIVE_PLATFORM" and just before returning it's preemted
>> - thread 1 wakes up and calls _eglGetNativePlatformFromEnv() which
>> returns _EGL_INVALID_PLATFORM because no env vars are set, updates
>> native_platform and then gets preemted again
>> - thread 2 wakes up and returns wrong _EGL_INVALID_PLATFORM
>>
> Afaict this/similar fix is necessary, yet let me put an alternative
> idea - add locking via _eglGlobal.Mutex.
> May be an bit of overkill, but it's what we consistently use
> throughout the code base.

It looks like that mutex is meant to protect _eglGlobal and not some
random variables, so I'd prefer the atomic version.

> Reverting the patch (as suggested by Grazvydas) does not fully address
> the problem, but only makes it less likely to hit.

Yeah I meant reverting instead of taking 1/2 and then applying this
fix on top. OTOH I'm also ok with this series as-is.

Gražvydas
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] util/rand_xor: add missing include statements

[Emil Velikov]

On 1 June 2017 at 15:15, Nicolas Dechesne  wrote:
> On Thu, Jun 1, 2017 at 3:27 PM, Eric Engestrom
>  wrote:
>> On Thursday, 2017-06-01 12:13:18 +0200, Nicolas Dechesne wrote:
>>> Fixes for:
>>>
>>> src/util/rand_xor.c:60:13: error: implicit declaration of function 'open' 
>>> [-Werror=implicit-function-declaration]
>>> int fd = open("/dev/urandom", O_RDONLY);
>>>  ^~~~
>>> src/util/rand_xor.c:60:34: error: 'O_RDONLY' undeclared (first use in this 
>>> function)
>>> int fd = open("/dev/urandom", O_RDONLY);
>>>   ^~~~
>>>
>>> Signed-off-by: Nicolas Dechesne 
>>
>> Thanks, r-b and pushed, and welcome to Mesa :)
>
> thanks , starting small ;)
>
> I would like this patch on the 17.1 branch as well. Is that possible?
> I actually found the issue on 17.1.1 release.

Should be handled now. See the documentation for future stable nominations [1].

-Emil
[1] https://www.mesa3d.org/submittingpatches.html#nominations
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] util/rand_xor: add missing include statements

[Eric Engestrom]

On Thursday, 2017-06-01 16:15:41 +0200, Nicolas Dechesne wrote:
> On Thu, Jun 1, 2017 at 3:27 PM, Eric Engestrom
>  wrote:
> > On Thursday, 2017-06-01 12:13:18 +0200, Nicolas Dechesne wrote:
> >> Fixes for:
> >>
> >> src/util/rand_xor.c:60:13: error: implicit declaration of function 'open' 
> >> [-Werror=implicit-function-declaration]
> >> int fd = open("/dev/urandom", O_RDONLY);
> >>  ^~~~
> >> src/util/rand_xor.c:60:34: error: 'O_RDONLY' undeclared (first use in this 
> >> function)
> >> int fd = open("/dev/urandom", O_RDONLY);
> >>   ^~~~
> >>
> >> Signed-off-by: Nicolas Dechesne 
> >
> > Thanks, r-b and pushed, and welcome to Mesa :)
> 
> thanks , starting small ;)
> 
> I would like this patch on the 17.1 branch as well. Is that possible?
> I actually found the issue on 17.1.1 release.

Actually, since this is a fix, I made a mistake: I should have marked
the commit that broke it with the `Fixes:` tag, before pushing it.

In this case:
Fixes: dd00a3c923ba94986efb ("util/rand_xor: add function to seed rand")

This would have automatically nominated the patch for any branch that
includes the offending commit (and is still maintained).

That said, even after the patch has been pushed, it is still possible to
nominate it for a stable branch; see:
https://www.mesa3d.org/submittingpatches.html#nominations

I've Bcc'ed this email to mesa-stable@, so this should be taken care of
now. (Bcc to avoid replies cluttering that ML.)

For reference, the commit on master is:
adadadc151fa8232ecd7 ("util/rand_xor: add missing include statements")

Cheers,
  Eric
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 08/11] intel/isl: Add a new layout for HiZ and stencil on Sandy Bridge

[Jason Ekstrand]

On Thu, Jun 1, 2017 at 2:40 AM, Pohjolainen, Topi <
topi.pohjolai...@gmail.com> wrote:

> On Tue, May 30, 2017 at 05:55:17PM -0700, Jason Ekstrand wrote:
> > ---
> >  src/intel/isl/isl.c | 162 ++
> --
> >  src/intel/isl/isl.h |  40 +
> >  2 files changed, 197 insertions(+), 5 deletions(-)
> >
> > diff --git a/src/intel/isl/isl.c b/src/intel/isl/isl.c
> > index 77b8a40..e4d7901 100644
> > --- a/src/intel/isl/isl.c
> > +++ b/src/intel/isl/isl.c
> > @@ -479,6 +479,12 @@ isl_choose_array_pitch_span(const struct
> isl_device *dev,
> > * compact QPitch possible in order to conserve memory.
> > */
> >return ISL_ARRAY_PITCH_SPAN_COMPACT;
> > +
> > +   case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
> > +  /* Each array image in the gen6 stencil of HiZ surface is compact
> in the
> > +   * sense that every LOD is a compact array of the same size as
> LOD0.
> > +   */
> > +  return ISL_ARRAY_PITCH_SPAN_COMPACT;
> > }
> >
> > unreachable("bad isl_dim_layout");
> > @@ -510,10 +516,15 @@ isl_choose_image_alignment_el(const struct
> isl_device *dev,
> >return;
> > } else if (info->format == ISL_FORMAT_HIZ) {
> >assert(ISL_DEV_GEN(dev) >= 6);
> > -  /* HiZ surfaces are always aligned to 16x8 pixels in the primary
> surface
> > -   * which works out to 2x2 HiZ elments.
> > -   */
> > -  *image_align_el = isl_extent3d(2, 2, 1);
> > +  if (ISL_DEV_GEN(dev) == 6) {
> > + /* HiZ surfaces on Sandy Bridge are packed tightly. */
> > + *image_align_el = isl_extent3d(1, 1, 1);
> > +  } else {
> > + /* On gen7+, HiZ surfaces are always aligned to 16x8 pixels in
> the
> > +  * primary surface which works out to 2x2 HiZ elments.
> > +  */
> > + *image_align_el = isl_extent3d(2, 2, 1);
> > +  }
> >return;
> > }
> >
> > @@ -540,6 +551,11 @@ isl_surf_choose_dim_layout(const struct isl_device
> *dev,
> > enum isl_surf_dim logical_dim,
> > enum isl_tiling tiling)
> >  {
> > +   /* Sandy bridge needs a special layout for HiZ and stencil. */
> > +   if (ISL_DEV_GEN(dev) == 6 &&
> > +   (tiling == ISL_TILING_W || tiling == ISL_TILING_HIZ))
> > +  return ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ;
> > +
> > if (ISL_DEV_GEN(dev) >= 9) {
> >switch (logical_dim) {
> >case ISL_SURF_DIM_1D:
> > @@ -608,6 +624,7 @@ isl_calc_phys_level0_extent_sa(const struct
> isl_device *dev,
> >
> >case ISL_DIM_LAYOUT_GEN9_1D:
> >case ISL_DIM_LAYOUT_GEN4_2D:
> > +  case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
> >   *phys_level0_sa = (struct isl_extent4d) {
> >  .w = isl_align_npot(info->width, fmtl->bw),
> >  .h = fmtl->bh,
> > @@ -619,7 +636,8 @@ isl_calc_phys_level0_extent_sa(const struct
> isl_device *dev,
> >break;
> >
> > case ISL_SURF_DIM_2D:
> > -  assert(dim_layout == ISL_DIM_LAYOUT_GEN4_2D);
> > +  assert(dim_layout == ISL_DIM_LAYOUT_GEN4_2D ||
> > + dim_layout == ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ);
> >
> >if (tiling == ISL_TILING_Ys && info->samples > 1)
> >   isl_finishme("%s:%s: multisample TileYs layout", __FILE__,
> __func__);
> > @@ -684,6 +702,7 @@ isl_calc_phys_level0_extent_sa(const struct
> isl_device *dev,
> >
> >switch (dim_layout) {
> >case ISL_DIM_LAYOUT_GEN9_1D:
> > +  case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
> >   unreachable("bad isl_dim_layout");
> >
> >case ISL_DIM_LAYOUT_GEN4_2D:
> > @@ -969,6 +988,67 @@ isl_calc_phys_total_extent_el_gen4_3d(
> >
> >  /**
> >   * A variant of isl_calc_phys_slice0_extent_sa() specific to
> > + * ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ.
> > + */
> > +static void
> > +isl_calc_phys_total_extent_el_gen6_stencil_hiz(
> > +  const struct isl_device *dev,
> > +  const struct isl_surf_init_info *restrict info,
> > +  const struct isl_tile_info *tile_info,
> > +  const struct isl_extent3d *image_align_sa,
> > +  const struct isl_extent4d *phys_level0_sa,
> > +  uint32_t *array_pitch_el_rows,
> > +  struct isl_extent2d *phys_total_el)
> > +{
> > +   const struct isl_format_layout *fmtl = isl_format_get_layout(info->
> format);
> > +
> > +   const struct isl_extent2d tile_extent_sa = {
> > +  .w = tile_info->logical_extent_el.w * fmtl->bw,
> > +  .h = tile_info->logical_extent_el.h * fmtl->bh,
> > +   };
> > +   /* Tile size is a multiple of image alignment */
> > +   assert(tile_extent_sa.w % image_align_sa->w == 0);
> > +   assert(tile_extent_sa.h % image_align_sa->h == 0);
> > +
> > +   const uint32_t W0 = phys_level0_sa->w;
> > +   const uint32_t H0 = phys_level0_sa->h;
> > +
> > +   /* Each image has the same height as LOD0 because the hardware thinks
> > +* everything is LOD0
> > +*/
> > +   const uint32_t H = isl_align(H0, image_align_sa->h) *
> phys_level0_sa->a;
> > +
> > 

[Mesa-dev] [PATCH] automake: Link all libGL.so variants with -Bsymbolic.

[Jose Fonseca]

We were linking src/glx with -Bsymbolic, but not the classic/gallium X11
libGL.so.

But it's always a good idea to build all libGL.so and all DRI drivers
with -Bsymbolic, otherwise they might resolve symbols from the 3rd party
application executable or shared libraries, which is _never_ what we
want.

In particular, this can happen when intercepting OpenGL calls with
apitrace, before
https://github.com/apitrace/apitrace/commit/63194b2573176ef34efce1a5c8b08e624b8dddf5

Cc: mesa-sta...@lists.freedesktop.org
---
 src/gallium/targets/libgl-xlib/Makefile.am | 1 +
 src/mesa/drivers/x11/Makefile.am   | 1 +
 2 files changed, 2 insertions(+)

diff --git a/src/gallium/targets/libgl-xlib/Makefile.am 
b/src/gallium/targets/libgl-xlib/Makefile.am
index d3a781848e..a29199fe78 100644
--- a/src/gallium/targets/libgl-xlib/Makefile.am
+++ b/src/gallium/targets/libgl-xlib/Makefile.am
@@ -52,6 +52,7 @@ lib@GL_LIB@_la_SOURCES = xlib.c
 lib@GL_LIB@_la_LDFLAGS = \
-no-undefined \
-version-number $(GL_MAJOR):$(GL_MINOR):$(GL_TINY) \
+   $(BSYMBOLIC) \
$(GC_SECTIONS) \
$(LD_NO_UNDEFINED)
 
diff --git a/src/mesa/drivers/x11/Makefile.am b/src/mesa/drivers/x11/Makefile.am
index 8c6b67d6a8..6e123ba1b3 100644
--- a/src/mesa/drivers/x11/Makefile.am
+++ b/src/mesa/drivers/x11/Makefile.am
@@ -74,6 +74,7 @@ lib@GL_LIB@_la_LIBADD = \
 lib@GL_LIB@_la_LDFLAGS = \
-no-undefined \
-version-number $(GL_MAJOR):$(GL_MINOR):$(GL_PATCH) \
+   $(BSYMBOLIC) \
$(GC_SECTIONS) \
$(LD_NO_UNDEFINED)
 
-- 
2.11.0

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 11/11] intel/blorp: Handle gen6 stencil/HiZ offsets in the back-end

[Jason Ekstrand]

On Thu, Jun 1, 2017 at 3:17 AM, Pohjolainen, Topi <
topi.pohjolai...@gmail.com> wrote:

> On Tue, May 30, 2017 at 05:55:20PM -0700, Jason Ekstrand wrote:
> > ---
> >  src/intel/blorp/blorp_genX_exec.h| 32 -
> >  src/intel/isl/isl_emit_depth_stencil.c   |  1 +
> >  src/mesa/drivers/dri/i965/brw_blorp.c| 61
> 
> >  src/mesa/drivers/dri/i965/brw_wm_surface_state.c |  3 +-
> >  src/mesa/drivers/dri/i965/intel_mipmap_tree.c| 17 ---
> >  src/mesa/drivers/dri/i965/intel_mipmap_tree.h|  2 +-
> >  6 files changed, 42 insertions(+), 74 deletions(-)
> >
> > diff --git a/src/intel/blorp/blorp_genX_exec.h
> b/src/intel/blorp/blorp_genX_exec.h
> > index 8b9b8d2..59c1d36 100644
> > --- a/src/intel/blorp/blorp_genX_exec.h
> > +++ b/src/intel/blorp/blorp_genX_exec.h
> > @@ -1384,9 +1384,23 @@ blorp_emit_depth_stencil_config(struct
> blorp_batch *batch,
> >if (info.hiz_usage == ISL_AUX_USAGE_HIZ) {
> >   info.hiz_surf = ¶ms->depth.aux_surf;
> >
> > + struct blorp_address hiz_address = params->depth.aux_addr;
> > +#if GEN_GEN == 6
> > + /* Sandy bridge hardware does not technically support
> mipmapped HiZ.
> > +  * However, we have a special layout that allows us to make it
> work
> > +  * anyway by manually offsetting to the specified miplevel.
> > +  */
> > + assert(info.hiz_surf->dim_layout ==
> ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ);
> > + uint32_t offset_B;
> > + isl_surf_get_image_offset_B_tile_sa(info.hiz_surf,
> > + info.view->base_level, 0,
> 0,
> > + &offset_B, NULL, NULL);
> > + hiz_address.offset += offset_B;
> > +#endif
> > +
> >   info.hiz_address =
> >  blorp_emit_reloc(batch, dw + isl_dev->ds.hiz_offset / 4,
> > - params->depth.aux_addr, 0);
> > + hiz_address, 0);
> >
> >   info.depth_clear_value = params->depth.clear_color.u32[0];
> >}
> > @@ -1395,9 +1409,23 @@ blorp_emit_depth_stencil_config(struct
> blorp_batch *batch,
> > if (params->stencil.enabled) {
> >info.stencil_surf = ¶ms->stencil.surf;
> >
> > +  struct blorp_address stencil_address = params->stencil.addr;
> > +#if GEN_GEN == 6
> > +  /* Sandy bridge hardware does not technically support mipmapped
> stencil.
> > +   * However, we have a special layout that allows us to make it
> work
> > +   * anyway by manually offsetting to the specified miplevel.
> > +   */
> > +  assert(info.stencil_surf->dim_layout ==
> ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ);
> > +  uint32_t offset_B;
> > +  isl_surf_get_image_offset_B_tile_sa(info.stencil_surf,
> > +  info.view->base_level, 0, 0,
> > +  &offset_B, NULL, NULL);
> > +  stencil_address.offset += offset_B;
> > +#endif
> > +
> >info.stencil_address =
> >   blorp_emit_reloc(batch, dw + isl_dev->ds.stencil_offset / 4,
> > -  params->stencil.addr, 0);
> > +  stencil_address, 0);
> > }
> >
> > isl_emit_depth_stencil_hiz_s(isl_dev, dw, &info);
> > diff --git a/src/intel/isl/isl_emit_depth_stencil.c
> b/src/intel/isl/isl_emit_depth_stencil.c
> > index 41a01be..04bed07 100644
> > --- a/src/intel/isl/isl_emit_depth_stencil.c
> > +++ b/src/intel/isl/isl_emit_depth_stencil.c
> > @@ -158,6 +158,7 @@ isl_genX(emit_depth_stencil_hiz_s)(const struct
> isl_device *dev, void *batch,
> >hiz.SurfaceBaseAddress = info->hiz_address;
> >hiz.HierarchicalDepthBufferMOCS = info->mocs;
> >hiz.SurfacePitch = info->hiz_surf->row_pitch - 1;
> > +
>
> Leftover?
>

Yup.  Fixed.


