freedreno: drop needs_rb_fbd
[mesa.git] / src / gallium / drivers / freedreno / a4xx / fd4_gmem.c
index e2115454d8682100c154b2688873c9cecee1faeb..32541fe06f02f8fe7b584e6d5e12f2bdd5856d28 100644 (file)
@@ -132,10 +132,10 @@ emit_mrt(struct fd_ringbuffer *ring, unsigned nr_bufs,
 }
 
 static bool
-use_hw_binning(struct fd_context *ctx)
+use_hw_binning(struct fd_batch *batch)
 {
-       struct fd_gmem_stateobj *gmem = &ctx->gmem;
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
+       struct fd_gmem_stateobj *gmem = &batch->ctx->gmem;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
 
        /* this seems to be a hw bug.. but this hack fixes piglit fbo-maxsize: */
        if ((pfb->width > 4096) && (pfb->height > 4096))
@@ -147,10 +147,10 @@ use_hw_binning(struct fd_context *ctx)
 /* transfer from gmem to system memory (ie. normal RAM) */
 
 static void
-emit_gmem2mem_surf(struct fd_context *ctx, bool stencil,
+emit_gmem2mem_surf(struct fd_batch *batch, bool stencil,
                uint32_t base, struct pipe_surface *psurf)
 {
-       struct fd_ringbuffer *ring = ctx->ring;
+       struct fd_ringbuffer *ring = batch->gmem;
        struct fd_resource *rsc = fd_resource(psurf->texture);
        enum pipe_format pformat = psurf->format;
        struct fd_resource_slice *slice;
@@ -180,17 +180,18 @@ emit_gmem2mem_surf(struct fd_context *ctx, bool stencil,
                        A4XX_RB_COPY_DEST_INFO_ENDIAN(ENDIAN_NONE) |
                        A4XX_RB_COPY_DEST_INFO_SWAP(fd4_pipe2swap(pformat)));
 
-       fd4_draw(ctx, ring, DI_PT_RECTLIST, IGNORE_VISIBILITY,
+       fd4_draw(batch, ring, DI_PT_RECTLIST, IGNORE_VISIBILITY,
                        DI_SRC_SEL_AUTO_INDEX, 2, 1, INDEX_SIZE_IGN, 0, 0, NULL);
 }
 
 static void
-fd4_emit_tile_gmem2mem(struct fd_context *ctx, struct fd_tile *tile)
+fd4_emit_tile_gmem2mem(struct fd_batch *batch, struct fd_tile *tile)
 {
+       struct fd_context *ctx = batch->ctx;
        struct fd4_context *fd4_ctx = fd4_context(ctx);
        struct fd_gmem_stateobj *gmem = &ctx->gmem;
-       struct fd_ringbuffer *ring = ctx->ring;
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
+       struct fd_ringbuffer *ring = batch->gmem;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
        struct fd4_emit emit = {
                        .debug = &ctx->debug,
                        .vtx = &fd4_ctx->solid_vbuf_state,
@@ -227,7 +228,7 @@ fd4_emit_tile_gmem2mem(struct fd_context *ctx, struct fd_tile *tile)
        OUT_PKT0(ring, REG_A4XX_GRAS_SU_MODE_CONTROL, 1);
        OUT_RING(ring, A4XX_GRAS_SU_MODE_CONTROL_LINEHALFWIDTH(0));
 
-       fd_wfi(ctx, ring);
+       fd_wfi(batch, ring);
 
        OUT_PKT0(ring, REG_A4XX_GRAS_CL_CLIP_CNTL, 1);
        OUT_RING(ring, 0x80000);      /* GRAS_CL_CLIP_CNTL */
@@ -269,22 +270,22 @@ fd4_emit_tile_gmem2mem(struct fd_context *ctx, struct fd_tile *tile)
        fd4_program_emit(ring, &emit, 0, NULL);
        fd4_emit_vertex_bufs(ring, &emit);
 
