From bb35000b4b6dfe60048b2f5d60bc102c4a7fd791 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Thu, 4 Mar 2010 15:47:19 -0800 Subject: [PATCH] intel: Remove non-kernel-exec-fencing support. Shaves 60k off the driver from removing the broken spans code. This means we now require 2.6.29, which seems fair given that it's a year old and we've removed support for non-KMS already in the last release of 2D. --- src/mesa/drivers/dri/i965/brw_curbe.c | 5 +- src/mesa/drivers/dri/i965/brw_draw_upload.c | 63 +-- src/mesa/drivers/dri/i965/brw_vs_surface_state.c | 4 +- src/mesa/drivers/dri/i965/brw_vtbl.c | 2 +- src/mesa/drivers/dri/i965/gen6_vs_state.c | 4 +- src/mesa/drivers/dri/i965/gen6_wm_state.c | 4 +- src/mesa/drivers/dri/intel/intel_blit.c | 8 +- src/mesa/drivers/dri/intel/intel_buffer_objects.c | 8 +- src/mesa/drivers/dri/intel/intel_context.c | 6 - src/mesa/drivers/dri/intel/intel_context.h | 21 - src/mesa/drivers/dri/intel/intel_depthtmp.h | 64 --- src/mesa/drivers/dri/intel/intel_fbo.c | 3 - src/mesa/drivers/dri/intel/intel_fbo.h | 3 - src/mesa/drivers/dri/intel/intel_mipmap_tree.c | 3 +- src/mesa/drivers/dri/intel/intel_regions.c | 6 +- src/mesa/drivers/dri/intel/intel_screen.c | 9 +- src/mesa/drivers/dri/intel/intel_screen.h | 1 - src/mesa/drivers/dri/intel/intel_span.c | 451 +++------------------- src/mesa/drivers/dri/intel/intel_spantmp.h | 67 ---- 19 files changed, 94 insertions(+), 638 deletions(-) delete mode 100644 src/mesa/drivers/dri/intel/intel_depthtmp.h delete mode 100644 src/mesa/drivers/dri/intel/intel_spantmp.h diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c index 6f2ead7..4e78b08 100644 --- a/src/mesa/drivers/dri/i965/brw_curbe.c +++ b/src/mesa/drivers/dri/i965/brw_curbe.c @@ -179,7 +179,6 @@ static GLfloat fixed_plane[6][4] = { */ static void prepare_constant_buffer(struct brw_context *brw) { - struct intel_context *intel = &brw->intel; GLcontext *ctx = &brw->intel.ctx; const struct brw_vertex_program *vp = brw_vertex_program_const(brw->vertex_program); @@ -307,7 +306,7 @@ static void prepare_constant_buffer(struct brw_context *brw) if (brw->curbe.curbe_bo != NULL && brw->curbe.curbe_next_offset + bufsz > brw->curbe.curbe_bo->size) { - intel_bo_unmap_gtt_preferred(intel, brw->curbe.curbe_bo); + drm_intel_gem_bo_unmap_gtt(brw->curbe.curbe_bo); dri_bo_unreference(brw->curbe.curbe_bo); brw->curbe.curbe_bo = NULL; } @@ -319,7 +318,7 @@ static void prepare_constant_buffer(struct brw_context *brw) brw->curbe.curbe_bo = dri_bo_alloc(brw->intel.bufmgr, "CURBE", 4096, 1 << 6); brw->curbe.curbe_next_offset = 0; - intel_bo_map_gtt_preferred(intel, brw->curbe.curbe_bo, GL_TRUE); + drm_intel_gem_bo_map_gtt(brw->curbe.curbe_bo); } brw->curbe.curbe_offset = brw->curbe.curbe_next_offset; diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c index f0a4e8a..71a4357 100644 --- a/src/mesa/drivers/dri/i965/brw_draw_upload.c +++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c @@ -276,7 +276,6 @@ copy_array_to_vbo_array( struct brw_context *brw, struct brw_vertex_element *element, GLuint dst_stride) { - struct intel_context *intel = &brw->intel; GLuint size = element->count * dst_stride; get_space(brw, size, &element->bo, &element->offset); @@ -289,52 +288,26 @@ copy_array_to_vbo_array( struct brw_context *brw, } if (dst_stride == element->glarray->StrideB) { - if (intel->intelScreen->kernel_exec_fencing) { - drm_intel_gem_bo_map_gtt(element->bo); - memcpy((char *)element->bo->virtual + element->offset, - element->glarray->Ptr, size); - drm_intel_gem_bo_unmap_gtt(element->bo); - } else { - dri_bo_subdata(element->bo, - element->offset, - size, - element->glarray->Ptr); - } + drm_intel_gem_bo_map_gtt(element->bo); + memcpy((char *)element->bo->virtual + element->offset, + element->glarray->Ptr, size); + drm_intel_gem_bo_unmap_gtt(element->bo); } else { char *dest; const unsigned char *src = element->glarray->Ptr; int i; - if (intel->intelScreen->kernel_exec_fencing) { - drm_intel_gem_bo_map_gtt(element->bo); - dest = element->bo->virtual; - dest += element->offset; - - for (i = 0; i < element->count; i++) { - memcpy(dest, src, dst_stride); - src += element->glarray->StrideB; - dest += dst_stride; - } - - drm_intel_gem_bo_unmap_gtt(element->bo); - } else { - void *data; - - data = malloc(dst_stride * element->count); - dest = data; - for (i = 0; i < element->count; i++) { - memcpy(dest, src, dst_stride); - src += element->glarray->StrideB; - dest += dst_stride; - } - - dri_bo_subdata(element->bo, - element->offset, - size, - data); + drm_intel_gem_bo_map_gtt(element->bo); + dest = element->bo->virtual; + dest += element->offset; - free(data); + for (i = 0; i < element->count; i++) { + memcpy(dest, src, dst_stride); + src += element->glarray->StrideB; + dest += dst_stride; } + + drm_intel_gem_bo_unmap_gtt(element->bo); } } @@ -646,13 +619,9 @@ static void brw_prepare_indices(struct brw_context *brw) /* Straight upload */ - if (intel->intelScreen->kernel_exec_fencing) { - drm_intel_gem_bo_map_gtt(bo); - memcpy((char *)bo->virtual + offset, index_buffer->ptr, ib_size); - drm_intel_gem_bo_unmap_gtt(bo); - } else { - dri_bo_subdata(bo, offset, ib_size, index_buffer->ptr); - } + drm_intel_gem_bo_map_gtt(bo); + memcpy((char *)bo->virtual + offset, index_buffer->ptr, ib_size); + drm_intel_gem_bo_unmap_gtt(bo); } else { offset = (GLuint) (unsigned long) index_buffer->ptr; brw->ib.start_vertex_offset = 0; diff --git a/src/mesa/drivers/dri/i965/brw_vs_surface_state.c b/src/mesa/drivers/dri/i965/brw_vs_surface_state.c index ead623f..950d075 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_surface_state.c +++ b/src/mesa/drivers/dri/i965/brw_vs_surface_state.c @@ -67,13 +67,13 @@ brw_vs_update_constant_buffer(struct brw_context *brw) */ _mesa_load_state_parameters(&brw->intel.ctx, vp->program.Base.Parameters); - intel_bo_map_gtt_preferred(intel, const_buffer, GL_TRUE); + drm_intel_gem_bo_map_gtt(const_buffer); for (i = 0; i < params->NumParameters; i++) { memcpy(const_buffer->virtual + i * 4 * sizeof(float), params->ParameterValues[i], 4 * sizeof(float)); } - intel_bo_unmap_gtt_preferred(intel, const_buffer); + drm_intel_gem_bo_unmap_gtt(const_buffer); return const_buffer; } diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c index 27a2a3e..96a44bf 100644 --- a/src/mesa/drivers/dri/i965/brw_vtbl.c +++ b/src/mesa/drivers/dri/i965/brw_vtbl.c @@ -144,7 +144,7 @@ static void brw_finish_batch(struct intel_context *intel) brw_emit_query_end(brw); if (brw->curbe.curbe_bo) { - intel_bo_unmap_gtt_preferred(intel, brw->curbe.curbe_bo); + drm_intel_gem_bo_unmap_gtt(brw->curbe.curbe_bo); drm_intel_bo_unreference(brw->curbe.curbe_bo); brw->curbe.curbe_bo = NULL; } diff --git a/src/mesa/drivers/dri/i965/gen6_vs_state.c b/src/mesa/drivers/dri/i965/gen6_vs_state.c index 211a623..fe597df 100644 --- a/src/mesa/drivers/dri/i965/gen6_vs_state.c +++ b/src/mesa/drivers/dri/i965/gen6_vs_state.c @@ -65,13 +65,13 @@ upload_vs_state(struct brw_context *brw) constant_bo = drm_intel_bo_alloc(intel->bufmgr, "VS constant_bo", nr_params * 4 * sizeof(float), 4096); - intel_bo_map_gtt_preferred(intel, constant_bo, GL_TRUE); + drm_intel_gem_bo_map_gtt(constant_bo); for (i = 0; i < nr_params; i++) { memcpy((char *)constant_bo->virtual + i * 4 * sizeof(float), vp->program.Base.Parameters->ParameterValues[i], 4 * sizeof(float)); } - intel_bo_unmap_gtt_preferred(intel, constant_bo); + drm_intel_gem_bo_unmap_gtt(constant_bo); BEGIN_BATCH(5); OUT_BATCH(CMD_3D_CONSTANT_VS_STATE << 16 | diff --git a/src/mesa/drivers/dri/i965/gen6_wm_state.c b/src/mesa/drivers/dri/i965/gen6_wm_state.c index e355261..1eb17ca 100644 --- a/src/mesa/drivers/dri/i965/gen6_wm_state.c +++ b/src/mesa/drivers/dri/i965/gen6_wm_state.c @@ -63,13 +63,13 @@ upload_wm_state(struct brw_context *brw) constant_bo = drm_intel_bo_alloc(intel->bufmgr, "WM constant_bo", nr_params * 4 * sizeof(float), 4096); - intel_bo_map_gtt_preferred(intel, constant_bo, GL_TRUE); + drm_intel_gem_bo_map_gtt(constant_bo); for (i = 0; i < nr_params; i++) { memcpy((char *)constant_bo->virtual + i * 4 * sizeof(float), fp->program.Base.Parameters->ParameterValues[i], 4 * sizeof(float)); } - intel_bo_unmap_gtt_preferred(intel, constant_bo); + drm_intel_gem_bo_unmap_gtt(constant_bo); BEGIN_BATCH(5); OUT_BATCH(CMD_3D_CONSTANT_PS_STATE << 16 | diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c index 6d6af86..f2769aa 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.c +++ b/src/mesa/drivers/dri/intel/intel_blit.c @@ -122,8 +122,8 @@ intelEmitCopyBlit(struct intel_context *intel, intel_prepare_render(intel); if (pass >= 2) { - intel_bo_map_gtt_preferred(intel, dst_buffer, GL_TRUE); - intel_bo_map_gtt_preferred(intel, src_buffer, GL_FALSE); + drm_intel_gem_bo_map_gtt(dst_buffer); + drm_intel_gem_bo_map_gtt(src_buffer); _mesa_copy_rect((GLubyte *)dst_buffer->virtual + dst_offset, cpp, dst_pitch, @@ -132,8 +132,8 @@ intelEmitCopyBlit(struct intel_context *intel, (GLubyte *)src_buffer->virtual + src_offset, src_pitch, src_x, src_y); - intel_bo_unmap_gtt_preferred(intel, src_buffer); - intel_bo_unmap_gtt_preferred(intel, dst_buffer); + drm_intel_gem_bo_unmap_gtt(src_buffer); + drm_intel_gem_bo_unmap_gtt(dst_buffer); return GL_TRUE; } diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c index 312866d..8fe3a0e 100644 --- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c +++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c @@ -285,7 +285,7 @@ intel_bufferobj_map(GLcontext * ctx, return NULL; } - if (write_only && intel->intelScreen->kernel_exec_fencing) { + if (write_only) { drm_intel_gem_bo_map_gtt(intel_obj->buffer); intel_obj->mapped_gtt = GL_TRUE; } else { @@ -379,8 +379,7 @@ intel_bufferobj_map_range(GLcontext * ctx, intel_obj->range_map_bo = drm_intel_bo_alloc(intel->bufmgr, "range map", length, 64); - if (!(access & GL_MAP_READ_BIT) && - intel->intelScreen->kernel_exec_fencing) { + if (!(access & GL_MAP_READ_BIT)) { drm_intel_gem_bo_map_gtt(intel_obj->range_map_bo); intel_obj->mapped_gtt = GL_TRUE; } else { @@ -393,8 +392,7 @@ intel_bufferobj_map_range(GLcontext * ctx, return obj->Pointer; } - if (!(access & GL_MAP_READ_BIT) && - intel->intelScreen->kernel_exec_fencing) { + if (!(access & GL_MAP_READ_BIT)) { drm_intel_gem_bo_map_gtt(intel_obj->buffer); intel_obj->mapped_gtt = GL_TRUE; } else { diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index ec379a7..d6a1ba6 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -755,12 +755,6 @@ intelInitContext(struct intel_context *intel, } intel->use_texture_tiling = driQueryOptionb(&intel->optionCache, "texture_tiling"); - if (intel->use_texture_tiling && - !intel->intelScreen->kernel_exec_fencing) { - fprintf(stderr, "No kernel support for execution fencing, " - "disabling texture tiling\n"); - intel->use_texture_tiling = GL_FALSE; - } intel->use_early_z = driQueryOptionb(&intel->optionCache, "early_z"); intel->prim.primitive = ~0; diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h index d20d444..22736a9 100644 --- a/src/mesa/drivers/dri/intel/intel_context.h +++ b/src/mesa/drivers/dri/intel/intel_context.h @@ -475,25 +475,4 @@ is_power_of_two(uint32_t value) return (value & (value - 1)) == 0; } -static INLINE void -intel_bo_map_gtt_preferred(struct intel_context *intel, - drm_intel_bo *bo, - GLboolean write) -{ - if (intel->intelScreen->kernel_exec_fencing) - drm_intel_gem_bo_map_gtt(bo); - else - drm_intel_bo_map(bo, write); -} - -static INLINE void -intel_bo_unmap_gtt_preferred(struct intel_context *intel, - drm_intel_bo *bo) -{ - if (intel->intelScreen->kernel_exec_fencing) - drm_intel_gem_bo_unmap_gtt(bo); - else - drm_intel_bo_unmap(bo); -} - #endif diff --git a/src/mesa/drivers/dri/intel/intel_depthtmp.h b/src/mesa/drivers/dri/intel/intel_depthtmp.h deleted file mode 100644 index a9c75d4..0000000 --- a/src/mesa/drivers/dri/intel/intel_depthtmp.h +++ /dev/null @@ -1,64 +0,0 @@ -/* - * Copyright © 2009 Intel Corporation - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * 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. - * - * Authors: - * Eric Anholt - * - */ - -/** - * Wrapper around the depthtmp.h macrofest to generate spans code for - * all the tiling styles. - */ - -#define VALUE_TYPE INTEL_VALUE_TYPE -#define WRITE_DEPTH(_x, _y, d) \ - (*(INTEL_VALUE_TYPE *)(irb->region->buffer->virtual + \ - NO_TILE(_x, _y)) = d) -#define READ_DEPTH(d, _x, _y) \ - d = *(INTEL_VALUE_TYPE *)(irb->region->buffer->virtual + \ - NO_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel_gttmap_##x) -#include "depthtmp.h" - -#define VALUE_TYPE INTEL_VALUE_TYPE -#define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(NO_TILE(_x, _y), d) -#define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(NO_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel##x) -#include "depthtmp.h" - -#define VALUE_TYPE INTEL_VALUE_TYPE -#define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(X_TILE(_x, _y), d) -#define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(X_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel_XTile_##x) -#include "depthtmp.h" - -#define VALUE_TYPE INTEL_VALUE_TYPE -#define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(Y_TILE(_x, _y), d) -#define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(Y_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel_YTile_##x) -#include "depthtmp.h" - -#undef INTEL_VALUE_TYPE -#undef INTEL_WRITE_DEPTH -#undef INTEL_READ_DEPTH -#undef INTEL_TAG diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c index b388a3d..a429f8d 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.c +++ b/src/mesa/drivers/dri/intel/intel_fbo.c @@ -70,9 +70,6 @@ intel_delete_renderbuffer(struct gl_renderbuffer *rb) ASSERT(irb); - if (irb->span_cache != NULL) - free(irb->span_cache); - if (intel && irb->region) { intel_region_release(&irb->region); } diff --git a/src/mesa/drivers/dri/intel/intel_fbo.h b/src/mesa/drivers/dri/intel/intel_fbo.h index 586dbbb..72413f7 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.h +++ b/src/mesa/drivers/dri/intel/intel_fbo.h @@ -40,9 +40,6 @@ struct intel_renderbuffer { struct gl_renderbuffer Base; struct intel_region *region; - - uint8_t *span_cache; - unsigned long span_cache_offset; }; diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c index a20ea5a..4f14946 100644 --- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c +++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c @@ -119,8 +119,7 @@ intel_miptree_create(struct intel_context *intel, struct intel_mipmap_tree *mt; uint32_t tiling; - if (intel->use_texture_tiling && compress_byte == 0 && - intel->intelScreen->kernel_exec_fencing) { + if (intel->use_texture_tiling && compress_byte == 0) { if (intel->gen >= 4 && (base_format == GL_DEPTH_COMPONENT || base_format == GL_DEPTH_STENCIL_EXT)) diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c index e3c0635..f042bcb 100644 --- a/src/mesa/drivers/dri/intel/intel_regions.c +++ b/src/mesa/drivers/dri/intel/intel_regions.c @@ -118,8 +118,7 @@ intel_region_map(struct intel_context *intel, struct intel_region *region) if (region->pbo) intel_region_cow(intel, region); - if (region->tiling != I915_TILING_NONE && - intel->intelScreen->kernel_exec_fencing) + if (region->tiling != I915_TILING_NONE) drm_intel_gem_bo_map_gtt(region->buffer); else dri_bo_map(region->buffer, GL_TRUE); @@ -134,8 +133,7 @@ intel_region_unmap(struct intel_context *intel, struct intel_region *region) { _DBG("%s %p\n", __FUNCTION__, region); if (!--region->map_refcount) { - if (region->tiling != I915_TILING_NONE && - intel->intelScreen->kernel_exec_fencing) + if (region->tiling != I915_TILING_NONE) drm_intel_gem_bo_unmap_gtt(region->buffer); else dri_bo_unmap(region->buffer); diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c index 959ccfb..9ff76c4 100644 --- a/src/mesa/drivers/dri/intel/intel_screen.c +++ b/src/mesa/drivers/dri/intel/intel_screen.c @@ -416,10 +416,11 @@ intel_init_bufmgr(struct intel_screen *intelScreen) return GL_FALSE; } - if (intel_get_param(spriv, I915_PARAM_NUM_FENCES_AVAIL, &num_fences)) - intelScreen->kernel_exec_fencing = !!num_fences; - else - intelScreen->kernel_exec_fencing = GL_FALSE; + if (!intel_get_param(spriv, I915_PARAM_NUM_FENCES_AVAIL, &num_fences) || + num_fences == 0) { + fprintf(stderr, "[%s: %u] Kernel 2.6.29 required.