#include "r600_cs.h"
#include "util/u_format.h"
#include "util/u_memory.h"
+#include "util/u_pack_color.h"
#include <errno.h>
#include <inttypes.h>
return;
}
- if (!rctx->dma_copy(ctx, dst, 0, 0, 0, 0,
- src, transfer->level,
- &transfer->box)) {
- ctx->resource_copy_region(ctx, dst, 0, 0, 0, 0,
- src, transfer->level, &transfer->box);
- }
+ rctx->dma_copy(ctx, dst, 0, 0, 0, 0, src, transfer->level,
+ &transfer->box);
}
/* Copy from a transfer's staging texture to a full GPU one. */
return;
}
- if (!rctx->dma_copy(ctx, dst, transfer->level,
- transfer->box.x, transfer->box.y, transfer->box.z,
- src, 0, &sbox)) {
- ctx->resource_copy_region(ctx, dst, transfer->level,
- transfer->box.x, transfer->box.y, transfer->box.z,
- src, 0, &sbox);
- }
+ rctx->dma_copy(ctx, dst, transfer->level,
+ transfer->box.x, transfer->box.y, transfer->box.z,
+ src, 0, &sbox);
}
static unsigned r600_texture_get_offset(struct r600_texture *rtex, unsigned level,
}
static int r600_init_surface(struct r600_common_screen *rscreen,
- struct radeon_surface *surface,
+ struct radeon_surf *surface,
const struct pipe_resource *ptex,
unsigned array_mode,
bool is_flushed_depth)
}
static boolean r600_texture_get_handle(struct pipe_screen* screen,
- struct pipe_resource *ptex,
- struct winsys_handle *whandle)
+ struct pipe_resource *ptex,
+ struct winsys_handle *whandle)
{
struct r600_texture *rtex = (struct r600_texture*)ptex;
struct r600_resource *resource = &rtex->resource;
- struct radeon_surface *surface = &rtex->surface;
+ struct radeon_surf *surface = &rtex->surface;
struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
rscreen->ws->buffer_set_tiling(resource->buf,
RADEON_LAYOUT_TILED : RADEON_LAYOUT_LINEAR,
surface->level[0].mode >= RADEON_SURF_MODE_2D ?
RADEON_LAYOUT_TILED : RADEON_LAYOUT_LINEAR,
+ surface->pipe_config,
surface->bankw, surface->bankh,
surface->tile_split,
surface->stencil_tile_split,
- surface->mtilea,
+ surface->mtilea, surface->num_banks,
surface->level[0].pitch_bytes,
(surface->flags & RADEON_SURF_SCANOUT) != 0);
if (rtex->flushed_depth_texture)
pipe_resource_reference((struct pipe_resource **)&rtex->flushed_depth_texture, NULL);
- pipe_resource_reference((struct pipe_resource**)&rtex->htile, NULL);
+ pipe_resource_reference((struct pipe_resource**)&rtex->htile_buffer, NULL);
if (rtex->cmask_buffer != &rtex->resource) {
pipe_resource_reference((struct pipe_resource**)&rtex->cmask_buffer, NULL);
}
+ pipe_resource_reference((struct pipe_resource**)&rtex->dcc_buffer, NULL);
pb_reference(&resource->buf, NULL);
FREE(rtex);
}
struct r600_fmask_info *out)
{
/* FMASK is allocated like an ordinary texture. */
- struct radeon_surface fmask = rtex->surface;
+ struct radeon_surf fmask = rtex->surface;
memset(out, 0, sizeof(*out));
fmask.nsamples = 1;
fmask.flags |= RADEON_SURF_FMASK;
+ /* Force 2D tiling if it wasn't set. This may occur when creating
+ * FMASK for MSAA resolve on R6xx. On R6xx, the single-sample
+ * destination buffer must have an FMASK too. */
+ fmask.flags = RADEON_SURF_CLR(fmask.flags, MODE);
+ fmask.flags |= RADEON_SURF_SET(RADEON_SURF_MODE_2D, MODE);
+
if (rscreen->chip_class >= SI) {
fmask.flags |= RADEON_SURF_HAS_TILE_MODE_INDEX;
}
out->slice_tile_max -= 1;
out->tile_mode_index = fmask.tiling_index[0];
- out->pitch = fmask.level[0].nblk_x;
+ out->pitch_in_pixels = fmask.level[0].nblk_x;
out->bank_height = fmask.bankh;
out->alignment = MAX2(256, fmask.bo_alignment);
out->size = fmask.bo_size;
unsigned cmask_tile_elements = cmask_tile_width * cmask_tile_height;
unsigned element_bits = 4;
unsigned cmask_cache_bits = 1024;
- unsigned num_pipes = rscreen->tiling_info.num_channels;
- unsigned pipe_interleave_bytes = rscreen->tiling_info.group_bytes;
+ unsigned num_pipes = rscreen->info.num_tile_pipes;
+ unsigned pipe_interleave_bytes = rscreen->info.pipe_interleave_bytes;
unsigned elements_per_macro_tile = (cmask_cache_bits / element_bits) * num_pipes;
unsigned pixels_per_macro_tile = elements_per_macro_tile * cmask_tile_elements;
assert(macro_tile_width % 128 == 0);
assert(macro_tile_height % 128 == 0);
+ out->pitch = pitch_elements;
