From 857ac1e817808f4b6bf985679162d0e3d709e5b5 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 26 Feb 2009 05:35:15 -0800 Subject: mesa: Resurrect SPARC asm code. This rewrites the sparc GLAPI code so that it's PIC friendly and works with all of the TLS/PTHREADS/64-bit/32-bit combinations properly. As a result we can turn SPARC asm back on. Currently it's only enabled on Linux, as that's the only place where I can test this stuff out. For the moment the cliptest SPARC asm routines are disabled as they are non-working. The problem is that they use register %g7 as a temporary which is where the threading libraries store the thread pointer on SPARC. I will fix that code up in a future change as it's a pretty important routine to optimize. Like x86 we do the runtime patch as a pthread once-invoked initializer in init_glapi_relocs(). Unlike x86, however, our GLAPI stubs on SPARC are just two instruction sequences that branch to a trampoline and put the GLAPI offset into a register. The trampoline is what we run-time patch. The stubs thus all look like: glFoo: ba __glapi_sparc_foo_stub sethi GLAPI_OFFSET(glFOO) * PTR_SIZE, %g3 This actually makes generate_entrypoint() a lot simpler on SPARC. For this case in generate_entrypoint() we generate stubs using a 'call' instead of the 'ba' above to make sure it can reach. In order to get a proper tail call going here, in the unpatched case, we do several tricks. To get the current PC, for example, we save the return address register into a temporary, do a call, save the return address register written by the call to another temporary, then restore the original return address register value. This is to avoid having to allocate a stack frame. This is necessary for PIC address formation. This new GLAPI scheme lets us get rid of the ugly SPARC GLAPI hacks in __glXInitialize() and one_time_init(). Signed-off-by: David S. Miller --- src/mesa/sparc/sparc.c | 38 ++------------------------------------ 1 file changed, 2 insertions(+), 36 deletions(-) (limited to 'src/mesa/sparc/sparc.c') diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c index 84e8ac6723..3bde98e34b 100644 --- a/src/mesa/sparc/sparc.c +++ b/src/mesa/sparc/sparc.c @@ -109,10 +109,10 @@ void _mesa_init_all_sparc_transform_asm(void) ASSIGN_XFORM_GROUP(sparc, 2) ASSIGN_XFORM_GROUP(sparc, 3) ASSIGN_XFORM_GROUP(sparc, 4) - +#if 0 _mesa_clip_tab[4] = _mesa_sparc_cliptest_points4; _mesa_clip_np_tab[4] = _mesa_sparc_cliptest_points4_np; - +#endif #if 0 /* disable these too. See bug 673938 */ _mesa_normal_tab[NORM_TRANSFORM | NORM_NORMALIZE] = @@ -140,38 +140,4 @@ void _mesa_init_all_sparc_transform_asm(void) #endif } -extern unsigned int _mesa_sparc_glapi_begin; -extern unsigned int _mesa_sparc_glapi_end; -extern void __glapi_sparc_icache_flush(unsigned int *); - -#endif /* USE_SPARC_ASM */ - - -void _mesa_init_sparc_glapi_relocs(void) -{ -#ifdef USE_SPARC_ASM - unsigned int *insn_ptr, *end_ptr; - unsigned long disp_addr; - - insn_ptr = &_mesa_sparc_glapi_begin; - end_ptr = &_mesa_sparc_glapi_end; - disp_addr = (unsigned long) &_glapi_Dispatch; - - while (insn_ptr < end_ptr) { -#ifdef __arch64__ - insn_ptr[0] |= (disp_addr >> (32 + 10)); - insn_ptr[1] |= ((disp_addr & 0xffffffff) >> 10); - __glapi_sparc_icache_flush(&insn_ptr[0]); - insn_ptr[2] |= ((disp_addr >> 32) & ((1 << 10) - 1)); - insn_ptr[3] |= (disp_addr & ((1 << 10) - 1)); - __glapi_sparc_icache_flush(&insn_ptr[2]); - insn_ptr += 11; -#else - insn_ptr[0] |= (disp_addr >> 10); - insn_ptr[1] |= (disp_addr & ((1 << 10) - 1)); - __glapi_sparc_icache_flush(&insn_ptr[0]); - insn_ptr += 5; -#endif - } #endif /* USE_SPARC_ASM */ -} -- cgit v1.2.3 From bde27b0d94dc2b3d0d4656e2c85ecbbd1245b3ac Mon Sep 17 00:00:00 2001 From: David Miller Date: Fri, 27 Feb 2009 02:38:37 -0800 Subject: mesa: Fix Sparc cliptest asm code and re-enable. Stop using register %g7 since that is used by the "system" (ie. the pthread implementation makes use of it). Also, the projection vector can be NULL and we shouldn't try to access it at all in _mesa_sparc_cliptest_points4_np(). ioquake3 would crash due to this bug. Finally, unconditionally emit the register directives and re-enable in _mesa_init_all_sparc_transform_asm(). Signed-off-by: David S. Miller --- src/mesa/sparc/clip.S | 29 +++++++++-------------------- src/mesa/sparc/sparc.c | 4 ++-- 2 files changed, 11 insertions(+), 22 deletions(-) (limited to 'src/mesa/sparc/sparc.c') diff --git a/src/mesa/sparc/clip.S b/src/mesa/sparc/clip.S index 58c228ed15..208843c606 100644 --- a/src/mesa/sparc/clip.S +++ b/src/mesa/sparc/clip.S @@ -25,12 +25,8 @@ #define VEC_SIZE_3 7 #define VEC_SIZE_4 15 -#if defined(SVR4) || defined(__SVR4) || defined(__svr4__) - /* Solaris requires this for 64-bit. */ .register %g2, #scratch .register %g3, #scratch - .register %g7, #scratch -#endif .text .align 64 @@ -78,7 +74,7 @@ _mesa_sparc_cliptest_points4: add %g1, 0x4, %g1 ld [%i0 + V4F_STRIDE], %l1 - ld [%i0 + V4F_COUNT], %g7 + ld [%i0 + V4F_COUNT], %l3 LDPTR [%i0 + V4F_START], %i0 LDPTR [%i1 + V4F_START], %i5 ldub [%i3], %g2 @@ -91,12 +87,12 @@ _mesa_sparc_cliptest_points4: st %g3, [%i1 + V4F_FLAGS] mov 3, %g3 st %g3, [%i1 + V4F_SIZE] - st %g7, [%i1 + V4F_COUNT] + st %l3, [%i1 + V4F_COUNT] clr %l2 clr %l0 /* l0: i - * g7: count + * l3: count * l1: stride * l2: c * g2: (tmpAndMask << 8) | tmpOrMask @@ -153,12 +149,12 @@ _mesa_sparc_cliptest_points4: 3: add %i5, 0x10, %i5 ! IEU1 add %l0, 1, %l0 ! IEU0 Group add %i2, 1, %i2 ! IEU0 Group - cmp %l0, %g7 ! IEU1 Group + cmp %l0, %l3 ! IEU1 Group bne 1b ! CTI add %i0, %l1, %i0 ! IEU0 Group stb %g2, [%i3] ! LSU srl %g2, 8, %g3 ! IEU0 Group - cmp %l2, %g7 ! IEU1 Group + cmp %l2, %l3 ! IEU1 Group bl,a 1f ! CTI clr %g3 ! IEU0 1: stb %g3, [%i4] ! LSU Group @@ -174,25 +170,18 @@ _mesa_sparc_cliptest_points4_np: add %g1, 0x4, %g1 ld [%i0 + V4F_STRIDE], %l1 - ld [%i0 + V4F_COUNT], %g7 + ld [%i0 + V4F_COUNT], %l3 LDPTR [%i0 + V4F_START], %i0 - LDPTR [%i1 + V4F_START], %i5 ldub [%i3], %g2 ldub [%i4], %g3 sll %g3, 8, %g3 or %g2, %g3, %g2 - ld [%i1 + V4F_FLAGS], %g3 - or %g3, VEC_SIZE_4, %g3 - st %g3, [%i1 + V4F_FLAGS] - mov 3, %g3 - st %g3, [%i1 + V4F_SIZE] - st %g7, [%i1 + V4F_COUNT] clr %l2 clr %l0 /* l0: i - * g7: count + * l3: count * l1: stride * l2: c * g2: (tmpAndMask << 8) | tmpOrMask @@ -230,12 +219,12 @@ _mesa_sparc_cliptest_points4_np: and %g2, %g4, %g2 ! IEU0 Group 2: add %l0, 1, %l0 ! IEU0 Group add %i2, 1, %i2 ! IEU0 Group - cmp %l0, %g7 ! IEU1 Group + cmp %l0, %l3 ! IEU1 Group bne 1b ! CTI add %i0, %l1, %i0 ! IEU0 Group stb %g2, [%i3] ! LSU srl %g2, 8, %g3 ! IEU0 Group - cmp %l2, %g7 ! IEU1 Group + cmp %l2, %l3 ! IEU1 Group bl,a 1f ! CTI clr %g3 ! IEU0 1: stb %g3, [%i4] ! LSU Group diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c index 3bde98e34b..fafb49b81c 100644 --- a/src/mesa/sparc/sparc.c +++ b/src/mesa/sparc/sparc.c @@ -109,10 +109,10 @@ void _mesa_init_all_sparc_transform_asm(void) ASSIGN_XFORM_GROUP(sparc, 2) ASSIGN_XFORM_GROUP(sparc, 3) ASSIGN_XFORM_GROUP(sparc, 4) -#if 0 + _mesa_clip_tab[4] = _mesa_sparc_cliptest_points4; _mesa_clip_np_tab[4] = _mesa_sparc_cliptest_points4_np; -#endif + #if 0 /* disable these too. See bug 673938 */ _mesa_normal_tab[NORM_TRANSFORM | NORM_NORMALIZE] = -- cgit v1.2.3 From cf89f063634ff89cbd732bf67950debc94897ed9 Mon Sep 17 00:00:00 2001 From: David Miller Date: Fri, 27 Feb 2009 02:38:40 -0800 Subject: mesa: Fix and re-enable sparc normal asm. Fix a bug reported in 2003 :-) The output vector has 4 entries, not 3. Unconditionally emit .register directives. Signed-off-by: David S. Miller --- src/mesa/sparc/norm.S | 