freedreno: rebind resource in all contexts
[mesa.git] / src / gallium / drivers / freedreno / freedreno_resource.c
index fcc8fc66189e0583b4d20f642023d6bf62f800b6..a4933acdf3bcacea3025ac730b19070bfe884ebc 100644 (file)
@@ -24,9 +24,9 @@
  *    Rob Clark <robclark@freedesktop.org>
  */
 
-#include "util/u_format.h"
-#include "util/u_format_rgtc.h"
-#include "util/u_format_zs.h"
+#include "util/format/u_format.h"
+#include "util/format/u_format_rgtc.h"
+#include "util/format/u_format_zs.h"
 #include "util/u_inlines.h"
 #include "util/u_transfer.h"
 #include "util/u_string.h"
 /**
  * Go through the entire state and see if the resource is bound
  * anywhere. If it is, mark the relevant state as dirty. This is
- * called on realloc_bo to ensure the neccessary state is re-
+ * called on realloc_bo to ensure the necessary state is re-
  * emitted so the GPU looks at the new backing bo.
  */
 static void
-rebind_resource(struct fd_context *ctx, struct pipe_resource *prsc)
+rebind_resource_in_ctx(struct fd_context *ctx, struct fd_resource *rsc)
 {
+       struct pipe_resource *prsc = &rsc->base;
+
        /* VBOs */
-       for (unsigned i = 0; i < ctx->vtx.vertexbuf.count && !(ctx->dirty & FD_DIRTY_VTXBUF); i++) {
-               if (ctx->vtx.vertexbuf.vb[i].buffer.resource == prsc)
-                       ctx->dirty |= FD_DIRTY_VTXBUF;
+       if (rsc->dirty & FD_DIRTY_VTXBUF) {
+               struct fd_vertexbuf_stateobj *vb = &ctx->vtx.vertexbuf;
+               for (unsigned i = 0; i < vb->count && !(ctx->dirty & FD_DIRTY_VTXBUF); i++) {
+                       if (vb->vb[i].buffer.resource == prsc)
+                               ctx->dirty |= FD_DIRTY_VTXBUF;
+               }
        }
 
+       const enum fd_dirty_3d_state per_stage_dirty =
+                       FD_DIRTY_CONST | FD_DIRTY_TEX | FD_DIRTY_IMAGE | FD_DIRTY_SSBO;
+
+       if (!(rsc->dirty & per_stage_dirty))
+               return;
+
        /* per-shader-stage resources: */
        for (unsigned stage = 0; stage < PIPE_SHADER_TYPES; stage++) {
                /* Constbufs.. note that constbuf[0] is normal uniforms emitted in
                 * cmdstream rather than by pointer..
                 */
-               const unsigned num_ubos = util_last_bit(ctx->constbuf[stage].enabled_mask);
-               for (unsigned i = 1; i < num_ubos; i++) {
-                       if (ctx->dirty_shader[stage] & FD_DIRTY_SHADER_CONST)
-                               break;
-                       if (ctx->constbuf[stage].cb[i].buffer == prsc)
-                               ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_CONST;
+               if ((rsc->dirty & FD_DIRTY_CONST) &&
+                               !(ctx->dirty_shader[stage] & FD_DIRTY_CONST)) {
+                       struct fd_constbuf_stateobj *cb = &ctx->constbuf[stage];
+                       const unsigned num_ubos = util_last_bit(cb->enabled_mask);
+                       for (unsigned i = 1; i < num_ubos; i++) {
+                               if (cb->cb[i].buffer == prsc) {
+                                       ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_CONST;
+                                       ctx->dirty |= FD_DIRTY_CONST;
+                                       break;
+                               }
+                       }
                }
 
                /* Textures */
-               for (unsigned i = 0; i < ctx->tex[stage].num_textures; i++) {
-                       if (ctx->dirty_shader[stage] & FD_DIRTY_SHADER_TEX)
-                               break;
-                       if (ctx->tex[stage].textures[i] && (ctx->tex[stage].textures[i]->texture == prsc))
-                               ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_TEX;
+               if ((rsc->dirty & FD_DIRTY_TEX) &&
+                               !(ctx->dirty_shader[stage] & FD_DIRTY_TEX)) {
+                       struct fd_texture_stateobj *tex = &ctx->tex[stage];
+                       for (unsigned i = 0; i < tex->num_textures; i++) {
+                               if (tex->textures[i] && (tex->textures[i]->texture == prsc)) {
+                                       ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_TEX;
+                                       ctx->dirty |= FD_DIRTY_TEX;
+                                       break;
+                               }
+                       }
                }
 