-       if (ctx->resolve & (FD_BUFFER_DEPTH | FD_BUFFER_STENCIL)) {
+       if (batch->resolve & (FD_BUFFER_DEPTH | FD_BUFFER_STENCIL)) {
                struct fd_resource *rsc = fd_resource(pfb->zsbuf->texture);
-               if (!rsc->stencil || (ctx->resolve & FD_BUFFER_DEPTH))
-                       emit_gmem2mem_surf(ctx, false, ctx->gmem.zsbuf_base[0], pfb->zsbuf);
-               if (rsc->stencil && (ctx->resolve & FD_BUFFER_STENCIL))
-                       emit_gmem2mem_surf(ctx, true, ctx->gmem.zsbuf_base[1], pfb->zsbuf);
+               if (!rsc->stencil || (batch->resolve & FD_BUFFER_DEPTH))
+                       emit_gmem2mem_surf(batch, false, ctx->gmem.zsbuf_base[0], pfb->zsbuf);
+               if (rsc->stencil && (batch->resolve & FD_BUFFER_STENCIL))
+                       emit_gmem2mem_surf(batch, true, ctx->gmem.zsbuf_base[1], pfb->zsbuf);
        }
 
-       if (ctx->resolve & FD_BUFFER_COLOR) {
+       if (batch->resolve & FD_BUFFER_COLOR) {
                unsigned i;
                for (i = 0; i < pfb->nr_cbufs; i++) {
                        if (!pfb->cbufs[i])
                                continue;
-                       if (!(ctx->resolve & (PIPE_CLEAR_COLOR0 << i)))
+                       if (!(batch->resolve & (PIPE_CLEAR_COLOR0 << i)))
                                continue;
-                       emit_gmem2mem_surf(ctx, false, gmem->cbuf_base[i], pfb->cbufs[i]);
+                       emit_gmem2mem_surf(batch, false, gmem->cbuf_base[i], pfb->cbufs[i]);
                }
        }
 
@@ -298,10 +299,10 @@ fd4_emit_tile_gmem2mem(struct fd_context *ctx, struct fd_tile *tile)
 /* transfer from system memory to gmem */
 
 static void
-emit_mem2gmem_surf(struct fd_context *ctx, uint32_t *bases,
+emit_mem2gmem_surf(struct fd_batch *batch, uint32_t *bases,
                struct pipe_surface **bufs, uint32_t nr_bufs, uint32_t bin_w)
 {
-       struct fd_ringbuffer *ring = ctx->ring;
+       struct fd_ringbuffer *ring = batch->gmem;
        struct pipe_surface *zsbufs[2];
 
        emit_mrt(ring, nr_bufs, bufs, bases, bin_w, false);
@@ -318,17 +319,18 @@ emit_mem2gmem_surf(struct fd_context *ctx, uint32_t *bases,
 
        fd4_emit_gmem_restore_tex(ring, nr_bufs, bufs);
 
-       fd4_draw(ctx, ring, DI_PT_RECTLIST, IGNORE_VISIBILITY,
+       fd4_draw(batch, ring, DI_PT_RECTLIST, IGNORE_VISIBILITY,
                        DI_SRC_SEL_AUTO_INDEX, 2, 1, INDEX_SIZE_IGN, 0, 0, NULL);
 }
 
 static void
-fd4_emit_tile_mem2gmem(struct fd_context *ctx, struct fd_tile *tile)
+fd4_emit_tile_mem2gmem(struct fd_batch *batch, struct fd_tile *tile)
 {
+       struct fd_context *ctx = batch->ctx;
        struct fd4_context *fd4_ctx = fd4_context(ctx);
        struct fd_gmem_stateobj *gmem = &ctx->gmem;
-       struct fd_ringbuffer *ring = ctx->ring;
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
+       struct fd_ringbuffer *ring = batch->gmem;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
        struct fd4_emit emit = {
                        .debug = &ctx->debug,
                        .vtx = &fd4_ctx->blit_vbuf_state,
@@ -455,14 +457,14 @@ fd4_emit_tile_mem2gmem(struct fd_context *ctx, struct fd_tile *tile)
        bin_w = gmem->bin_w;
        bin_h = gmem->bin_h;
 