25 +++++++++++-------------- src/mesa/sparc/sparc.c | 3 --- 2 files changed, 11 insertions(+), 17 deletions(-) (limited to 'src/mesa/sparc/sparc.c') diff --git a/src/mesa/sparc/norm.S b/src/mesa/sparc/norm.S index 44950a10a5..117d36fa22 100644 --- a/src/mesa/sparc/norm.S +++ b/src/mesa/sparc/norm.S @@ -1,11 +1,8 @@ #include "sparc_matrix.h" -#if defined(SVR4) || defined(__SVR4) || defined(__svr4__) - /* Solaris requires this for 64-bit. */ .register %g2, #scratch .register %g3, #scratch -#endif .text @@ -98,7 +95,7 @@ _mesa_sparc_transform_normalize_normals: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer ba 7f nop @@ -154,7 +151,7 @@ _mesa_sparc_transform_normalize_normals: cmp %o4, %g1 ! continue if (i < count) bl 5b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -225,7 +222,7 @@ _mesa_sparc_transform_normalize_normals_no_rot: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer ba 7f nop @@ -263,7 +260,7 @@ _mesa_sparc_transform_normalize_normals_no_rot: cmp %o4, %g1 ! continue if (i < count) bl 5b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -314,7 +311,7 @@ _mesa_sparc_transform_rescale_normals_no_rot: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -379,7 +376,7 @@ _mesa_sparc_transform_rescale_normals: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -421,7 +418,7 @@ _mesa_sparc_transform_normals_no_rot: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -471,7 +468,7 @@ _mesa_sparc_transform_normals: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -529,7 +526,7 @@ _mesa_sparc_normalize_normals: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer ba 7f nop @@ -557,7 +554,7 @@ _mesa_sparc_normalize_normals: cmp %o4, %g1 ! continue if (i < count) bl 5b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop @@ -602,7 +599,7 @@ _mesa_sparc_rescale_normals: cmp %o4, %g1 ! continue if (i < count) bl 1b - add %g3, 0x0c, %g3 ! advance out vector pointer + add %g3, 0x10, %g3 ! advance out vector pointer 7: retl nop diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c index fafb49b81c..d2286a2c83 100644 --- a/src/mesa/sparc/sparc.c +++ b/src/mesa/sparc/sparc.c @@ -113,8 +113,6 @@ void _mesa_init_all_sparc_transform_asm(void) _mesa_clip_tab[4] = _mesa_sparc_cliptest_points4; _mesa_clip_np_tab[4] = _mesa_sparc_cliptest_points4_np; -#if 0 - /* disable these too. See bug 673938 */ _mesa_normal_tab[NORM_TRANSFORM | NORM_NORMALIZE] = _mesa_sparc_transform_normalize_normals; _mesa_normal_tab[NORM_TRANSFORM_NO_ROT | NORM_NORMALIZE] = @@ -131,7 +129,6 @@ void _mesa_init_all_sparc_transform_asm(void) _mesa_sparc_normalize_normals; _mesa_normal_tab[NORM_RESCALE] = _mesa_sparc_rescale_normals; -#endif #ifdef DEBUG_MATH _math_test_all_transform_functions("sparc"); -- cgit v1.2.3 From b4922b533155cc139ebafb111502bb55d2ad2ccf Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Wed, 26 Aug 2009 09:51:15 -0700 Subject: mesa: Add support for ARB_depth_clamp. This currently doesn't include fixing up the cliptests in the assembly paths to support ARB_depth_clamp, so enabling depth_clamp forces the C path. --- src/mesa/drivers/dri/swrast/swrast.c | 1 + src/mesa/drivers/x11/xm_api.c | 1 + src/mesa/glapi/ARB_depth_clamp.xml | 12 ++++++++++ src/mesa/glapi/Makefile | 1 + src/mesa/glapi/gl_API.xml | 2 ++ src/mesa/main/attrib.c | 7 ++++++ src/mesa/main/enable.c | 20 ++++++++++++++++ src/mesa/main/extensions.c | 2 ++ src/mesa/main/get_gen.py | 4 ++++ src/mesa/main/mtypes.h | 2 ++ src/mesa/math/m_clip_tmp.h | 44 ++++++++++++++++++++++++------------ src/mesa/math/m_debug_clip.c | 9 +++++--- src/mesa/math/m_xform.h | 6 +++-- src/mesa/sparc/clip.S | 3 ++- src/mesa/sparc/sparc.c | 6 +++-- src/mesa/swrast/s_depth.c | 27 ++++++++++++++++++++++ src/mesa/swrast/s_depth.h | 2 ++ src/mesa/swrast/s_span.c | 3 +++ src/mesa/tnl/t_rasterpos.c | 25 ++++++++++++-------- src/mesa/tnl/t_vb_program.c | 6 +++-- src/mesa/tnl/t_vb_vertex.c | 6 +++-- src/mesa/x86/x86_xform.c | 9 +++++--- 22 files changed, 158 insertions(+), 40 deletions(-) create mode 100644 src/mesa/glapi/ARB_depth_clamp.xml (limited to 'src/mesa/sparc/sparc.c') diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c index a858af30c1..9a130d61cf 100644 --- a/src/mesa/drivers/dri/swrast/swrast.c +++ b/src/mesa/drivers/dri/swrast/swrast.c @@ -95,6 +95,7 @@ const struct dri_extension card_extensions[] = { "GL_EXT_histogram", GL_EXT_histogram_functions }, { "GL_SGI_color_table", GL_SGI_color_table_functions }, + { "GL_ARB_depth_clamp", NULL }, { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, { "GL_ARB_vertex_array_object", GL_ARB_vertex_array_object_functions }, { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c index 2c7be9f182..78545d0e17 100644 --- a/src/mesa/drivers/x11/xm_api.c +++ b/src/mesa/drivers/x11/xm_api.c @@ -1347,6 +1347,7 @@ const struct dri_extension card_extensions[] = { "GL_EXT_histogram", GL_EXT_histogram_functions }, { "GL_SGI_color_table", GL_SGI_color_table_functions }, + { "GL_ARB_depth_clamp", NULL }, { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, { "GL_ARB_sync", GL_ARB_sync_functions }, { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, diff --git a/src/mesa/glapi/ARB_depth_clamp.xml b/src/mesa/glapi/ARB_depth_clamp.xml new file mode 100644 index 0000000000..157c9a86b1 --- /dev/null +++ b/src/mesa/glapi/ARB_depth_clamp.xml @@ -0,0 +1,12 @@ + + + + + + + + + + + + diff --git a/src/mesa/glapi/Makefile b/src/mesa/glapi/Makefile index 65edab7cec..22f65b74c2 100644 --- a/src/mesa/glapi/Makefile +++ b/src/mesa/glapi/Makefile @@ -48,6 +48,7 @@ SERVER_OUTPUTS = \ API_XML = gl_API.xml \ EXT_framebuffer_object.xml \ ARB_copy_buffer.xml \ + ARB_depth_clamp.xml \ ARB_framebuffer_object.xml \ ARB_map_buffer_range.xml \ ARB_seamless_cube_map.xml \ diff --git a/src/mesa/glapi/gl_API.xml b/src/mesa/glapi/gl_API.xml index 1dfd92be08..920ce80c45 100644 --- a/src/mesa/glapi/gl_API.xml +++ b/src/mesa/glapi/gl_API.xml @@ -7950,6 +7950,8 @@ + + diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c index ab99ca1c64..0fb8fa3bba 100644 --- a/src/mesa/main/attrib.c +++ b/src/mesa/main/attrib.c @@ -74,6 +74,7 @@ struct gl_enable_attrib GLboolean Convolution2D; GLboolean Separable2D; GLboolean CullFace; + GLboolean DepthClamp; GLboolean DepthTest; GLboolean Dither; GLboolean Fog; @@ -265,6 +266,7 @@ _mesa_PushAttrib(GLbitfield mask) attr->Convolution2D = ctx->Pixel.Convolution2DEnabled; attr->Separable2D = ctx->Pixel.Separable2DEnabled; attr->CullFace = ctx->Polygon.CullFlag; + attr->DepthClamp = ctx->Transform.DepthClamp; attr->DepthTest = ctx->Depth.Test; attr->Dither = ctx->Color.DitherFlag; attr->Fog = ctx->Fog.Enabled; @@ -514,6 +516,8 @@ pop_enable_group(GLcontext *ctx, const struct gl_enable_attrib *enable) enable->ColorTable[COLORTABLE_POSTCOLORMATRIX], GL_POST_COLOR_MATRIX_COLOR_TABLE); TEST_AND_UPDATE(ctx->Polygon.CullFlag, enable->CullFace, GL_CULL_FACE); + TEST_AND_UPDATE(ctx->Transform.DepthClamp, enable->DepthClamp, + GL_DEPTH_CLAMP); TEST_AND_UPDATE(ctx->Depth.Test, enable->DepthTest, GL_DEPTH_TEST); TEST_AND_UPDATE(ctx->Color.DitherFlag, enable->Dither, GL_DITHER); TEST_AND_UPDATE(ctx->Pixel.Convolution1DEnabled, enable->Convolution1D, @@ -1221,6 +1225,9 @@ _mesa_PopAttrib(void) if (xform->RescaleNormals != ctx->Transform.RescaleNormals) _mesa_set_enable(ctx, GL_RESCALE_NORMAL_EXT, ctx->Transform.RescaleNormals); + if (xform->DepthClamp != ctx->Transform.DepthClamp) + _mesa_set_enable(ctx, GL_DEPTH_CLAMP, + ctx->Transform.DepthClamp); } break; case GL_TEXTURE_BIT: diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c index 4bc54771e9..d066153fc2 100644 --- a/src/mesa/main/enable.c +++ b/src/mesa/main/enable.c @@ -37,6 +37,7 @@ #include "mtypes.h" #include "enums.h" #include "math/m_matrix.h" +#include "math/m_xform.h" #include "api_arrayelt.h" @@ -947,6 +948,20 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state) ctx->Depth.BoundsTest = state; break; + case GL_DEPTH_CLAMP: + if (ctx->Transform.DepthClamp == state) + return; + /* Neither the x86 nor sparc asm cliptest functions have been updated + * for ARB_depth_clamp, so force the C paths. + */ + if (state) + init_c_cliptest(); + + CHECK_EXTENSION(ARB_depth_clamp, cap); + FLUSH_VERTICES(ctx, _NEW_TRANSFORM); + ctx->Transform.DepthClamp = state; + break; + #if FEATURE_ATI_fragment_shader case GL_FRAGMENT_SHADER_ATI: CHECK_EXTENSION(ATI_fragment_shader, cap); @@ -1395,6 +1410,11 @@ _mesa_IsEnabled( GLenum cap ) CHECK_EXTENSION(EXT_depth_bounds_test); return ctx->Depth.BoundsTest; + /* GL_ARB_depth_clamp */ + case GL_DEPTH_CLAMP: + CHECK_EXTENSION(ARB_depth_clamp); + return ctx->Transform.DepthClamp; + #if FEATURE_ATI_fragment_shader case GL_FRAGMENT_SHADER_ATI: CHECK_EXTENSION(ATI_fragment_shader); diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c index e3070b1547..ea67f820af 100644 --- a/src/mesa/main/extensions.c +++ b/src/mesa/main/extensions.c @@ -47,6 +47,7 @@ static const struct { } default_extensions[] = { { OFF, "GL_ARB_copy_buffer", F(ARB_copy_buffer) }, { OFF, "GL_ARB_depth_texture", F(ARB_depth_texture) }, + { OFF, "GL_ARB_depth_clamp", F(ARB_depth_clamp) }, { ON, "GL_ARB_draw_buffers", F(ARB_draw_buffers) }, { OFF, "GL_ARB_fragment_program", F(ARB_fragment_program) }, { OFF, "GL_ARB_fragment_program_shadow", F(ARB_fragment_program_shadow) }, @@ -192,6 +193,7 @@ void _mesa_enable_sw_extensions(GLcontext *ctx) { ctx->Extensions.ARB_copy_buffer = GL_TRUE; + ctx->Extensions.ARB_depth_clamp = GL_TRUE; ctx->Extensions.ARB_depth_texture = GL_TRUE; /*ctx->Extensions.ARB_draw_buffers = GL_TRUE;*/ #if FEATURE_ARB_fragment_program diff --git a/src/mesa/main/get_gen.py b/src/mesa/main/get_gen.py index 2878c1b552..364d8c55c4 100644 --- a/src/mesa/main/get_gen.py +++ b/src/mesa/main/get_gen.py @@ -905,6 +905,10 @@ StateVars = [ ["ctx->Depth.BoundsMin", "ctx->Depth.BoundsMax"], "", ["EXT_depth_bounds_test"] ), + # GL_ARB_depth_clamp + ( "GL_DEPTH_CLAMP", GLboolean, ["ctx->Transform.DepthClamp"], "", + ["ARB_depth_clamp"] ), + # GL_ARB_draw_buffers ( "GL_MAX_DRAW_BUFFERS_ARB", GLint, ["ctx->Const.MaxDrawBuffers"], "", None ), diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h index 6b64bf8139..20cd3aa5c0 100644 --- a/src/mesa/main/mtypes.h +++ b/src/mesa/main/mtypes.h @@ -1482,6 +1482,7 @@ struct gl_transform_attrib GLboolean Normalize; /**< Normalize all normals? */ GLboolean RescaleNormals; /**< GL_EXT_rescale_normal */ GLboolean RasterPositionUnclipped; /**< GL_IBM_rasterpos_clip */ + GLboolean DepthClamp; /**< GL_ARB_depth_clamp */ GLboolean CullVertexFlag; /**< True if GL_CULL_VERTEX_EXT is enabled */ GLfloat CullEyePos[4]; @@ -2475,6 +2476,7 @@ struct gl_extensions GLboolean dummy; /* don't remove this! */ GLboolean ARB_copy_buffer; GLboolean ARB_depth_texture; + GLboolean ARB_depth_clamp; GLboolean ARB_draw_buffers; GLboolean ARB_fragment_program; GLboolean ARB_fragment_program_shadow; diff --git a/src/mesa/math/m_clip_tmp.h b/src/mesa/math/m_clip_tmp.h index f3a589be05..2e30964057 100644 --- a/src/mesa/math/m_clip_tmp.h +++ b/src/mesa/math/m_clip_tmp.h @@ -44,7 +44,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points4)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ) + GLubyte *andMask, + GLboolean viewport_z_clip ) { const GLuint stride = clip_vec->stride; const GLfloat *from = (GLfloat *)clip_vec->start; @@ -66,16 +67,20 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points4)( GLvector4f *clip_vec, mask |= (((cw < -cx) << CLIP_LEFT_SHIFT)); mask |= (((cw < cy) << CLIP_TOP_SHIFT)); mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT)); - mask |= (((cw < cz) << CLIP_FAR_SHIFT)); - mask |= (((cw < -cz) << CLIP_NEAR_SHIFT)); + if (viewport_z_clip) { + mask |= (((cw < cz) << CLIP_FAR_SHIFT)); + mask |= (((cw < -cz) << CLIP_NEAR_SHIFT)); + } #else /* !defined(macintosh)) */ GLubyte mask = 0; if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT; if ( cx + cw < 0) mask |= CLIP_LEFT_BIT; if (-cy + cw < 0) mask |= CLIP_TOP_BIT; if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT; - if (-cz + cw < 0) mask |= CLIP_FAR_BIT; - if ( cz + cw < 0) mask |= CLIP_NEAR_BIT; + if (viewport_z_clip) { + if (-cz + cw < 0) mask |= CLIP_FAR_BIT; + if ( cz + cw < 0) mask |= CLIP_NEAR_BIT; + } #endif /* defined(macintosh) */ clipMask[i] = mask; @@ -119,7 +124,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_np_points4)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ) + GLubyte *andMask, + GLboolean viewport_z_clip ) { const GLuint stride = clip_vec->stride; const GLuint count = clip_vec->count; @@ -141,16 +147,20 @@ static GLvector4f * _XFORMAPI TAG(cliptest_np_points4)( GLvector4f *clip_vec, mask |= (((cw < -cx) << CLIP_LEFT_SHIFT)); mask |= (((cw < cy) << CLIP_TOP_SHIFT)); mask |= (((cw < -cy) << CLIP_BOTTOM_SHIFT)); - mask |= (((cw < cz) << CLIP_FAR_SHIFT)); - mask |= (((cw < -cz) << CLIP_NEAR_SHIFT)); + if (viewport_z_clip) { + mask |= (((cw < cz) << CLIP_FAR_SHIFT)); + mask |= (((cw < -cz) << CLIP_NEAR_SHIFT)); + } #else /* !defined(macintosh)) */ GLubyte mask = 0; if (-cx + cw < 0) mask |= CLIP_RIGHT_BIT; if ( cx + cw < 0) mask |= CLIP_LEFT_BIT; if (-cy + cw < 0) mask |= CLIP_TOP_BIT; if ( cy + cw < 0) mask |= CLIP_BOTTOM_BIT; - if (-cz + cw < 0) mask |= CLIP_FAR_BIT; - if ( cz + cw < 0) mask |= CLIP_NEAR_BIT; + if (viewport_z_clip) { + if (-cz + cw < 0) mask |= CLIP_FAR_BIT; + if ( cz + cw < 0) mask |= CLIP_NEAR_BIT; + } #endif /* defined(macintosh) */ clipMask[i] = mask; @@ -171,7 +181,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points3)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ) + GLubyte *andMask, + GLboolean viewport_z_clip ) { const GLuint stride = clip_vec->stride; const GLuint count = clip_vec->count; @@ -187,8 +198,10 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points3)( GLvector4f *clip_vec, else if (cx < -1.0) mask |= CLIP_LEFT_BIT; if (cy > 1.0) mask |= CLIP_TOP_BIT; else if (cy < -1.0) mask |= CLIP_BOTTOM_BIT; - if (cz > 1.0) mask |= CLIP_FAR_BIT; - else if (cz < -1.0) mask |= CLIP_NEAR_BIT; + if (viewport_z_clip) { + if (cz > 1.0) mask |= CLIP_FAR_BIT; + else if (cz < -1.0) mask |= CLIP_NEAR_BIT; + } clipMask[i] = mask; tmpOrMask |= mask; tmpAndMask &= mask; @@ -204,7 +217,8 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points2)( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ) + GLubyte *andMask, + GLboolean viewport_z_clip ) { const GLuint stride = clip_vec->stride; const GLuint count = clip_vec->count; @@ -231,7 +245,7 @@ static GLvector4f * _XFORMAPI TAG(cliptest_points2)( GLvector4f *clip_vec, } -static void TAG(init_c_cliptest)( void ) +void TAG(init_c_cliptest)( void ) { _mesa_clip_tab[4] = TAG(cliptest_points4); _mesa_clip_tab[3] = TAG(cliptest_points3); diff --git a/src/mesa/math/m_debug_clip.c b/src/mesa/math/m_debug_clip.c index 460fed4a75..f2b757a91b 100644 --- a/src/mesa/math/m_debug_clip.c +++ b/src/mesa/math/m_debug_clip.c @@ -67,7 +67,8 @@ static GLvector4f *ref_cliptest_points4( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ) + GLubyte *andMask, + GLboolean viewport_z_clip ) { const GLuint stride = clip_vec->stride; const GLuint count = clip_vec->count; @@ -87,8 +88,10 @@ static GLvector4f *ref_cliptest_points4( GLvector4f *clip_vec, if ( cx + cw < 0 ) mask |= CLIP_LEFT_BIT; if ( -cy + cw < 0 ) mask |= CLIP_TOP_BIT; if ( cy + cw < 0 ) mask |= CLIP_BOTTOM_BIT; - if ( -cz + cw < 0 ) mask |= CLIP_FAR_BIT; - if ( cz + cw < 0 ) mask |= CLIP_NEAR_BIT; + if (viewport_z_clip) { + if ( -cz + cw < 0 ) mask |= CLIP_FAR_BIT; + if ( cz + cw < 0 ) mask |= CLIP_NEAR_BIT; + } clipMask[i] = mask; if ( mask ) { c++; diff --git a/src/mesa/math/m_xform.h b/src/mesa/math/m_xform.h index 7ef76e0b92..33421ad1c0 100644 --- a/src/mesa/math/m_xform.h +++ b/src/mesa/math/m_xform.h @@ -43,7 +43,8 @@ extern void _math_init_transformation(void); - +extern void +init_c_cliptest(void); /* KW: Clip functions now do projective divide as well. The projected * coordinates are very useful to us because they let us cull @@ -102,7 +103,8 @@ typedef GLvector4f * (_XFORMAPIP clip_func)( GLvector4f *vClip, GLvector4f *vProj, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ); + GLubyte *andMask, + GLboolean viewport_z_clip ); typedef void (*dotprod_func)( GLfloat *out, GLuint out_stride, diff --git a/src/mesa/sparc/clip.S b/src/mesa/sparc/clip.S index 208843c606..dc239171ff 100644 --- a/src/mesa/sparc/clip.S +++ b/src/mesa/sparc/clip.S @@ -58,7 +58,8 @@ clip_table: .byte 31, 29, 31, 30, 27, 25, 27, 26 /* GLvector4f *clip_vec, GLvector4f *proj_vec, - GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask */ + GLubyte clipMask[], GLubyte *orMask, GLubyte *andMask, + GLboolean viewport_z_enable */ .align 64 __pc_tramp: diff --git a/src/mesa/sparc/sparc.c b/src/mesa/sparc/sparc.c index d2286a2c83..cea0c7cecf 100644 --- a/src/mesa/sparc/sparc.c +++ b/src/mesa/sparc/sparc.c @@ -78,13 +78,15 @@ extern GLvector4f *_mesa_sparc_cliptest_points4(GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask); + GLubyte *andMask, + GLboolean viewport_z_clip); extern GLvector4f *_mesa_sparc_cliptest_points4_np(GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask); + GLubyte *andMask, + GLboolean viewport_z_clip); #define NORM_ARGS const GLmatrix *mat, \ GLfloat scale, \ diff --git a/src/mesa/swrast/s_depth.c b/src/mesa/swrast/s_depth.c index 26e23f02d5..1a428fb1a2 100644 --- a/src/mesa/swrast/s_depth.c +++ b/src/mesa/swrast/s_depth.c @@ -497,6 +497,33 @@ depth_test_span32( GLcontext *ctx, GLuint n, return passed; } +/* Apply ARB_depth_clamp to span of fragments. */ +void +_swrast_depth_clamp_span( GLcontext *ctx, SWspan *span ) +{ + struct gl_framebuffer *fb = ctx->DrawBuffer; + struct gl_renderbuffer *rb = fb->_DepthBuffer; + const GLuint count = span->end; + GLuint *zValues = span->array->z; + GLuint near, far; + int i; + + if (rb->DataType == GL_UNSIGNED_SHORT) { + near = FLOAT_TO_UINT(ctx->Viewport.Near); + far = FLOAT_TO_UINT(ctx->Viewport.Far); + } else { + assert(rb->DataType == GL_UNSIGNED_INT); + CLAMPED_FLOAT_TO_USHORT(near, ctx->Viewport.Near); + CLAMPED_FLOAT_TO_USHORT(far, ctx->Viewport.Far); + } + for (i = 0; i < count; i++) { + if (zValues[i] < near) + zValues[i] = near; + if (zValues[i] > far) + zValues[i] = far; + } +} + /* diff --git a/src/mesa/swrast/s_depth.h b/src/mesa/swrast/s_depth.h index 3688625683..7eae366742 100644 --- a/src/mesa/swrast/s_depth.h +++ b/src/mesa/swrast/s_depth.h @@ -33,6 +33,8 @@ extern GLuint _swrast_depth_test_span( GLcontext *ctx, SWspan *span); +extern void +_swrast_depth_clamp_span( GLcontext *ctx, SWspan *span ); extern GLboolean _swrast_depth_bounds_test( GLcontext *ctx, SWspan *span ); diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c index 0e2793b474..a45eac438e 100644 --- a/src/mesa/swrast/s_span.c +++ b/src/mesa/swrast/s_span.c @@ -880,6 +880,9 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span) stipple_polygon_span(ctx, span); } + if (ctx->Transform.DepthClamp) + _swrast_depth_clamp_span(ctx, span); + /* Stencil and Z testing */ if (ctx->Stencil._Enabled || ctx->Depth.Test) { if (!