Re: [Mesa-dev] [PATCH] nouveau: Use targ in nv50_ir_generate_code

2015-08-31 Thread Marcos Paulo de souza 

ping :)

Em 27-08-2015 12:59, Marcos Paulo de Souza escreveu:

instead of call prog->getTarget(), since the target never change in prog.

Signed-off-by: Marcos Paulo de Souza <marcos.souza@gmail.com>
---
  src/gallium/drivers/nouveau/codegen/nv50_ir.cpp | 6 +++---
  1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp 
b/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp
index cce6055..1dad098 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp
@@ -1195,7 +1195,7 @@ nv50_ir_generate_code(struct nv50_ir_prog_info *info)
prog->print();
  
 targ->parseDriverInfo(info);

-   prog->getTarget()->runLegalizePass(prog, nv50_ir::CG_STAGE_PRE_SSA);
+   targ->runLegalizePass(prog, nv50_ir::CG_STAGE_PRE_SSA);
  
 prog->convertToSSA();
  
@@ -1203,7 +1203,7 @@ nv50_ir_generate_code(struct nv50_ir_prog_info *info)

prog->print();
  
 prog->optimizeSSA(info->optLevel);

-   prog->getTarget()->runLegalizePass(prog, nv50_ir::CG_STAGE_SSA);
+   targ->runLegalizePass(prog, nv50_ir::CG_STAGE_SSA);
  
 if (prog->dbgFlags & NV50_IR_DEBUG_BASIC)

prog->print();
@@ -1212,7 +1212,7 @@ nv50_ir_generate_code(struct nv50_ir_prog_info *info)
ret = -4;
goto out;
 }
-   prog->getTarget()->runLegalizePass(prog, nv50_ir::CG_STAGE_POST_RA);
+   targ->runLegalizePass(prog, nv50_ir::CG_STAGE_POST_RA);
  
 prog->optimizePostRA(info->optLevel);
  


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

Re: [Mesa-dev] [PATCH] nouveau: Use targ in nv50_ir_generate_code

2015-08-31 Thread Marcos Paulo de souza 



Em 31-08-2015 13:31, Ilia Mirkin escreveu:

On Mon, Aug 31, 2015 at 12:26 PM, Tobias Klausmann
<tobias.johannes.klausm...@mni.thm.de> wrote:


On 31.08.2015 18:23, Ilia Mirkin wrote:

On Mon, Aug 31, 2015 at 12:21 PM, Tobias Klausmann
<tobias.johannes.klausm...@mni.thm.de> wrote:


On 31.08.2015 17:19, Marcos Paulo de souza wrote:

ping :)

Em 27-08-2015 12:59, Marcos Paulo de Souza escreveu:

instead of call prog->getTarget(), since the target never change in
prog.


It is not changing now, don't know if it will in the future. Using
prog->getTarget() may save us from hassling with this piece of code
again.
Plus it is more explicit to read in my eyes.

but... targ gets used *right above* each one of those instances.


Kill it alltogether, meaning lets use prog->getTarget() as a default
pattern.

Strongly disagree, it's shorter and clearer to just use 'targ' there,
instead of wondering what it is that prog->getTarget() does.
To keep everybody happy, maybe we can add some methods in prog and use 
the target member of program.


What do you think?


   -ilia


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

[Mesa-dev] [PATCH] nouveau: Use targ in nv50_ir_generate_code

2015-08-27 Thread Marcos Paulo de Souza 
instead of call prog-getTarget(), since the target never change in prog.

Signed-off-by: Marcos Paulo de Souza marcos.souza@gmail.com
---
 src/gallium/drivers/nouveau/codegen/nv50_ir.cpp | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp 
b/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp
index cce6055..1dad098 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir.cpp
@@ -1195,7 +1195,7 @@ nv50_ir_generate_code(struct nv50_ir_prog_info *info)
   prog-print();
 
targ-parseDriverInfo(info);
-   prog-getTarget()-runLegalizePass(prog, nv50_ir::CG_STAGE_PRE_SSA);
+   targ-runLegalizePass(prog, nv50_ir::CG_STAGE_PRE_SSA);
 
prog-convertToSSA();
 