-       if (fd_gmem_needs_restore(ctx, tile, FD_BUFFER_COLOR)) {
+       if (fd_gmem_needs_restore(batch, tile, FD_BUFFER_COLOR)) {
                emit.prog = &ctx->blit_prog[pfb->nr_cbufs - 1];
                emit.fp = NULL;      /* frag shader changed so clear cache */
                fd4_program_emit(ring, &emit, pfb->nr_cbufs, pfb->cbufs);
-               emit_mem2gmem_surf(ctx, gmem->cbuf_base, pfb->cbufs, pfb->nr_cbufs, bin_w);
+               emit_mem2gmem_surf(batch, gmem->cbuf_base, pfb->cbufs, pfb->nr_cbufs, bin_w);
        }
 
-       if (fd_gmem_needs_restore(ctx, tile, FD_BUFFER_DEPTH | FD_BUFFER_STENCIL)) {
+       if (fd_gmem_needs_restore(batch, tile, FD_BUFFER_DEPTH | FD_BUFFER_STENCIL)) {
                switch (pfb->zsbuf->format) {
                case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT:
                case PIPE_FORMAT_Z32_FLOAT:
@@ -493,7 +495,7 @@ fd4_emit_tile_mem2gmem(struct fd_context *ctx, struct fd_tile *tile)
                }
                emit.fp = NULL;      /* frag shader changed so clear cache */
                fd4_program_emit(ring, &emit, 1, &pfb->zsbuf);
-               emit_mem2gmem_surf(ctx, gmem->zsbuf_base, &pfb->zsbuf, 1, bin_w);
+               emit_mem2gmem_surf(batch, gmem->zsbuf_base, &pfb->zsbuf, 1, bin_w);
        }
 
        OUT_PKT0(ring, REG_A4XX_GRAS_SC_CONTROL, 1);
@@ -508,24 +510,24 @@ fd4_emit_tile_mem2gmem(struct fd_context *ctx, struct fd_tile *tile)
 }
 
 static void
-patch_draws(struct fd_context *ctx, enum pc_di_vis_cull_mode vismode)
+patch_draws(struct fd_batch *batch, enum pc_di_vis_cull_mode vismode)
 {
        unsigned i;
-       for (i = 0; i < fd_patch_num_elements(&ctx->draw_patches); i++) {
-               struct fd_cs_patch *patch = fd_patch_element(&ctx->draw_patches, i);
+       for (i = 0; i < fd_patch_num_elements(&batch->draw_patches); i++) {
+               struct fd_cs_patch *patch = fd_patch_element(&batch->draw_patches, i);
                *patch->cs = patch->val | DRAW4(0, 0, 0, vismode);
        }
-       util_dynarray_resize(&ctx->draw_patches, 0);
+       util_dynarray_resize(&batch->draw_patches, 0);
 }
 
 /* for rendering directly to system memory: */
 static void
-fd4_emit_sysmem_prep(struct fd_context *ctx)
+fd4_emit_sysmem_prep(struct fd_batch *batch)
 {
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
-       struct fd_ringbuffer *ring = ctx->ring;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
+       struct fd_ringbuffer *ring = batch->gmem;
 
-       fd4_emit_restore(ctx);
+       fd4_emit_restore(batch, ring);
 
        OUT_PKT0(ring, REG_A4XX_RB_FRAME_BUFFER_DIMENSION, 1);
        OUT_RING(ring, A4XX_RB_FRAME_BUFFER_DIMENSION_WIDTH(pfb->width) |
@@ -552,14 +554,15 @@ fd4_emit_sysmem_prep(struct fd_context *ctx)
        OUT_PKT0(ring, REG_A4XX_RB_RENDER_CONTROL, 1);
        OUT_RING(ring, 0x8);
 