> >  #if GEN_GEN >= 8
> >/* From the SKL PRM Vol2a:
> > *
> > diff --git a/src/mesa/drivers/dri/i965/brw_blorp.c
> b/src/mesa/drivers/dri/i965/brw_blorp.c
> > index 61c6bda..28be620 100644
> > --- a/src/mesa/drivers/dri/i965/brw_blorp.c
> > +++ b/src/mesa/drivers/dri/i965/brw_blorp.c
> > @@ -118,36 +118,6 @@ brw_blorp_init(struct brw_context *brw)
> >  }
> >
> >  static void
> > -apply_gen6_stencil_hiz_offset(struct isl_surf *surf,
> > -  struct intel_mipmap_tree *mt,
> > -  uint32_t lod,
> > -  uint32_t *offset)
> > -{
> > -   assert(mt->array_layout == GEN6_HIZ_STENCIL);
> > -
> > -   if (mt->format == MESA_FORMAT_S_UINT8) {
> > -  /* Note: we can't compute the stencil offset using
> > -   * intel_miptree_get_aligned_offset(), because the miptree
> > -   * claims that the region is untiled even though it's W tiled.
> > -   */
> > -  *offset = mt->level[lod].level_y * mt->pitch +
> > -mt->level[lod].level_x * 64;
> > -   } else {
> > -  *offset = intel_miptree_get_aligned_of

Re: [Mesa-dev] [PATCH] automake: Link all libGL.so variants with -Bsymbolic.

[Brian Paul]

On 06/01/2017 09:46 AM, Jose Fonseca wrote:

We were linking src/glx with -Bsymbolic, but not the classic/gallium X11
libGL.so.

But it's always a good idea to build all libGL.so and all DRI drivers
with -Bsymbolic, otherwise they might resolve symbols from the 3rd party
application executable or shared libraries, which is _never_ what we
want.

In particular, this can happen when intercepting OpenGL calls with
apitrace, before
https://github.com/apitrace/apitrace/commit/63194b2573176ef34efce1a5c8b08e624b8dddf5

Cc: mesa-sta...@lists.freedesktop.org
---
  src/gallium/targets/libgl-xlib/Makefile.am | 1 +
  src/mesa/drivers/x11/Makefile.am   | 1 +
  2 files changed, 2 insertions(+)

diff --git a/src/gallium/targets/libgl-xlib/Makefile.am 
b/src/gallium/targets/libgl-xlib/Makefile.am
index d3a781848e..a29199fe78 100644
--- a/src/gallium/targets/libgl-xlib/Makefile.am
+++ b/src/gallium/targets/libgl-xlib/Makefile.am
@@ -52,6 +52,7 @@ lib@GL_LIB@_la_SOURCES = xlib.c
  lib@GL_LIB@_la_LDFLAGS = \
-no-undefined \
-version-number $(GL_MAJOR):$(GL_MINOR):$(GL_TINY) \
+   $(BSYMBOLIC) \
$(GC_SECTIONS) \
$(LD_NO_UNDEFINED)

diff --git a/src/mesa/drivers/x11/Makefile.am b/src/mesa/drivers/x11/Makefile.am
index 8c6b67d6a8..6e123ba1b3 100644
--- a/src/mesa/drivers/x11/Makefile.am
+++ b/src/mesa/drivers/x11/Makefile.am
@@ -74,6 +74,7 @@ lib@GL_LIB@_la_LIBADD = \
  lib@GL_LIB@_la_LDFLAGS = \
-no-undefined \
-version-number $(GL_MAJOR):$(GL_MINOR):$(GL_PATCH) \
+   $(BSYMBOLIC) \
$(GC_SECTIONS) \
$(LD_NO_UNDEFINED)




LGTM.

Reviewed-by: Brian Paul 

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [Bug 98604] [VDPAU, DRI3] Fullscreen flash video fails when hardware acceleration is enabled.

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=98604

--- Comment #18 from joeri.exelm...@gmail.com ---
Created attachment 131644
  --> https://bugs.freedesktop.org/attachment.cgi?id=131644&action=edit
vdpauinfo

-- 
You are receiving this mail because:
You are on the CC list for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [Bug 98604] [VDPAU, DRI3] Fullscreen flash video fails when hardware acceleration is enabled.

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=98604

--- Comment #19 from joeri.exelm...@gmail.com ---
Sorry, ignore the 'vdpau' attachment I just uploaded, this belongs to another
bug (wrong browser tab!)

-- 
You are receiving this mail because:
You are on the CC list for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 5/5] r100: Silence numerous unused this or that warnings

[Marek Olšák]

For the series:

Reviewed-by: Marek Olšák 

About your r100-r200 coding style question, I don't have a specific
answer. It's up to you what you wanna do with it.

Marek


On Wed, May 31, 2017 at 2:40 AM, Ian Romanick  wrote:
> From: Ian Romanick 
>
> radeon_fbo.c: In function ‘radeon_map_renderbuffer_s8z24’:
> radeon_fbo.c:147:50: warning: unused parameter ‘ctx’ [-Wunused-parameter]
>  radeon_map_renderbuffer_s8z24(struct gl_context *ctx,
>   ^~~
> radeon_fbo.c: In function ‘radeon_map_renderbuffer_z16’:
> radeon_fbo.c:186:48: warning: unused parameter ‘ctx’ [-Wunused-parameter]
>  radeon_map_renderbuffer_z16(struct gl_context *ctx,
> ^~~
> radeon_fbo.c: In function ‘radeon_unmap_renderbuffer_s8z24’:
> radeon_fbo.c:344:52: warning: unused parameter ‘ctx’ [-Wunused-parameter]
>  radeon_unmap_renderbuffer_s8z24(struct gl_context *ctx,
> ^~~
> radeon_fbo.c: In function ‘radeon_unmap_renderbuffer_z16’:
> radeon_fbo.c:377:50: warning: unused parameter ‘ctx’ [-Wunused-parameter]
>  radeon_unmap_renderbuffer_z16(struct gl_context *ctx,
>   ^~~
> radeon_fbo.c: In function ‘radeon_nop_alloc_storage’:
> radeon_fbo.c:624:75: warning: unused parameter ‘rb’ [-Wunused-parameter]
>  radeon_nop_alloc_storage(struct gl_context * ctx, struct gl_renderbuffer *rb,
>^~
> radeon_fbo.c:625:12: warning: unused parameter ‘internalFormat’ 
> [-Wunused-parameter]
>  GLenum internalFormat, GLuint width, GLuint height)
> ^~
> radeon_fbo.c:625:35: warning: unused parameter ‘width’ [-Wunused-parameter]
>  GLenum internalFormat, GLuint width, GLuint height)
>^
> radeon_fbo.c:625:49: warning: unused parameter ‘height’ [-Wunused-parameter]
>  GLenum internalFormat, GLuint width, GLuint height)
>  ^~
> radeon_fbo.c: In function ‘radeon_bind_framebuffer’:
> radeon_fbo.c:696:74: warning: unused parameter ‘fbread’ [-Wunused-parameter]
> struct gl_framebuffer *fb, struct gl_framebuffer 
> *fbread)
>   
> ^~
> radeon_fbo.c: In function ‘radeon_validate_framebuffer’:
> radeon_fbo.c:832:19: warning: unused variable ‘radeon’ [-Wunused-variable]
>   radeonContextPtr radeon = RADEON_CONTEXT(ctx);
>^~
>
> Signed-off-by: Ian Romanick 
> ---
>  src/mesa/drivers/dri/radeon/radeon_fbo.c | 17 +
>  1 file changed, 9 insertions(+), 8 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c 
> b/src/mesa/drivers/dri/radeon/radeon_fbo.c
> index 89ea776..37c9c3f 100644
> --- a/src/mesa/drivers/dri/radeon/radeon_fbo.c
> +++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c
> @@ -144,8 +144,7 @@ static GLuint get_depth_z16(const struct 
> radeon_renderbuffer *rrb,
>  #endif
>
>  static void
> -radeon_map_renderbuffer_s8z24(struct gl_context *ctx,
> -  struct gl_renderbuffer *rb,
> +radeon_map_renderbuffer_s8z24(struct gl_renderbuffer *rb,
>GLuint x, GLuint y, GLuint w, GLuint h,
>GLbitfield mode,
>GLubyte **out_map,
> @@ -183,8 +182,7 @@ radeon_map_renderbuffer_s8z24(struct gl_context *ctx,
>  }
>
>  static void
> -radeon_map_renderbuffer_z16(struct gl_context *ctx,
> -   struct gl_renderbuffer *rb,
> +radeon_map_renderbuffer_z16(struct gl_renderbuffer *rb,
> GLuint x, GLuint y, GLuint w, GLuint h,
> GLbitfield mode,
> GLubyte **out_map,
> @@ -307,12 +305,12 @@ radeon_map_renderbuffer(struct gl_context *ctx,
>
> if ((rmesa->radeonScreen->chip_flags & RADEON_CHIPSET_DEPTH_ALWAYS_TILED) 
> && !rrb->has_surface) {
> if (rb->Format == MESA_FORMAT_Z24_UNORM_S8_UINT || rb->Format == 
> MESA_FORMAT_Z24_UNORM_X8_UINT) {
> -  radeon_map_renderbuffer_s8z24(ctx, rb, x, y, w, h,
> +  radeon_map_renderbuffer_s8z24(rb, x, y, w, h,
>  mode, out_map, out_stride);
>return;
> }
> if (rb->Format == MESA_FORMAT_Z_UNORM16) {
> -  radeon_map_renderbuffer_z16(ctx, rb, x, y, w, h,
> +  radeon_map_renderbuffer_z16(rb, x, y, w, h,
>mode, out_map, out_stride);
>return;
> }
> @@ -621,8 +619,11 @@ radeon_alloc_window_storage(struct gl_context * ctx, 
> struct gl_renderbuffer *rb,
>
>  /** Dummy function for gl_renderbuffer::AllocStorage() */
>  static GLboolean
> -radeon_nop_alloc_storage(struct gl_context * ctx, struct gl_renderbuffer *rb,
> -GLenum internalFormat, 

[Mesa-dev] [Bug 101071] compiling glsl fails with undefined reference to `pthread_create'

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=101071

--- Comment #10 from war...@o2.pl ---
Emil,
Unfortunately minimizing build options doesn't help me with successful
bisecting.
I'm always failing in the middle with following erorr:

//home/piotro/minimyth-dev/images/build/bin/bash ../../libtool  --tag=CC  
--mode=link x86_64-minimyth-linux-gnu-gcc -pthread -Werror=pointer-arith
-Werror=vla -fvisibility=hidden -ffat-lto-objects -pipe -flto -pipe
-march=x86-64 -mtune=generic -O3 -mfpmath=sse -flto -m64 -Wall -std=c99
-Werror=implicit-function-declaration -Werror=missing-prototypes
-fno-math-errno -fno-trapping-math  -no-undefined -Wl,--gc-sections
-Wl,--no-undefined -Wl,--as-needed -pipe -flto -pipe -march=x86-64
-mtune=generic -O3 -mfpmath=sse -flto -m64 -o shared-glapi/libglapi.la -rpath
/usr/lib shared_glapi_libglapi_la-entry.lo
shared_glapi_libglapi_la-mapi_glapi.lo shared_glapi_libglapi_la-stub.lo
shared_glapi_libglapi_la-table.lo shared_glapi_libglapi_la-u_current.lo
shared_glapi_libglapi_la-u_execmem.lo -lpthread
libtool: link: x86_64-minimyth-linux-gnu-gcc -shared  -fPIC -DPIC 
.libs/shared_glapi_libglapi_la-entry.o
.libs/shared_glapi_libglapi_la-mapi_glapi.o
.libs/shared_glapi_libglapi_la-stub.o .libs/shared_glapi_libglapi_la-table.o
.libs/shared_glapi_libglapi_la-u_current.o
.libs/shared_glapi_libglapi_la-u_execmem.o   -lpthread  -pthread -flto
-march=x86-64 -mtune=generic -O3 -mfpmath=sse -flto -m64 -Wl,--gc-sections
-Wl,--no-undefined -Wl,--as-needed -flto -march=x86-64 -mtune=generic -O3
-mfpmath=sse -flto -m64   -pthread -Wl,-soname -Wl,libglapi.so.0 -o
shared-glapi/.libs/libglapi.so.0.0.0
libtool: link: (cd "shared-glapi/.libs" && rm -f "libglapi.so.0" && ln -s
"libglapi.so.0.0.0" "libglapi.so.0")
libtool: link: (cd "shared-glapi/.libs" && rm -f "libglapi.so" && ln -s
"libglapi.so.0.0.0" "libglapi.so")
libtool: link: ( cd "shared-glapi/.libs" && rm -f "libglapi.la" && ln -s
"../libglapi.la" "libglapi.la" )
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p ../../lib; \
for f in shared-glapi/.libs/libglapi.so*; do \
if test -h .libs/$f; then   \
cp -d $f ../../lib; \
else\
ln -f $f ../../lib; \
fi; \
done && touch .install-mesa-links
make[6]: Leaving directory
`/home/piotro/minimyth-dev/script/xorg/Mesa/work/main.d/mesa-17.1.1/src/mapi'
make[5]: Leaving directory
`/home/piotro/minimyth-dev/script/xorg/Mesa/work/main.d/mesa-17.1.1/src/mapi'
make[4]: Leaving directory
`/home/piotro/minimyth-dev/script/xorg/Mesa/work/main.d/mesa-17.1.1/src/mapi'
Making all in compiler
make[4]: Entering directory
`/home/piotro/minimyth-dev/script/xorg/Mesa/work/main.d/mesa-17.1.1/src/compiler'
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl/glcpp
bison  -o glsl/glsl_parser.cpp -p "_mesa_glsl_" --defines=./glsl/glsl_parser.h
./glsl/glsl_parser.yy
flex  -o glsl/glsl_lexer.cpp ./glsl/glsl_lexer.ll
python  ./glsl/ir_expression_operation.py strings >
glsl/ir_expression_operation_strings.h || (rm -f
glsl/ir_expression_operation_strings.h; false)
python  ./glsl/ir_expression_operation.py enum > glsl/ir_expression_operation.h
|| (rm -f glsl/ir_expression_operation.h; false)
python  ./glsl/ir_expression_operation.py constant >
glsl/ir_expression_operation_constant.h || (rm -f
glsl/ir_expression_operation_constant.h; false)
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p glsl/glcpp
bison  -o glsl/glcpp/glcpp-parse.c -p "glcpp_parser_"
--defines=./glsl/glcpp/glcpp-parse.h ./glsl/glcpp/glcpp-parse.y
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p nir
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p nir
flex  -o glsl/glcpp/glcpp-lex.c ./glsl/glcpp/glcpp-lex.l
//home/piotro/minimyth-dev/images/build/usr/bin/mkdir -p nir
python  ./nir/nir_builder_opcodes_h.py > nir/nir_builder_opcodes.h || (rm -f
nir/nir_builder_opcodes.h; false)
python  ./nir/nir_constant_expressions.py > nir/nir_constant_expressions.c ||
(rm -f nir/nir_constant_expressions.c; false)
python  ./nir/nir_opcodes_c.py > nir/nir_opcodes.c || (rm -f nir/nir_opcodes.c;
false)
Traceback (most recent call last):
  File "./nir/nir_builder_opcodes_h.py", line 46, in 
from nir_opcodes import opcodes
  File
"/home/piotro/minimyth-dev/script/xorg/Mesa/work/main.d/mesa-17.1.1/src/compiler/nir/nir_opcodes.py",
line 178, in 
unop_convert("{}2{}{}".format(src_t[0], dst_t[0], bit_size),
ValueError: zero length field name in format
Traceback (most recent call last):
  File "./nir/nir_constant_expressions.py", line 405, in 
from nir_opcodes import

Re: [Mesa-dev] [PATCH V2] i965: Rename brw_format variable names to isl_format

[Anuj Phogat]