@@ -1203,7 +1203,7 @@ nv50_ir_generate_code(struct nv50_ir_prog_info *info)
   prog-print();
 
prog-optimizeSSA(info-optLevel);
-   prog-getTarget()-runLegalizePass(prog, nv50_ir::CG_STAGE_SSA);
+   targ-runLegalizePass(prog, nv50_ir::CG_STAGE_SSA);
 
if (prog-dbgFlags  NV50_IR_DEBUG_BASIC)
   prog-print();
@@ -1212,7 +1212,7 @@ nv50_ir_generate_code(struct nv50_ir_prog_info *info)
   ret = -4;
   goto out;
}
-   prog-getTarget()-runLegalizePass(prog, nv50_ir::CG_STAGE_POST_RA);
+   targ-runLegalizePass(prog, nv50_ir::CG_STAGE_POST_RA);
 
prog-optimizePostRA(info-optLevel);
 
-- 
2.4.3

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

[Mesa-dev] [PATCH 1/3] tgsi: set implicit array size for tess stages

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

---
 src/gallium/auxiliary/tgsi/tgsi_text.c | 6 +-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c 
b/src/gallium/auxiliary/tgsi/tgsi_text.c
index a6675c5..24e2dbd 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -259,7 +259,7 @@ struct translate_ctx
struct tgsi_token *tokens_end;
struct tgsi_header *header;
unsigned processor : 4;
-   int implied_array_size : 5;
+   unsigned implied_array_size : 6;
unsigned num_immediates;
 };
 
@@ -1623,6 +1623,10 @@ static boolean translate( struct translate_ctx *ctx )
if (!parse_header( ctx ))
   return FALSE;
 
+   if (ctx-processor == TGSI_PROCESSOR_TESS_CTRL ||
+   ctx-processor == TGSI_PROCESSOR_TESS_EVAL)
+   ctx-implied_array_size = 32 ;
+
while (*ctx-cur != '\0') {
   uint label_val = 0;
   if (!eat_white( ctx-cur )) {
-- 
2.4.3

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

[Mesa-dev] [PATCH 3/3] nouveau: recognize tess stages in nouveau_compiler

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

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

diff --git a/src/gallium/drivers/nouveau/nouveau_compiler.c 
b/src/gallium/drivers/nouveau/nouveau_compiler.c
index 8660498..495450b 100644
--- a/src/gallium/drivers/nouveau/nouveau_compiler.c
+++ b/src/gallium/drivers/nouveau/nouveau_compiler.c
@@ -190,6 +190,10 @@ main(int argc, char *argv[])
   type = PIPE_SHADER_GEOMETRY;
else if (!strncmp(text, COMP, 4))
   type = PIPE_SHADER_COMPUTE;
+   else if (!strncmp(text, TESS_CTRL, 9))
+  type = PIPE_SHADER_TESS_CTRL;
+   else if (!strncmp(text, TESS_EVAL, 9))
+  type = PIPE_SHADER_TESS_EVAL;
else {
   _debug_printf(Unrecognized TGSI header\n);
   return 1;
-- 
2.4.3

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

[Mesa-dev] [PATCH 2/3] tgsi: Fix index printed in tgsi_dump and dst outputs

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

Before this patch, the tgsi_dumps was printing declaration as:
DCL IN[][0][0], GENERIC[0]

and now it is parsed correctly:
DCL IN[][0], GENERIC[0]

In the same way, for tess stages, the output addr now is parsed correctly, 
doing like src parser from:
LRP OUT[0][3], TEMP[1]., TEMP[3], TEMP[2]