-       patch_draws(ctx, IGNORE_VISIBILITY);
+       patch_draws(batch, IGNORE_VISIBILITY);
 }
 
 static void
-update_vsc_pipe(struct fd_context *ctx)
+update_vsc_pipe(struct fd_batch *batch)
 {
+       struct fd_context *ctx = batch->ctx;
        struct fd4_context *fd4_ctx = fd4_context(ctx);
-       struct fd_ringbuffer *ring = ctx->ring;
+       struct fd_ringbuffer *ring = batch->gmem;
        int i;
 
        OUT_PKT0(ring, REG_A4XX_VSC_SIZE_ADDRESS, 1);
@@ -592,11 +595,12 @@ update_vsc_pipe(struct fd_context *ctx)
 }
 
 static void
-emit_binning_pass(struct fd_context *ctx)
+emit_binning_pass(struct fd_batch *batch)
 {
+       struct fd_context *ctx = batch->ctx;
        struct fd_gmem_stateobj *gmem = &ctx->gmem;
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
-       struct fd_ringbuffer *ring = ctx->ring;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
+       struct fd_ringbuffer *ring = batch->gmem;
        int i;
 
        uint32_t x1 = gmem->minx;
@@ -635,10 +639,10 @@ emit_binning_pass(struct fd_context *ctx)
        }
 
        /* emit IB to binning drawcmds: */
-       ctx->emit_ib(ring, ctx->binning_start, ctx->binning_end);
+       ctx->emit_ib(ring, batch->binning);
 
-       fd_reset_wfi(ctx);
-       fd_wfi(ctx, ring);
+       fd_reset_wfi(batch);
+       fd_wfi(batch, ring);
 
        /* and then put stuff back the way it was: */
 
@@ -651,26 +655,32 @@ emit_binning_pass(struct fd_context *ctx)
                        A4XX_GRAS_SC_CONTROL_MSAA_SAMPLES(MSAA_ONE) |
                        A4XX_GRAS_SC_CONTROL_RASTER_MODE(0));
 
-       fd_event_write(ctx, ring, CACHE_FLUSH);
-       fd_wfi(ctx, ring);
+       fd_event_write(batch, ring, CACHE_FLUSH);
+       fd_wfi(batch, ring);
 }
 
 /* before first tile */
 static void
-fd4_emit_tile_init(struct fd_context *ctx)
+fd4_emit_tile_init(struct fd_batch *batch)
 {
-       struct fd_ringbuffer *ring = ctx->ring;
-       struct fd_gmem_stateobj *gmem = &ctx->gmem;
+       struct fd_ringbuffer *ring = batch->gmem;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
+       struct fd_gmem_stateobj *gmem = &batch->ctx->gmem;
 
-       fd4_emit_restore(ctx);
+       fd4_emit_restore(batch, ring);
 
        OUT_PKT0(ring, REG_A4XX_VSC_BIN_SIZE, 1);
        OUT_RING(ring, A4XX_VSC_BIN_SIZE_WIDTH(gmem->bin_w) |
                        A4XX_VSC_BIN_SIZE_HEIGHT(gmem->bin_h));
 
-       update_vsc_pipe(ctx);
+       update_vsc_pipe(batch);
+
+       fd_wfi(batch, ring);
+       OUT_PKT0(ring, REG_A4XX_RB_FRAME_BUFFER_DIMENSION, 1);
+       OUT_RING(ring, A4XX_RB_FRAME_BUFFER_DIMENSION_WIDTH(pfb->width) |
+                       A4XX_RB_FRAME_BUFFER_DIMENSION_HEIGHT(pfb->height));
 