+ out->height = height;
+ out->xalign = macro_tile_width;
+ out->yalign = macro_tile_height;
out->slice_tile_max = ((pitch_elements * height) / (128*128)) - 1;
out->alignment = MAX2(256, base_align);
- out->size = rtex->surface.array_size * align(slice_bytes, base_align);
+ out->size = (util_max_layer(&rtex->resource.b.b, 0) + 1) *
+ align(slice_bytes, base_align);
}
static void si_texture_get_cmask_info(struct r600_common_screen *rscreen,
struct r600_texture *rtex,
struct r600_cmask_info *out)
{
- unsigned pipe_interleave_bytes = rscreen->tiling_info.group_bytes;
- unsigned num_pipes = rscreen->tiling_info.num_channels;
+ unsigned pipe_interleave_bytes = rscreen->info.pipe_interleave_bytes;
+ unsigned num_pipes = rscreen->info.num_tile_pipes;
unsigned cl_width, cl_height;
switch (num_pipes) {
/* Each element of CMASK is a nibble. */
unsigned slice_bytes = slice_elements / 2;
+ out->pitch = width;
+ out->height = height;
+ out->xalign = cl_width * 8;
+ out->yalign = cl_height * 8;
out->slice_tile_max = (width * height) / (128*128);
if (out->slice_tile_max)
out->slice_tile_max -= 1;
out->alignment = MAX2(256, base_align);
- out->size = rtex->surface.array_size * align(slice_bytes, base_align);
+ out->size = (util_max_layer(&rtex->resource.b.b, 0) + 1) *
+ align(slice_bytes, base_align);
}
static void r600_texture_allocate_cmask(struct r600_common_screen *rscreen,
rtex->cmask.offset = align(rtex->size, rtex->cmask.alignment);
rtex->size = rtex->cmask.offset + rtex->cmask.size;
+
+ if (rscreen->chip_class >= SI)
+ rtex->cb_color_info |= SI_S_028C70_FAST_CLEAR(1);
+ else
+ rtex->cb_color_info |= EG_S_028C70_FAST_CLEAR(1);
}
-void r600_texture_init_cmask(struct r600_common_screen *rscreen,
- struct r600_texture *rtex)
+static void r600_texture_alloc_cmask_separate(struct r600_common_screen *rscreen,
+ struct r600_texture *rtex)
{
+ if (rtex->cmask_buffer)
+ return;
+
assert(rtex->cmask.size == 0);
- r600_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
+ if (rscreen->chip_class >= SI) {
+ si_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
+ } else {
+ r600_texture_get_cmask_info(rscreen, rtex, &rtex->cmask);
+ }
rtex->cmask_buffer = (struct r600_resource *)
pipe_buffer_create(&rscreen->b, PIPE_BIND_CUSTOM,
- PIPE_USAGE_STATIC, rtex->cmask.size);
+ PIPE_USAGE_DEFAULT, rtex->cmask.size);
if (rtex->cmask_buffer == NULL) {
rtex->cmask.size = 0;
+ return;
}
+
+ /* update colorbuffer state bits */
+ rtex->cmask.base_address_reg = rtex->cmask_buffer->gpu_address >> 8;
+
+ if (rscreen->chip_class >= SI)
+ rtex->cb_color_info |= SI_S_028C70_FAST_CLEAR(1);
+ else
+ rtex->cb_color_info |= EG_S_028C70_FAST_CLEAR(1);
}
-static void r600_texture_allocate_htile(struct r600_common_screen *rscreen,
- struct r600_texture *rtex)
+static void vi_texture_alloc_dcc_separate(struct r600_common_screen *rscreen,
+ struct r600_texture *rtex)
{
- unsigned sw = rtex->surface.level[0].nblk_x * rtex->surface.blk_w;
- unsigned sh = rtex->surface.level[0].nblk_y * rtex->surface.blk_h;
- unsigned htile_size;
- unsigned npipes = rscreen->info.r600_num_tile_pipes;
+ if (rscreen->debug_flags & DBG_NO_DCC)
+ return;
- /* XXX also use it for other texture targets */
- if (rscreen->info.drm_minor < 26 ||
- rtex->resource.b.b.target != PIPE_TEXTURE_2D ||
- rtex->surface.level[0].nblk_x < 32 ||
- rtex->surface.level[0].nblk_y < 32) {
+ rtex->dcc_buffer = (struct r600_resource *)
+ r600_aligned_buffer_create(&rscreen->b, PIPE_BIND_CUSTOM,
+ PIPE_USAGE_DEFAULT, rtex->surface.dcc_size, rtex->surface.dcc_alignment);
+ if (rtex->dcc_buffer == NULL) {
return;
}
- /* this alignment and htile size only apply to linear htile buffer */
- sw = align(sw, 16 << 3);
- sh = align(sh, npipes << 3);
- htile_size = (sw >> 3) * (sh >> 3) * 4;
- /* must be aligned with 2K * npipes */
- htile_size = align(htile_size, (2 << 10) * npipes);
+ r600_screen_clear_buffer(rscreen, &rtex->dcc_buffer->b.b, 0, rtex->surface.dcc_size,
+ 0xFFFFFFFF, true);
+
+ rtex->cb_color_info |= VI_S_028C70_DCC_ENABLE(1);
+}
+
+static unsigned r600_texture_get_htile_size(struct r600_common_screen *rscreen,
+ struct r600_texture *rtex)