\n", __func__, __LINE__); + return GL_FALSE; + } drm_intel_bufmgr_gem_enable_fenced_relocs(intelScreen->bufmgr); diff --git a/src/mesa/drivers/dri/intel/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h index 1ce476d..5863093 100644 --- a/src/mesa/drivers/dri/intel/intel_screen.h +++ b/src/mesa/drivers/dri/intel/intel_screen.h @@ -46,7 +46,6 @@ struct intel_screen GLboolean no_vbo; dri_bufmgr *bufmgr; - GLboolean kernel_exec_fencing; struct _mesa_HashTable *named_regions; /** diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 2fcb2d8..fb5c01b 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -43,218 +43,6 @@ static void intel_set_span_functions(struct intel_context *intel, struct gl_renderbuffer *rb); -#define SPAN_CACHE_SIZE 4096 - -static void -get_span_cache(struct intel_renderbuffer *irb, uint32_t offset) -{ - if (irb->span_cache == NULL) { - irb->span_cache = malloc(SPAN_CACHE_SIZE); - irb->span_cache_offset = -1; - } - - if ((offset & ~(SPAN_CACHE_SIZE - 1)) != irb->span_cache_offset) { - irb->span_cache_offset = offset & ~(SPAN_CACHE_SIZE - 1); - dri_bo_get_subdata(irb->region->buffer, irb->span_cache_offset, - SPAN_CACHE_SIZE, irb->span_cache); - } -} - -static void -clear_span_cache(struct intel_renderbuffer *irb) -{ - irb->span_cache_offset = -1; -} - -static uint32_t -pread_32(struct intel_renderbuffer *irb, uint32_t offset) -{ - get_span_cache(irb, offset); - - return *(uint32_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1))); -} - -static uint32_t -pread_xrgb8888(struct intel_renderbuffer *irb, uint32_t offset) -{ - get_span_cache(irb, offset); - - return *(uint32_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1))) | - 0xff000000; -} - -static uint16_t -pread_16(struct intel_renderbuffer *irb, uint32_t offset) -{ - get_span_cache(irb, offset); - - return *(uint16_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1))); -} - -static uint8_t -pread_8(struct intel_renderbuffer *irb, uint32_t offset) -{ - get_span_cache(irb, offset); - - return *(uint8_t *)(irb->span_cache + (offset & (SPAN_CACHE_SIZE - 1))); -} - -static void -pwrite_32(struct intel_renderbuffer *irb, uint32_t offset, uint32_t val) -{ - clear_span_cache(irb); - - dri_bo_subdata(irb->region->buffer, offset, 4, &val); -} - -static void -pwrite_xrgb8888(struct intel_renderbuffer *irb, uint32_t offset, uint32_t val) -{ - clear_span_cache(irb); - - dri_bo_subdata(irb->region->buffer, offset, 3, &val); -} - -static void -pwrite_16(struct intel_renderbuffer *irb, uint32_t offset, uint16_t val) -{ - clear_span_cache(irb); - - dri_bo_subdata(irb->region->buffer, offset, 2, &val); -} - -static void -pwrite_8(struct intel_renderbuffer *irb, uint32_t offset, uint8_t val) -{ - clear_span_cache(irb); - - dri_bo_subdata(irb->region->buffer, offset, 1, &val); -} - -static uint32_t no_tile_swizzle(struct intel_renderbuffer *irb, - int x, int y) -{ - return (y * irb->region->pitch + x) * irb->region->cpp; -} - -/* - * Deal with tiled surfaces - */ - -static uint32_t x_tile_swizzle(struct intel_renderbuffer *irb, - int x, int y) -{ - int tile_stride; - int xbyte; - int x_tile_off, y_tile_off; - int x_tile_number, y_tile_number; - int tile_off, tile_base; - - x += irb->region->draw_x; - y += irb->region->draw_y; - - tile_stride = (irb->region->pitch * irb->region->cpp) << 3; - - xbyte = x * irb->region->cpp; - - x_tile_off = xbyte & 0x1ff; - y_tile_off = y & 7; - - x_tile_number = xbyte >> 9; - y_tile_number = y >> 3; - - tile_off = (y_tile_off << 9) + x_tile_off; - - switch (irb->region->bit_6_swizzle) { - case I915_BIT_6_SWIZZLE_NONE: - break; - case I915_BIT_6_SWIZZLE_9: - tile_off ^= ((tile_off >> 3) & 64); - break; - case I915_BIT_6_SWIZZLE_9_10: - tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64); - break; - case I915_BIT_6_SWIZZLE_9_11: - tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 5) & 64); - break; - case I915_BIT_6_SWIZZLE_9_10_11: - tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64) ^ - ((tile_off >> 5) & 64); - break; - default: - fprintf(stderr, "Unknown tile swizzling mode %d\n", - irb->region->bit_6_swizzle); - exit(1); - } - - tile_base = (x_tile_number << 12) + y_tile_number * tile_stride; - -#if 0 - printf("(%d,%d) -> %d + %d = %d (pitch = %d, tstride = %d)\n", - x, y, tile_off, tile_base, - tile_off + tile_base, - irb->region->pitch, tile_stride); -#endif - - return tile_base + tile_off; -} - -static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb, - int x, int y) -{ - int tile_stride; - int xbyte; - int x_tile_off, y_tile_off; - int x_tile_number, y_tile_number; - int tile_off, tile_base; - - x += irb->region->draw_x; - y += irb->region->draw_y; - - tile_stride = (irb->region->pitch * irb->region->cpp) << 5; - - xbyte = x * irb->region->cpp; - - x_tile_off = xbyte & 0x7f; - y_tile_off = y & 0x1f; - - x_tile_number = xbyte >> 7; - y_tile_number = y >> 5; - - tile_off = ((x_tile_off & ~0xf) << 5) + (y_tile_off << 4) + - (x_tile_off & 0xf); - - switch (irb->region->bit_6_swizzle) { - case I915_BIT_6_SWIZZLE_NONE: - break; - case I915_BIT_6_SWIZZLE_9: - tile_off ^= ((tile_off >> 3) & 64); - break; - case I915_BIT_6_SWIZZLE_9_10: - tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64); - break; - case I915_BIT_6_SWIZZLE_9_11: - tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 5) & 64); - break; - case I915_BIT_6_SWIZZLE_9_10_11: - tile_off ^= ((tile_off >> 3) & 64) ^ ((tile_off >> 4) & 64) ^ - ((tile_off >> 5) & 64); - break; - default: - fprintf(stderr, "Unknown tile swizzling mode %d\n", - irb->region->bit_6_swizzle); - exit(1); - } - - tile_base = (x_tile_number << 12) + y_tile_number * tile_stride; - - return tile_base + tile_off; -} - -/* - break intelWriteRGBASpan_ARGB8888 -*/ - #undef DBG #define DBG 0 @@ -280,50 +68,43 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb, #define HW_UNLOCK() -/* Convenience macros to avoid typing the swizzle argument over and over */ -#define NO_TILE(_X, _Y) no_tile_swizzle(irb, (_X), (_Y)) -#define X_TILE(_X, _Y) x_tile_swizzle(irb, (_X), (_Y)) -#define Y_TILE(_X, _Y) y_tile_swizzle(irb, (_X), (_Y)) +/* Convenience macros to avoid typing the address argument over and over */ +#define NO_TILE(_X, _Y) (((_Y) * irb->region->pitch + (_X)) * irb->region->cpp) /* r5g6b5 color span and pixel functions */ -#define INTEL_PIXEL_FMT GL_RGB -#define INTEL_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5 -#define INTEL_READ_VALUE(offset) pread_16(irb, offset) -#define INTEL_WRITE_VALUE(offset, v) pwrite_16(irb, offset, v) -#define INTEL_TAG(x) x##_RGB565 -#include "intel_spantmp.h" +#define SPANTMP_PIXEL_FMT GL_RGB +#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5 +#define TAG(x) intel_##x##_RGB565 +#define TAG2(x,y) intel_##x##y_RGB565 +#include "spantmp2.h" /* a4r4g4b4 color span and pixel functions */ -#define INTEL_PIXEL_FMT GL_BGRA -#define INTEL_PIXEL_TYPE GL_UNSIGNED_SHORT_4_4_4_4_REV -#define INTEL_READ_VALUE(offset) pread_16(irb, offset) -#define INTEL_WRITE_VALUE(offset, v) pwrite_16(irb, offset, v) -#define INTEL_TAG(x) x##_ARGB4444 -#include "intel_spantmp.h" +#define SPANTMP_PIXEL_FMT GL_BGRA +#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_4_4_4_4_REV +#define TAG(x) intel_##x##_ARGB4444 +#define TAG2(x,y) intel_##x##y_ARGB4444 +#include "spantmp2.h" /* a1r5g5b5 color span and pixel functions */ -#define INTEL_PIXEL_FMT GL_BGRA -#define INTEL_PIXEL_TYPE GL_UNSIGNED_SHORT_1_5_5_5_REV -#define INTEL_READ_VALUE(offset) pread_16(irb, offset) -#define INTEL_WRITE_VALUE(offset, v) pwrite_16(irb, offset, v) -#define INTEL_TAG(x) x##_ARGB1555 -#include "intel_spantmp.h" +#define SPANTMP_PIXEL_FMT GL_BGRA +#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_1_5_5_5_REV +#define TAG(x) intel_##x##_ARGB1555 +#define TAG2(x,y) intel_##x##y##_ARGB1555 +#include "spantmp2.h" /* a8r8g8b8 color span and pixel functions */ -#define INTEL_PIXEL_FMT GL_BGRA -#define INTEL_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV -#define INTEL_READ_VALUE(offset) pread_32(irb, offset) -#define INTEL_WRITE_VALUE(offset, v) pwrite_32(irb, offset, v) -#define INTEL_TAG(x) x##_ARGB8888 -#include "intel_spantmp.h" +#define SPANTMP_PIXEL_FMT GL_BGRA +#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV +#define TAG(x) intel_##x##_ARGB8888 +#define TAG2(x,y) intel_##x##y##_ARGB8888 +#include "spantmp2.h" /* x8r8g8b8 color span and pixel functions */ -#define INTEL_PIXEL_FMT GL_BGR -#define INTEL_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV -#define INTEL_READ_VALUE(offset) pread_xrgb8888(irb, offset) -#define INTEL_WRITE_VALUE(offset, v) pwrite_xrgb8888(irb, offset, v) -#define INTEL_TAG(x) x##_xRGB8888 -#include "intel_spantmp.h" +#define SPANTMP_PIXEL_FMT GL_BGR +#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV +#define TAG(x) intel_##x##_xRGB8888 +#define TAG2(x,y) intel_##x##y##_xRGB8888 +#include "spantmp2.h" #define LOCAL_DEPTH_VARS \ struct intel_renderbuffer *irb = intel_renderbuffer(rb); \ @@ -339,18 +120,22 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb, #define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS /* z16 depthbuffer functions. */ -#define INTEL_VALUE_TYPE GLushort -#define INTEL_WRITE_DEPTH(offset, d) pwrite_16(irb, offset, d) -#define INTEL_READ_DEPTH(offset) pread_16(irb, offset) -#define INTEL_TAG(name) name##_z16 -#include "intel_depthtmp.h" - -/* z24x8 depthbuffer functions. */ -#define INTEL_VALUE_TYPE GLuint -#define INTEL_WRITE_DEPTH(offset, d) pwrite_32(irb, offset, d) -#define INTEL_READ_DEPTH(offset) pread_32(irb, offset) -#define INTEL_TAG(name) name##_z24_x8 -#include "intel_depthtmp.h" +#define VALUE_TYPE GLushort +#define WRITE_DEPTH(_x, _y, d) \ + (*(uint16_t *)(irb->region->buffer->virtual + NO_TILE(_x, _y)) = d) +#define READ_DEPTH(d, _x, _y) \ + d = *(uint16_t *)(irb->region->buffer->virtual + NO_TILE(_x, _y)) +#define TAG(x) intel_##x##_z16 +#include "depthtmp.h" + +/* z24_s8 and z24_x8 depthbuffer functions. */ +#define VALUE_TYPE GLuint +#define WRITE_DEPTH(_x, _y, d) \ + (*(uint32_t *)(irb->region->buffer->virtual + NO_TILE(_x, _y)) = d) +#define READ_DEPTH(d, _x, _y) \ + d = *(uint32_t *)(irb->region->buffer->virtual + NO_TILE(_x, _y)) +#define TAG(x) intel_##x##_z24_x8 +#include "depthtmp.h" void intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb) @@ -360,8 +145,7 @@ intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb) if (irb == NULL || irb->region == NULL) return; - if (intel->intelScreen->kernel_exec_fencing) - drm_intel_gem_bo_map_gtt(irb->region->buffer); + drm_intel_gem_bo_map_gtt(irb->region->buffer); intel_set_span_functions(intel, rb); } @@ -375,10 +159,7 @@ intel_renderbuffer_unmap(struct intel_context *intel, if (irb == NULL || irb->region == NULL) return; - if (intel->intelScreen->kernel_exec_fencing) - drm_intel_gem_bo_unmap_gtt(irb->region->buffer); - else - clear_span_cache(irb); + drm_intel_gem_bo_unmap_gtt(irb->region->buffer); rb->GetRow = NULL; rb->PutRow = NULL; @@ -558,158 +339,34 @@ intel_set_span_functions(struct intel_context *intel, struct gl_renderbuffer *rb) { struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb; - uint32_t tiling = irb->region->tiling; - - if (intel->intelScreen->kernel_exec_fencing) { - switch (irb->Base.Format) { - case MESA_FORMAT_RGB565: - intel_gttmap_InitPointers_RGB565(rb); - break; - case MESA_FORMAT_ARGB4444: - intel_gttmap_InitPointers_ARGB4444(rb); - break; - case MESA_FORMAT_ARGB1555: - intel_gttmap_InitPointers_ARGB1555(rb); - break; - case MESA_FORMAT_XRGB8888: - intel_gttmap_InitPointers_xRGB8888(rb); - break; - case MESA_FORMAT_ARGB8888: - intel_gttmap_InitPointers_ARGB8888(rb); - break; - case MESA_FORMAT_Z16: - intel_gttmap_InitDepthPointers_z16(rb); - break; - case MESA_FORMAT_X8_Z24: - case MESA_FORMAT_S8_Z24: - intel_gttmap_InitDepthPointers_z24_x8(rb); - break; - default: - _mesa_problem(NULL, - "Unexpected MesaFormat %d in intelSetSpanFunctions", - irb->Base.Format); - break; - } - return; - } - /* If in GEM mode, we need to do the tile address swizzling ourselves, - * instead of the fence registers handling it. - */ switch (irb->Base.Format) { case MESA_FORMAT_RGB565: - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitPointers_RGB565(rb); - break; - case I915_TILING_X: - intel_XTile_InitPointers_RGB565(rb); - break; - case I915_TILING_Y: - intel_YTile_InitPointers_RGB565(rb); - break; - } + intel_InitPointers_RGB565(rb); break; case MESA_FORMAT_ARGB4444: - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitPointers_ARGB4444(rb); - break; - case I915_TILING_X: - intel_XTile_InitPointers_ARGB4444(rb); - break; - case I915_TILING_Y: - intel_YTile_InitPointers_ARGB4444(rb); - break; - } + intel_InitPointers_ARGB4444(rb); break; case MESA_FORMAT_ARGB1555: - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitPointers_ARGB1555(rb); - break; - case I915_TILING_X: - intel_XTile_InitPointers_ARGB1555(rb); - break; - case I915_TILING_Y: - intel_YTile_InitPointers_ARGB1555(rb); - break; - } + intel_InitPointers_ARGB1555(rb); break; case MESA_FORMAT_XRGB8888: - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitPointers_xRGB8888(rb); - break; - case I915_TILING_X: - intel_XTile_InitPointers_xRGB8888(rb); - break; - case I915_TILING_Y: - intel_YTile_InitPointers_xRGB8888(rb); - break; - } + intel_InitPointers_xRGB8888(rb); break; case MESA_FORMAT_ARGB8888: - /* 8888 RGBA */ - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitPointers_ARGB8888(rb); - break; - case I915_TILING_X: - intel_XTile_InitPointers_ARGB8888(rb); - break; - case I915_TILING_Y: - intel_YTile_InitPointers_ARGB8888(rb); - break; - } + intel_InitPointers_ARGB8888(rb); break; case MESA_FORMAT_Z16: - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitDepthPointers_z16(rb); - break; - case I915_TILING_X: - intel_XTile_InitDepthPointers_z16(rb); - break; - case I915_TILING_Y: - intel_YTile_InitDepthPointers_z16(rb); - break; - } + intel_InitDepthPointers_z16(rb); break; case MESA_FORMAT_X8_Z24: case MESA_FORMAT_S8_Z24: - /* There are a few different ways SW asks us to access the S8Z24 data: - * Z24 depth-only depth reads - * S8Z24 depth reads - * S8Z24 stencil reads. - */ - if (rb->Format == MESA_FORMAT_S8_Z24) { - switch (tiling) { - case I915_TILING_NONE: - default: - intelInitDepthPointers_z24_x8(rb); - break; - case I915_TILING_X: - intel_XTile_InitDepthPointers_z24_x8(rb); - break; - case I915_TILING_Y: - intel_YTile_InitDepthPointers_z24_x8(rb); - break; - } - } else { - _mesa_problem(NULL, - "Unexpected ActualFormat in intelSetSpanFunctions"); - } + intel_InitDepthPointers_z24_x8(rb); break; default: _mesa_problem(NULL, - "Unexpected MesaFormat in intelSetSpanFunctions"); + "Unexpected MesaFormat %d in intelSetSpanFunctions", + irb->Base.Format); break; } } diff --git a/src/mesa/drivers/dri/intel/intel_spantmp.h b/src/mesa/drivers/dri/intel/intel_spantmp.h deleted file mode 100644 index bad0339..0000000 --- a/src/mesa/drivers/dri/intel/intel_spantmp.h +++ /dev/null @@ -1,67 +0,0 @@ -/* - * Copyright © 2009 Intel Corporation - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * 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. - * - * Authors: - * Eric Anholt - * - */ - -/** - * Wrapper around the spantmp.h macrofest to generate spans code for - * all the tiling styles. - */ - -#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT -#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE -#define TAG(x) INTEL_TAG(intel_gttmap_##x) -#define TAG2(x, y) INTEL_TAG(intel_gttmap_##x##y) -#include "spantmp2.h" - -#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT -#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE -#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(NO_TILE(_x, _y), v) -#define GET_VALUE(_x, _y) INTEL_READ_VALUE(NO_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel##x) -#define TAG2(x, y) INTEL_TAG(intel##x)##y -#include "spantmp2.h" - -#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT -#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE -#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(X_TILE(_x, _y), v) -#define GET_VALUE(_x, _y) INTEL_READ_VALUE(X_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel_XTile_##x) -#define TAG2(x, y) INTEL_TAG(intel_XTile_##x)##y -#include "spantmp2.h" - -#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT -#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE -#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(Y_TILE(_x, _y), v) -#define GET_VALUE(_x, _y) INTEL_READ_VALUE(Y_TILE(_x, _y)) -#define TAG(x) INTEL_TAG(intel_YTile_##x) -#define TAG2(x, y) INTEL_TAG(intel_YTile_##x)##y -#include "spantmp2.h" - -#undef INTEL_PIXEL_FMT -#undef INTEL_PIXEL_TYPE -#undef INTEL_WRITE_VALUE -#undef INTEL_READ_VALUE -#undef INTEL_TAG -- 2.7.4