On Tue, May 23, 2017 at 2:35 PM, Anuj Phogat  wrote:
> This patch makes non functional changes. Renaming is just to
> make the code more readable.
>
> V2: update the types to "enum isl_format"
Jason, do you have any other questions ? r-b ?
>
> Signed-off-by: Anuj Phogat 
> Cc: Jason Ekstrand 
> ---
>  src/mesa/drivers/dri/i965/brw_context.c  | 5 +++--
>  src/mesa/drivers/dri/i965/brw_wm_surface_state.c | 6 +++---
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.c| 5 +++--
>  3 files changed, 9 insertions(+), 7 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_context.c 
> b/src/mesa/drivers/dri/i965/brw_context.c
> index d3ed871..cb6a76a 100644
> --- a/src/mesa/drivers/dri/i965/brw_context.c
> +++ b/src/mesa/drivers/dri/i965/brw_context.c
> @@ -205,9 +205,10 @@ intel_texture_view_requires_resolve(struct brw_context 
> *brw,
> !intel_miptree_is_lossless_compressed(brw, intel_tex->mt))
>   return false;
>
> -   const uint32_t brw_format = 
> brw_isl_format_for_mesa_format(intel_tex->_Format);
> +   const enum isl_format isl_format =
> +  brw_isl_format_for_mesa_format(intel_tex->_Format);
>
> -   if (isl_format_supports_ccs_e(&brw->screen->devinfo, brw_format))
> +   if (isl_format_supports_ccs_e(&brw->screen->devinfo, isl_format))
>return false;
>
> perf_debug("Incompatible sampling format (%s) for rbc (%s)\n",
> diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c 
> b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> index a0fed60..05e41dc 100644
> --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> @@ -686,7 +686,7 @@ brw_update_buffer_texture_surface(struct gl_context *ctx,
> uint32_t size = tObj->BufferSize;
> struct brw_bo *bo = NULL;
> mesa_format format = tObj->_BufferObjectFormat;
> -   uint32_t brw_format = brw_isl_format_for_mesa_format(format);
> +   const enum isl_format isl_format = brw_isl_format_for_mesa_format(format);
> int texel_size = _mesa_get_format_bytes(format);
>
> if (intel_obj) {
> @@ -712,14 +712,14 @@ brw_update_buffer_texture_surface(struct gl_context 
> *ctx,
>  */
> size = MIN2(size, ctx->Const.MaxTextureBufferSize * (unsigned) 
> texel_size);
>
> -   if (brw_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
> +   if (isl_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
>_mesa_problem(NULL, "bad format %s for texture buffer\n",
> _mesa_get_format_name(format));
> }
>
> brw_emit_buffer_surface_state(brw, surf_offset, bo,
>   tObj->BufferOffset,
> - brw_format,
> + isl_format,
>   size,
>   texel_size,
>   false /* rw */);
> diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c 
> b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> index db0a397..a92e3cb 100644
> --- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> +++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> @@ -209,8 +209,9 @@ intel_miptree_supports_non_msrt_fast_clear(struct 
> brw_context *brw,
>
> if (brw->gen >= 9) {
>mesa_format linear_format = _mesa_get_srgb_format_linear(mt->format);
> -  const uint32_t brw_format = 
> brw_isl_format_for_mesa_format(linear_format);
> -  return isl_format_supports_ccs_e(&brw->screen->devinfo, brw_format);
> +  const enum isl_format isl_format =
> + brw_isl_format_for_mesa_format(linear_format);
> +  return isl_format_supports_ccs_e(&brw->screen->devinfo, isl_format);
> } else
>return true;
>  }
> --
> 2.9.3
>
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH V2] i965: Rename brw_format variable names to isl_format

[Jason Ekstrand]

On Thu, Jun 1, 2017 at 9:41 AM, Anuj Phogat  wrote:

> On Tue, May 23, 2017 at 2:35 PM, Anuj Phogat 
> wrote:
> > This patch makes non functional changes. Renaming is just to
> > make the code more readable.
> >
> > V2: update the types to "enum isl_format"
> Jason, do you have any other questions ? r-b ?
>

Patch 1/6 of the series Chad sent out yesterday also does this and a bit
more.  I don't think he saw yours.


> >
> > Signed-off-by: Anuj Phogat 
> > Cc: Jason Ekstrand 
> > ---
> >  src/mesa/drivers/dri/i965/brw_context.c  | 5 +++--
> >  src/mesa/drivers/dri/i965/brw_wm_surface_state.c | 6 +++---
> >  src/mesa/drivers/dri/i965/intel_mipmap_tree.c| 5 +++--
> >  3 files changed, 9 insertions(+), 7 deletions(-)
> >
> > diff --git a/src/mesa/drivers/dri/i965/brw_context.c
> b/src/mesa/drivers/dri/i965/brw_context.c
> > index d3ed871..cb6a76a 100644
> > --- a/src/mesa/drivers/dri/i965/brw_context.c
> > +++ b/src/mesa/drivers/dri/i965/brw_context.c
> > @@ -205,9 +205,10 @@ intel_texture_view_requires_resolve(struct
> brw_context *brw,
> > !intel_miptree_is_lossless_compressed(brw, intel_tex->mt))
> >   return false;
> >
> > -   const uint32_t brw_format = brw_isl_format_for_mesa_
> format(intel_tex->_Format);
> > +   const enum isl_format isl_format =
> > +  brw_isl_format_for_mesa_format(intel_tex->_Format);
> >
> > -   if (isl_format_supports_ccs_e(&brw->screen->devinfo, brw_format))
> > +   if (isl_format_supports_ccs_e(&brw->screen->devinfo, isl_format))
> >return false;
> >
> > perf_debug("Incompatible sampling format (%s) for rbc (%s)\n",
> > diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> > index a0fed60..05e41dc 100644
> > --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> > +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
> > @@ -686,7 +686,7 @@ brw_update_buffer_texture_surface(struct gl_context
> *ctx,
> > uint32_t size = tObj->BufferSize;
> > struct brw_bo *bo = NULL;
> > mesa_format format = tObj->_BufferObjectFormat;
> > -   uint32_t brw_format = brw_isl_format_for_mesa_format(format);
> > +   const enum isl_format isl_format = brw_isl_format_for_mesa_
> format(format);
> > int texel_size = _mesa_get_format_bytes(format);
> >
> > if (intel_obj) {
> > @@ -712,14 +712,14 @@ brw_update_buffer_texture_surface(struct
> gl_context *ctx,
> >  */
> > size = MIN2(size, ctx->Const.MaxTextureBufferSize * (unsigned)
> texel_size);
> >
> > -   if (brw_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
> > +   if (isl_format == 0 && format != MESA_FORMAT_RGBA_FLOAT32) {
> >_mesa_problem(NULL, "bad format %s for texture buffer\n",
> > _mesa_get_format_name(format));
> > }
> >
> > brw_emit_buffer_surface_state(brw, surf_offset, bo,
> >   tObj->BufferOffset,
> > - brw_format,
> > + isl_format,
> >   size,
> >   texel_size,
> >   false /* rw */);
> > diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > index db0a397..a92e3cb 100644
> > --- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > +++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > @@ -209,8 +209,9 @@ intel_miptree_supports_non_msrt_fast_clear(struct
> brw_context *brw,
> >
> > if (brw->gen >= 9) {
> >mesa_format linear_format = _mesa_get_srgb_format_linear(
> mt->format);
> > -  const uint32_t brw_format = brw_isl_format_for_mesa_
> format(linear_format);
> > -  return isl_format_supports_ccs_e(&brw->screen->devinfo,
> brw_format);
> > +  const enum isl_format isl_format =
> > + brw_isl_format_for_mesa_format(linear_format);
> > +  return isl_format_supports_ccs_e(&brw->screen->devinfo,
> isl_format);
> > } else
> >return true;
> >  }
> > --
> > 2.9.3
> >
>
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] automake: Link all libGL.so variants with -Bsymbolic.

[Emil Velikov]

On 1 June 2017 at 16:46, Jose Fonseca  wrote:
> We were linking src/glx with -Bsymbolic, but not the classic/gallium X11
> libGL.so.
>
> But it's always a good idea to build all libGL.so and all DRI drivers
> with -Bsymbolic, otherwise they might resolve symbols from the 3rd party
> application executable or shared libraries, which is _never_ what we
> want.
>
> In particular, this can happen when intercepting OpenGL calls with
> apitrace, before
> https://github.com/apitrace/apitrace/commit/63194b2573176ef34efce1a5c8b08e624b8dddf5
>
> Cc: mesa-sta...@lists.freedesktop.org
Yes please
Reviewed-by: Emil Velikov 

FWIW there was some concerns -Bsymbolic and --dynamic-list
interactions for gallium drivers.
But that's orthogonal to this particular patch.

-Emil
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 2/2] gallium/u_threaded: remove 16 bytes from tc_batch

[Marek Olšák]

From: Marek Olšák 

All other sentinels occupy what is otherwise unused space.
---
 src/gallium/auxiliary/util/u_threaded_context.c | 2 --
 src/gallium/auxiliary/util/u_threaded_context.h | 1 -
 2 files changed, 3 deletions(-)

diff --git a/src/gallium/auxiliary/util/u_threaded_context.c 
b/src/gallium/auxiliary/util/u_threaded_context.c
index 34206bf..71211e6 100644
--- a/src/gallium/auxiliary/util/u_threaded_context.c
+++ b/src/gallium/auxiliary/util/u_threaded_context.c
@@ -60,21 +60,20 @@ enum tc_call_id {
 };
 
 typedef void (*tc_execute)(struct pipe_context *pipe, union tc_payload 
*payload);
 
 static const tc_execute execute_func[TC_NUM_CALLS];
 
 static void
 tc_batch_check(struct tc_batch *batch)
 {
tc_assert(batch->sentinel == TC_SENTINEL);
-   tc_assert(batch->sentinel2 == TC_SENTINEL);
tc_assert(batch->num_total_call_slots <= TC_CALLS_PER_BATCH);
 }
 
 static void
 tc_debug_check(struct threaded_context *tc)
 {
for (unsigned i = 0; i < TC_MAX_BATCHES; i++) {
   tc_batch_check(&tc->batch_slots[i]);
   tc_assert(tc->batch_slots[i].pipe == tc->pipe);
}
@@ -2200,21 +2199,20 @@ threaded_context_create(struct pipe_context *pipe,
 
/* The queue size is the number of batches "waiting". Batches are removed
 * from the queue before being executed, so keep one tc_batch slot for that
 * execution. Also, keep one unused slot for an unflushed batch.
 */
if (!util_queue_init(&tc->queue, "gallium_drv", TC_MAX_BATCHES - 2, 1))
   goto fail;
 
for (unsigned i = 0; i < TC_MAX_BATCHES; i++) {
   tc->batch_slots[i].sentinel = TC_SENTINEL;
-  tc->batch_slots[i].sentinel2 = TC_SENTINEL;
   tc->batch_slots[i].pipe = pipe;
   util_queue_fence_init(&tc->batch_slots[i].fence);
}
 
LIST_INITHEAD(&tc->unflushed_queries);
 
slab_create_child(&tc->pool_transfers, parent_transfer_pool);
 
 #define CTX_INIT(_member) \
tc->base._member = tc->pipe->_member ? tc_##_member : NULL
diff --git a/src/gallium/auxiliary/util/u_threaded_context.h 
b/src/gallium/auxiliary/util/u_threaded_context.h
index 5d2a10c..2e7e301 100644
--- a/src/gallium/auxiliary/util/u_threaded_context.h
+++ b/src/gallium/auxiliary/util/u_threaded_context.h
@@ -286,21 +286,20 @@ struct ALIGN16 tc_call {
ushort call_id;
union tc_payload payload;
 };
 
 struct tc_batch {
struct pipe_context *pipe;
unsigned sentinel;
unsigned num_total_call_slots;
struct util_queue_fence fence;
struct tc_call call[TC_CALLS_PER_BATCH];
-   unsigned sentinel2;
 };
 
 struct threaded_context {
struct pipe_context base;
struct pipe_context *pipe;
struct slab_child_pool pool_transfers;
tc_replace_buffer_storage_func replace_buffer_storage;
unsigned map_buffer_alignment;
 
struct list_head unflushed_queries;
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH] st/mesa: don't load cached TGSI shaders on demand

[Marek Olšák]

From: Marek Olšák 

This fixes a performance issue with the shader cache that delayed Gallium
shader create calls until draw calls.

I'd like this in stable, but it's not a showstopper.

Cc: 17.1 
---
 src/mesa/state_tracker/st_shader_cache.c | 7 ++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/src/mesa/state_tracker/st_shader_cache.c 
b/src/mesa/state_tracker/st_shader_cache.c
index 31c3430..305435f 100644
--- a/src/mesa/state_tracker/st_shader_cache.c
+++ b/src/mesa/state_tracker/st_shader_cache.c
@@ -15,21 +15,21 @@
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
  * DEALINGS IN THE SOFTWARE.
  */
 
 #include 
-
+#include "st_debug.h"
 #include "st_program.h"
 #include "st_shader_cache.h"
 #include "compiler/glsl/program.h"
 #include "pipe/p_shader_tokens.h"
 #include "program/ir_to_mesa.h"
 #include "util/u_memory.h"
 
 static void
 write_stream_out_to_cache(struct blob *blob,
   struct pipe_shader_state *tgsi)
@@ -360,20 +360,25 @@ st_load_tgsi_from_disk_cache(struct gl_context *ctx,
  if (ctx->_Shader->Flags & GLSL_CACHE_INFO) {
 _mesa_sha1_format(sha1_buf, sha1);
 fprintf(stderr, "%s tgsi_tokens retrieved from cache: %s\n",
 _mesa_shader_stage_to_string(i), sha1_buf);
  }
 
  st_set_prog_affected_state_flags(glprog);
  _mesa_associate_uniform_storage(ctx, prog, glprog->Parameters,
  false);
 
+ /* Create Gallium shaders now instead of on demand. */
+ if (ST_DEBUG & DEBUG_PRECOMPILE ||
+ st->shader_has_one_variant[glprog->info.stage])
+st_precompile_shader_variant(st, glprog);
+
  free(buffer);
   } else {
  /* Failed to find a matching cached shader so fallback to recompile.
   */
  if (ctx->_Shader->Flags & GLSL_CACHE_INFO) {
 fprintf(stderr, "TGSI cache item not found.\n");
  }
 
  goto fallback_recompile;
   }
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH] tgsi/scan: fix scanning fragment shaders with PrimID and Position/Face

[Marek Olšák]

From: Marek Olšák 

Not relevant to radeonsi, because Position/Face are system values
with radeonsi, while this codepath is for drivers where Position and
Face are ordinary inputs.
---
 src/gallium/auxiliary/tgsi/tgsi_scan.c | 17 ++---
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.c 
b/src/gallium/auxiliary/tgsi/tgsi_scan.c
index 847f4fc..018ca4b 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_scan.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_scan.c
@@ -550,27 +550,30 @@ scan_declaration(struct tgsi_shader_info *info,
   case TGSI_FILE_INPUT:
  info->input_semantic_name[reg] = (ubyte) semName;
  info->input_semantic_index[reg] = (ubyte) semIndex;
  info->input_interpolate[reg] = (ubyte)fulldecl->Interp.Interpolate;
  info->input_interpolate_loc[reg] = (ubyte)fulldecl->Interp.Location;
  info->input_cylindrical_wrap[reg] = 
(ubyte)fulldecl->Interp.CylindricalWrap;
 
  /* Vertex shaders can have inputs with holes between them. */
  info->num_inputs = MAX2(info->num_inputs, reg + 1);
 
- if (semName == TGSI_SEMANTIC_PRIMID)
-info->uses_primid = TRUE;
- else if (procType == PIPE_SHADER_FRAGMENT) {
-if (semName == TGSI_SEMANTIC_POSITION)
-   info->reads_position = TRUE;
-else if (semName == TGSI_SEMANTIC_FACE)
-   info->uses_frontface = TRUE;
+ switch (semName) {
+ case TGSI_SEMANTIC_PRIMID:
+info->uses_primid = true;
+break;
+ case TGSI_SEMANTIC_POSITION:
+info->reads_position = true;
+break;
+ case TGSI_SEMANTIC_FACE:
+info->uses_frontface = true;
+break;
  }
  break;
 
   case TGSI_FILE_SYSTEM_VALUE:
  index = fulldecl->Range.First;
 
  info->system_value_semantic_name[index] = semName;
  info->num_system_values = MAX2(info->num_system_values, index + 1);
 
  switch (semName) {
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 1/2] gallium/u_threaded: align batches and call slots to 16 bytes