to:
LRP OUT[ADDR[1].x][3], TEMP[1]., TEMP[3], TEMP[2]
---
 src/gallium/auxiliary/tgsi/tgsi_text.c | 16 +++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c 
b/src/gallium/auxiliary/tgsi/tgsi_text.c
index 24e2dbd..141a3b1 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -684,7 +684,13 @@ parse_register_dcl(
* input primitive. so we want to declare just
* the index relevant to the semantics which is in
* the second bracket */
-  if (ctx-processor == TGSI_PROCESSOR_GEOMETRY  *file == 
TGSI_FILE_INPUT) {
+
+  /* tessellation has similar constraints to geometry shader */
+  bool is_in = *file == TGSI_FILE_INPUT;
+  bool is_out = *file == TGSI_FILE_OUTPUT;
+  if ((ctx-processor == TGSI_PROCESSOR_GEOMETRY  is_i) ||
+  (ctx-processor == TGSI_PROCESSOR_TESS_EVAL  is_in) ||
+  (ctx-processor == TGSI_PROCESSOR_TESS_CTRL  (is_in || is_out))) {
  brackets[0] = brackets[1];
  *num_brackets = 1;
   } else {
@@ -740,6 +746,14 @@ parse_dst_operand(
   dst-Dimension.Indirect = 0;
   dst-Dimension.Dimension = 0;
   dst-Dimension.Index = bracket[0].index;
+
+  if (bracket[0].ind_file != TGSI_FILE_NULL) {
+ dst-Dimension.Indirect = 1;
+ dst-DimIndirect.File = bracket[0].ind_file;
+ dst-DimIndirect.Index = bracket[0].ind_index;
+ dst-DimIndirect.Swizzle = bracket[0].ind_comp;
+ dst-DimIndirect.ArrayID = bracket[0].ind_array;
+  }
   bracket[0] = bracket[1];
}
dst-Register.Index = bracket[0].index;
-- 
2.4.3

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

[Mesa-dev] [PATCH 3/3] nouveau: recognize tess stages in nouveau_compiler

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

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

diff --git a/src/gallium/drivers/nouveau/nouveau_compiler.c 
b/src/gallium/drivers/nouveau/nouveau_compiler.c
index 8660498..495450b 100644
--- a/src/gallium/drivers/nouveau/nouveau_compiler.c
+++ b/src/gallium/drivers/nouveau/nouveau_compiler.c
@@ -190,6 +190,10 @@ main(int argc, char *argv[])
   type = PIPE_SHADER_GEOMETRY;
else if (!strncmp(text, COMP, 4))
   type = PIPE_SHADER_COMPUTE;
+   else if (!strncmp(text, TESS_CTRL, 9))
+  type = PIPE_SHADER_TESS_CTRL;
+   else if (!strncmp(text, TESS_EVAL, 9))
+  type = PIPE_SHADER_TESS_EVAL;
else {
   _debug_printf(Unrecognized TGSI header\n);
   return 1;
-- 
2.4.3

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

[Mesa-dev] [PATCH 2/3] tgsi: Fix index printed in tgsi_dump and dst outputs

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

Before this patch, the tgsi_dumps was printing declaration as:
DCL IN[][0][0], GENERIC[0]

and now it is parsed correctly:
DCL IN[][0], GENERIC[0]

In the same way, for tess stages, the output addr now is parsed correctly, 
doing like src parser from:
LRP OUT[0][3], TEMP[1]., TEMP[3], TEMP[2]

to:
LRP OUT[ADDR[1].x][3], TEMP[1]., TEMP[3], TEMP[2]
---
 src/gallium/auxiliary/tgsi/tgsi_text.c | 16 +++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c 
b/src/gallium/auxiliary/tgsi/tgsi_text.c
index 24e2dbd..075f2cb 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -684,7 +684,13 @@ parse_register_dcl(
* input primitive. so we want to declare just
* the index relevant to the semantics which is in
* the second bracket */
-  if (ctx-processor == TGSI_PROCESSOR_GEOMETRY  *file == 
TGSI_FILE_INPUT) {
+
+  /* tessellation has similar constraints to geometry shader */
+  bool is_in = *file == TGSI_FILE_INPUT;
+  bool is_out = *file == TGSI_FILE_OUTPUT;
+  if ((ctx-processor == TGSI_PROCESSOR_GEOMETRY  is_in) ||
+  (ctx-processor == TGSI_PROCESSOR_TESS_EVAL  is_in) ||
+  (ctx-processor == TGSI_PROCESSOR_TESS_CTRL  (is_in || is_out))) {
  brackets[0] = brackets[1];
  *num_brackets = 1;
   } else {
@@ -740,6 +746,14 @@ parse_dst_operand(
   dst-Dimension.Indirect = 0;
   dst-Dimension.Dimension = 0;
   dst-Dimension.Index = bracket[0].index;
+
+  if (bracket[0].ind_file != TGSI_FILE_NULL) {
+ dst-Dimension.Indirect = 1;
+ dst-DimIndirect.File = bracket[0].ind_file;
+ dst-DimIndirect.Index = bracket[0].ind_index;
+ dst-DimIndirect.Swizzle = bracket[0].ind_comp;
+ dst-DimIndirect.ArrayID = bracket[0].ind_array;
+  }
   bracket[0] = bracket[1];
}
dst-Register.Index = bracket[0].index;
-- 
2.4.3

