From dc9e604ef16ee19cf7480325100e6edd768dbb16 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Mathias=20Fr=C3=B6hlich?= Date: Sun, 22 May 2016 14:10:19 +0200 Subject: [PATCH] mesa: Use bitmask/ffs to iterate enabled clip planes. MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit Replaces an iterate and test bit in a bitmask loop by a loop only iterating over the bits set in the bitmask. v2: Use _mesa_bit_scan{,64} instead of open coding. v3: Use u_bit_scan{,64} instead of _mesa_bit_scan{,64}. Reviewed-by: Brian Paul Signed-off-by: Mathias Fröhlich --- src/mesa/drivers/common/meta.c | 22 +++++++------- src/mesa/main/matrix.c | 19 ++++++------ src/mesa/main/rastpos.c | 21 +++++++------- src/mesa/tnl/t_vb_cliptmp.h | 51 ++++++++++++++++----------------- src/mesa/tnl/t_vb_program.c | 65 +++++++++++++++++++++--------------------- src/mesa/tnl/t_vb_render.c | 1 + src/mesa/tnl/t_vb_vertex.c | 59 +++++++++++++++++++------------------- 7 files changed, 117 insertions(+), 121 deletions(-) diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c index 3c86305..be671b4 100644 --- a/src/mesa/drivers/common/meta.c +++ b/src/mesa/drivers/common/meta.c @@ -85,6 +85,7 @@ #include "drivers/common/meta.h" #include "main/enums.h" #include "main/glformats.h" +#include "util/bitscan.h" #include "util/ralloc.h" /** Return offset in bytes of the field within a vertex struct */ @@ -682,12 +683,12 @@ _mesa_meta_begin(struct gl_context *ctx, GLbitfield state) } if (state & MESA_META_CLIP) { + GLbitfield mask; save->ClipPlanesEnabled = ctx->Transform.ClipPlanesEnabled; - if (ctx->Transform.ClipPlanesEnabled) { - GLuint i; - for (i = 0; i < ctx->Const.MaxClipPlanes; i++) { - _mesa_set_enable(ctx, GL_CLIP_PLANE0 + i, GL_FALSE); - } + mask = ctx->Transform.ClipPlanesEnabled; + while (mask) { + const int i = u_bit_scan(&mask); + _mesa_set_enable(ctx, GL_CLIP_PLANE0 + i, GL_FALSE); } } @@ -1090,13 +1091,10 @@ _mesa_meta_end(struct gl_context *ctx) } if (state & MESA_META_CLIP) { - if (save->ClipPlanesEnabled) { - GLuint i; - for (i = 0; i < ctx->Const.MaxClipPlanes; i++) { - if (save->ClipPlanesEnabled & (1 << i)) { - _mesa_set_enable(ctx, GL_CLIP_PLANE0 + i, GL_TRUE); - } - } + GLbitfield mask = save->ClipPlanesEnabled; + while (mask) { + const int i = u_bit_scan(&mask); + _mesa_set_enable(ctx, GL_CLIP_PLANE0 + i, GL_TRUE); } } diff --git a/src/mesa/main/matrix.c b/src/mesa/main/matrix.c index 5ff5ac5..293d50c 100644 --- a/src/mesa/main/matrix.c +++ b/src/mesa/main/matrix.c @@ -43,6 +43,7 @@ #include "matrix.h" #include "mtypes.h" #include "math/m_matrix.h" +#include "util/bitscan.h" /** @@ -554,20 +555,20 @@ _mesa_MultTransposeMatrixd( const GLdouble *m ) static void update_projection( struct gl_context *ctx ) { + GLbitfield mask; + _math_matrix_analyse( ctx->ProjectionMatrixStack.Top ); /* Recompute clip plane positions in clipspace. This is also done * in _mesa_ClipPlane(). */ - if (ctx->Transform.ClipPlanesEnabled) { - GLuint p; - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { - _mesa_transform_vector( ctx->Transform._ClipUserPlane[p], - ctx->Transform.EyeUserPlane[p], - ctx->ProjectionMatrixStack.Top->inv ); - } - } + mask = ctx->Transform.ClipPlanesEnabled; + while (mask) { + const int p = u_bit_scan(&mask); + + _mesa_transform_vector( ctx->Transform._ClipUserPlane[p], + ctx->Transform.EyeUserPlane[p], + ctx->ProjectionMatrixStack.Top->inv ); } } diff --git a/src/mesa/main/rastpos.c b/src/mesa/main/rastpos.c index 8f971f5b..4fddad1 100644 --- a/src/mesa/main/rastpos.c +++ b/src/mesa/main/rastpos.c @@ -91,17 +91,16 @@ viewclip_point_z( const GLfloat v[] ) static GLuint userclip_point( struct gl_context *ctx, const GLfloat v[] ) { - GLuint p; - - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { - GLfloat dot = v[0] * ctx->Transform._ClipUserPlane[p][0] - + v[1] * ctx->Transform._ClipUserPlane[p][1] - + v[2] * ctx->Transform._ClipUserPlane[p][2] - + v[3] * ctx->Transform._ClipUserPlane[p][3]; - if (dot < 0.0F) { - return 0; - } + GLbitfield mask = ctx->Transform.ClipPlanesEnabled; + while (mask) { + const int p = u_bit_scan(&mask); + GLfloat dot = v[0] * ctx->Transform._ClipUserPlane[p][0] + + v[1] * ctx->Transform._ClipUserPlane[p][1] + + v[2] * ctx->Transform._ClipUserPlane[p][2] + + v[3] * ctx->Transform._ClipUserPlane[p][3]; + + if (dot < 0.0F) { + return 0; } } diff --git a/src/mesa/tnl/t_vb_cliptmp.h b/src/mesa/tnl/t_vb_cliptmp.h index 12181f0..c654666 100644 --- a/src/mesa/tnl/t_vb_cliptmp.h +++ b/src/mesa/tnl/t_vb_cliptmp.h @@ -124,7 +124,6 @@ TAG(clip_line)( struct gl_context *ctx, GLuint v0, GLuint v1, GLubyte mask ) GLuint newvert = VB->Count; GLfloat t0 = 0; GLfloat t1 = 0; - GLuint p; const GLuint v0_orig = v0; if (mask & CLIP_FRUSTUM_BITS) { @@ -137,14 +136,14 @@ TAG(clip_line)( struct gl_context *ctx, GLuint v0, GLuint v1, GLubyte mask ) } if (mask & CLIP_USER_BIT) { - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { - const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; - const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; - const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; - const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; - LINE_CLIP( CLIP_USER_BIT, a, b, c, d ); - } + GLbitfield enabled = ctx->Transform.ClipPlanesEnabled; + while (enabled) { + const int p = u_bit_scan(&enabled); + const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; + const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; + const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; + const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; + LINE_CLIP( CLIP_USER_BIT, a, b, c, d ); } } @@ -194,7 +193,6 @@ TAG(clip_tri)( struct gl_context *ctx, GLuint v0, GLuint v1, GLuint v2, GLubyte GLuint pv = v2; GLuint vlist[2][MAX_CLIPPED_VERTICES]; GLuint *inlist = vlist[0], *outlist = vlist[1]; - GLuint p; GLuint n = 3; ASSIGN_3V(inlist, v2, v0, v1 ); /* pv rotated to slot zero */ @@ -226,14 +224,14 @@ TAG(clip_tri)( struct gl_context *ctx, GLuint v0, GLuint v1, GLuint v2, GLubyte } if (mask & CLIP_USER_BIT) { - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { - const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; - const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; - const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; - const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; - POLY_CLIP( CLIP_USER_BIT, a, b, c, d ); - } + GLbitfield enabled = ctx->Transform.ClipPlanesEnabled; + while (enabled) { + const int p = u_bit_scan(&enabled); + const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; + const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; + const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; + const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; + POLY_CLIP( CLIP_USER_BIT, a, b, c, d ); } } @@ -274,7 +272,6 @@ TAG(clip_quad)( struct gl_context *ctx, GLuint v0, GLuint v1, GLuint v2, GLuint GLuint pv = v3; GLuint vlist[2][MAX_CLIPPED_VERTICES]; GLuint *inlist = vlist[0], *outlist = vlist[1]; - GLuint p; GLuint n = 4; ASSIGN_4V(inlist, v3, v0, v1, v2 ); /* pv rotated to slot zero */ @@ -289,14 +286,14 @@ TAG(clip_quad)( struct gl_context *ctx, GLuint v0, GLuint v1, GLuint v2, GLuint } if (mask & CLIP_USER_BIT) { - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { - const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; - const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; - const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; - const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; - POLY_CLIP( CLIP_USER_BIT, a, b, c, d ); - } + GLbitfield enabled = ctx->Transform.ClipPlanesEnabled; + while (enabled) { + const int p = u_bit_scan(&enabled); + const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; + const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; + const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; + const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; + POLY_CLIP( CLIP_USER_BIT, a, b, c, d ); } } diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c index 21fd6cd..55c44d1 100644 --- a/src/mesa/tnl/t_vb_program.c +++ b/src/mesa/tnl/t_vb_program.c @@ -40,6 +40,7 @@ #include "program/prog_statevars.h" #include "program/prog_execute.h" #include "swrast/s_context.h" +#include "util/bitscan.h" #include "tnl/tnl.h" #include "tnl/t_context.h" @@ -84,40 +85,38 @@ userclip( struct gl_context *ctx, GLubyte *clipormask, GLubyte *clipandmask ) { - GLuint p; - - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { - GLuint nr, i; - const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; - const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; - const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; - const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; - GLfloat *coord = (GLfloat *)clip->data; - GLuint stride = clip->stride; - GLuint count = clip->count; - - for (nr = 0, i = 0 ; i < count ; i++) { - GLfloat dp = (coord[0] * a + - coord[1] * b + - coord[2] * c + - coord[3] * d); - - if (dp < 0) { - nr++; - clipmask[i] |= CLIP_USER_BIT; - } - - STRIDE_F(coord, stride); - } + GLbitfield mask = ctx->Transform.ClipPlanesEnabled; + while (mask) { + const int p = u_bit_scan(&mask); + GLuint nr, i; + const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; + const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; + const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; + const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; + GLfloat *coord = (GLfloat *)clip->data; + GLuint stride = clip->stride; + GLuint count = clip->count; + + for (nr = 0, i = 0 ; i < count ; i++) { + GLfloat dp = (coord[0] * a + + coord[1] * b + + coord[2] * c + + coord[3] * d); + + if (dp < 0) { + nr++; + clipmask[i] |= CLIP_USER_BIT; + } + + STRIDE_F(coord, stride); + } - if (nr > 0) { - *clipormask |= CLIP_USER_BIT; - if (nr == count) { - *clipandmask |= CLIP_USER_BIT; - return; - } - } + if (nr > 0) { + *clipormask |= CLIP_USER_BIT; + if (nr == count) { + *clipandmask |= CLIP_USER_BIT; + return; + } } } } diff --git a/src/mesa/tnl/t_vb_render.c b/src/mesa/tnl/t_vb_render.c index 03e8fcf..9ff1f18 100644 --- a/src/mesa/tnl/t_vb_render.c +++ b/src/mesa/tnl/t_vb_render.c @@ -46,6 +46,7 @@ #include "main/imports.h" #include "main/mtypes.h" #include "math/m_xform.h" +#include "util/bitscan.h" #include "t_pipeline.h" diff --git a/src/mesa/tnl/t_vb_vertex.c b/src/mesa/tnl/t_vb_vertex.c index b56d680..71a32b4 100644 --- a/src/mesa/tnl/t_vb_vertex.c +++ b/src/mesa/tnl/t_vb_vertex.c @@ -33,6 +33,8 @@ #include "math/m_xform.h" +#include "util/bitscan.h" + #include "t_context.h" #include "t_pipeline.h" @@ -63,40 +65,39 @@ static void NAME( struct gl_context *ctx, \ GLubyte *clipormask, \ GLubyte *clipandmask ) \ { \ - GLuint p; \ - \ - for (p = 0; p < ctx->Const.MaxClipPlanes; p++) \ - if (ctx->Transform.ClipPlanesEnabled & (1 << p)) { \ - GLuint nr, i; \ - const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; \ - const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; \ - const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; \ - const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; \ - GLfloat *coord = (GLfloat *)clip->data; \ - GLuint stride = clip->stride; \ - GLuint count = clip->count; \ + GLbitfield mask = ctx->Transform.ClipPlanesEnabled; \ + while (mask) { \ + const int p = u_bit_scan(&mask); \ + GLuint nr, i; \ + const GLfloat a = ctx->Transform._ClipUserPlane[p][0]; \ + const GLfloat b = ctx->Transform._ClipUserPlane[p][1]; \ + const GLfloat c = ctx->Transform._ClipUserPlane[p][2]; \ + const GLfloat d = ctx->Transform._ClipUserPlane[p][3]; \ + GLfloat *coord = (GLfloat *)clip->data; \ + GLuint stride = clip->stride; \ + GLuint count = clip->count; \ \ - for (nr = 0, i = 0 ; i < count ; i++) { \ - GLfloat dp = coord[0] * a + coord[1] * b; \ - if (SZ > 2) dp += coord[2] * c; \ - if (SZ > 3) dp += coord[3] * d; else dp += d; \ + for (nr = 0, i = 0 ; i < count ; i++) { \ + GLfloat dp = coord[0] * a + coord[1] * b; \ + if (SZ > 2) dp += coord[2] * c; \ + if (SZ > 3) dp += coord[3] * d; else dp += d; \ \ - if (dp < 0) { \ - nr++; \ - clipmask[i] |= CLIP_USER_BIT; \ - } \ + if (dp < 0) { \ + nr++; \ + clipmask[i] |= CLIP_USER_BIT; \ + } \ \ - STRIDE_F(coord, stride); \ - } \ + STRIDE_F(coord, stride); \ + } \ \ - if (nr > 0) { \ - *clipormask |= CLIP_USER_BIT; \ - if (nr == count) { \ - *clipandmask |= CLIP_USER_BIT; \ - return; \ - } \ - } \ + if (nr > 0) { \ + *clipormask |= CLIP_USER_BIT; \ + if (nr == count) { \ + *clipandmask |= CLIP_USER_BIT; \ + return; \ + } \ } \ + } \ } -- 2.7.4