[Marek Olšák]

From: Marek Olšák 

not sure if this helps
---
 src/gallium/auxiliary/util/u_threaded_context.c | 11 +--
 src/gallium/auxiliary/util/u_threaded_context.h |  9 -
 2 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/src/gallium/auxiliary/util/u_threaded_context.c 
b/src/gallium/auxiliary/util/u_threaded_context.c
index 8ea7f8a..34206bf 100644
--- a/src/gallium/auxiliary/util/u_threaded_context.c
+++ b/src/gallium/auxiliary/util/u_threaded_context.c
@@ -2118,21 +2118,21 @@ tc_destroy(struct pipe_context *_pipe)
 
if (tc->base.const_uploader &&
tc->base.stream_uploader != tc->base.const_uploader)
   u_upload_destroy(tc->base.const_uploader);
 
if (tc->base.stream_uploader)
   u_upload_destroy(tc->base.stream_uploader);
 
slab_destroy_child(&tc->pool_transfers);
pipe->destroy(pipe);
-   FREE(tc);
+   os_free_aligned(tc);
 }
 
 static const tc_execute execute_func[TC_NUM_CALLS] = {
 #define CALL(name) tc_call_##name,
 #include "u_threaded_context_calls.h"
 #undef CALL
 };
 
 /**
  * Wrap an existing pipe_context into a threaded_context.
@@ -2158,25 +2158,32 @@ threaded_context_create(struct pipe_context *pipe,
STATIC_ASSERT(sizeof(struct tc_call) <= 16);
 
if (!pipe)
   return NULL;
 
util_cpu_detect();
 
if (!debug_get_bool_option("GALLIUM_THREAD", util_cpu_caps.nr_cpus > 1))
   return pipe;
 
-   tc = CALLOC_STRUCT(threaded_context);
+   tc = os_malloc_aligned(sizeof(struct threaded_context), 16);
if (!tc) {
   pipe->destroy(pipe);
   return NULL;
}
+   memset(tc, 0, sizeof(*tc));
+
+   assert((uintptr_t)tc % 16 == 0);
+   STATIC_ASSERT(offsetof(struct threaded_context, batch_slots[0]) % 16 == 0);
+   STATIC_ASSERT(offsetof(struct threaded_context, batch_slots[0].call[0]) % 
16 == 0);
+   STATIC_ASSERT(offsetof(struct threaded_context, batch_slots[0].call[1]) % 
16 == 0);
+   STATIC_ASSERT(offsetof(struct threaded_context, batch_slots[1].call[0]) % 
16 == 0);
 
/* The driver context isn't wrapped, so set its "priv" to NULL. */
pipe->priv = NULL;
 
tc->pipe = pipe;
tc->replace_buffer_storage = replace_buffer;
tc->map_buffer_alignment =
   pipe->screen->get_param(pipe->screen, PIPE_CAP_MIN_MAP_BUFFER_ALIGNMENT);
tc->base.priv = pipe; /* priv points to the wrapped driver context */
tc->base.screen = pipe->screen;
diff --git a/src/gallium/auxiliary/util/u_threaded_context.h 
b/src/gallium/auxiliary/util/u_threaded_context.h
index f139230..5d2a10c 100644
--- a/src/gallium/auxiliary/util/u_threaded_context.h
+++ b/src/gallium/auxiliary/util/u_threaded_context.h
@@ -266,21 +266,28 @@ struct threaded_query {
  * Most calls will typecast this to the type they need, typically larger
  * than 8 bytes.
  */
 union tc_payload {
struct pipe_query *query;
struct pipe_resource *resource;
struct pipe_transfer *transfer;
uint64_t __use_8_bytes;
 };
 
-struct tc_call {
+#ifdef _MSC_VER
+#define ALIGN16 __declspec(align(16))
+#else
+#define ALIGN16 __attribute__((aligned(16)))
+#endif
+
+/* Each call slot should be aligned to its own size for optimal cache usage. */
+struct ALIGN16 tc_call {
unsigned sentinel;
ushort num_call_slots;
ushort call_id;
union tc_payload payload;
 };
 
 struct tc_batch {
struct pipe_context *pipe;
unsigned sentinel;
unsigned num_total_call_slots;
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 03/11] radeonsi: decrease the number of compiler threads to num CPUs - 1

[Marek Olšák]

From: Marek Olšák 

Reserve one core for other things (like draw calls).
---
 src/gallium/drivers/radeonsi/si_pipe.c | 5 -
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/gallium/drivers/radeonsi/si_pipe.c 
b/src/gallium/drivers/radeonsi/si_pipe.c
index de4e5da..4704304 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/radeonsi/si_pipe.c
@@ -874,22 +874,25 @@ struct pipe_screen *radeonsi_screen_create(struct 
radeon_winsys *ws)
 
si_init_screen_state_functions(sscreen);
 
if (!r600_common_screen_init(&sscreen->b, ws) ||
!si_init_gs_info(sscreen) ||
!si_init_shader_cache(sscreen)) {
FREE(sscreen);
return NULL;
}
 
-   /* Only enable as many threads as we have target machines and CPUs. */
+   /* Only enable as many threads as we have target machines, but at most
+* the number of CPUs - 1 if there is more than one.
+*/
num_cpus = sysconf(_SC_NPROCESSORS_ONLN);
+   num_cpus = MAX2(1, num_cpus - 1);
num_compiler_threads = MIN2(num_cpus, ARRAY_SIZE(sscreen->tm));
 
if (!util_queue_init(&sscreen->shader_compiler_queue, "si_shader",
 32, num_compiler_threads)) {
si_destroy_shader_cache(sscreen);
FREE(sscreen);
return NULL;
}
 
si_handle_env_var_force_family(sscreen);
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 05/11] radeonsi: use a compiler queue with a low priority for optimized shaders

[Marek Olšák]

From: Marek Olšák 

---
 src/gallium/drivers/radeonsi/si_pipe.c  | 31 +
 src/gallium/drivers/radeonsi/si_pipe.h  |  3 +++
 src/gallium/drivers/radeonsi/si_state_shaders.c |  8 +++
 3 files changed, 34 insertions(+), 8 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_pipe.c 
b/src/gallium/drivers/radeonsi/si_pipe.c
index 8bf6fd9..082ba99 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/radeonsi/si_pipe.c
@@ -734,25 +734,30 @@ static void si_destroy_screen(struct pipe_screen* pscreen)
sscreen->gs_prologs,
sscreen->ps_prologs,
sscreen->ps_epilogs
};
unsigned i;
 
if (!sscreen->b.ws->unref(sscreen->b.ws))
return;
 
util_queue_destroy(&sscreen->shader_compiler_queue);
+   util_queue_destroy(&sscreen->shader_compiler_queue_low_priority);
 
for (i = 0; i < ARRAY_SIZE(sscreen->tm); i++)
if (sscreen->tm[i])
LLVMDisposeTargetMachine(sscreen->tm[i]);
 
+   for (i = 0; i < ARRAY_SIZE(sscreen->tm_low_priority); i++)
+   if (sscreen->tm_low_priority[i])
+   LLVMDisposeTargetMachine(sscreen->tm_low_priority[i]);
+
/* Free shader parts. */
for (i = 0; i < ARRAY_SIZE(parts); i++) {
while (parts[i]) {
struct si_shader_part *part = parts[i];
 
parts[i] = part->next;
radeon_shader_binary_clean(&part->binary);
FREE(part);
}
}
@@ -852,21 +857,21 @@ static void si_test_vmfault(struct si_screen *sscreen)
if (sscreen->b.debug_flags & DBG_TEST_VMFAULT_SHADER) {
util_test_constant_buffer(ctx, buf);
puts("VM fault test: Shader - done.");
}
exit(0);
 }
 
 struct pipe_screen *radeonsi_screen_create(struct radeon_winsys *ws)
 {
struct si_screen *sscreen = CALLOC_STRUCT(si_screen);
-   unsigned num_cpus, num_compiler_threads, i;
+   unsigned num_threads, num_compiler_threads, 
num_compiler_threads_lowprio, i;
 
if (!sscreen) {
return NULL;
}
 
/* Set functions first. */
sscreen->b.b.context_create = si_pipe_create_context;
sscreen->b.b.destroy = si_destroy_screen;
sscreen->b.b.get_param = si_get_param;
sscreen->b.b.get_shader_param = si_get_shader_param;
@@ -877,31 +882,47 @@ struct pipe_screen *radeonsi_screen_create(struct 
radeon_winsys *ws)
if (!r600_common_screen_init(&sscreen->b, ws) ||
!si_init_gs_info(sscreen) ||
!si_init_shader_cache(sscreen)) {
FREE(sscreen);
return NULL;
}
 
/* Only enable as many threads as we have target machines, but at most
 * the number of CPUs - 1 if there is more than one.
 */
-   num_cpus = sysconf(_SC_NPROCESSORS_ONLN);
-   num_cpus = MAX2(1, num_cpus - 1);
-   num_compiler_threads = MIN2(num_cpus, ARRAY_SIZE(sscreen->tm));
+   num_threads = sysconf(_SC_NPROCESSORS_ONLN);
+   num_threads = MAX2(1, num_threads - 1);
+   num_compiler_threads = MIN2(num_threads, ARRAY_SIZE(sscreen->tm));
+   num_compiler_threads_lowprio =
+   MIN2(num_threads, ARRAY_SIZE(sscreen->tm_low_priority));
 
if (!util_queue_init(&sscreen->shader_compiler_queue, "si_shader",
 32, num_compiler_threads, 0)) {
si_destroy_shader_cache(sscreen);
FREE(sscreen);
return NULL;
}
 
+   /* The queue must be large enough so that adding optimized shaders
+* doesn't stall draw calls when the queue is full. Especially varying
+* packing generates a very high volume of optimized shader compilation
+* jobs.
+*/
+   if (!util_queue_init(&sscreen->shader_compiler_queue_low_priority,
+"si_shader_low",
+1024, num_compiler_threads,
+UTIL_QUEUE_INIT_USE_MINIMUM_PRIORITY)) {
+  si_destroy_shader_cache(sscreen);
+  FREE(sscreen);
+  return NULL;
+   }
+
si_handle_env_var_force_family(sscreen);
 
if (!debug_get_bool_option("RADEON_DISABLE_PERFCOUNTERS", false))
si_init_perfcounters(sscreen);
 
/* Hawaii has a bug with offchip buffers > 256 that can be worked
 * around by setting 4K granularity.
 */
sscreen->tess_offchip_block_dw_size =
sscreen->b.family == CHIP_HAWAII ? 4096 : 8192;
@@ -951,20 +972,22 @@ struct pipe_screen *radeonsi_screen_create(struct 
radeon_winsys *ws)
sscreen->b.barrier_flags.cp_to_L2 = SI_CONTEXT_INV_SMEM_L1 |
SI_CONTEXT_INV_VMEM_L1 |

[Mesa-dev] [PATCH 01/11] util/u_queue: add a way to remove a job when we just want to destroy it

[Marek Olšák]

From: Marek Olšák 

---
 src/util/u_queue.c | 50 --
 src/util/u_queue.h |  2 ++
 2 files changed, 46 insertions(+), 6 deletions(-)

diff --git a/src/util/u_queue.c b/src/util/u_queue.c
index 8db09b0..3834b6f 100644
--- a/src/util/u_queue.c
+++ b/src/util/u_queue.c
@@ -173,27 +173,29 @@ util_queue_thread_func(void *input)
   if (job.job) {
  job.execute(job.job, thread_index);
  util_queue_fence_signal(job.fence);
  if (job.cleanup)
 job.cleanup(job.job, thread_index);
   }
}
 
/* signal remaining jobs before terminating */
mtx_lock(&queue->lock);
-   while (queue->jobs[queue->read_idx].job) {
-  util_queue_fence_signal(queue->jobs[queue->read_idx].fence);
-
-  queue->jobs[queue->read_idx].job = NULL;
-  queue->read_idx = (queue->read_idx + 1) % queue->max_jobs;
+   for (unsigned i = queue->read_idx; i != queue->write_idx;
+i = (i + 1) % queue->max_jobs) {
+  if (queue->jobs[i].job) {
+ util_queue_fence_signal(queue->jobs[i].fence);
+ queue->jobs[i].job = NULL;
+  }
}
-   queue->num_queued = 0; /* reset this when exiting the thread */
+   queue->read_idx = (queue->read_idx + queue->num_queued) % queue->max_jobs;
+   queue->num_queued = 0;
mtx_unlock(&queue->lock);
return 0;
 }
 
 bool
 util_queue_init(struct util_queue *queue,
 const char *name,
 unsigned max_jobs,
 unsigned num_threads)
 {
@@ -322,19 +324,55 @@ util_queue_add_job(struct util_queue *queue,
ptr->fence = fence;
ptr->execute = execute;
ptr->cleanup = cleanup;
queue->write_idx = (queue->write_idx + 1) % queue->max_jobs;
 
queue->num_queued++;
cnd_signal(&queue->has_queued_cond);
mtx_unlock(&queue->lock);
 }
 
+/**
+ * Remove a queued job. If the job hasn't started execution, it's removed from
+ * the queue. If the job has started execution, the function waits for it to
+ * complete.
+ *
+ * In all cases, the fence is signalled when the function returns.
+ *
+ * The function can be used when destroying an object associated with the job
+ * when you don't care about the job completion state.
+ */
+void
+util_queue_drop_job(struct util_queue *queue, struct util_queue_fence *fence)
+{
+   bool removed = false;
+
+   if (util_queue_fence_is_signalled(fence))
+  return;
+
+   mtx_lock(&queue->lock);
+   for (unsigned i = queue->read_idx; i != queue->write_idx;
+i = (i + 1) % queue->max_jobs) {
+  if (queue->jobs[i].fence == fence) {
+ /* Just clear it. The threads will drop it. */
+ memset(&queue->jobs[i], 0, sizeof(queue->jobs[i]));
+ removed = true;
+ break;
+  }
+   }
+   mtx_unlock(&queue->lock);
+
+   if (removed)
+  util_queue_fence_signal(fence);
+   else
+  util_queue_fence_wait(fence);
+}
+
 int64_t
 util_queue_get_thread_time_nano(struct util_queue *queue, unsigned 
thread_index)
 {
/* Allow some flexibility by not raising an error. */
if (thread_index >= queue->num_threads)
   return 0;
 
return u_thread_get_time_nano(queue->threads[thread_index]);
 }
diff --git a/src/util/u_queue.h b/src/util/u_queue.h
index 4aec1f2..9876865 100644
--- a/src/util/u_queue.h
+++ b/src/util/u_queue.h
@@ -85,20 +85,22 @@ bool util_queue_init(struct util_queue *queue,
 void util_queue_destroy(struct util_queue *queue);
 void util_queue_fence_init(struct util_queue_fence *fence);
 void util_queue_fence_destroy(struct util_queue_fence *fence);
 
 /* optional cleanup callback is called after fence is signaled: */
 void util_queue_add_job(struct util_queue *queue,
 void *job,
 struct util_queue_fence *fence,
 util_queue_execute_func execute,
 util_queue_execute_func cleanup);
+void util_queue_drop_job(struct util_queue *queue,
+ struct util_queue_fence *fence);
 
 void util_queue_fence_wait(struct util_queue_fence *fence);
 int64_t util_queue_get_thread_time_nano(struct util_queue *queue,
 unsigned thread_index);
 
 /* util_queue needs to be cleared to zeroes for this to work */
 static inline bool
 util_queue_is_initialized(struct util_queue *queue)
 {
return queue->threads != NULL;
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 00/11] RadeonSI: Shader compiler queue improvements & GFX9 fixes

[Marek Olšák]

Hi,

The beginning of this series improves our shader compiler queues,
mainly the optimized compilations now run at the lowest thread
priority so as not to affect rendering performance.

It's followed by cleanups, and then fixes for GFX9.

Please review.