                /* Images */
-               const unsigned num_images = util_last_bit(ctx->shaderimg[stage].enabled_mask);
-               for (unsigned i = 0; i < num_images; i++) {
-                       if (ctx->dirty_shader[stage] & FD_DIRTY_SHADER_IMAGE)
-                               break;
-                       if (ctx->shaderimg[stage].si[i].resource == prsc)
-                               ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_IMAGE;
+               if ((rsc->dirty & FD_DIRTY_IMAGE) &&
+                               !(ctx->dirty_shader[stage] & FD_DIRTY_IMAGE)) {
+                       struct fd_shaderimg_stateobj *si = &ctx->shaderimg[stage];
+                       const unsigned num_images = util_last_bit(si->enabled_mask);
+                       for (unsigned i = 0; i < num_images; i++) {
+                               if (si->si[i].resource == prsc) {
+                                       ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_IMAGE;
+                                       ctx->dirty |= FD_DIRTY_IMAGE;
+                                       break;
+                               }
+                       }
                }
 
                /* SSBOs */
-               const unsigned num_ssbos = util_last_bit(ctx->shaderbuf[stage].enabled_mask);
-               for (unsigned i = 0; i < num_ssbos; i++) {
-                       if (ctx->dirty_shader[stage] & FD_DIRTY_SHADER_SSBO)
-                               break;
-                       if (ctx->shaderbuf[stage].sb[i].buffer == prsc)
-                               ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_SSBO;
+               if ((rsc->dirty & FD_DIRTY_SSBO) &&
+                               !(ctx->dirty_shader[stage] & FD_DIRTY_SSBO)) {
+                       struct fd_shaderbuf_stateobj *sb = &ctx->shaderbuf[stage];
+                       const unsigned num_ssbos = util_last_bit(sb->enabled_mask);
+                       for (unsigned i = 0; i < num_ssbos; i++) {
+                               if (sb->sb[i].buffer == prsc) {
+                                       ctx->dirty_shader[stage] |= FD_DIRTY_SHADER_SSBO;
+                                       ctx->dirty |= FD_DIRTY_SSBO;
+                                       break;
+                               }
+                       }
                }
        }
 }
 
+static void
+rebind_resource(struct fd_resource *rsc)
+{
+       struct fd_screen *screen = fd_screen(rsc->base.screen);
+
+       mtx_lock(&screen->lock);
+       fd_resource_lock(rsc);
+
+       if (rsc->dirty)
+               list_for_each_entry (struct fd_context, ctx, &screen->context_list, node)
+                       rebind_resource_in_ctx(ctx, rsc);
+
+       fd_resource_unlock(rsc);
+       mtx_unlock(&screen->lock);
+}
+
 static void
 realloc_bo(struct fd_resource *rsc, uint32_t size)
 {
@@ -132,7 +179,20 @@ realloc_bo(struct fd_resource *rsc, uint32_t size)
                fd_bo_del(rsc->bo);
 
        rsc->bo = fd_bo_new(screen->dev, size, flags, "%ux%ux%u@%u:%x",
-                       prsc->width0, prsc->height0, prsc->depth0, rsc->cpp, prsc->bind);
+                       prsc->width0, prsc->height0, prsc->depth0, rsc->layout.cpp, prsc->bind);
+
+       /* Zero out the UBWC area on allocation.  This fixes intermittent failures
+        * with UBWC, which I suspect are due to the HW having a hard time
+        * interpreting arbitrary values populating the flags buffer when the BO
+        * was recycled through the bo cache (instead of fresh allocations from
+        * the kernel, which are zeroed).  sleep(1) in this spot didn't work
+        * around the issue, but any memset value seems to.
+        */
+       if (rsc->layout.ubwc) {
+               void *buf = fd_bo_map(rsc->bo);
+               memset(buf, 0, rsc->layout.slices[0].offset);
+       }
+
        rsc->seqno = p_atomic_inc_return(&screen->rsc_seqno);
        util_range_set_empty(&rsc->valid_buffer_range);
        fd_bc_invalidate_resource(rsc, true);
@@ -223,10 +283,7 @@ fd_try_shadow_resource(struct fd_context *ctx, struct fd_resource *rsc,
        /* TODO valid_buffer_range?? */
        swap(rsc->bo,        shadow->bo);
        swap(rsc->write_batch,   shadow->write_batch);
-       swap(rsc->offset, shadow->offset);
-       swap(rsc->ubwc_offset, shadow->ubwc_offset);
-       swap(rsc->ubwc_pitch, shadow->ubwc_pitch);
-       swap(rsc->ubwc_size, shadow->ubwc_size);
+       swap(rsc->layout, shadow->layout);
        rsc->seqno = p_atomic_inc_return(&ctx->screen->rsc_seqno);
 