+{
+ unsigned cl_width, cl_height, width, height;
+ unsigned slice_elements, slice_bytes, pipe_interleave_bytes, base_align;
+ unsigned num_pipes = rscreen->info.num_tile_pipes;
+
+ if (rscreen->chip_class <= EVERGREEN &&
+ rscreen->info.drm_major == 2 && rscreen->info.drm_minor < 26)
+ return 0;
+
+ /* HW bug on R6xx. */
+ if (rscreen->chip_class == R600 &&
+ (rtex->surface.level[0].npix_x > 7680 ||
+ rtex->surface.level[0].npix_y > 7680))
+ return 0;
+
+ /* HTILE is broken with 1D tiling on old kernels and CIK. */
+ if (rscreen->chip_class >= CIK &&
+ rtex->surface.level[0].mode == RADEON_SURF_MODE_1D &&
+ rscreen->info.drm_major == 2 && rscreen->info.drm_minor < 38)
+ return 0;
+
+ /* Overalign HTILE on Stoney to fix piglit/depthstencil-render-miplevels 585. */
+ if (rscreen->family == CHIP_STONEY)
+ num_pipes = 4;
+
+ switch (num_pipes) {
+ case 1:
+ cl_width = 32;
+ cl_height = 16;
+ break;
+ case 2:
+ cl_width = 32;
+ cl_height = 32;
+ break;
+ case 4:
+ cl_width = 64;
+ cl_height = 32;
+ break;
+ case 8:
+ cl_width = 64;
+ cl_height = 64;
+ break;
+ case 16:
+ cl_width = 128;
+ cl_height = 64;
+ break;
+ default:
+ assert(0);
+ return 0;
+ }
+
+ width = align(rtex->surface.npix_x, cl_width * 8);
+ height = align(rtex->surface.npix_y, cl_height * 8);
+
+ slice_elements = (width * height) / (8 * 8);
+ slice_bytes = slice_elements * 4;
- /* XXX don't allocate it separately */
- rtex->htile = (struct r600_resource*)pipe_buffer_create(&rscreen->b, PIPE_BIND_CUSTOM,
- PIPE_USAGE_STATIC, htile_size);
- if (rtex->htile == NULL) {
+ pipe_interleave_bytes = rscreen->info.pipe_interleave_bytes;
+ base_align = num_pipes * pipe_interleave_bytes;
+
+ rtex->htile.pitch = width;
+ rtex->htile.height = height;
+ rtex->htile.xalign = cl_width * 8;
+ rtex->htile.yalign = cl_height * 8;
+
+ return (util_max_layer(&rtex->resource.b.b, 0) + 1) *
+ align(slice_bytes, base_align);
+}
+
+static void r600_texture_allocate_htile(struct r600_common_screen *rscreen,
+ struct r600_texture *rtex)
+{
+ unsigned htile_size = r600_texture_get_htile_size(rscreen, rtex);
+
+ if (!htile_size)
+ return;
+
+ rtex->htile_buffer = (struct r600_resource*)
+ pipe_buffer_create(&rscreen->b, PIPE_BIND_CUSTOM,
+ PIPE_USAGE_DEFAULT, htile_size);
+ if (rtex->htile_buffer == NULL) {
/* this is not a fatal error as we can still keep rendering
- * without htile buffer
- */
- R600_ERR("r600: failed to create bo for htile buffers\n");
+ * without htile buffer */
+ R600_ERR("Failed to create buffer object for htile buffer.\n");
} else {
- r600_screen_clear_buffer(rscreen, &rtex->htile->b.b, 0, htile_size, 0);
+ r600_screen_clear_buffer(rscreen, &rtex->htile_buffer->b.b, 0,
+ htile_size, 0, true);
}
}
+void r600_print_texture_info(struct r600_texture *rtex, FILE *f)
+{
+ int i;
+
+ fprintf(f, " Info: npix_x=%u, npix_y=%u, npix_z=%u, blk_w=%u, "
+ "blk_h=%u, blk_d=%u, array_size=%u, last_level=%u, "
+ "bpe=%u, nsamples=%u, flags=0x%x, %s\n",
+ rtex->surface.npix_x, rtex->surface.npix_y,
+ rtex->surface.npix_z, rtex->surface.blk_w,
+ rtex->surface.blk_h, rtex->surface.blk_d,
+ rtex->surface.array_size, rtex->surface.last_level,
+ rtex->surface.bpe, rtex->surface.nsamples,
+ rtex->surface.flags, util_format_short_name(rtex->resource.b.b.format));
+
+ fprintf(f, " Layout: size=%"PRIu64", alignment=%"PRIu64", bankw=%u, "
+ "bankh=%u, nbanks=%u, mtilea=%u, tilesplit=%u, pipeconfig=%u, scanout=%u\n",
+ rtex->surface.bo_size, rtex->surface.bo_alignment, rtex->surface.bankw,
+ rtex->surface.bankh, rtex->surface.num_banks, rtex->surface.mtilea,
+ rtex->surface.tile_split, rtex->surface.pipe_config,
+ (rtex->surface.flags & RADEON_SURF_SCANOUT) != 0);
+
+ if (rtex->fmask.size)
+ fprintf(f, " FMask: offset=%u, size=%u, alignment=%u, pitch_in_pixels=%u, "
+ "bankh=%u, slice_tile_max=%u, tile_mode_index=%u\n",
+ rtex->fmask.offset, rtex->fmask.size, rtex->fmask.alignment,
+ rtex->fmask.pitch_in_pixels, rtex->fmask.bank_height,
+ rtex->fmask.slice_tile_max, rtex->fmask.tile_mode_index);
+
+ if (rtex->cmask.size)
+ fprintf(f, " CMask: offset=%u, size=%u, alignment=%u, pitch=%u, "