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

[Mesa-dev] [PATCHv2 0/3] Add tess stages to mesa/nouveau_compiler

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

Hi guys,

first of all, sorry for the flood... my last bunch of patches were sent as a 
mistake... but
now the compiling problem was solved :)

I'm happy o send my first patches to mesa, and I have to say a huge thank you 
to Ilia Mirkin for all the
help and tips to fix this problem in tess stages.

I hope you like these patches!

Marcos Paulo de Souza (3):
  tgsi: set implicit array size for tess stages
  tgsi: Fix index printed in tgsi_dump and dst outputs
  nouveau: recognize tess stages in nouveau_compiler

 src/gallium/auxiliary/tgsi/tgsi_text.c | 22 --
 src/gallium/drivers/nouveau/nouveau_compiler.c |  4 
 2 files changed, 24 insertions(+), 2 deletions(-)

-- 
2.4.3

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

[Mesa-dev] [PATCH 1/3] tgsi: set implicit array size for tess stages

2015-08-16 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

---
 src/gallium/auxiliary/tgsi/tgsi_text.c | 6 +-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c 
b/src/gallium/auxiliary/tgsi/tgsi_text.c
index a6675c5..24e2dbd 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -259,7 +259,7 @@ struct translate_ctx
struct tgsi_token *tokens_end;
struct tgsi_header *header;
unsigned processor : 4;
-   int implied_array_size : 5;
+   unsigned implied_array_size : 6;
unsigned num_immediates;
 };
 
@@ -1623,6 +1623,10 @@ static boolean translate( struct translate_ctx *ctx )
if (!parse_header( ctx ))
   return FALSE;
 
+   if (ctx-processor == TGSI_PROCESSOR_TESS_CTRL ||
+   ctx-processor == TGSI_PROCESSOR_TESS_EVAL)
+   ctx-implied_array_size = 32 ;
+
while (*ctx-cur != '\0') {
   uint label_val = 0;
   if (!eat_white( ctx-cur )) {
-- 
2.4.3

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

Re: [Mesa-dev] [PATCH] tgsi/nouveau: Add support for tesselation ctrl and tesselation eval

2015-08-16 Thread Marcos Paulo de souza 

Hi Ilia,

Em 14-08-2015 01:45, Ilia Mirkin escreveu:

On Fri, Aug 14, 2015 at 12:43 AM, Marcos Souza
marcos.souza@gmail.com wrote:

HI Ilia

2015-08-14 1:31 GMT-03:00 Ilia Mirkin imir...@alum.mit.edu:

On Fri, Aug 14, 2015 at 12:25 AM, Marcos Souza
marcos.souza@gmail.com wrote:

Hi Ilia,

2015-08-14 1:02 GMT-03:00 Ilia Mirkin imir...@alum.mit.edu:

On Thu, Aug 13, 2015 at 11:55 PM, Marcos Souza
marcos.souza@gmail.com wrote:

Hi Ilia,

So i found the point here it addrs that double brackets, and the
following
patch solves it, but this is a right solution? If someone could guide
me
here, I could fix it :)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c
b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index 8ceb5b4..046471e 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -302,10 +302,14 @@ iter_declaration(
TXT([]);
 }

-   if (decl-Declaration.Dimension) {

The issue is that the declaration is getting a dimension set by the
parser, which in turn is causing it to print funny. It shouldn't be
getting a dimension in the first place for those.


The following patch fix the problem, is it the right place to put it?