        /* at this point, the newly created shadow buffer is not referenced
@@ -268,7 +325,10 @@ fd_try_shadow_resource(struct fd_context *ctx, struct fd_resource *rsc,
                set_box(box.height, u_minify(prsc->height0, l));
                set_box(box.depth,  u_minify(prsc->depth0, l));
 
-               do_blit(ctx, &blit, fallback);
+               for (int i = 0; i < prsc->array_size; i++) {
+                       set_box(box.z, i);
+                       do_blit(ctx, &blit, fallback);
+               }
        }
 
        /* deal w/ current level specially, since we might need to split
@@ -327,17 +387,7 @@ fd_resource_uncompress(struct fd_context *ctx, struct fd_resource *rsc)
        /* shadow should not fail in any cases where we need to uncompress: */
        debug_assert(success);
 
-       /*
-        * TODO what if rsc is used in other contexts, we don't currently
-        * have a good way to rebind_resource() in other contexts.  And an
-        * app that is reading one resource in multiple contexts, isn't
-        * going to expect that the resource is modified.
-        *
-        * Hopefully the edge cases where we need to uncompress are rare
-        * enough that they mostly only show up in deqp.
-        */
-
-       rebind_resource(ctx, &rsc->base);
+       rebind_resource(rsc);
 }
 
 static struct fd_resource *
@@ -419,7 +469,7 @@ static void fd_resource_transfer_flush_region(struct pipe_context *pctx,
        struct fd_resource *rsc = fd_resource(ptrans->resource);
 
        if (ptrans->resource->target == PIPE_BUFFER)
-               util_range_add(&rsc->valid_buffer_range,
+               util_range_add(&rsc->base, &rsc->valid_buffer_range,
                                           ptrans->box.x + box->x,
                                           ptrans->box.x + box->x + box->width);
 }
@@ -449,15 +499,14 @@ flush_resource(struct fd_context *ctx, struct fd_resource *rsc, unsigned usage)
                mtx_unlock(&ctx->screen->lock);
 
                foreach_batch(batch, &ctx->screen->batch_cache, batch_mask)
-                       fd_batch_flush(batch, false);
+                       fd_batch_flush(batch);
 
                foreach_batch(batch, &ctx->screen->batch_cache, batch_mask) {
-                       fd_batch_sync(batch);
                        fd_batch_reference(&batches[batch->idx], NULL);
                }
                assert(rsc->batch_mask == 0);
        } else if (write_batch) {
-               fd_batch_flush(write_batch, true);
+               fd_batch_flush(write_batch);
        }
 
        fd_batch_reference(&write_batch, NULL);
@@ -489,7 +538,7 @@ fd_resource_transfer_unmap(struct pipe_context *pctx,
                fd_bo_cpu_fini(rsc->bo);
        }
 
-       util_range_add(&rsc->valid_buffer_range,
+       util_range_add(&rsc->base, &rsc->valid_buffer_range,
                                   ptrans->box.x,
                                   ptrans->box.x + ptrans->box.width);
 
@@ -506,7 +555,7 @@ fd_resource_transfer_map(struct pipe_context *pctx,
 {
        struct fd_context *ctx = fd_context(pctx);
        struct fd_resource *rsc = fd_resource(prsc);
-       struct fd_resource_slice *slice = fd_resource_slice(rsc, level);
+       struct fdl_slice *slice = fd_resource_slice(rsc, level);
        struct fd_transfer *trans;
        struct pipe_transfer *ptrans;
        enum pipe_format format = prsc->format;
@@ -530,8 +579,8 @@ fd_resource_transfer_map(struct pipe_context *pctx,
        ptrans->level = level;
        ptrans->usage = usage;
        ptrans->box = *box;
-       ptrans->stride = util_format_get_nblocksx(format, slice->pitch) * rsc->cpp;
-       ptrans->layer_stride = rsc->layer_first ? rsc->layer_size : slice->size0;
+       ptrans->stride = slice->pitch;
+       ptrans->layer_stride = fd_resource_layer_stride(rsc, level);
 