Thanks,
Marek
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 06/11] radeonsi: move handling of DBG_NO_OPT_VARIANT into si_shader_selector_key

[Marek Olšák]

From: Marek Olšák 

---
 src/gallium/drivers/radeonsi/si_state_shaders.c | 7 +++
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c 
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 5a22add..55e881c 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -1431,20 +1431,23 @@ static inline void si_shader_selector_key(struct 
pipe_context *ctx,
sel->info.uses_linear_sample > 1;
}
}
 
key->part.ps.epilog.alpha_func = si_get_alpha_test_func(sctx);
break;
}
default:
assert(0);
}
+
+   if (unlikely(sctx->screen->b.debug_flags & DBG_NO_OPT_VARIANT))
+   memset(&key->opt, 0, sizeof(key->opt));
 }
 
 static void si_build_shader_variant(void *job, int thread_index)
 {
struct si_shader *shader = (struct si_shader *)job;
struct si_shader_selector *sel = shader->selector;
struct si_screen *sscreen = sel->screen;
LLVMTargetMachineRef tm;
struct pipe_debug_callback *debug = &shader->compiler_ctx_state.debug;
int r;
@@ -1526,24 +1529,20 @@ static int si_shader_select_with_key(struct si_screen 
*sscreen,
 struct si_shader_ctx_state *state,
 struct si_compiler_ctx_state 
*compiler_state,
 struct si_shader_key *key,
 int thread_index)
 {
struct si_shader_selector *sel = state->cso;
struct si_shader_selector *previous_stage_sel = NULL;
struct si_shader *current = state->current;
struct si_shader *iter, *shader = NULL;
 
-   if (unlikely(sscreen->b.debug_flags & DBG_NO_OPT_VARIANT)) {
-   memset(&key->opt, 0, sizeof(key->opt));
-   }
-
 again:
/* Check if we don't need to change anything.
 * This path is also used for most shaders that don't need multiple
 * variants, it will cost just a computation of the key and this
 * test. */
if (likely(current &&
   memcmp(¤t->key, key, sizeof(*key)) == 0 &&
   (!current->is_optimized ||
util_queue_fence_is_signalled(¤t->optimized_ready
return current->compilation_failed ? -1 : 0;
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 04/11] util/u_queue: add an option to set the minimum thread priority

[Marek Olšák]

From: Marek Olšák 

---
 src/gallium/auxiliary/util/u_threaded_context.c   |  2 +-
 src/gallium/drivers/freedreno/freedreno_batch.c   |  2 +-
 src/gallium/drivers/radeonsi/si_pipe.c|  2 +-
 src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c |  2 +-
 src/gallium/winsys/radeon/drm/radeon_drm_winsys.c |  2 +-
 src/util/disk_cache.c |  2 +-
 src/util/u_queue.c| 19 ++-
 src/util/u_queue.h|  6 +-
 8 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/src/gallium/auxiliary/util/u_threaded_context.c 
b/src/gallium/auxiliary/util/u_threaded_context.c
index 71211e6..554cc88 100644
--- a/src/gallium/auxiliary/util/u_threaded_context.c
+++ b/src/gallium/auxiliary/util/u_threaded_context.c
@@ -2194,21 +2194,21 @@ threaded_context_create(struct pipe_context *pipe,
else
   tc->base.const_uploader = u_upload_clone(&tc->base, 
pipe->const_uploader);
 
if (!tc->base.stream_uploader || !tc->base.const_uploader)
   goto fail;
 
/* The queue size is the number of batches "waiting". Batches are removed
 * from the queue before being executed, so keep one tc_batch slot for that
 * execution. Also, keep one unused slot for an unflushed batch.
 */
-   if (!util_queue_init(&tc->queue, "gallium_drv", TC_MAX_BATCHES - 2, 1))
+   if (!util_queue_init(&tc->queue, "gallium_drv", TC_MAX_BATCHES - 2, 1, 0))
   goto fail;
 
for (unsigned i = 0; i < TC_MAX_BATCHES; i++) {
   tc->batch_slots[i].sentinel = TC_SENTINEL;
   tc->batch_slots[i].pipe = pipe;
   util_queue_fence_init(&tc->batch_slots[i].fence);
}
 
LIST_INITHEAD(&tc->unflushed_queries);
 
diff --git a/src/gallium/drivers/freedreno/freedreno_batch.c 
b/src/gallium/drivers/freedreno/freedreno_batch.c
index 5783ee8..33b6240 100644
--- a/src/gallium/drivers/freedreno/freedreno_batch.c
+++ b/src/gallium/drivers/freedreno/freedreno_batch.c
@@ -265,21 +265,21 @@ batch_flush(struct fd_batch *batch)
fd_batch_set_stage(batch, FD_STAGE_NULL);
 
fd_context_all_dirty(batch->ctx);
batch_flush_reset_dependencies(batch, true);
 
if (batch->ctx->screen->reorder) {
struct fd_batch *tmp = NULL;
fd_batch_reference(&tmp, batch);
 
if (!util_queue_is_initialized(&batch->ctx->flush_queue))
-   util_queue_init(&batch->ctx->flush_queue, 
"flush_queue", 16, 1);
+   util_queue_init(&batch->ctx->flush_queue, 
"flush_queue", 16, 1, 0);
 
util_queue_add_job(&batch->ctx->flush_queue,
batch, &batch->flush_fence,
batch_flush_func, batch_cleanup_func);
} else {
fd_gmem_render_tiles(batch);
batch_reset_resources(batch);
}
 
debug_assert(batch->reference.count > 0);
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c 
b/src/gallium/drivers/radeonsi/si_pipe.c
index 4704304..8bf6fd9 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/radeonsi/si_pipe.c
@@ -882,21 +882,21 @@ struct pipe_screen *radeonsi_screen_create(struct 
radeon_winsys *ws)
}
 
/* Only enable as many threads as we have target machines, but at most
 * the number of CPUs - 1 if there is more than one.
 */
num_cpus = sysconf(_SC_NPROCESSORS_ONLN);
num_cpus = MAX2(1, num_cpus - 1);
num_compiler_threads = MIN2(num_cpus, ARRAY_SIZE(sscreen->tm));
 
if (!util_queue_init(&sscreen->shader_compiler_queue, "si_shader",
-32, num_compiler_threads)) {
+32, num_compiler_threads, 0)) {
si_destroy_shader_cache(sscreen);
FREE(sscreen);
return NULL;
}
 
si_handle_env_var_force_family(sscreen);
 
if (!debug_get_bool_option("RADEON_DISABLE_PERFCOUNTERS", false))
si_init_perfcounters(sscreen);
 
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c 
b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
index c8bd60e..43f2ed2 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
@@ -298,21 +298,21 @@ amdgpu_winsys_create(int fd, radeon_screen_create_t 
screen_create)
ws->base.read_registers = amdgpu_read_registers;
 
amdgpu_bo_init_functions(ws);
amdgpu_cs_init_functions(ws);
amdgpu_surface_init_functions(ws);
 
LIST_INITHEAD(&ws->global_bo_list);
(void) mtx_init(&ws->global_bo_list_lock, mtx_plain);
(void) mtx_init(&ws->bo_fence_lock, mtx_plain);
 
-   if (!util_queue_init(&ws->cs_queue, "amdgpu_cs", 8, 1)) {
+   if (!util_queue_init(&ws->cs_queue, "amdgpu_cs", 8, 1, 0)) {
   amdgpu_winsys_destroy(&ws->base);
   mtx_unlock(&dev_tab_mutex);
   return NULL;
}
 
/* Create the screen at the end. The wins

[Mesa-dev] [PATCH 07/11] radeonsi: remove dead code in declare_input_fs

[Marek Olšák]

From: Marek Olšák 

Colors are interpolated in the PS prolog. This was never used.
---
 src/gallium/drivers/radeonsi/si_shader.c | 5 -
 1 file changed, 5 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_shader.c 
b/src/gallium/drivers/radeonsi/si_shader.c
index 3a86c0b..5c7deeb 100644
--- a/src/gallium/drivers/radeonsi/si_shader.c
+++ b/src/gallium/drivers/radeonsi/si_shader.c
@@ -1360,25 +1360,20 @@ static void declare_input_fs(
}
 
interp_param_idx = lookup_interp_param_index(decl->Interp.Interpolate,
 decl->Interp.Location);
if (interp_param_idx == -1)
return;
else if (interp_param_idx) {
interp_param = LLVMGetParam(ctx->main_fn, interp_param_idx);
}
 
-   if (decl->Semantic.Name == TGSI_SEMANTIC_COLOR &&
-   decl->Interp.Interpolate == TGSI_INTERPOLATE_COLOR &&
-   ctx->shader->key.part.ps.prolog.flatshade_colors)
-   interp_param = NULL; /* load the constant color */
-
interp_fs_input(ctx, input_index, decl->Semantic.Name,
decl->Semantic.Index, shader->selector->info.num_inputs,
shader->selector->info.colors_read, interp_param,
LLVMGetParam(main_fn, SI_PARAM_PRIM_MASK),
LLVMGetParam(main_fn, SI_PARAM_FRONT_FACE),
&out[0]);
 }
 
 static LLVMValueRef get_sample_id(struct si_shader_context *ctx)
 {
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 08/11] radeonsi: move streamout state update out of si_update_shaders

[Marek Olšák]

From: Marek Olšák 

---
 src/gallium/drivers/radeonsi/si_shader.h|  1 +
 src/gallium/drivers/radeonsi/si_state_shaders.c | 40 +++--
 2 files changed, 25 insertions(+), 16 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_shader.h 
b/src/gallium/drivers/radeonsi/si_shader.h
index aab902b..7112621 100644
--- a/src/gallium/drivers/radeonsi/si_shader.h
+++ b/src/gallium/drivers/radeonsi/si_shader.h
@@ -329,20 +329,21 @@ struct si_shader_selector {
 
/* GS parameters. */
unsignedesgs_itemsize;
unsignedgs_input_verts_per_prim;
unsignedgs_output_prim;
unsignedgs_max_out_vertices;
unsignedgs_num_invocations;
unsignedmax_gs_stream; /* count - 1 */
unsignedgsvs_vertex_size;
unsignedmax_gsvs_emit_size;
+   unsignedenabled_streamout_buffer_mask;
 
/* PS parameters. */
unsignedcolor_attr_index[2];
unsigneddb_shader_control;
/* Set 0xf or 0x0 (4 bits) per each written output.
 * ANDed with spi_shader_col_format.
 */
unsignedcolors_written_4bit;
 
/* CS parameters */
diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c 
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 55e881c..5cbb91b 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -1949,20 +1949,27 @@ static void *si_create_shader_selector(struct 
pipe_context *ctx,
}
 
sel->so = state->stream_output;
tgsi_scan_shader(state->tokens, &sel->info);
sel->type = sel->info.processor;
p_atomic_inc(&sscreen->b.num_shaders_created);
si_get_active_slot_masks(&sel->info,
 &sel->active_const_and_shader_buffers,
 &sel->active_samplers_and_images);
 
+   /* Record which streamout buffers are enabled. */
+   for (i = 0; i < sel->so.num_outputs; i++) {
+   sel->enabled_streamout_buffer_mask |=
+   (1 << sel->so.output[i].output_buffer) <<
+   (sel->so.output[i].stream * 4);
+   }
+
/* The prolog is a no-op if there are no inputs. */
sel->vs_needs_prolog = sel->type == PIPE_SHADER_VERTEX &&
   sel->info.num_inputs;
 
/* Set which opcode uses which (i,j) pair. */
if (sel->info.uses_persp_opcode_interp_centroid)
sel->info.uses_persp_centroid = true;
 
if (sel->info.uses_linear_opcode_interp_centroid)
sel->info.uses_linear_centroid = true;
@@ -2130,34 +2137,49 @@ static void *si_create_shader_selector(struct 
pipe_context *ctx,
r600_can_dump_shader(&sscreen->b, sel->info.processor))
si_init_shader_selector_async(sel, -1);
else
util_queue_add_job(&sscreen->shader_compiler_queue, sel,
&sel->ready, si_init_shader_selector_async,
NULL);
 
return sel;
 }
 
+static void si_update_streamout_state(struct si_context *sctx)
+{
+   struct si_shader_selector *shader_with_so =
+   sctx->gs_shader.cso ? sctx->gs_shader.cso :
+   sctx->tes_shader.cso ? sctx->tes_shader.cso :
+  sctx->vs_shader.cso;
+   if (!shader_with_so)
+   return;
+
+   sctx->b.streamout.enabled_stream_buffers_mask =
+   shader_with_so->enabled_streamout_buffer_mask;
+   sctx->b.streamout.stride_in_dw = shader_with_so->so.stride;
+}
+
 static void si_bind_vs_shader(struct pipe_context *ctx, void *state)
 {
struct si_context *sctx = (struct si_context *)ctx;
struct si_shader_selector *sel = state;
 
if (sctx->vs_shader.cso == sel)
return;
 
sctx->vs_shader.cso = sel;
sctx->vs_shader.current = sel ? sel->first_variant : NULL;
sctx->do_update_shaders = true;
si_mark_atom_dirty(sctx, &sctx->clip_regs);
r600_update_vs_writes_viewport_index(&sctx->b, si_get_vs_info(sctx));
si_set_active_descriptors_for_shader(sctx, sel);
+   si_update_streamout_state(sctx);
 }
 
 static void si_update_tess_uses_prim_id(struct si_context *sctx)
 {
sctx->ia_multi_vgt_param_key.u.tess_uses_prim_id =
(sctx->tes_shader.cso &&
 sctx->tes_shader.cso->info.uses_primid) ||
(sctx->tcs_shader.cso &&
 sctx->tcs_shader.cso->info.uses_primid) ||
(sctx->gs_shader.cso &&
@@ -2182,20 +2204,21 @@ static void si_bind_gs_shader(struct pipe_context *ctx, 
void *state)
si_mark_atom_dirty(sctx, &sctx->clip_regs);
sctx->last_rast_prim = -1; /* reset this so that it gets updated */
 
if (enable_changed) {
si_shader_

[Mesa-dev] [PATCH 02/11] radeonsi: drop unfinished shader compilations when destroying shaders

[Marek Olšák]

From: Marek Olšák 

If we enqueue too many jobs and destroy the GL context, it may take
several seconds before the jobs finish. Just drop them instead.
---
 src/gallium/drivers/radeonsi/si_compute.c   | 3 ++-
 src/gallium/drivers/radeonsi/si_state_shaders.c | 5 +++--
 src/util/u_queue.c  | 5 -
 3 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_compute.c 
b/src/gallium/drivers/radeonsi/si_compute.c
index 4c98066..0338b8a 100644
--- a/src/gallium/drivers/radeonsi/si_compute.c
+++ b/src/gallium/drivers/radeonsi/si_compute.c
@@ -853,21 +853,22 @@ static void si_launch_grid(
 
 static void si_delete_compute_state(struct pipe_context *ctx, void* state){
struct si_compute *program = (struct si_compute *)state;
struct si_context *sctx = (struct si_context*)ctx;
 
if (!state) {
return;
}
 
if (program->ir_type == PIPE_SHADER_IR_TGSI) {
-   util_queue_fence_wait(&program->ready);
+   util_queue_drop_job(&sctx->screen->shader_compiler_queue,
+   &program->ready);
util_queue_fence_destroy(&program->ready);
}
 
if (program == sctx->cs_shader_state.program)
sctx->cs_shader_state.program = NULL;
 
if (program == sctx->cs_shader_state.emitted_program)
sctx->cs_shader_state.emitted_program = NULL;
 
si_shader_destroy(&program->shader);
diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c 
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 8ac4309..62bb221 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -2251,21 +2251,22 @@ static void si_bind_ps_shader(struct pipe_context *ctx, 
void *state)
sctx->do_update_shaders = true;
if (sel && sctx->ia_multi_vgt_param_key.u.uses_tess)
si_update_tess_uses_prim_id(sctx);
si_mark_atom_dirty(sctx, &sctx->cb_render_state);
si_set_active_descriptors_for_shader(sctx, sel);
 }
 
 static void si_delete_shader(struct si_context *sctx, struct si_shader *shader)
 {
if (shader->is_optimized) {
-   util_queue_fence_wait(&shader->optimized_ready);
+   util_queue_drop_job(&sctx->screen->shader_compiler_queue,
+   &shader->optimized_ready);
util_queue_fence_destroy(&shader->optimized_ready);
}
 
if (shader->pm4) {
switch (shader->selector->type) {
case PIPE_SHADER_VERTEX:
if (shader->key.as_ls) {
assert(sctx->b.chip_class <= VI);
si_pm4_delete_state(sctx, ls, shader->pm4);
} else if (shader->key.as_es) {
@@ -2308,21 +2309,21 @@ static void si_destroy_shader_selector(struct 
si_context *sctx,
 {
struct si_shader *p = sel->first_variant, *c;
struct si_shader_ctx_state *current_shader[SI_NUM_SHADERS] = {
[PIPE_SHADER_VERTEX] = &sctx->vs_shader,
[PIPE_SHADER_TESS_CTRL] = &sctx->tcs_shader,
[PIPE_SHADER_TESS_EVAL] = &sctx->tes_shader,
[PIPE_SHADER_GEOMETRY] = &sctx->gs_shader,
[PIPE_SHADER_FRAGMENT] = &sctx->ps_shader,
};
 