I don't think so. Just glanced at the code, look at

parse_register_dcl

   /* for geometry shader we don't really care about
* the first brackets it's always the size of the
* input primitive. so we want to declare just
* the index relevant to the semantics which is in
* the second bracket */
   if (ctx-processor == TGSI_PROCESSOR_GEOMETRY  *file ==
TGSI_FILE_INPUT) {
  brackets[0] = brackets[1];
  *num_brackets = 1;
   }

Basically you need to extend this logic to similarly exclude

(a) tess ctrl inputs and outputs
(b) tess eval inputs

Technically you need to exclude patch/tessinner/tessouter from that,
but in practice they won't have an extra set of brackets either.


Sorry for flooding the list, but I'm relaly excited about it :)

So, this is the change you asked. It also solved the problem:
diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c
b/src/gallium/auxiliary/tgsi/tgsi_text.c
index 8647e4e..95c1daf 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -684,7 +684,12 @@ parse_register_dcl(
 * input primitive. so we want to declare just
 * the index relevant to the semantics which is in
 * the second bracket */
-  if (ctx-processor == TGSI_PROCESSOR_GEOMETRY  *file ==
TGSI_FILE_INPUT) {
+
+  /* similarly from tessalation */

tessellation

OK.



+  int exclude = (ctx-processor == TGSI_PROCESSOR_TESS_EVAL  *file ==
TGSI_FILE_INPUT) ||
+  (ctx-processor == TGSI_PROCESSOR_TESS_CTRL  (*file ==
TGSI_FILE_INPUT ||
+ *file == TGSI_FILE_OUTPUT));

Why is this separate from the geometry thing?
I just separated because it's easier to read, since we have a lot of 
ANDs and ORs. But, if you think it's better, I could put it all together 
will geometry.



+  if ((ctx-processor == TGSI_PROCESSOR_GEOMETRY  *file ==
TGSI_FILE_INPUT) || exclude) {
   brackets[0] = brackets[1];
   *num_brackets = 1;
} else {

What do you think Ilia?

Generally sounds good.

Now I'll take a look about the last problem of LRP and MOV.


   -ilia


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

[Mesa-dev] [PATCH] tgsi/nouveau: Add support for tesselation ctrl and tesselation eval

2015-08-12 Thread Marcos Paulo de Souza 
From: Marcos Paulo de Souza marcos.souza@gmail.com

Signed-off-by: Marcos Paulo de Souza marcos.souza.org
Suggested-by: Ilia Mirkin imir...@alum.mit.edu
---
 src/gallium/auxiliary/tgsi/tgsi_text.c | 6 +-
 src/gallium/drivers/nouveau/nouveau_compiler.c | 4 
 2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c 
b/src/gallium/auxiliary/tgsi/tgsi_text.c
index a6675c5..8647e4e 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -259,7 +259,7 @@ struct translate_ctx
struct tgsi_token *tokens_end;
struct tgsi_header *header;
unsigned processor : 4;
-   int implied_array_size : 5;
+   int implied_array_size : 6;
unsigned num_immediates;
 };
 
@@ -1623,6 +1623,10 @@ static boolean translate( struct translate_ctx *ctx )
if (!parse_header( ctx ))
   return FALSE;
 
+   if (ctx-processor == TGSI_PROCESSOR_TESS_CTRL ||
+   ctx-processor == TGSI_PROCESSOR_TESS_EVAL)
+   ctx-implied_array_size = 32 ;
+
while (*ctx-cur != '\0') {
   uint label_val = 0;
   if (!eat_white( ctx-cur )) {
diff --git a/src/gallium/drivers/nouveau/nouveau_compiler.c 
b/src/gallium/drivers/nouveau/nouveau_compiler.c
index 8660498..495450b 100644
--- a/src/gallium/drivers/nouveau/nouveau_compiler.c
+++ b/src/gallium/drivers/nouveau/nouveau_compiler.c
@@ -190,6 +190,10 @@ main(int argc, char *argv[])
   type = PIPE_SHADER_GEOMETRY;
else if (!strncmp(text, COMP, 4))
   type = PIPE_SHADER_COMPUTE;
+   else if (!strncmp(text, TESS_CTRL, 9))
+  type = PIPE_SHADER_TESS_CTRL;
+   else if (!strncmp(text, TESS_EVAL, 9))
+  type = PIPE_SHADER_TESS_EVAL;
else {
   _debug_printf(Unrecognized TGSI header\n);
   return 1;
-- 
2.4.3

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