        /* we always need a staging texture for tiled buffers:
         *
@@ -539,17 +588,17 @@ fd_resource_transfer_map(struct pipe_context *pctx,
         * splitting a batch.. for ex, mid-frame texture uploads to a tiled
         * texture.
         */
-       if (rsc->tile_mode) {
+       if (rsc->layout.tile_mode) {
                struct fd_resource *staging_rsc;
 
                staging_rsc = fd_alloc_staging(ctx, rsc, level, box);
                if (staging_rsc) {
+                       struct fdl_slice *staging_slice =
+                               fd_resource_slice(staging_rsc, 0);
                        // TODO for PIPE_TRANSFER_READ, need to do untiling blit..
                        trans->staging_prsc = &staging_rsc->base;
-                       trans->base.stride = util_format_get_nblocksx(format,
-                               staging_rsc->slices[0].pitch) * staging_rsc->cpp;
-                       trans->base.layer_stride = staging_rsc->layer_first ?
-                               staging_rsc->layer_size : staging_rsc->slices[0].size0;
+                       trans->base.stride = staging_slice->pitch;
+                       trans->base.layer_stride = fd_resource_layer_stride(staging_rsc, 0);
                        trans->staging_box = *box;
                        trans->staging_box.x = 0;
                        trans->staging_box.y = 0;
@@ -558,21 +607,6 @@ fd_resource_transfer_map(struct pipe_context *pctx,
                        if (usage & PIPE_TRANSFER_READ) {
                                fd_blit_to_staging(ctx, trans);
 
-                               struct fd_batch *batch = NULL;
-
-                               fd_context_lock(ctx);
-                               fd_batch_reference_locked(&batch, staging_rsc->write_batch);
-                               fd_context_unlock(ctx);
-
-                               /* we can't fd_bo_cpu_prep() until the blit to staging
-                                * is submitted to kernel.. in that case write_batch
-                                * wouldn't be NULL yet:
-                                */
-                               if (batch) {
-                                       fd_batch_sync(batch);
-                                       fd_batch_reference(&batch, NULL);
-                               }
-
                                fd_bo_cpu_prep(staging_rsc->bo, ctx->pipe,
                                                DRM_FREEDRENO_PREP_READ);
                        }
@@ -597,9 +631,13 @@ fd_resource_transfer_map(struct pipe_context *pctx,
        if (usage & PIPE_TRANSFER_WRITE)
                op |= DRM_FREEDRENO_PREP_WRITE;
 
+       bool needs_flush = pending(rsc, !!(usage & PIPE_TRANSFER_WRITE));
+
        if (usage & PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE) {
-               realloc_bo(rsc, fd_bo_size(rsc->bo));
-               rebind_resource(ctx, prsc);
+               if (needs_flush || fd_resource_busy(rsc, op)) {
+                       realloc_bo(rsc, fd_bo_size(rsc->bo));
+                       rebind_resource(rsc);
+               }
        } else if ((usage & PIPE_TRANSFER_WRITE) &&
                           prsc->target == PIPE_BUFFER &&
                           !util_ranges_intersect(&rsc->valid_buffer_range,
@@ -624,9 +662,7 @@ fd_resource_transfer_map(struct pipe_context *pctx,
                /* If the GPU is writing to the resource, or if it is reading from the
                 * resource and we're trying to write to it, flush the renders.
                 */
-               bool needs_flush = pending(rsc, !!(usage & PIPE_TRANSFER_WRITE));
-               bool busy = needs_flush || (0 != fd_bo_cpu_prep(rsc->bo,
-                               ctx->pipe, op | DRM_FREEDRENO_PREP_NOSYNC));
+               bool busy = needs_flush || fd_resource_busy(rsc, op);
 