-   util_queue_fence_wait(&sel->ready);
+   util_queue_drop_job(&sctx->screen->shader_compiler_queue, &sel->ready);
 
if (current_shader[sel->type]->cso == sel) {
current_shader[sel->type]->cso = NULL;
current_shader[sel->type]->current = NULL;
}
 
while (p) {
c = p->next_variant;
si_delete_shader(sctx, p);
p = c;
diff --git a/src/util/u_queue.c b/src/util/u_queue.c
index 3834b6f..99de34c 100644
--- a/src/util/u_queue.c
+++ b/src/util/u_queue.c
@@ -346,21 +346,24 @@ util_queue_drop_job(struct util_queue *queue, struct 
util_queue_fence *fence)
 {
bool removed = false;
 
if (util_queue_fence_is_signalled(fence))
   return;
 
mtx_lock(&queue->lock);
for (unsigned i = queue->read_idx; i != queue->write_idx;
 i = (i + 1) % queue->max_jobs) {
   if (queue->jobs[i].fence == fence) {
- /* Just clear it. The threads will drop it. */
+ if (queue->jobs[i].cleanup)
+queue->jobs[i].cleanup(queue->jobs[i].job, -1);
+
+ /* Just clear it. The threads will treat as a no-op job. */
  memset(&queue->jobs[i], 0, sizeof(queue->jobs[i]));
  removed = true;
  break;
   }
}
mtx_unlock(&queue->lock);
 
if (removed)
   util_queue_fence_signal(fence);
else
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 10/11] radeonsi/gfx9: wait for main part compilation of 1st shaders or merged shaders

[Marek Olšák]

From: Marek Olšák 

---
 src/gallium/drivers/radeonsi/si_state_shaders.c | 4 
 1 file changed, 4 insertions(+)

diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c 
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 631272e..727bf4b 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -1595,20 +1595,24 @@ again:
shader->selector = sel;
shader->key = *key;
shader->compiler_ctx_state = *compiler_state;
 
/* If this is a merged shader, get the first shader's selector. */
if (sscreen->b.chip_class >= GFX9) {
if (sel->type == PIPE_SHADER_TESS_CTRL)
previous_stage_sel = key->part.tcs.ls;
else if (sel->type == PIPE_SHADER_GEOMETRY)
previous_stage_sel = key->part.gs.es;
+
+   /* We need to wait for the previous shader. */
+   if (previous_stage_sel && thread_index < 0)
+   util_queue_fence_wait(&previous_stage_sel->ready);
}
 
/* Compile the main shader part if it doesn't exist. This can happen
 * if the initial guess was wrong. */
bool is_pure_monolithic =
sscreen->use_monolithic_shaders ||
memcmp(&key->mono, &zeroed.mono, sizeof(key->mono)) != 0;
 
if (!is_pure_monolithic) {
bool ok;
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 11/11] radeonsi/gfx9: prevent a race when the previous shader's main part is missing

[Marek Olšák]

From: Marek Olšák 

---
 src/gallium/drivers/radeonsi/si_state_shaders.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c 
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 727bf4b..08d647b 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -1627,23 +1627,25 @@ again:
if (previous_stage_sel) {
struct si_shader_key shader1_key = zeroed;
 
if (sel->type == PIPE_SHADER_TESS_CTRL)
shader1_key.as_ls = 1;
else if (sel->type == PIPE_SHADER_GEOMETRY)
shader1_key.as_es = 1;
else
assert(0);
 
+   mtx_lock(&previous_stage_sel->mutex);
ok = si_check_missing_main_part(sscreen,
previous_stage_sel,
compiler_state, 
&shader1_key);
+   mtx_unlock(&previous_stage_sel->mutex);
} else {
ok = si_check_missing_main_part(sscreen, sel,
compiler_state, key);
}
if (!ok) {
FREE(shader);
mtx_unlock(&sel->mutex);
return -ENOMEM; /* skip the draw call */
}
}
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 09/11] radeonsi/gfx9: fix LS scratch buffer support without TCS for GFX9

[Marek Olšák]

From: Marek Olšák 

LS is merged into TCS. If there is no TCS, LS is merged into fixed-func
TCS. The problem is the fixed-func TCS was ignored by scratch update
functions, so LS didn't have the scratch buffer set up.

Note that Mesa 17.1 doesn't have merged shaders.
---
 src/gallium/drivers/radeonsi/si_state_shaders.c | 21 ++---
 1 file changed, 18 insertions(+), 3 deletions(-)

diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c 
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 5cbb91b..631272e 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -2704,58 +2704,73 @@ static int si_update_scratch_buffer(struct si_context 
*sctx,
 static unsigned si_get_current_scratch_buffer_size(struct si_context *sctx)
 {
return sctx->scratch_buffer ? sctx->scratch_buffer->b.b.width0 : 0;
 }
 
 static unsigned si_get_scratch_buffer_bytes_per_wave(struct si_shader *shader)
 {
return shader ? shader->config.scratch_bytes_per_wave : 0;
 }
 
+static struct si_shader *si_get_tcs_current(struct si_context *sctx)
+{
+   if (!sctx->tes_shader.cso)
+   return NULL; /* tessellation disabled */
+
+   return sctx->tcs_shader.cso ? sctx->tcs_shader.current :
+ sctx->fixed_func_tcs_shader.current;
+}
+
 static unsigned si_get_max_scratch_bytes_per_wave(struct si_context *sctx)
 {
unsigned bytes = 0;
 
bytes = MAX2(bytes, 
si_get_scratch_buffer_bytes_per_wave(sctx->ps_shader.current));
bytes = MAX2(bytes, 
si_get_scratch_buffer_bytes_per_wave(sctx->gs_shader.current));
bytes = MAX2(bytes, 
si_get_scratch_buffer_bytes_per_wave(sctx->vs_shader.current));
-   bytes = MAX2(bytes, 
si_get_scratch_buffer_bytes_per_wave(sctx->tcs_shader.current));
bytes = MAX2(bytes, 
si_get_scratch_buffer_bytes_per_wave(sctx->tes_shader.current));
+
+   if (sctx->tes_shader.cso) {
+   struct si_shader *tcs = si_get_tcs_current(sctx);
+
+   bytes = MAX2(bytes, si_get_scratch_buffer_bytes_per_wave(tcs));
+   }
return bytes;
 }
 
 static bool si_update_scratch_relocs(struct si_context *sctx)
 {
+   struct si_shader *tcs = si_get_tcs_current(sctx);
int r;
 
/* Update the shaders, so that they are using the latest scratch.
 * The scratch buffer may have been changed since these shaders were
 * last used, so we still need to try to update them, even if they
 * require scratch buffers smaller than the current size.
 */
r = si_update_scratch_buffer(sctx, sctx->ps_shader.current);
if (r < 0)
return false;
if (r == 1)
si_pm4_bind_state(sctx, ps, sctx->ps_shader.current->pm4);
 
r = si_update_scratch_buffer(sctx, sctx->gs_shader.current);
if (r < 0)
return false;
if (r == 1)
si_pm4_bind_state(sctx, gs, sctx->gs_shader.current->pm4);
 
-   r = si_update_scratch_buffer(sctx, sctx->tcs_shader.current);
+   r = si_update_scratch_buffer(sctx, tcs);
if (r < 0)
return false;
if (r == 1)
-   si_pm4_bind_state(sctx, hs, sctx->tcs_shader.current->pm4);
+   si_pm4_bind_state(sctx, hs, tcs->pm4);
 
/* VS can be bound as LS, ES, or VS. */
r = si_update_scratch_buffer(sctx, sctx->vs_shader.current);
if (r < 0)
return false;
if (r == 1) {
if (sctx->tes_shader.current)
si_pm4_bind_state(sctx, ls, 
sctx->vs_shader.current->pm4);
else if (sctx->gs_shader.current)
si_pm4_bind_state(sctx, es, 
sctx->vs_shader.current->pm4);
-- 
2.7.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [Bug 101071] compiling glsl fails with undefined reference to `pthread_create'

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=101071

--- Comment #11 from Emil Velikov  ---
(In reply to warpme from comment #10)
> Emil,
> Unfortunately minimizing build options doesn't help me with successful
> bisecting.
Minimising the build options was aimed to a) reduce the surface area and b)
speed up the build, hence the bisection.

> I'm always failing in the middle with following erorr:

> python  ./nir/nir_opcodes_c.py > nir/nir_opcodes.c || (rm -f
> nir/nir_opcodes.c; false)
> Traceback (most recent call last):
>   File "./nir/nir_builder_opcodes_h.py", line 46, in 
> from nir_opcodes import opcodes
These python errors seem new and unrelated. I would simply "git bisect skip"
them.

To isolate any other bugs you could do:
$ git bisect start .
$ build & test
$ git bisect good/bad
$ git clean -fxd // warning: it purge _all_ files that are not checked in git
$ rm -rf $builddir // if building out of tree
$ build & test
...

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [Bug 100613] Regression in Mesa 17 on s390x (zSystems)

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=100613

--- Comment #34 from Stefan Dirsch  ---
(In reply to intermedi...@hotmail.com from comment #33)
> Hi Stefan sorry for asking this small ot,
> how is the situation there on RadeonSi Mesa EGL?  
> here on Qoriq and on PowerMac G5  glamor crashes if invoche egl_radeonsi and
> made the Xorg not usable.
> There there is the same situation?

I haven't heard of such issues yet.

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 11/12] i965/cnl: Properly handle l3 configuration

[Anuj Phogat]

On Thu, May 11, 2017 at 8:31 AM, Ben Widawsky  wrote:
> On 17-05-02 11:51:28, Francisco Jerez wrote:
>>
>> Anuj Phogat  writes:
>>
>>> On Mon, Apr 24, 2017 at 9:15 PM, Ben Widawsky  wrote:
>>>
 On 17-04-18 18:18:39, Francisco Jerez wrote:

 Most, if not all of the unrelated changes that snuck in were due to
 rebase.
 Anuj, would you mind fixing those? I tried my best to address the rest,
 but I'm
 admittedly stumbling my way through some of the l3 programming.

 Anuj Phogat  writes:
>
>
> From: Ben Widawsky 
>>
>>
>> V2: Squash the changes in one patch and rebased on master (Anuj).
>>
>> Signed-off-by: Ben Widawsky 
>> Signed-off-by: Anuj Phogat 
>> ---
>>  src/intel/common/gen_l3_config.c | 43 ++
>> --
>>  1 file changed, 37 insertions(+), 6 deletions(-)
>>
>> diff --git a/src/intel/common/gen_l3_config.c
>> b/src/intel/common/gen_l3_config.c
>> index 4fe3503..f3e8793 100644
>> --- a/src/intel/common/gen_l3_config.c
>> +++ b/src/intel/common/gen_l3_config.c
>> @@ -102,6 +102,26 @@ static const struct gen_l3_config
>> chv_l3_configs[]
>> = {
>>  };
>>
>>  /**
>> + * On CNL, RO clients are merged and shared with read/write space. As
>> a
>> result
>> + * we have fewer allocation parameters.
>>
>
> The two sentences above make it sound like RO clients haven't been part
> of the same partition until CNL.  They have.  I'd drop this.
>
>
 So the difference I was trying to spell out is that the previous "IS"
 "C"
 and
 "T" fields do not exist in a programmable way.

 Also, programming does not require any
>>
>> + * back scaling. Programming simply works in 2k increments and is
>> scaled by the
>> + * hardware.
>>
>
> That's basically the case (up to the specific scale factor) on all
> hardware, I'd drop this too.
>
>
 I personally think the existing code isn't as self-documenting to me as
 it
 is to
 you, and so I was trying to spell it out. I was trying to document, not
 show
 differentiation. In either event, I don't care if we keep this or leave
 it.

 + */
>>
>> +static const struct gen_l3_config cnl_l3_configs[] = {
>> +   /* SLM URB Rest  DC  RO */
>>
>
> s/Rest/ALL/ (these are L3 partition enum labels), and align to the
> column boundaries below.
>
>
 Sure.


 +   {{  0, 64, 64,  0,  0 }},
>>
>> +   {{  0, 64,  0, 16, 48 }},
>> +   {{  0, 48,  0, 16, 64 }},
>> +   {{  0, 32,  0,  0, 96 }},
>> +   {{  0, 32, 96,  0,  0 }},
>> +   {{  0, 32,  0, 16, 80 }},
>> +   {{ 32, 16, 80,  0,  0 }},
>> +   {{ 32, 16,  0, 64, 16 }},
>> +   {{ 32,  0, 96,  0,  0 }},
>> +   {{ 0 }}
>> +};
>> +
>> +/**
>>   * Return a zero-terminated array of validated L3 configurations for
>> the
>>   * specified device.
>>   */
>> @@ -116,9 +136,11 @@ get_l3_configs(const struct gen_device_info
>> *devinfo)
>>return (devinfo->is_cherryview ? chv_l3_configs :
>> bdw_l3_configs);
>>
>> case 9:
>> -   case 10:
>>return chv_l3_configs;
>>
>> +   case 10:
>> +  return cnl_l3_configs;
>> +
>> default:
>>unreachable("Not implemented");
>> }
>> @@ -258,13 +280,19 @@ get_l3_way_size(const struct gen_device_info
>> *devinfo)
>> if (devinfo->is_baytrail)
>>return 2;
>>
>> -   else if (devinfo->gt == 1 ||
>> -devinfo->is_cherryview ||
>> -devinfo->is_broxton)
>>
>
> Unrelated change sneaked in.
>
>
 See above reply (not sure how this got in other than rebase).

 +   /* Way size is actually 6 * num_slices, because it's 2k per bank,
 and
>>
>> +* normally 3 banks per slice. However, on CNL+ this information
>> isn't
>> +* needed to setup the URB/l3 configuration. We fudge the answer
>> here
>> +* and then use the scaling to fix it up later.
>> +*/
>>
>
> The comment makes it sound like you're lying to the caller and
> returning
> a bogus way size you're going to fix up later.  That's not the case
> though, the value you're returning below is accurate for all CNL
> configs.  6 * num_slices OTOH *would* be inaccurate.  I'd drop the
> comment.
>
>
 Anuj, would you mind doing what Curro asks?

 +   if (devinfo->gen >= 10)
>>
>> +  return 2 * devinfo->l3_banks;
>> +
>>
>
> It would be nice if we could use the 'l3_banks' devinfo field you just
> added on previous gens too in order to simplify things.  I'm okay if
> you
> don't feel like doing the clean up right away but maybe add a short
> FINIS

Re: [Mesa-dev] [Mesa-stable] [PATCH 6/6] i965/dri: Fix bad GL error in intel_create_winsys_renderbuffer()

[Chad Versace]

On Wed 31 May 2017, Ian Romanick wrote:
> On 05/31/2017 04:43 PM, Chad Versace wrote:
> > This function never occurs in the callchain of a GL function. It occurs
> > only in the callchain of eglCreate*Surface and the analogous paths for
> > GLX.  Therefore, even if a  thread does have a bound GL context,
> > emitting a GL error here is wrong. A misplaced GL error, when no GL
> > call is made, can confuse clients.
> 
> This seems right.  What seems wrong, however, is that the callers ignore
> the potentially NULL return.  intelCreateBuffer (intel_screen.c) could
> easily return false when it gets NULL, but it merrily plugs along.

Yes, that seems to be a problem throughout the DRI layer. iirc, some DRI
vfunc signatures even lack return codes. I'll explore the
intelCreateBuffer callgraph and reply with some follow-up patches.

> This patch clearly makes things better, but I'd love to see a follow up
> that fixes the other pre-existing problems.
> 
> Reviewed-by: Ian Romanick 

Thanks.

> > 
> > Cc: mesa-sta...@lists.freedesktop.org
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH V2] i965: Rename brw_format variable names to isl_format

[Chad Versace]

On Thu 01 Jun 2017, Jason Ekstrand wrote:
> On Thu, Jun 1, 2017 at 9:41 AM, Anuj Phogat  wrote:
> 
> > On Tue, May 23, 2017 at 2:35 PM, Anuj Phogat 
> > wrote:
> > > This patch makes non functional changes. Renaming is just to
> > > make the code more readable.
> > >
> > > V2: update the types to "enum isl_format"
> > Jason, do you have any other questions ? r-b ?
> >
> 
> Patch 1/6 of the series Chad sent out yesterday also does this and a bit
> more.  I don't think he saw yours.