(span->arrayMask & SPAN_Z)) diff --git a/src/mesa/tnl/t_rasterpos.c b/src/mesa/tnl/t_rasterpos.c index f1fdddf0f5..99b6787455 100644 --- a/src/mesa/tnl/t_rasterpos.c +++ b/src/mesa/tnl/t_rasterpos.c @@ -46,11 +46,10 @@ * \return zero if outside view volume, or one if inside. */ static GLuint -viewclip_point( const GLfloat v[] ) +viewclip_point_xy( const GLfloat v[] ) { if ( v[0] > v[3] || v[0] < -v[3] - || v[1] > v[3] || v[1] < -v[3] - || v[2] > v[3] || v[2] < -v[3] ) { + || v[1] > v[3] || v[1] < -v[3] ) { return 0; } else { @@ -408,18 +407,18 @@ _tnl_RasterPos(GLcontext *ctx, const GLfloat vObj[4]) /* apply projection matrix: clip = Proj * eye */ TRANSFORM_POINT( clip, ctx->ProjectionMatrixStack.Top->m, eye ); - /* clip to view volume */ - if (ctx->Transform.RasterPositionUnclipped) { - /* GL_IBM_rasterpos_clip: only clip against Z */ + /* clip to view volume. */ + if (!ctx->Transform.DepthClamp) { if (viewclip_point_z(clip) == 0) { ctx->Current.RasterPosValid = GL_FALSE; return; } } - else if (viewclip_point(clip) == 0) { - /* Normal OpenGL behaviour */ - ctx->Current.RasterPosValid = GL_FALSE; - return; + if (!ctx->Transform.RasterPositionUnclipped) { + if (viewclip_point_xy(clip) == 0) { + ctx->Current.RasterPosValid = GL_FALSE; + return; + } } /* clip to user clipping planes */ @@ -443,6 +442,12 @@ _tnl_RasterPos(GLcontext *ctx, const GLfloat vObj[4]) / ctx->DrawBuffer->_DepthMaxF; ctx->Current.RasterPos[3] = clip[3]; + if (ctx->Transform.DepthClamp) { + ctx->Current.RasterPos[3] = CLAMP(ctx->Current.RasterPos[3], + ctx->Viewport.Near, + ctx->Viewport.Far); + } + /* compute raster distance */ if (ctx->Fog.FogCoordinateSource == GL_FOG_COORDINATE_EXT) ctx->Current.RasterDistance = ctx->Current.Attrib[VERT_ATTRIB_FOG][0]; diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c index dc954bcba1..5d89f8bc31 100644 --- a/src/mesa/tnl/t_vb_program.c +++ b/src/mesa/tnl/t_vb_program.c @@ -137,7 +137,8 @@ do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store) &store->ndcCoords, store->clipmask, &store->ormask, - &store->andmask ); + &store->andmask, + !ctx->Transform.DepthClamp ); } else { VB->NdcPtr = NULL; @@ -145,7 +146,8 @@ do_ndc_cliptest(GLcontext *ctx, struct vp_stage_data *store) NULL, store->clipmask, &store->ormask, - &store->andmask ); + &store->andmask, + !ctx->Transform.DepthClamp ); } if (store->andmask) { diff --git a/src/mesa/tnl/t_vb_vertex.c b/src/mesa/tnl/t_vb_vertex.c index 30aa7c4086..6a746417c8 100644 --- a/src/mesa/tnl/t_vb_vertex.c +++ b/src/mesa/tnl/t_vb_vertex.c @@ -173,7 +173,8 @@ static GLboolean run_vertex_stage( GLcontext *ctx, &store->proj, store->clipmask, &store->ormask, - &store->andmask ); + &store->andmask, + !ctx->Transform.DepthClamp ); } else { VB->NdcPtr = NULL; @@ -181,7 +182,8 @@ static GLboolean run_vertex_stage( GLcontext *ctx, NULL, store->clipmask, &store->ormask, - &store->andmask ); + &store->andmask, + !ctx->Transform.DepthClamp ); } if (store->andmask) diff --git a/src/mesa/x86/x86_xform.c b/src/mesa/x86/x86_xform.c index 16b2b26bcc..52f6b25d81 100644 --- a/src/mesa/x86/x86_xform.c +++ b/src/mesa/x86/x86_xform.c @@ -60,21 +60,24 @@ _mesa_x86_cliptest_points4( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ); + GLubyte *andMask, + GLboolean viewport_z_clip ); extern GLvector4f * _ASMAPI _mesa_x86_cliptest_points4_np( GLvector4f *clip_vec, GLvector4f *proj_vec, GLubyte clipMask[], GLubyte *orMask, - GLubyte *andMask ); + GLubyte *andMask, + GLboolean viewport_z_clip ); extern void _ASMAPI _mesa_v16_x86_cliptest_points4( GLfloat *first_vert, GLfloat *last_vert, GLubyte *or_mask, GLubyte *and_mask, - GLubyte *clip_mask ); + GLubyte *clip_mask, + GLboolean viewport_z_clip ); extern void _ASMAPI _mesa_v16_x86_general_xform( GLfloat *dest, -- cgit v1.2.3