                /* if we need to flush/stall, see if we can make a shadow buffer
                 * to avoid this:
@@ -643,7 +679,7 @@ fd_resource_transfer_map(struct pipe_context *pctx,
                        if (needs_flush && fd_try_shadow_resource(ctx, rsc, level,
                                                        box, DRM_FORMAT_MOD_LINEAR)) {
                                needs_flush = busy = false;
-                               rebind_resource(ctx, prsc);
+                               rebind_resource(rsc);
                                ctx->stats.shadow_uploads++;
                        } else {
                                struct fd_resource *staging_rsc;
@@ -660,11 +696,12 @@ fd_resource_transfer_map(struct pipe_context *pctx,
                                 */
                                staging_rsc = fd_alloc_staging(ctx, rsc, level, box);
                                if (staging_rsc) {
+                                       struct fdl_slice *staging_slice =
+                                               fd_resource_slice(staging_rsc, 0);
                                        trans->staging_prsc = &staging_rsc->base;
-                                       trans->base.stride = util_format_get_nblocksx(format,
-                                               staging_rsc->slices[0].pitch) * staging_rsc->cpp;
-                                       trans->base.layer_stride = staging_rsc->layer_first ?
-                                               staging_rsc->layer_size : staging_rsc->slices[0].size0;
+                                       trans->base.stride = staging_slice->pitch;
+                                       trans->base.layer_stride =
+                                               fd_resource_layer_stride(staging_rsc, 0);
                                        trans->staging_box = *box;
                                        trans->staging_box.x = 0;
                                        trans->staging_box.y = 0;
@@ -704,7 +741,7 @@ fd_resource_transfer_map(struct pipe_context *pctx,
        buf = fd_bo_map(rsc->bo);
        offset =
                box->y / util_format_get_blockheight(format) * ptrans->stride +
-               box->x / util_format_get_blockwidth(format) * rsc->cpp +
+               box->x / util_format_get_blockwidth(format) * rsc->layout.cpp +
                fd_resource_offset(rsc, level, box->z);
 
        if (usage & PIPE_TRANSFER_WRITE)
@@ -731,16 +768,17 @@ fd_resource_destroy(struct pipe_screen *pscreen,
                renderonly_scanout_destroy(rsc->scanout, fd_screen(pscreen)->ro);
 
        util_range_destroy(&rsc->valid_buffer_range);
+       simple_mtx_destroy(&rsc->lock);
        FREE(rsc);
 }
 
 static uint64_t
 fd_resource_modifier(struct fd_resource *rsc)
 {
-       if (!rsc->tile_mode)
+       if (!rsc->layout.tile_mode)
                return DRM_FORMAT_MOD_LINEAR;
 
-       if (rsc->ubwc_size)
+       if (rsc->layout.ubwc_layer_size)
                return DRM_FORMAT_MOD_QCOM_COMPRESSED;
 
        /* TODO invent a modifier for tiled but not UBWC buffers: */
@@ -759,7 +797,7 @@ fd_resource_get_handle(struct pipe_screen *pscreen,
        handle->modifier = fd_resource_modifier(rsc);
 
        return fd_screen_bo_get_handle(pscreen, rsc->bo, rsc->scanout,
-                       rsc->slices[0].pitch * rsc->cpp, handle);
+                       fd_resource_slice(rsc, 0)->pitch, handle);
 }
 
 static uint32_t
@@ -776,17 +814,17 @@ setup_slices(struct fd_resource *rsc, uint32_t alignment, enum pipe_format forma
        /* in layer_first layout, the level (slice) contains just one
         * layer (since in fact the layer contains the slices)
         */
-       uint32_t layers_in_level = rsc->layer_first ? 1 : prsc->array_size;
+       uint32_t layers_in_level = rsc->layout.layer_first ? 1 : prsc->array_size;
 
        for (level = 0; level <= prsc->last_level; level++) {
-               struct fd_resource_slice *slice = fd_resource_slice(rsc, level);
+               struct fdl_slice *slice = fd_resource_slice(rsc, level);
                uint32_t blocks;
 