Right. I didn't see Anuj's patch.

Anuj, your patch looks good to me, and it's
Reviewed-by: Chad Versace 

If your patch lands before mine, it's no trouble to rebase my patches.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH V2] i965: Rename brw_format variable names to isl_format

[Anuj Phogat]

On Thu, Jun 1, 2017 at 12:00 PM, Chad Versace  wrote:
> On Thu 01 Jun 2017, Jason Ekstrand wrote:
>> On Thu, Jun 1, 2017 at 9:41 AM, Anuj Phogat  wrote:
>>
>> > On Tue, May 23, 2017 at 2:35 PM, Anuj Phogat 
>> > wrote:
>> > > This patch makes non functional changes. Renaming is just to
>> > > make the code more readable.
>> > >
>> > > V2: update the types to "enum isl_format"
>> > Jason, do you have any other questions ? r-b ?
>> >
>>
>> Patch 1/6 of the series Chad sent out yesterday also does this and a bit
>> more.  I don't think he saw yours.
>
> Right. I didn't see Anuj's patch.
>
> Anuj, your patch looks good to me, and it's
> Reviewed-by: Chad Versace 
>
> If your patch lands before mine, it's no trouble to rebase my patches.
Thanks Chad.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [Bug 100988] glXGetCurrentDisplay() no longer works for FakeGLX contexts?

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=100988

--- Comment #3 from Brian Paul  ---
(In reply to Tom Hudson from comment #2)
> Awesome, unless I'm missing something this fixes all our tests, and is a lot
> simpler than anything we'd tried.

Great.  I'll put Tested-by: Tom Hudson  on the patch
and push it soon.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 4/6] i965: Replace 0 with ISL_FORMAT_UNSUPPORTED in format table

[Chad Versace]

On Wed 31 May 2017, Jason Ekstrand wrote:
> On May 31, 2017 9:32:23 PM Ian Romanick  wrote:
> 
> > Having the unsupported format value not be zero isn't very safe.  The
> > C99 rules say that any field missing an initializer is implicitly
> > initialized to zero.  If a MESA_FORMAT_ value is added but is not added
> > to the array initializer, we'll have this same problem... but in a way
> > that is much harder to detect.
> 
> Good point.  Chad, how would you feel about leaving things zero and then
> having a loop that walks the array, knows about the one special case, and
> sets all unsupported things to ISL_FORMAT_UNSUPPORTED?  Then we get the best
> of both worlds.

My patch is garbage. We all agree.

I prefer Matt's approach that preserves the array's static constness. If
that doesn't work, then I'll write a loop. I'll reply back after testing
it.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH] i965: Replace 0 with ISL_FORMAT_UNSUPPORTED in format table (v2)

[Chad Versace]

From: Chad Versace 

When given an *unsupported* mesa_format,
brw_isl_format_for_mesa_format() returned 0, a *valid* isl_format,
ISL_FORMAT_R32G32B32A32_FLOAT.  The problem is that
brw_isl_format_for_mesa_format's inner table used 0 instead of
ISL_FORMAT_UNSUPPORTED to indicate unsupported mesa formats.

Some callers of brw_isl_format_for_mesa_format() were aware of this
weirdness, and worked around it. This patch removes those workarounds.

Tested on Broadwell as below, no regressions:

> deqp-egl --deqp-case='dEQP-EGL.functional.image.modify.*'
Test run totals:
  Passed:24/37 (64.9%)
  Failed:0/37 (0.0%)
  Not supported: 13/37 (35.1%)
  Warnings:  0/37 (0.0%)

> deqp-gles3 --deqp-case='dEQP-GLES3.functional.texture.format.*'
Test run totals:
  Passed:530/530 (100.0%)
  Failed:0/530 (0.0%)
  Not supported: 0/530 (0.0%)
  Warnings:  0/530 (0.0%)


v2: Ensure that all array elements are initialized to
  ISL_FORMAT_UNSUPPORTED, even when new formats are added to enum
  mesa_format, by using an designated range initializer.
---

 src/mesa/drivers/dri/i965/brw_surface_formats.c  | 96 ++--
 src/mesa/drivers/dri/i965/brw_wm_surface_state.c |  2 +-
 2 files changed, 6 insertions(+), 92 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_surface_formats.c 
b/src/mesa/drivers/dri/i965/brw_surface_formats.c
index 52d3acb..f878317 100644
--- a/src/mesa/drivers/dri/i965/brw_surface_formats.c
+++ b/src/mesa/drivers/dri/i965/brw_surface_formats.c
@@ -36,31 +36,19 @@ brw_isl_format_for_mesa_format(mesa_format mesa_format)
 * staying in sync, so we initialize to 0 even though
 * ISL_FORMAT_R32G32B32A32_FLOAT happens to also be 0.
 */
-   static const uint32_t table[MESA_FORMAT_COUNT] =
-   {
-  [MESA_FORMAT_A8B8G8R8_UNORM] = 0,
+   static const enum isl_format table[MESA_FORMAT_COUNT] = {
+  [0 ... MESA_FORMAT_COUNT-1] = ISL_FORMAT_UNSUPPORTED,
+
   [MESA_FORMAT_R8G8B8A8_UNORM] = ISL_FORMAT_R8G8B8A8_UNORM,
   [MESA_FORMAT_B8G8R8A8_UNORM] = ISL_FORMAT_B8G8R8A8_UNORM,
-  [MESA_FORMAT_A8R8G8B8_UNORM] = 0,
-  [MESA_FORMAT_X8B8G8R8_UNORM] = 0,
   [MESA_FORMAT_R8G8B8X8_UNORM] = ISL_FORMAT_R8G8B8X8_UNORM,
   [MESA_FORMAT_B8G8R8X8_UNORM] = ISL_FORMAT_B8G8R8X8_UNORM,
-  [MESA_FORMAT_X8R8G8B8_UNORM] = 0,
-  [MESA_FORMAT_BGR_UNORM8] = 0,
   [MESA_FORMAT_RGB_UNORM8] = ISL_FORMAT_R8G8B8_UNORM,
   [MESA_FORMAT_B5G6R5_UNORM] = ISL_FORMAT_B5G6R5_UNORM,
-  [MESA_FORMAT_R5G6B5_UNORM] = 0,
   [MESA_FORMAT_B4G4R4A4_UNORM] = ISL_FORMAT_B4G4R4A4_UNORM,
-  [MESA_FORMAT_A4R4G4B4_UNORM] = 0,
-  [MESA_FORMAT_A1B5G5R5_UNORM] = 0,
   [MESA_FORMAT_B5G5R5A1_UNORM] = ISL_FORMAT_B5G5R5A1_UNORM,
-  [MESA_FORMAT_A1R5G5B5_UNORM] = 0,
-  [MESA_FORMAT_L4A4_UNORM] = 0,
   [MESA_FORMAT_L8A8_UNORM] = ISL_FORMAT_L8A8_UNORM,
-  [MESA_FORMAT_A8L8_UNORM] = 0,
   [MESA_FORMAT_L16A16_UNORM] = ISL_FORMAT_L16A16_UNORM,
-  [MESA_FORMAT_A16L16_UNORM] = 0,
-  [MESA_FORMAT_B2G3R3_UNORM] = 0,
   [MESA_FORMAT_A_UNORM8] = ISL_FORMAT_A8_UNORM,
   [MESA_FORMAT_A_UNORM16] = ISL_FORMAT_A16_UNORM,
   [MESA_FORMAT_L_UNORM8] = ISL_FORMAT_L8_UNORM,
@@ -71,29 +59,16 @@ brw_isl_format_for_mesa_format(mesa_format mesa_format)
   [MESA_FORMAT_YCBCR] = ISL_FORMAT_YCRCB_SWAPUVY,
   [MESA_FORMAT_R_UNORM8] = ISL_FORMAT_R8_UNORM,
   [MESA_FORMAT_R8G8_UNORM] = ISL_FORMAT_R8G8_UNORM,
-  [MESA_FORMAT_G8R8_UNORM] = 0,
   [MESA_FORMAT_R_UNORM16] = ISL_FORMAT_R16_UNORM,
   [MESA_FORMAT_R16G16_UNORM] = ISL_FORMAT_R16G16_UNORM,
-  [MESA_FORMAT_G16R16_UNORM] = 0,
   [MESA_FORMAT_B10G10R10A2_UNORM] = ISL_FORMAT_B10G10R10A2_UNORM,
-  [MESA_FORMAT_S8_UINT_Z24_UNORM] = 0,
-  [MESA_FORMAT_Z24_UNORM_S8_UINT] = 0,
-  [MESA_FORMAT_Z_UNORM16] = 0,
-  [MESA_FORMAT_Z24_UNORM_X8_UINT] = 0,
-  [MESA_FORMAT_X8_UINT_Z24_UNORM] = 0,
-  [MESA_FORMAT_Z_UNORM32] = 0,
   [MESA_FORMAT_S_UINT8] = ISL_FORMAT_R8_UINT,
 
-  [MESA_FORMAT_BGR_SRGB8] = 0,
-  [MESA_FORMAT_A8B8G8R8_SRGB] = 0,
   [MESA_FORMAT_B8G8R8A8_SRGB] = ISL_FORMAT_B8G8R8A8_UNORM_SRGB,
-  [MESA_FORMAT_A8R8G8B8_SRGB] = 0,
   [MESA_FORMAT_R8G8B8A8_SRGB] = ISL_FORMAT_R8G8B8A8_UNORM_SRGB,
-  [MESA_FORMAT_X8R8G8B8_SRGB] = 0,
   [MESA_FORMAT_B8G8R8X8_SRGB] = ISL_FORMAT_B8G8R8X8_UNORM_SRGB,
   [MESA_FORMAT_L_SRGB8] = ISL_FORMAT_L8_UNORM_SRGB,
   [MESA_FORMAT_L8A8_SRGB] = ISL_FORMAT_L8A8_UNORM_SRGB,
-  [MESA_FORMAT_A8L8_SRGB] = 0,
   [MESA_FORMAT_SRGB_DXT1] = ISL_FORMAT_BC1_UNORM_SRGB,
   [MESA_FORMAT_SRGBA_DXT1] = ISL_FORMAT_BC1_UNORM_SRGB,
   [MESA_FORMAT_SRGBA_DXT3] = ISL_FORMAT_BC2_UNORM_SRGB,
@@ -109,7 +84,6 @@ brw_isl_format_for_mesa_format(mesa_format mesa_format)
   [MESA_FORMAT_RGBA_FLOAT32] = ISL_FORMAT_R32G32B32A32_FLOAT,
   [MESA_FORMAT_RGBA_FLOAT16] = ISL_FORMAT_R16G16B16A16_FLOAT,
   [MESA_FORMAT_RGB_FLO

[Mesa-dev] [Bug 100613] Regression in Mesa 17 on s390x (zSystems)

[bugzilla-daemon]

https://bugs.freedesktop.org/show_bug.cgi?id=100613

--- Comment #35 from intermedi...@hotmail.com  ---
just for information 
here my story about with Michel Danze reply
https://bugs.freedesktop.org/show_bug.cgi?id=99859#c19

But there are many post on many big Endian hardware,
was only curious if on s390 there was the same issue or not.

Thanks
Luigi

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 2/2] i965: Simplify l3 way size computations

[Anuj Phogat]

By making use of l3_banks field in gen_device_info struct
l3_way_size for gen7+ = 2 * l3_banks.

Suggested-by: Francisco Jerez 
Signed-off-by: Anuj Phogat 
Cc: Francisco Jerez 
---
 src/intel/common/gen_l3_config.c | 21 ++---
 1 file changed, 2 insertions(+), 19 deletions(-)

diff --git a/src/intel/common/gen_l3_config.c b/src/intel/common/gen_l3_config.c
index 0783217..f0da0f4 100644
--- a/src/intel/common/gen_l3_config.c
+++ b/src/intel/common/gen_l3_config.c
@@ -249,24 +249,6 @@ gen_get_l3_config(const struct gen_device_info *devinfo,
 }
 
 /**
- * Return the size of an L3 way in KB.
- */
-static unsigned
-get_l3_way_size(const struct gen_device_info *devinfo)
-{
-   if (devinfo->is_baytrail)
-  return 2;
-
-   else if (devinfo->gt == 1 ||
-devinfo->is_cherryview ||
-devinfo->is_broxton)
-  return 4;
-
-   else
-  return 8 * devinfo->num_slices;
-}
-
-/**
  * Return the unit brw_context::urb::size is expressed in, in KB.  \sa
  * gen_device_info::urb::size.
  */
@@ -288,8 +270,9 @@ gen_get_l3_config_urb_size(const struct gen_device_info 
*devinfo,
 * allocation of the L3 data array to provide 3*384KB=1152KB for URB, but
 * only 1008KB of this will be used."
 */
+   const unsigned l3_way_size = 2 * devinfo->l3_banks; /* KB */
const unsigned max = (devinfo->gen == 9 ? 1008 : ~0);
-   return MIN2(max, cfg->n[GEN_L3P_URB] * get_l3_way_size(devinfo)) /
+   return MIN2(max, cfg->n[GEN_L3P_URB] * l3_way_size) /
   get_urb_size_scale(devinfo);
 }
 
-- 
2.9.3

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

[Mesa-dev] [PATCH 1/2] i965: Add and initialize l3_banks field for gen7+

[Anuj Phogat]

This new field helps simplify l3 way size computations
in next patch.

Suggested-by: Francisco Jerez 
Signed-off-by: Anuj Phogat 
Cc: Francisco Jerez 
---
 src/intel/common/gen_device_info.c | 23 ++-
 src/intel/common/gen_device_info.h |  1 +
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/src/intel/common/gen_device_info.c 
b/src/intel/common/gen_device_info.c
index 209b293..7c1f9b4 100644
--- a/src/intel/common/gen_device_info.c
+++ b/src/intel/common/gen_device_info.c
@@ -132,6 +132,7 @@ static const struct gen_device_info gen_device_info_snb_gt2 
= {
 static const struct gen_device_info gen_device_info_ivb_gt1 = {
GEN7_FEATURES, .is_ivybridge = true, .gt = 1,
.num_slices = 1,
+   .l3_banks = 2,
.max_vs_threads = 36,
.max_tcs_threads = 36,
.max_tes_threads = 36,
@@ -156,6 +157,7 @@ static const struct gen_device_info gen_device_info_ivb_gt1 
= {
 static const struct gen_device_info gen_device_info_ivb_gt2 = {
GEN7_FEATURES, .is_ivybridge = true, .gt = 2,
.num_slices = 1,
+   .l3_banks = 4,
.max_vs_threads = 128,
.max_tcs_threads = 128,
.max_tes_threads = 128,
@@ -180,6 +182,7 @@ static const struct gen_device_info gen_device_info_ivb_gt2 
= {
 static const struct gen_device_info gen_device_info_byt = {
GEN7_FEATURES, .is_baytrail = true, .gt = 1,
.num_slices = 1,
+   .l3_banks = 1,
.has_llc = false,
.max_vs_threads = 36,
.max_tcs_threads = 36,
@@ -211,6 +214,7 @@ static const struct gen_device_info gen_device_info_byt = {
 static const struct gen_device_info gen_device_info_hsw_gt1 = {
HSW_FEATURES, .gt = 1,
.num_slices = 1,
+   .l3_banks = 2,
.max_vs_threads = 70,
.max_tcs_threads = 70,
.max_tes_threads = 70,
@@ -235,6 +239,7 @@ static const struct gen_device_info gen_device_info_hsw_gt1 
= {
 static const struct gen_device_info gen_device_info_hsw_gt2 = {
HSW_FEATURES, .gt = 2,
.num_slices = 1,
+   .l3_banks = 4,
.max_vs_threads = 280,
.max_tcs_threads = 256,
.max_tes_threads = 280,
@@ -259,6 +264,7 @@ static const struct gen_device_info gen_device_info_hsw_gt2 
= {
 static const struct gen_device_info gen_device_info_hsw_gt3 = {
HSW_FEATURES, .gt = 3,
.num_slices = 2,
+   .l3_banks = 8,
.max_vs_threads = 280,
.max_tcs_threads = 256,
.max_tes_threads = 280,
@@ -299,6 +305,7 @@ static const struct gen_device_info gen_device_info_hsw_gt3 
= {
 static const struct gen_device_info gen_device_info_bdw_gt1 = {
GEN8_FEATURES, .gt = 1,
.num_slices = 1,
+   .l3_banks = 2,
.max_cs_threads = 42,
.urb = {
   .size = 192,
@@ -318,6 +325,7 @@ static const struct gen_device_info gen_device_info_bdw_gt1 
= {
 static const struct gen_device_info gen_device_info_bdw_gt2 = {
GEN8_FEATURES, .gt = 2,
.num_slices = 1,
+   .l3_banks = 4,
.max_cs_threads = 56,
.urb = {
   .size = 384,
@@ -337,6 +345,7 @@ static const struct gen_device_info gen_device_info_bdw_gt2 
= {
 static const struct gen_device_info gen_device_info_bdw_gt3 = {
GEN8_FEATURES, .gt = 3,
.num_slices = 2,
+   .l3_banks = 8,
.max_cs_threads = 56,
.urb = {
   .size = 384,
@@ -357,6 +366,7 @@ static const struct gen_device_info gen_device_info_chv = {
GEN8_FEATURES, .is_cherryview = 1, .gt = 1,
.has_llc = false,
.num_slices = 1,
+   .l3_banks = 2,
.max_vs_threads = 80,
.max_tcs_threads = 80,
.max_tes_threads = 80,
@@ -413,6 +423,7 @@ static const struct gen_device_info gen_device_info_chv = {
.gt = 1,\
.has_llc = false,   \
.num_slices = 1,\
+   .l3_banks = 2,  \
.max_vs_threads = 112,  \
.max_tcs_threads = 112, \
.max_tes_threads = 112, \
@@ -457,22 +468,26 @@ static const struct gen_device_info gen_device_info_chv = 
{
 static const struct gen_device_info gen_device_info_skl_gt1 = {
GEN9_FEATURES, .gt = 1,
.num_slices = 1,
+   .l3_banks = 2,
.urb.size = 192,
 };
 