-       if (use_hw_binning(ctx)) {
+       if (use_hw_binning(batch)) {
                OUT_PKT0(ring, REG_A4XX_RB_MODE_CONTROL, 1);
                OUT_RING(ring, A4XX_RB_MODE_CONTROL_WIDTH(gmem->bin_w) |
                                A4XX_RB_MODE_CONTROL_HEIGHT(gmem->bin_h));
@@ -681,11 +691,11 @@ fd4_emit_tile_init(struct fd_context *ctx)
                                0x8);
 
                /* emit hw binning pass: */
-               emit_binning_pass(ctx);
+               emit_binning_pass(batch);
 
-               patch_draws(ctx, USE_VISIBILITY);
+               patch_draws(batch, USE_VISIBILITY);
        } else {
-               patch_draws(ctx, IGNORE_VISIBILITY);
+               patch_draws(batch, IGNORE_VISIBILITY);
        }
 
        OUT_PKT0(ring, REG_A4XX_RB_MODE_CONTROL, 1);
@@ -696,10 +706,11 @@ fd4_emit_tile_init(struct fd_context *ctx)
 
 /* before mem2gmem */
 static void
-fd4_emit_tile_prep(struct fd_context *ctx, struct fd_tile *tile)
+fd4_emit_tile_prep(struct fd_batch *batch, struct fd_tile *tile)
 {
-       struct fd_ringbuffer *ring = ctx->ring;
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
+       struct fd_context *ctx = batch->ctx;
+       struct fd_ringbuffer *ring = batch->gmem;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
        struct fd_gmem_stateobj *gmem = &ctx->gmem;
 
        if (pfb->zsbuf) {
@@ -739,37 +750,30 @@ fd4_emit_tile_prep(struct fd_context *ctx, struct fd_tile *tile)
        } else {
                OUT_RING(ring, A4XX_GRAS_DEPTH_CONTROL_FORMAT(DEPTH4_NONE));
        }
-
-       if (ctx->needs_rb_fbd) {
-               fd_wfi(ctx, ring);
-               OUT_PKT0(ring, REG_A4XX_RB_FRAME_BUFFER_DIMENSION, 1);
-               OUT_RING(ring, A4XX_RB_FRAME_BUFFER_DIMENSION_WIDTH(pfb->width) |
-                               A4XX_RB_FRAME_BUFFER_DIMENSION_HEIGHT(pfb->height));
-               ctx->needs_rb_fbd = false;
-       }
 }
 
 /* before IB to rendering cmds: */
 static void
-fd4_emit_tile_renderprep(struct fd_context *ctx, struct fd_tile *tile)
+fd4_emit_tile_renderprep(struct fd_batch *batch, struct fd_tile *tile)
 {
+       struct fd_context *ctx = batch->ctx;
        struct fd4_context *fd4_ctx = fd4_context(ctx);
-       struct fd_ringbuffer *ring = ctx->ring;
+       struct fd_ringbuffer *ring = batch->gmem;
        struct fd_gmem_stateobj *gmem = &ctx->gmem;
-       struct pipe_framebuffer_state *pfb = &ctx->framebuffer;
+       struct pipe_framebuffer_state *pfb = &batch->framebuffer;
 
        uint32_t x1 = tile->xoff;
        uint32_t y1 = tile->yoff;
        uint32_t x2 = tile->xoff + tile->bin_w - 1;
        uint32_t y2 = tile->yoff + tile->bin_h - 1;
 
-       if (use_hw_binning(ctx)) {
+       if (use_hw_binning(batch)) {
                struct fd_vsc_pipe *pipe = &ctx->pipe[tile->p];
 
                assert(pipe->w * pipe->h);
 
-               fd_event_write(ctx, ring, HLSQ_FLUSH);
-               fd_wfi(ctx, ring);
+               fd_event_write(batch, ring, HLSQ_FLUSH);
+               fd_wfi(batch, ring);
 
                OUT_PKT0(ring, REG_A4XX_PC_VSTREAM_CONTROL, 1);
                OUT_RING(ring, A4XX_PC_VSTREAM_CONTROL_SIZE(pipe->w * pipe->h) |