                if (layout == UTIL_FORMAT_LAYOUT_ASTC)
-                       slice->pitch = width =
-                               util_align_npot(width, pitchalign * util_format_get_blockwidth(format));
+                       width = util_align_npot(width, pitchalign * util_format_get_blockwidth(format));
                else
-                       slice->pitch = width = align(width, pitchalign);
+                       width = align(width, pitchalign);
+               slice->pitch = util_format_get_nblocksx(format, width) * rsc->layout.cpp;
                slice->offset = size;
                blocks = util_format_get_nblocks(format, width, height);
                /* 1d array and 2d array textures must all have the same layer size
@@ -797,12 +835,12 @@ setup_slices(struct fd_resource *rsc, uint32_t alignment, enum pipe_format forma
                 */
                if (prsc->target == PIPE_TEXTURE_3D && (
                                        level == 1 ||
-                                       (level > 1 && rsc->slices[level - 1].size0 > 0xf000)))
-                       slice->size0 = align(blocks * rsc->cpp, alignment);
-               else if (level == 0 || rsc->layer_first || alignment == 1)
-                       slice->size0 = align(blocks * rsc->cpp, alignment);
+                                       (level > 1 && fd_resource_slice(rsc, level - 1)->size0 > 0xf000)))
+                       slice->size0 = align(blocks * rsc->layout.cpp, alignment);
+               else if (level == 0 || rsc->layout.layer_first || alignment == 1)
+                       slice->size0 = align(blocks * rsc->layout.cpp, alignment);
                else
-                       slice->size0 = rsc->slices[level - 1].size0;
+                       slice->size0 = fd_resource_slice(rsc, level - 1)->size0;
 
                size += slice->size0 * depth * layers_in_level;
 
@@ -848,10 +886,10 @@ fd_setup_slices(struct fd_resource *rsc)
        if (is_a4xx(screen)) {
                switch (rsc->base.target) {
                case PIPE_TEXTURE_3D:
-                       rsc->layer_first = false;
+                       rsc->layout.layer_first = false;
                        break;
                default:
-                       rsc->layer_first = true;
+                       rsc->layout.layer_first = true;
                        alignment = 1;
                        break;
                }
@@ -874,23 +912,19 @@ fd_resource_resize(struct pipe_resource *prsc, uint32_t sz)
        realloc_bo(rsc, fd_screen(prsc->screen)->setup_slices(rsc));
 }
 
-// TODO common helper?
-static bool
-has_depth(enum pipe_format format)
+static void
+fd_resource_layout_init(struct pipe_resource *prsc)
 {
-       switch (format) {
-       case PIPE_FORMAT_Z16_UNORM:
-       case PIPE_FORMAT_Z32_UNORM:
-       case PIPE_FORMAT_Z32_FLOAT:
-       case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT:
-       case PIPE_FORMAT_Z24_UNORM_S8_UINT:
-       case PIPE_FORMAT_S8_UINT_Z24_UNORM:
-       case PIPE_FORMAT_Z24X8_UNORM:
-       case PIPE_FORMAT_X8Z24_UNORM:
-               return true;
-       default:
-               return false;
-       }
+       struct fd_resource *rsc = fd_resource(prsc);
+       struct fdl_layout *layout = &rsc->layout;
+
+       layout->width0 = prsc->width0;
+       layout->height0 = prsc->height0;
+       layout->depth0 = prsc->depth0;
+
+       layout->cpp = util_format_get_blocksize(prsc->format);
+       layout->cpp *= fd_resource_nr_samples(prsc);
+       layout->cpp_shift = ffs(layout->cpp) - 1;
 }
 
 /**
@@ -953,6 +987,7 @@ fd_resource_create_with_modifiers(struct pipe_screen *pscreen,
                return NULL;
 
        *prsc = *tmpl;
+       fd_resource_layout_init(prsc);
 
 #define LINEAR \
        (PIPE_BIND_SCANOUT | \
@@ -963,6 +998,9 @@ fd_resource_create_with_modifiers(struct pipe_screen *pscreen,
        if (tmpl->bind & LINEAR)
                linear = true;
 
+       if (fd_mesa_debug & FD_DBG_NOTILE)
+               linear = true;
+
        /* Normally, for non-shared buffers, allow buffer compression if
         * not shared, otherwise only allow if QCOM_COMPRESSED modifier
         * is requested:
@@ -984,48 +1022,25 @@ fd_resource_create_with_modifiers(struct pipe_screen *pscreen,
        if (screen->tile_mode &&
                        (tmpl->target != PIPE_BUFFER) &&
                        !linear) {
-               rsc->tile_mode = screen->tile_mode(prsc);
+               rsc->layout.tile_mode = screen->tile_mode(prsc);
        }
 
        util_range_init(&rsc->valid_buffer_range);
 
-       rsc->internal_format = format;
-       rsc->cpp = util_format_get_blocksize(format);
-       rsc->cpp *= fd_resource_nr_samples(prsc);
-
-       assert(rsc->cpp);
-
-       // XXX probably need some extra work if we hit rsc shadowing path w/ lrz..
-       if ((is_a5xx(screen) || is_a6xx(screen)) &&
-                (fd_mesa_debug & FD_DBG_LRZ) && has_depth(format)) {
-               const uint32_t flags = DRM_FREEDRENO_GEM_CACHE_WCOMBINE |
-                               DRM_FREEDRENO_GEM_TYPE_KMEM; /* TODO */
-               unsigned lrz_pitch  = align(DIV_ROUND_UP(tmpl->width0, 8), 64);
-               unsigned lrz_height = DIV_ROUND_UP(tmpl->height0, 8);
-
-               /* LRZ buffer is super-sampled: */
-               switch (prsc->nr_samples) {
-               case 4:
-                       lrz_pitch *= 2;
-               case 2:
-                       lrz_height *= 2;
-               }
+       simple_mtx_init(&rsc->lock, mtx_plain);
 
-               unsigned size = lrz_pitch * lrz_height * 2;
+       rsc->internal_format = format;
 
-               size += 0x1000; /* for GRAS_LRZ_FAST_CLEAR_BUFFER */
+       rsc->layout.ubwc = rsc->layout.tile_mode && is_a6xx(screen) && allow_ubwc;
 
-               rsc->lrz_height = lrz_height;
-               rsc->lrz_width = lrz_pitch;
-               rsc->lrz_pitch = lrz_pitch;
-               rsc->lrz = fd_bo_new(screen->dev, size, flags, "lrz");
+       if (prsc->target == PIPE_BUFFER) {
+               assert(prsc->format == PIPE_FORMAT_R8_UNORM);
+               size = prsc->width0;
+               fdl_layout_buffer(&rsc->layout, size);
+       } else {
+               size = screen->setup_slices(rsc);
        }
 
-       size = screen->setup_slices(rsc);
-
-       if (allow_ubwc && screen->fill_ubwc_buffer_sizes && rsc->tile_mode)
-               size += screen->fill_ubwc_buffer_sizes(rsc);
-
        /* special case for hw-query buffer, which we need to allocate before we
         * know the size:
         */
@@ -1035,11 +1050,15 @@ fd_resource_create_with_modifiers(struct pipe_screen *pscreen,
                return prsc;
        }
 
-       if (rsc->layer_first) {
-               rsc->layer_size = align(size, 4096);
-               size = rsc->layer_size * prsc->array_size;
+       /* Set the layer size if the (non-a6xx) backend hasn't done so. */
+       if (rsc->layout.layer_first && !rsc->layout.layer_size) {
+               rsc->layout.layer_size = align(size, 4096);
+               size = rsc->layout.layer_size * prsc->array_size;
        }
 
+       if (fd_mesa_debug & FD_DBG_LAYOUT)
+               fdl_dump_layout(&rsc->layout);
+
        realloc_bo(rsc, size);
        if (!rsc->bo)
                goto fail;
@@ -1058,26 +1077,6 @@ fd_resource_create(struct pipe_screen *pscreen,
        return fd_resource_create_with_modifiers(pscreen, tmpl, &mod, 1);
 }
 
-static bool
-is_supported_modifier(struct pipe_screen *pscreen, enum pipe_format pfmt,
-               uint64_t mod)
-{
-       int count;
-
-       /* Get the count of supported modifiers: */
-       pscreen->query_dmabuf_modifiers(pscreen, pfmt, 0, NULL, NULL, &count);
-
-       /* Get the supported modifiers: */
-       uint64_t modifiers[count];
-       pscreen->query_dmabuf_modifiers(pscreen, pfmt, count, modifiers, NULL, &count);
-
-       for (int i = 0; i < count; i++)
-               if (modifiers[i] == mod)
-                       return true;
-
-       return false;
-}
-
 /**
  * Create a texture from a winsys_handle. The handle is often created in
  * another process by first creating a pipe texture and then calling
@@ -1090,9 +1089,9 @@ fd_resource_from_handle(struct pipe_screen *pscreen,
 {
        struct fd_screen *screen = fd_screen(pscreen);
        struct fd_resource *rsc = CALLOC_STRUCT(fd_resource);
-       struct fd_resource_slice *slice = &rsc->slices[0];
+       struct fdl_slice *slice = fd_resource_slice(rsc, 0);
        struct pipe_resource *prsc = &rsc->base;
-       uint32_t pitchalign = fd_screen(pscreen)->gmem_alignw;
+       uint32_t pitchalign = fd_screen(pscreen)->gmem_alignw * rsc->layout.cpp;
 
        DBG("target=%d, format=%s, %ux%ux%u, array_size=%u, last_level=%u, "
                        "nr_samples=%u, usage=%u, bind=%x, flags=%x",
@@ -1105,6 +1104,7 @@ fd_resource_from_handle(struct pipe_screen *pscreen,
                return NULL;
 
        *prsc = *tmpl;
+       fd_resource_layout_init(prsc);
 
        pipe_reference_init(&prsc->reference, 1);
 
@@ -1112,35 +1112,25 @@ fd_resource_from_handle(struct pipe_screen *pscreen,
 
        util_range_init(&rsc->valid_buffer_range);
 
+       simple_mtx_init(&rsc->lock, mtx_plain);
+
        rsc->bo = fd_screen_bo_from_handle(pscreen, handle);
        if (!rsc->bo)
                goto fail;
 
        rsc->internal_format = tmpl->format;
-       rsc->cpp = util_format_get_blocksize(tmpl->format);
-       rsc->cpp *= fd_resource_nr_samples(prsc);
-       slice->pitch = handle->stride / rsc->cpp;
+       slice->pitch = handle->stride;
        slice->offset = handle->offset;
        slice->size0 = handle->stride * prsc->height0;
 
-       if ((slice->pitch < align(prsc->width0, pitchalign)) ||
+       if ((slice->pitch < align(prsc->width0 * rsc->layout.cpp, pitchalign)) ||
                        (slice->pitch & (pitchalign - 1)))
                goto fail;
 
-       if (handle->modifier == DRM_FORMAT_MOD_QCOM_COMPRESSED) {
-               if (!is_supported_modifier(pscreen, tmpl->format,
-                               DRM_FORMAT_MOD_QCOM_COMPRESSED)) {
-                       DBG("bad modifier: %"PRIx64, handle->modifier);
-                       goto fail;
-               }
-               debug_assert(screen->fill_ubwc_buffer_sizes);
-               screen->fill_ubwc_buffer_sizes(rsc);
-       } else if (handle->modifier &&
-                       (handle->modifier != DRM_FORMAT_MOD_INVALID)) {
-               goto fail;
-       }
+       assert(rsc->layout.cpp);
 
-       assert(rsc->cpp);
+       if (screen->layout_resource_for_modifier(rsc, handle->modifier) < 0)
+               goto fail;
 
        if (screen->ro) {
                rsc->scanout =
@@ -1243,6 +1233,21 @@ static const struct u_transfer_vtbl transfer_vtbl = {
                .get_stencil              = fd_resource_get_stencil,
 };
 
+static const uint64_t supported_modifiers[] = {
+       DRM_FORMAT_MOD_LINEAR,
+};
+
+static int
+fd_layout_resource_for_modifier(struct fd_resource *rsc, uint64_t modifier)
+{
+       switch (modifier) {
+       case DRM_FORMAT_MOD_LINEAR:
+               return 0;
+       default:
+               return -1;
+       }
+}
+
 void
 fd_resource_screen_init(struct pipe_screen *pscreen)
 {
@@ -1263,6 +1268,12 @@ fd_resource_screen_init(struct pipe_screen *pscreen)
 
        if (!screen->setup_slices)
                screen->setup_slices = fd_setup_slices;
+       if (!screen->layout_resource_for_modifier)
+               screen->layout_resource_for_modifier = fd_layout_resource_for_modifier;
+       if (!screen->supported_modifiers) {
+               screen->supported_modifiers = supported_modifiers;
+               screen->num_supported_modifiers = ARRAY_SIZE(supported_modifiers);
+       }
 }
 
 static void