 static const struct gen_device_info gen_device_info_skl_gt2 = {
GEN9_FEATURES, .gt = 2,
.num_slices = 1,
+   .l3_banks = 4,
 };
 
 static const struct gen_device_info gen_device_info_skl_gt3 = {
GEN9_FEATURES, .gt = 3,
.num_slices = 2,
+   .l3_banks = 8,
 };
 
 static const struct gen_device_info gen_device_info_skl_gt4 = {
GEN9_FEATURES, .gt = 4,
.num_slices = 3,
+   .l3_banks = 12,
/* From the "L3 Allocation and Programming" documentation:
 *
 * "URB is limited to 1008KB due to programming restrictions.  This is not a
@@ -489,7 +504,8 @@ static const struct gen_device_info gen_device_info_bxt = {
 };
 
 static const struct gen_device_info gen_device_info_bxt_2x6 = {
-   GEN9_LP_FEATURES_2X6
+   GEN9_LP_FEATURES_2X6,
+   .l3_banks = 1,
 };
 /*
  * Note: for all KBL SKUs, the PRM says SKL for

[Mesa-dev] [PATCH 1/1] automake: r600 should only depend on libamd_common if opencl is enabled

[Jan Vesely]

Signed-off-by: Jan Vesely 
---
Hi guys,

this is the first step towards dropping libamd_common dependency.
It's based on Emil's patches 3/5 and 4/5.
Enabling opencl still falls back to the old way of requiring libamd_common.
I'll try to address that in the next step (no time estimate, feel free to beat 
me to it). I think we can drop part of those functions rather than just copying 
them.

Emil, I didn't find anything in android build that enabled opencl, so I dropped 
it entirely.

regards,
Jan

 configure.ac| 3 ++-
 src/gallium/drivers/r600/Android.mk | 5 -
 src/gallium/drivers/r600/Automake.inc   | 2 +-
 src/gallium/targets/pipe-loader/Makefile.am | 2 +-
 4 files changed, 4 insertions(+), 8 deletions(-)

diff --git a/configure.ac b/configure.ac
index 5caf316..e0996a0 100644
--- a/configure.ac
+++ b/configure.ac
@@ -2631,7 +2631,8 @@ AM_CONDITIONAL(HAVE_SWRAST_DRI, test x$HAVE_SWRAST_DRI = 
xyes)
 AM_CONDITIONAL(HAVE_RADEON_VULKAN, test "x$HAVE_RADEON_VULKAN" = xyes)
 AM_CONDITIONAL(HAVE_INTEL_VULKAN, test "x$HAVE_INTEL_VULKAN" = xyes)
 
-AM_CONDITIONAL(HAVE_AMD_DRIVERS, test "x$HAVE_GALLIUM_R600" = xyes -o \
+AM_CONDITIONAL(HAVE_AMD_DRIVERS, test \( "x$HAVE_GALLIUM_R600" = xyes -a \
+  "x$enable_opencl" = xyes \) -o \
   "x$HAVE_GALLIUM_RADEONSI" = xyes -o \
   "x$HAVE_RADEON_VULKAN" = xyes)
 
diff --git a/src/gallium/drivers/r600/Android.mk 
b/src/gallium/drivers/r600/Android.mk
index 87f433d..18c5bb6 100644
--- a/src/gallium/drivers/r600/Android.mk
+++ b/src/gallium/drivers/r600/Android.mk
@@ -30,11 +30,7 @@ include $(CLEAR_VARS)
 
 LOCAL_SRC_FILES := $(C_SOURCES) $(CXX_SOURCES)
 
-ifeq ($(MESA_ENABLE_LLVM),true)
-LOCAL_STATIC_LIBRARIES := libmesa_amd_common
-else
 LOCAL_C_INCLUDES += $(MESA_TOP)/src/amd/common
-endif
 
 LOCAL_SHARED_LIBRARIES := libdrm_radeon
 LOCAL_MODULE := libmesa_pipe_r600
@@ -45,7 +41,6 @@ include $(BUILD_STATIC_LIBRARY)
 ifneq ($(HAVE_GALLIUM_R600),)
 $(eval GALLIUM_LIBS += \
$(LOCAL_MODULE) \
-   $(LOCAL_STATIC_LIBRARIES) \
libmesa_winsys_radeon)
 $(eval GALLIUM_SHARED_LIBS += $(LOCAL_SHARED_LIBRARIES))
 endif
diff --git a/src/gallium/drivers/r600/Automake.inc 
b/src/gallium/drivers/r600/Automake.inc
index fa45735..c96fe74 100644
--- a/src/gallium/drivers/r600/Automake.inc
+++ b/src/gallium/drivers/r600/Automake.inc
@@ -13,7 +13,7 @@ TARGET_RADEON_WINSYS = \
 TARGET_RADEON_COMMON = \
$(top_builddir)/src/gallium/drivers/radeon/libradeon.la
 
-if HAVE_GALLIUM_LLVM
+if HAVE_AMD_DRIVERS
 TARGET_RADEON_COMMON += \
$(top_builddir)/src/amd/common/libamd_common.la
 endif
diff --git a/src/gallium/targets/pipe-loader/Makefile.am 
b/src/gallium/targets/pipe-loader/Makefile.am
index 5f629a2..c991533 100644
--- a/src/gallium/targets/pipe-loader/Makefile.am
+++ b/src/gallium/targets/pipe-loader/Makefile.am
@@ -131,7 +131,7 @@ pipe_r600_la_LIBADD = \
$(LIBDRM_LIBS) \
$(RADEON_LIBS)
 
-if HAVE_GALLIUM_LLVM
+if HAVE_AMD_DRIVERS
 pipe_r600_la_LIBADD += \
$(top_builddir)/src/amd/common/libamd_common.la
 endif
-- 
2.9.4

___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH] i965: Replace 0 with ISL_FORMAT_UNSUPPORTED in format table (v2)

[Matt Turner]

On Thu, Jun 1, 2017 at 12:55 PM, Chad Versace  wrote:
> From: Chad Versace 
>
> When given an *unsupported* mesa_format,
> brw_isl_format_for_mesa_format() returned 0, a *valid* isl_format,
> ISL_FORMAT_R32G32B32A32_FLOAT.  The problem is that
> brw_isl_format_for_mesa_format's inner table used 0 instead of
> ISL_FORMAT_UNSUPPORTED to indicate unsupported mesa formats.
>
> Some callers of brw_isl_format_for_mesa_format() were aware of this
> weirdness, and worked around it. This patch removes those workarounds.
>
> Tested on Broadwell as below, no regressions:
>
> > deqp-egl --deqp-case='dEQP-EGL.functional.image.modify.*'
> Test run totals:
>   Passed:24/37 (64.9%)
>   Failed:0/37 (0.0%)
>   Not supported: 13/37 (35.1%)
>   Warnings:  0/37 (0.0%)
>
> > deqp-gles3 --deqp-case='dEQP-GLES3.functional.texture.format.*'
> Test run totals:
>   Passed:530/530 (100.0%)
>   Failed:0/530 (0.0%)
>   Not supported: 0/530 (0.0%)
>   Warnings:  0/530 (0.0%)
>
>
> v2: Ensure that all array elements are initialized to
>   ISL_FORMAT_UNSUPPORTED, even when new formats are added to enum
>   mesa_format, by using an designated range initializer.
> ---
>
>  src/mesa/drivers/dri/i965/brw_surface_formats.c  | 96 
> ++--
>  src/mesa/drivers/dri/i965/brw_wm_surface_state.c |  2 +-
>  2 files changed, 6 insertions(+), 92 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_surface_formats.c 
> b/src/mesa/drivers/dri/i965/brw_surface_formats.c
> index 52d3acb..f878317 100644
> --- a/src/mesa/drivers/dri/i965/brw_surface_formats.c
> +++ b/src/mesa/drivers/dri/i965/brw_surface_formats.c
> @@ -36,31 +36,19 @@ brw_isl_format_for_mesa_format(mesa_format mesa_format)
>  * staying in sync, so we initialize to 0 even though
>  * ISL_FORMAT_R32G32B32A32_FLOAT happens to also be 0.
>  */
> -   static const uint32_t table[MESA_FORMAT_COUNT] =
> -   {
> -  [MESA_FORMAT_A8B8G8R8_UNORM] = 0,
> +   static const enum isl_format table[MESA_FORMAT_COUNT] = {
> +  [0 ... MESA_FORMAT_COUNT-1] = ISL_FORMAT_UNSUPPORTED,
> +

Awesome.

Reviewed-by: Matt Turner 
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Re: [Mesa-dev] [PATCH 1/1] automake: r600 should only depend on libamd_common if opencl is enabled

[Aaron Watry]

On Thu, Jun 1, 2017 at 3:28 PM, Jan Vesely  wrote:

> Signed-off-by: Jan Vesely 
> ---
> Hi guys,
>
> this is the first step towards dropping libamd_common dependency.
> It's based on Emil's patches 3/5 and 4/5.
> Enabling opencl still falls back to the old way of requiring libamd_common.
> I'll try to address that in the next step (no time estimate, feel free to
> beat me to it). I think we can drop part of those functions rather than
> just copying them.
>

I did a build test of r600 on my laptop with an old libdrm before/after
this patch with/without opencl enabled, and got:
before, no CL: Build Failure
after, no CL: Build succeeded
before, with CL:  Build Failure
after, with CL: Build Failure

I then went and updated libdrm/amdgpu, and the with/without CL build for
r600 works again there.  I then built this on my r600 box and
double-checked that clinfo and piglit's cl-program-tester ran and could
execute a simple kernel (tests/cl/program/execute/builtin/geometric/
length.cl), so I'd say that this is:
Tested-By: Aaron Watry 

And yes, I agree that we should remove the amdgpu dependency from
libamd_common (or at least the parts that r600 depends on), but I'd be ok
with doing that in a follow-up... That, or we also extend vinson's patch
[0] and wrap it in a check for if opencl is enabled so that we get a
configure-time failure if R600 is being built with CL and amdgpu's drm
headers are missing/old.

--Aaron

[0] https://lists.freedesktop.org/archives/mesa-dev/2017-May/157327.html


>
> Emil, I didn't find anything in android build that enabled opencl, so I
> dropped it entirely.
>
> regards,
> Jan
>
>  configure.ac| 3 ++-
>  src/gallium/drivers/r600/Android.mk | 5 -
>  src/gallium/drivers/r600/Automake.inc   | 2 +-
>  src/gallium/targets/pipe-loader/Makefile.am | 2 +-
>  4 files changed, 4 insertions(+), 8 deletions(-)
>
> diff --git a/configure.ac b/configure.ac
> index 5caf316..e0996a0 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -2631,7 +2631,8 @@ AM_CONDITIONAL(HAVE_SWRAST_DRI, test
> x$HAVE_SWRAST_DRI = xyes)
>  AM_CONDITIONAL(HAVE_RADEON_VULKAN, test "x$HAVE_RADEON_VULKAN" = xyes)
>  AM_CONDITIONAL(HAVE_INTEL_VULKAN, test "x$HAVE_INTEL_VULKAN" = xyes)
>
> -AM_CONDITIONAL(HAVE_AMD_DRIVERS, test "x$HAVE_GALLIUM_R600" = xyes -o \
> +AM_CONDITIONAL(HAVE_AMD_DRIVERS, test \( "x$HAVE_GALLIUM_R600" = xyes -a
> \
> +  "x$enable_opencl" = xyes \) -o \
>"x$HAVE_GALLIUM_RADEONSI" = xyes -o
> \
>"x$HAVE_RADEON_VULKAN" = xyes)
>
> diff --git a/src/gallium/drivers/r600/Android.mk
> b/src/gallium/drivers/r600/Android.mk
> index 87f433d..18c5bb6 100644
> --- a/src/gallium/drivers/r600/Android.mk
> +++ b/src/gallium/drivers/r600/Android.mk
> @@ -30,11 +30,7 @@ include $(CLEAR_VARS)
>
>  LOCAL_SRC_FILES := $(C_SOURCES) $(CXX_SOURCES)
>
> -ifeq ($(MESA_ENABLE_LLVM),true)
> -LOCAL_STATIC_LIBRARIES := libmesa_amd_common
> -else
>  LOCAL_C_INCLUDES += $(MESA_TOP)/src/amd/common
> -endif
>
>  LOCAL_SHARED_LIBRARIES := libdrm_radeon
>  LOCAL_MODULE := libmesa_pipe_r600
> @@ -45,7 +41,6 @@ include $(BUILD_STATIC_LIBRARY)
>  ifneq ($(HAVE_GALLIUM_R600),)
>  $(eval GALLIUM_LIBS += \
> $(LOCAL_MODULE) \
> -   $(LOCAL_STATIC_LIBRARIES) \
> libmesa_winsys_radeon)
>  $(eval GALLIUM_SHARED_LIBS += $(LOCAL_SHARED_LIBRARIES))
>  endif
> diff --git a/src/gallium/drivers/r600/Automake.inc
> b/src/gallium/drivers/r600/Automake.inc
> index fa45735..c96fe74 100644
> --- a/src/gallium/drivers/r600/Automake.inc
> +++ b/src/gallium/drivers/r600/Automake.inc
> @@ -13,7 +13,7 @@ TARGET_RADEON_WINSYS = \
>  TARGET_RADEON_COMMON = \
> $(top_builddir)/src/gallium/drivers/radeon/libradeon.la
>
> -if HAVE_GALLIUM_LLVM
> +if HAVE_AMD_DRIVERS
>  TARGET_RADEON_COMMON += \
> $(top_builddir)/src/amd/common/libamd_common.la
>  endif
> diff --git a/src/gallium/targets/pipe-loader/Makefile.am
> b/src/gallium/targets/pipe-loader/Makefile.am
> index 5f629a2..c991533 100644
> --- a/src/gallium/targets/pipe-loader/Makefile.am
> +++ b/src/gallium/targets/pipe-loader/Makefile.am
> @@ -131,7 +131,7 @@ pipe_r600_la_LIBADD = \
> $(LIBDRM_LIBS) \
> $(RADEON_LIBS)
>
> -if HAVE_GALLIUM_LLVM
> +if HAVE_AMD_DRIVERS
>  pipe_r600_la_LIBADD += \
> $(top_builddir)/src/amd/common/libamd_common.la
>  endif
> --
> 2.9.4
>
>
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev