/* * Copyright © 2015 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. */ #include #include #include #include #include #include "anv_private.h" #include "vk_format_info.h" #include "genxml/gen_macros.h" #include "genxml/genX_pack.h" #if GEN_GEN == 7 && !GEN_IS_HASWELL static int64_t clamp_int64(int64_t x, int64_t min, int64_t max) { if (x < min) return min; else if (x < max) return x; else return max; } void gen7_cmd_buffer_emit_scissor(struct anv_cmd_buffer *cmd_buffer) { struct anv_framebuffer *fb = cmd_buffer->state.framebuffer; uint32_t count = cmd_buffer->state.gfx.dynamic.scissor.count; const VkRect2D *scissors = cmd_buffer->state.gfx.dynamic.scissor.scissors; struct anv_state scissor_state = anv_cmd_buffer_alloc_dynamic_state(cmd_buffer, count * 8, 32); for (uint32_t i = 0; i < count; i++) { const VkRect2D *s = &scissors[i]; /* Since xmax and ymax are inclusive, we have to have xmax < xmin or * ymax < ymin for empty clips. In case clip x, y, width height are all * 0, the clamps below produce 0 for xmin, ymin, xmax, ymax, which isn't * what we want. Just special case empty clips and produce a canonical * empty clip. */ static const struct GEN7_SCISSOR_RECT empty_scissor = { .ScissorRectangleYMin = 1, .ScissorRectangleXMin = 1, .ScissorRectangleYMax = 0, .ScissorRectangleXMax = 0 }; const int max = 0xffff; uint32_t y_min = s->offset.y; uint32_t x_min = s->offset.x; uint32_t y_max = s->offset.y + s->extent.height - 1; uint32_t x_max = s->offset.x + s->extent.width - 1; /* Do this math using int64_t so overflow gets clamped correctly. */ if (cmd_buffer->level == VK_COMMAND_BUFFER_LEVEL_PRIMARY) { y_min = clamp_int64((uint64_t) y_min, cmd_buffer->state.render_area.offset.y, max); x_min = clamp_int64((uint64_t) x_min, cmd_buffer->state.render_area.offset.x, max); y_max = clamp_int64((uint64_t) y_max, 0, cmd_buffer->state.render_area.offset.y + cmd_buffer->state.render_area.extent.height - 1); x_max = clamp_int64((uint64_t) x_max, 0, cmd_buffer->state.render_area.offset.x + cmd_buffer->state.render_area.extent.width - 1); } else if (fb) { y_min = clamp_int64((uint64_t) y_min, 0, max); x_min = clamp_int64((uint64_t) x_min, 0, max); y_max = clamp_int64((uint64_t) y_max, 0, fb->height - 1); x_max = clamp_int64((uint64_t) x_max, 0, fb->width - 1); } struct GEN7_SCISSOR_RECT scissor = { .ScissorRectangleYMin = y_min, .ScissorRectangleXMin = x_min, .ScissorRectangleYMax = y_max, .ScissorRectangleXMax = x_max }; if (s->extent.width <= 0 || s->extent.height <= 0) { GEN7_SCISSOR_RECT_pack(NULL, scissor_state.map + i * 8, &empty_scissor); } else { GEN7_SCISSOR_RECT_pack(NULL, scissor_state.map + i * 8, &scissor); } } anv_batch_emit(&cmd_buffer->batch, GEN7_3DSTATE_SCISSOR_STATE_POINTERS, ssp) { ssp.ScissorRectPointer = scissor_state.offset; } } #endif static const uint32_t vk_to_gen_index_type[] = { [VK_INDEX_TYPE_UINT16] = INDEX_WORD, [VK_INDEX_TYPE_UINT32] = INDEX_DWORD, }; static const uint32_t restart_index_for_type[] = { [VK_INDEX_TYPE_UINT16] = UINT16_MAX, [VK_INDEX_TYPE_UINT32] = UINT32_MAX, }; void genX(CmdBindIndexBuffer)( VkCommandBuffer commandBuffer, VkBuffer _buffer, VkDeviceSize offset, VkIndexType indexType) { ANV_FROM_HANDLE(anv_cmd_buffer, cmd_buffer, commandBuffer); ANV_FROM_HANDLE(anv_buffer, buffer, _buffer); cmd_buffer->state.gfx.dirty |= ANV_CMD_DIRTY_INDEX_BUFFER; if (GEN_IS_HASWELL) cmd_buffer->state.restart_index = restart_index_for_type[indexType]; cmd_buffer->state.gfx.gen7.index_buffer = buffer; cmd_buffer->state.gfx.gen7.index_type = vk_to_gen_index_type[indexType]; cmd_buffer->state.gfx.gen7.index_offset = offset; } static uint32_t get_depth_format(struct anv_cmd_buffer *cmd_buffer) { const struct anv_render_pass *pass = cmd_buffer->state.pass; const struct anv_subpass *subpass = cmd_buffer->state.subpass; if (!subpass->depth_stencil_attachment) return D16_UNORM; struct anv_render_pass_attachment *att = &pass->attachments[subpass->depth_stencil_attachment->attachment]; switch (att->format) { case VK_FORMAT_D16_UNORM: case VK_FORMAT_D16_UNORM_S8_UINT: return D16_UNORM; case VK_FORMAT_X8_D24_UNORM_PACK32: case VK_FORMAT_D24_UNORM_S8_UINT: return D24_UNORM_X8_UINT; case VK_FORMAT_D32_SFLOAT: case VK_FORMAT_D32_SFLOAT_S8_UINT: return D32_FLOAT; default: return D16_UNORM; } } void genX(cmd_buffer_flush_dynamic_state)(struct anv_cmd_buffer *cmd_buffer) { struct anv_pipeline *pipeline = cmd_buffer->state.gfx.base.pipeline; struct anv_dynamic_state *d = &cmd_buffer->state.gfx.dynamic; if (cmd_buffer->state.gfx.dirty & (ANV_CMD_DIRTY_PIPELINE | ANV_CMD_DIRTY_RENDER_TARGETS | ANV_CMD_DIRTY_DYNAMIC_LINE_WIDTH | ANV_CMD_DIRTY_DYNAMIC_DEPTH_BIAS)) { uint32_t sf_dw[GENX(3DSTATE_SF_length)]; struct GENX(3DSTATE_SF) sf = { GENX(3DSTATE_SF_header), .DepthBufferSurfaceFormat = get_depth_format(cmd_buffer), .LineWidth = d->line_width, .GlobalDepthOffsetConstant = d->depth_bias.bias, .GlobalDepthOffsetScale = d->depth_bias.slope, .GlobalDepthOffsetClamp = d->depth_bias.clamp }; GENX(3DSTATE_SF_pack)(NULL, sf_dw, &sf); anv_batch_emit_merge(&cmd_buffer->batch, sf_dw, pipeline->gen7.sf); } if (cmd_buffer->state.gfx.dirty & (ANV_CMD_DIRTY_DYNAMIC_BLEND_CONSTANTS | ANV_CMD_DIRTY_DYNAMIC_STENCIL_REFERENCE)) { struct anv_state cc_state = anv_cmd_buffer_alloc_dynamic_state(cmd_buffer, GENX(COLOR_CALC_STATE_length) * 4, 64); struct GENX(COLOR_CALC_STATE) cc = { .BlendConstantColorRed = d->blend_constants[0], .BlendConstantColorGreen = d->blend_constants[1], .BlendConstantColorBlue = d->blend_constants[2], .BlendConstantColorAlpha = d->blend_constants[3], .StencilReferenceValue = d->stencil_reference.front & 0xff, .BackfaceStencilReferenceValue = d->stencil_reference.back & 0xff, }; GENX(COLOR_CALC_STATE_pack)(NULL, cc_state.map, &cc); anv_batch_emit(&cmd_buffer->batch, GENX(3DSTATE_CC_STATE_POINTERS), ccp) { ccp.ColorCalcStatePointer = cc_state.offset; } } if (cmd_buffer->state.gfx.dirty & (ANV_CMD_DIRTY_PIPELINE | ANV_CMD_DIRTY_RENDER_TARGETS | ANV_CMD_DIRTY_DYNAMIC_STENCIL_COMPARE_MASK | ANV_CMD_DIRTY_DYNAMIC_STENCIL_WRITE_MASK)) { uint32_t depth_stencil_dw[GENX(DEPTH_STENCIL_STATE_length)]; struct GENX(DEPTH_STENCIL_STATE) depth_stencil = { .StencilTestMask = d->stencil_compare_mask.front & 0xff, .StencilWriteMask = d->stencil_write_mask.front & 0xff, .BackfaceStencilTestMask = d->stencil_compare_mask.back & 0xff, .BackfaceStencilWriteMask = d->stencil_write_mask.back & 0xff, .StencilBufferWriteEnable = (d->stencil_write_mask.front || d->stencil_write_mask.back) && pipeline->writes_stencil, }; GENX(DEPTH_STENCIL_STATE_pack)(NULL, depth_stencil_dw, &depth_stencil); struct anv_state ds_state = anv_cmd_buffer_merge_dynamic(cmd_buffer, depth_stencil_dw, pipeline->gen7.depth_stencil_state, GENX(DEPTH_STENCIL_STATE_length), 64); anv_batch_emit(&cmd_buffer->batch, GENX(3DSTATE_DEPTH_STENCIL_STATE_POINTERS), dsp) { dsp.PointertoDEPTH_STENCIL_STATE = ds_state.offset; } } if (cmd_buffer->state.gfx.gen7.index_buffer && cmd_buffer->state.gfx.dirty & (ANV_CMD_DIRTY_PIPELINE | ANV_CMD_DIRTY_INDEX_BUFFER)) { struct anv_buffer *buffer = cmd_buffer->state.gfx.gen7.index_buffer; uint32_t offset = cmd_buffer->state.gfx.gen7.index_offset; #if GEN_IS_HASWELL anv_batch_emit(&cmd_buffer->batch, GEN75_3DSTATE_VF, vf) { vf.IndexedDrawCutIndexEnable = pipeline->primitive_restart; vf.CutIndex = cmd_buffer->state.restart_index; } #endif anv_batch_emit(&cmd_buffer->batch, GENX(3DSTATE_INDEX_BUFFER), ib) { #if !GEN_IS_HASWELL ib.CutIndexEnable = pipeline->primitive_restart; #endif ib.IndexFormat = cmd_buffer->state.gfx.gen7.index_type; ib.MOCS = anv_mocs_for_bo(cmd_buffer->device, buffer->address.bo); ib.BufferStartingAddress = anv_address_add(buffer->address, offset); ib.BufferEndingAddress = anv_address_add(buffer->address, buffer->size); } } cmd_buffer->state.gfx.dirty = 0; } void genX(cmd_buffer_enable_pma_fix)(struct anv_cmd_buffer *cmd_buffer, bool enable) { /* The NP PMA fix doesn't exist on gen7 */ } void genX(CmdSetEvent)( VkCommandBuffer commandBuffer, VkEvent event, VkPipelineStageFlags stageMask) { anv_finishme("Implement events on gen7"); } void genX(CmdResetEvent)( VkCommandBuffer commandBuffer, VkEvent event, VkPipelineStageFlags stageMask) { anv_finishme("Implement events on gen7"); } void genX(CmdWaitEvents)( VkCommandBuffer commandBuffer, uint32_t eventCount, const VkEvent* pEvents, VkPipelineStageFlags srcStageMask, VkPipelineStageFlags destStageMask, uint32_t memoryBarrierCount, const VkMemoryBarrier* pMemoryBarriers, uint32_t bufferMemoryBarrierCount, const VkBufferMemoryBarrier* pBufferMemoryBarriers, uint32_t imageMemoryBarrierCount, const VkImageMemoryBarrier* pImageMemoryBarriers) { anv_finishme("Implement events on gen7"); genX(CmdPipelineBarrier)(commandBuffer, srcStageMask, destStageMask, false, /* byRegion */ memoryBarrierCount, pMemoryBarriers, bufferMemoryBarrierCount, pBufferMemoryBarriers, imageMemoryBarrierCount, pImageMemoryBarriers); }