+ "height=%u, xalign=%u, yalign=%u, slice_tile_max=%u\n",
+ rtex->cmask.offset, rtex->cmask.size, rtex->cmask.alignment,
+ rtex->cmask.pitch, rtex->cmask.height, rtex->cmask.xalign,
+ rtex->cmask.yalign, rtex->cmask.slice_tile_max);
+
+ if (rtex->htile_buffer)
+ fprintf(f, " HTile: size=%u, alignment=%u, pitch=%u, height=%u, "
+ "xalign=%u, yalign=%u\n",
+ rtex->htile_buffer->b.b.width0,
+ rtex->htile_buffer->buf->alignment, rtex->htile.pitch,
+ rtex->htile.height, rtex->htile.xalign, rtex->htile.yalign);
+
+ if (rtex->dcc_buffer) {
+ fprintf(f, " DCC: size=%u, alignment=%u\n",
+ rtex->dcc_buffer->b.b.width0,
+ rtex->dcc_buffer->buf->alignment);
+ for (i = 0; i <= rtex->surface.last_level; i++)
+ fprintf(f, " DCCLevel[%i]: offset=%"PRIu64"\n",
+ i, rtex->surface.level[i].dcc_offset);
+ }
+
+ for (i = 0; i <= rtex->surface.last_level; i++)
+ fprintf(f, " Level[%i]: offset=%"PRIu64", slice_size=%"PRIu64", "
+ "npix_x=%u, npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
+ "nblk_z=%u, pitch_bytes=%u, mode=%u\n",
+ i, rtex->surface.level[i].offset,
+ rtex->surface.level[i].slice_size,
+ u_minify(rtex->resource.b.b.width0, i),
+ u_minify(rtex->resource.b.b.height0, i),
+ u_minify(rtex->resource.b.b.depth0, i),
+ rtex->surface.level[i].nblk_x,
+ rtex->surface.level[i].nblk_y,
+ rtex->surface.level[i].nblk_z,
+ rtex->surface.level[i].pitch_bytes,
+ rtex->surface.level[i].mode);
+
+ if (rtex->surface.flags & RADEON_SURF_SBUFFER) {
+ for (i = 0; i <= rtex->surface.last_level; i++) {
+ fprintf(f, " StencilLayout: tilesplit=%u\n",
+ rtex->surface.stencil_tile_split);
+ fprintf(f, " StencilLevel[%i]: offset=%"PRIu64", "
+ "slice_size=%"PRIu64", npix_x=%u, "
+ "npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
+ "nblk_z=%u, pitch_bytes=%u, mode=%u\n",
+ i, rtex->surface.stencil_level[i].offset,
+ rtex->surface.stencil_level[i].slice_size,
+ u_minify(rtex->resource.b.b.width0, i),
+ u_minify(rtex->resource.b.b.height0, i),
+ u_minify(rtex->resource.b.b.depth0, i),
+ rtex->surface.stencil_level[i].nblk_x,
+ rtex->surface.stencil_level[i].nblk_y,
+ rtex->surface.stencil_level[i].nblk_z,
+ rtex->surface.stencil_level[i].pitch_bytes,
+ rtex->surface.stencil_level[i].mode);
+ }
+ }
+}
+
+/* Common processing for r600_texture_create and r600_texture_from_handle */
static struct r600_texture *
r600_texture_create_object(struct pipe_screen *screen,
const struct pipe_resource *base,
unsigned pitch_in_bytes_override,
struct pb_buffer *buf,
- struct radeon_surface *surface)
+ struct radeon_surf *surface)
{
struct r600_texture *rtex;
struct r600_resource *resource;
struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
- int r;
rtex = CALLOC_STRUCT(r600_texture);
- if (rtex == NULL)
+ if (!rtex)
return NULL;
resource = &rtex->resource;
resource->b.vtbl = &r600_texture_vtbl;
pipe_reference_init(&resource->b.b.reference, 1);
resource->b.b.screen = screen;
- rtex->pitch_override = pitch_in_bytes_override;
/* don't include stencil-only formats which we don't support for rendering */
rtex->is_depth = util_format_has_depth(util_format_description(rtex->resource.b.b.format));
rtex->surface = *surface;
- r = r600_setup_surface(screen, rtex, pitch_in_bytes_override);
- if (r) {
+ if (r600_setup_surface(screen, rtex, pitch_in_bytes_override)) {
FREE(rtex);
return NULL;
}
* Applies to R600-Cayman. */
rtex->non_disp_tiling = rtex->is_depth && rtex->surface.level[0].mode >= RADEON_SURF_MODE_1D;
- if (base->nr_samples > 1 && !rtex->is_depth && !buf) {
- r600_texture_allocate_fmask(rscreen, rtex);
- r600_texture_allocate_cmask(rscreen, rtex);
- rtex->cmask_buffer = &rtex->resource;
- }
-
- if (!rtex->is_depth && base->nr_samples > 1 &&
- (!rtex->fmask.size || !rtex->cmask.size)) {
- FREE(rtex);
- return NULL;
- }
+ if (rtex->is_depth) {
+ if (!(base->flags & (R600_RESOURCE_FLAG_TRANSFER |
+ R600_RESOURCE_FLAG_FLUSHED_DEPTH)) &&
+ !(rscreen->debug_flags & DBG_NO_HYPERZ)) {
- if (rtex->is_depth &&
- !(base->flags & (R600_RESOURCE_FLAG_TRANSFER |
- R600_RESOURCE_FLAG_FLUSHED_DEPTH)) &&
- !(rscreen->debug_flags & DBG_NO_HYPERZ)) {
- if (rscreen->chip_class >= SI) {
- /* XXX implement Hyper-Z for SI.
- * Reuse the CMASK allocator, which is almost the same as HTILE. */
- } else {
r600_texture_allocate_htile(rscreen, rtex);
}
+ } else {
+ if (base->nr_samples > 1) {
+ if (!buf) {
+ r600_texture_allocate_fmask(rscreen, rtex);
+ r600_texture_allocate_cmask(rscreen, rtex);
+ rtex->cmask_buffer = &rtex->resource;
+ }
+ if (!rtex->fmask.size || !rtex->cmask.size) {
+ FREE(rtex);
+ return NULL;
+ }
+ }
+ if (rtex->surface.dcc_size)
+ vi_texture_alloc_dcc_separate(rscreen, rtex);
}
/* Now create the backing buffer. */
if (!buf) {
- unsigned base_align = rtex->surface.bo_alignment;
- unsigned usage = rtex->surface.level[0].mode >= RADEON_SURF_MODE_1D ?
- PIPE_USAGE_STATIC : base->usage;
-
- if (!r600_init_resource(rscreen, resource, rtex->size, base_align, FALSE, usage)) {
+ if (!r600_init_resource(rscreen, resource, rtex->size,
+ rtex->surface.bo_alignment, TRUE)) {
FREE(rtex);
return NULL;
}
} else {
resource->buf = buf;
- resource->cs_buf = rscreen->ws->buffer_get_cs_handle(buf);
- resource->domains = RADEON_DOMAIN_GTT | RADEON_DOMAIN_VRAM;
+ resource->gpu_address = rscreen->ws->buffer_get_virtual_address(resource->buf);
+ resource->domains = rscreen->ws->buffer_get_initial_domain(resource->buf);
}
if (rtex->cmask.size) {
/* Initialize the cmask to 0xCC (= compressed state). */
r600_screen_clear_buffer(rscreen, &rtex->cmask_buffer->b.b,
- rtex->cmask.offset, rtex->cmask.size, 0xCCCCCCCC);
+ rtex->cmask.offset, rtex->cmask.size,
+ 0xCCCCCCCC, true);
}
+ /* Initialize the CMASK base register value. */
+ rtex->cmask.base_address_reg =
+ (rtex->resource.gpu_address + rtex->cmask.offset) >> 8;
+
if (rscreen->debug_flags & DBG_VM) {
- fprintf(stderr, "VM start=0x%"PRIu64" end=0x%"PRIu64" | Texture %ix%ix%i, %i levels, %i samples, %s\n",
- r600_resource_va(screen, &rtex->resource.b.b),
- r600_resource_va(screen, &rtex->resource.b.b) + rtex->resource.buf->size,
+ fprintf(stderr, "VM start=0x%"PRIX64" end=0x%"PRIX64" | Texture %ix%ix%i, %i levels, %i samples, %s\n",
+ rtex->resource.gpu_address,
+ rtex->resource.gpu_address + rtex->resource.buf->size,
base->width0, base->height0, util_max_layer(base, 0)+1, base->last_level+1,
base->nr_samples ? base->nr_samples : 1, util_format_short_name(base->format));
}
- if (rscreen->debug_flags & DBG_TEX ||
- (rtex->resource.b.b.last_level > 0 && rscreen->debug_flags & DBG_TEXMIP)) {
- printf("Texture: npix_x=%u, npix_y=%u, npix_z=%u, blk_w=%u, "
- "blk_h=%u, blk_d=%u, array_size=%u, last_level=%u, "
- "bpe=%u, nsamples=%u, flags=0x%x, %s\n",
- rtex->surface.npix_x, rtex->surface.npix_y,
- rtex->surface.npix_z, rtex->surface.blk_w,
- rtex->surface.blk_h, rtex->surface.blk_d,
- rtex->surface.array_size, rtex->surface.last_level,
- rtex->surface.bpe, rtex->surface.nsamples,
- rtex->surface.flags, util_format_short_name(base->format));
- for (int i = 0; i <= rtex->surface.last_level; i++) {
- printf(" L %i: offset=%"PRIu64", slice_size=%"PRIu64", npix_x=%u, "
- "npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
- "nblk_z=%u, pitch_bytes=%u, mode=%u\n",
- i, rtex->surface.level[i].offset,
- rtex->surface.level[i].slice_size,
- u_minify(rtex->resource.b.b.width0, i),
- u_minify(rtex->resource.b.b.height0, i),
- u_minify(rtex->resource.b.b.depth0, i),
- rtex->surface.level[i].nblk_x,
- rtex->surface.level[i].nblk_y,
- rtex->surface.level[i].nblk_z,
- rtex->surface.level[i].pitch_bytes,
- rtex->surface.level[i].mode);
- }
- if (rtex->surface.flags & RADEON_SURF_SBUFFER) {
- for (int i = 0; i <= rtex->surface.last_level; i++) {
- printf(" S %i: offset=%"PRIu64", slice_size=%"PRIu64", npix_x=%u, "
- "npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
- "nblk_z=%u, pitch_bytes=%u, mode=%u\n",
- i, rtex->surface.stencil_level[i].offset,
- rtex->surface.stencil_level[i].slice_size,
- u_minify(rtex->resource.b.b.width0, i),
- u_minify(rtex->resource.b.b.height0, i),
- u_minify(rtex->resource.b.b.depth0, i),
- rtex->surface.stencil_level[i].nblk_x,
- rtex->surface.stencil_level[i].nblk_y,
- rtex->surface.stencil_level[i].nblk_z,
- rtex->surface.stencil_level[i].pitch_bytes,
- rtex->surface.stencil_level[i].mode);
- }
- }
+ if (rscreen->debug_flags & DBG_TEX) {
+ puts("Texture:");
+ r600_print_texture_info(rtex, stdout);
}
+
return rtex;
}
const struct pipe_resource *templ)
{
const struct util_format_description *desc = util_format_description(templ->format);
+ bool force_tiling = templ->flags & R600_RESOURCE_FLAG_FORCE_TILING;
/* MSAA resources must be 2D tiled. */
if (templ->nr_samples > 1)
if (templ->flags & R600_RESOURCE_FLAG_TRANSFER)
return RADEON_SURF_MODE_LINEAR_ALIGNED;
+ /* r600g: force tiling on TEXTURE_2D and TEXTURE_3D compute resources. */
+ if (rscreen->chip_class >= R600 && rscreen->chip_class <= CAYMAN &&
+ (templ->bind & PIPE_BIND_COMPUTE_RESOURCE) &&
+ (templ->target == PIPE_TEXTURE_2D ||
+ templ->target == PIPE_TEXTURE_3D))
+ force_tiling = true;
+
/* Handle common candidates for the linear mode.
* Compressed textures must always be tiled. */
- if (!(templ->flags & R600_RESOURCE_FLAG_FORCE_TILING) &&
- !util_format_is_compressed(templ->format)) {
- /* Tiling doesn't work with the 422 (SUBSAMPLED) formats on R600-Cayman. */
- if (rscreen->chip_class <= CAYMAN &&
- desc->layout == UTIL_FORMAT_LAYOUT_SUBSAMPLED)
+ if (!force_tiling && !util_format_is_compressed(templ->format)) {
+ /* Not everything can be linear, so we cannot enforce it
+ * for all textures. */
+ if ((rscreen->debug_flags & DBG_NO_TILING) &&
+ (!util_format_is_depth_or_stencil(templ->format) ||
+ !(templ->flags & R600_RESOURCE_FLAG_FLUSHED_DEPTH)))
+ return RADEON_SURF_MODE_LINEAR_ALIGNED;
+
+ /* Tiling doesn't work with the 422 (SUBSAMPLED) formats on R600+. */
+ if (desc->layout == UTIL_FORMAT_LAYOUT_SUBSAMPLED)
return RADEON_SURF_MODE_LINEAR_ALIGNED;
/* Cursors are linear on SI.
}
/* Make small textures 1D tiled. */
- if (templ->width0 <= 16 || templ->height0 <= 16)
+ if (templ->width0 <= 16 || templ->height0 <= 16 ||
+ (rscreen->debug_flags & DBG_NO_2D_TILING))
return RADEON_SURF_MODE_1D;
/* The allocator will switch to 1D if needed. */
const struct pipe_resource *templ)
{
struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
- struct radeon_surface surface = {0};
+ struct radeon_surf surface = {0};
int r;
r = r600_init_surface(rscreen, &surface, templ,
0, NULL, &surface);
}
-struct pipe_resource *r600_texture_from_handle(struct pipe_screen *screen,
- const struct pipe_resource *templ,
- struct winsys_handle *whandle)
+static struct pipe_resource *r600_texture_from_handle(struct pipe_screen *screen,
+ const struct pipe_resource *templ,
+ struct winsys_handle *whandle)
{
struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
struct pb_buffer *buf = NULL;
unsigned stride = 0;
unsigned array_mode;
enum radeon_bo_layout micro, macro;
- struct radeon_surface surface;
+ struct radeon_surf surface;
bool scanout;
int r;
resource.array_size = texture->array_size;
resource.last_level = texture->last_level;
resource.nr_samples = texture->nr_samples;
- resource.usage = staging ? PIPE_USAGE_STAGING : PIPE_USAGE_STATIC;
+ resource.usage = staging ? PIPE_USAGE_STAGING : PIPE_USAGE_DEFAULT;
resource.bind = texture->bind & ~PIPE_BIND_DEPTH_STENCIL;
resource.flags = texture->flags | R600_RESOURCE_FLAG_FLUSHED_DEPTH;
res->height0 = box->height;
res->depth0 = 1;
res->array_size = 1;
- res->usage = flags & R600_RESOURCE_FLAG_TRANSFER ? PIPE_USAGE_STAGING : PIPE_USAGE_STATIC;
+ res->usage = flags & R600_RESOURCE_FLAG_TRANSFER ? PIPE_USAGE_STAGING : PIPE_USAGE_DEFAULT;
res->flags = flags;
/* We must set the correct texture target and dimensions for a 3D box. */
* the CPU is much happier reading out of cached system memory
* than uncached VRAM.
*/
- if (rtex->surface.level[level].mode >= RADEON_SURF_MODE_1D)
+ if (rtex->surface.level[0].mode >= RADEON_SURF_MODE_1D) {
use_staging_texture = TRUE;
-
- /* Untiled buffers in VRAM, which is slow for CPU reads */
- if ((usage & PIPE_TRANSFER_READ) && !(usage & PIPE_TRANSFER_MAP_DIRECTLY) &&
+ } else if ((usage & PIPE_TRANSFER_READ) && !(usage & PIPE_TRANSFER_MAP_DIRECTLY) &&
(rtex->resource.domains == RADEON_DOMAIN_VRAM)) {
+ /* Untiled buffers in VRAM, which is slow for CPU reads */
use_staging_texture = TRUE;
- }
-
- /* Use a staging texture for uploads if the underlying BO is busy. */
- if (!(usage & PIPE_TRANSFER_READ) &&
- (r600_rings_is_buffer_referenced(rctx, rtex->resource.cs_buf, RADEON_USAGE_READWRITE) ||
- rctx->ws->buffer_is_busy(rtex->resource.buf, RADEON_USAGE_READWRITE))) {
+ } else if (!(usage & PIPE_TRANSFER_READ) &&
+ (r600_rings_is_buffer_referenced(rctx, rtex->resource.buf, RADEON_USAGE_READWRITE) ||
+ !rctx->ws->buffer_wait(rtex->resource.buf, 0, RADEON_USAGE_READWRITE))) {
+ /* Use a staging texture for uploads if the underlying BO is busy. */
use_staging_texture = TRUE;
}
}
trans = CALLOC_STRUCT(r600_transfer);
- if (trans == NULL)
+ if (!trans)
return NULL;
trans->transfer.resource = texture;
trans->transfer.level = level;
if (usage & PIPE_TRANSFER_READ) {
struct pipe_resource *temp = ctx->screen->resource_create(ctx->screen, &resource);
+ if (!temp) {
+ R600_ERR("failed to create a temporary depth texture\n");
+ FREE(trans);
+ return NULL;
+ }
r600_copy_region_with_blit(ctx, temp, 0, 0, 0, 0, texture, level, box);
rctx->blit_decompress_depth(ctx, (struct r600_texture*)temp, staging_depth,
0, 0, 0, box->depth, 0, 0);
- pipe_resource_reference((struct pipe_resource**)&temp, NULL);
+ pipe_resource_reference(&temp, NULL);
}
}
else {
r600_init_temp_resource_from_box(&resource, texture, box, level,
R600_RESOURCE_FLAG_TRANSFER);
+ resource.usage = (usage & PIPE_TRANSFER_READ) ?
+ PIPE_USAGE_STAGING : PIPE_USAGE_STREAM;
/* Create the temporary texture. */
staging = (struct r600_texture*)ctx->screen->resource_create(ctx->screen, &resource);
- if (staging == NULL) {
+ if (!staging) {
R600_ERR("failed to create temporary texture to hold untiled copy\n");
FREE(trans);
return NULL;
if (trans->staging) {
buf = trans->staging;
+ if (!rtex->is_depth && !(usage & PIPE_TRANSFER_READ))
+ usage |= PIPE_TRANSFER_UNSYNCHRONIZED;
} else {
buf = &rtex->resource;
}
struct pipe_transfer* transfer)
{
struct r600_transfer *rtransfer = (struct r600_transfer*)transfer;
- struct r600_common_context *rctx = (struct r600_common_context*)ctx;
- struct radeon_winsys_cs_handle *buf;
struct pipe_resource *texture = transfer->resource;
struct r600_texture *rtex = (struct r600_texture*)texture;
- if (rtransfer->staging) {
- buf = rtransfer->staging->cs_buf;
- } else {
- buf = r600_resource(transfer->resource)->cs_buf;
- }
- rctx->ws->buffer_unmap(buf);
-
if ((transfer->usage & PIPE_TRANSFER_WRITE) && rtransfer->staging) {
if (rtex->is_depth && rtex->resource.b.b.nr_samples <= 1) {
ctx->resource_copy_region(ctx, texture, transfer->level,
static const struct u_resource_vtbl r600_texture_vtbl =
{
- r600_texture_get_handle, /* get_handle */
+ NULL, /* get_handle */
r600_texture_destroy, /* resource_destroy */
r600_texture_transfer_map, /* transfer_map */
- NULL, /* transfer_flush_region */
+ u_default_transfer_flush_region, /* transfer_flush_region */
r600_texture_transfer_unmap, /* transfer_unmap */
NULL /* transfer_inline_write */
};
+
+struct pipe_surface *r600_create_surface_custom(struct pipe_context *pipe,
+ struct pipe_resource *texture,
+ const struct pipe_surface *templ,
+ unsigned width, unsigned height)
+{
+ struct r600_surface *surface = CALLOC_STRUCT(r600_surface);
+
+ if (!surface)
+ return NULL;
+
+ assert(templ->u.tex.first_layer <= util_max_layer(texture, templ->u.tex.level));
+ assert(templ->u.tex.last_layer <= util_max_layer(texture, templ->u.tex.level));
+
+ pipe_reference_init(&surface->base.reference, 1);
+ pipe_resource_reference(&surface->base.texture, texture);
+ surface->base.context = pipe;
+ surface->base.format = templ->format;
+ surface->base.width = width;
+ surface->base.height = height;
+ surface->base.u = templ->u;
+ return &surface->base;
+}
+
+static struct pipe_surface *r600_create_surface(struct pipe_context *pipe,
+ struct pipe_resource *tex,
+ const struct pipe_surface *templ)
+{
+ unsigned level = templ->u.tex.level;
+ unsigned width = u_minify(tex->width0, level);
+ unsigned height = u_minify(tex->height0, level);
+
+ if (tex->target != PIPE_BUFFER && templ->format != tex->format) {
+ const struct util_format_description *tex_desc
+ = util_format_description(tex->format);
+ const struct util_format_description *templ_desc
+ = util_format_description(templ->format);
+
+ assert(tex_desc->block.bits == templ_desc->block.bits);
+
+ /* Adjust size of surface if and only if the block width or
+ * height is changed. */
+ if (tex_desc->block.width != templ_desc->block.width ||
+ tex_desc->block.height != templ_desc->block.height) {
+ unsigned nblks_x = util_format_get_nblocksx(tex->format, width);
+ unsigned nblks_y = util_format_get_nblocksy(tex->format, height);
+
+ width = nblks_x * templ_desc->block.width;
+ height = nblks_y * templ_desc->block.height;
+ }
+ }
+
+ return r600_create_surface_custom(pipe, tex, templ, width, height);
+}
+
+static void r600_surface_destroy(struct pipe_context *pipe,
+ struct pipe_surface *surface)
+{
+ struct r600_surface *surf = (struct r600_surface*)surface;
+ pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_fmask, NULL);
+ pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_cmask, NULL);
+ pipe_resource_reference(&surface->texture, NULL);
+ FREE(surface);
+}
+
+unsigned r600_translate_colorswap(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+#define HAS_SWIZZLE(chan,swz) (desc->swizzle[chan] == UTIL_FORMAT_SWIZZLE_##swz)
+
+ if (format == PIPE_FORMAT_R11G11B10_FLOAT) /* isn't plain */
+ return V_0280A0_SWAP_STD;
+
+ if (desc->layout != UTIL_FORMAT_LAYOUT_PLAIN)
+ return ~0U;
+
+ switch (desc->nr_channels) {
+ case 1:
+ if (HAS_SWIZZLE(0,X))
+ return V_0280A0_SWAP_STD; /* X___ */
+ else if (HAS_SWIZZLE(3,X))
+ return V_0280A0_SWAP_ALT_REV; /* ___X */
+ break;
+ case 2:
+ if ((HAS_SWIZZLE(0,X) && HAS_SWIZZLE(1,Y)) ||
+ (HAS_SWIZZLE(0,X) && HAS_SWIZZLE(1,NONE)) ||
+ (HAS_SWIZZLE(0,NONE) && HAS_SWIZZLE(1,Y)))
+ return V_0280A0_SWAP_STD; /* XY__ */
+ else if ((HAS_SWIZZLE(0,Y) && HAS_SWIZZLE(1,X)) ||
+ (HAS_SWIZZLE(0,Y) && HAS_SWIZZLE(1,NONE)) ||
+ (HAS_SWIZZLE(0,NONE) && HAS_SWIZZLE(1,X)))
+ return V_0280A0_SWAP_STD_REV; /* YX__ */
+ else if (HAS_SWIZZLE(0,X) && HAS_SWIZZLE(3,Y))
+ return V_0280A0_SWAP_ALT; /* X__Y */
+ else if (HAS_SWIZZLE(0,Y) && HAS_SWIZZLE(3,X))
+ return V_0280A0_SWAP_ALT_REV; /* Y__X */
+ break;
+ case 3:
+ if (HAS_SWIZZLE(0,X))
+ return V_0280A0_SWAP_STD; /* XYZ */
+ else if (HAS_SWIZZLE(0,Z))
+ return V_0280A0_SWAP_STD_REV; /* ZYX */
+ break;
+ case 4:
+ /* check the middle channels, the 1st and 4th channel can be NONE */
+ if (HAS_SWIZZLE(1,Y) && HAS_SWIZZLE(2,Z))
+ return V_0280A0_SWAP_STD; /* XYZW */
+ else if (HAS_SWIZZLE(1,Z) && HAS_SWIZZLE(2,Y))
+ return V_0280A0_SWAP_STD_REV; /* WZYX */
+ else if (HAS_SWIZZLE(1,Y) && HAS_SWIZZLE(2,X))
+ return V_0280A0_SWAP_ALT; /* ZYXW */
+ else if (HAS_SWIZZLE(1,X) && HAS_SWIZZLE(2,Y))
+ return V_0280A0_SWAP_ALT_REV; /* WXYZ */
+ break;
+ }
+ return ~0U;
+}
+
+static void evergreen_set_clear_color(struct r600_texture *rtex,
+ enum pipe_format surface_format,
+ const union pipe_color_union *color)
+{
+ union util_color uc;
+
+ memset(&uc, 0, sizeof(uc));
+
+ if (util_format_is_pure_uint(surface_format)) {
+ util_format_write_4ui(surface_format, color->ui, 0, &uc, 0, 0, 0, 1, 1);
+ } else if (util_format_is_pure_sint(surface_format)) {
+ util_format_write_4i(surface_format, color->i, 0, &uc, 0, 0, 0, 1, 1);
+ } else {
+ util_pack_color(color->f, surface_format, &uc);
+ }
+
+ memcpy(rtex->color_clear_value, &uc, 2 * sizeof(uint32_t));
+}
+
+static void vi_get_fast_clear_parameters(enum pipe_format surface_format,
+ const union pipe_color_union *color,
+ uint32_t* reset_value,
+ bool* clear_words_needed)
+{
+ bool values[4] = {};
+ int i;
+ bool main_value = false;
+ bool extra_value = false;
+ int extra_channel;
+ const struct util_format_description *desc = util_format_description(surface_format);
+
+ *clear_words_needed = true;
+ *reset_value = 0x20202020U;
+
+ /* If we want to clear without needing a fast clear eliminate step, we
+ * can set each channel to 0 or 1 (or 0/max for integer formats). We
+ * have two sets of flags, one for the last or first channel(extra) and
+ * one for the other channels(main).
+ */
+
+ if (surface_format == PIPE_FORMAT_R11G11B10_FLOAT ||
+ surface_format == PIPE_FORMAT_B5G6R5_UNORM ||
+ surface_format == PIPE_FORMAT_B5G6R5_SRGB) {
+ extra_channel = -1;
+ } else if (desc->layout == UTIL_FORMAT_LAYOUT_PLAIN) {
+ if(r600_translate_colorswap(surface_format) <= 1)
+ extra_channel = desc->nr_channels - 1;
+ else
+ extra_channel = 0;
+ } else
+ return;
+
+ for (i = 0; i < 4; ++i) {
+ int index = desc->swizzle[i] - UTIL_FORMAT_SWIZZLE_X;
+
+ if (desc->swizzle[i] < UTIL_FORMAT_SWIZZLE_X ||
+ desc->swizzle[i] > UTIL_FORMAT_SWIZZLE_W)
+ continue;
+
+ if (util_format_is_pure_sint(surface_format)) {
+ values[i] = color->i[i] != 0;
+ if (color->i[i] != 0 && color->i[i] != INT32_MAX)
+ return;
+ } else if (util_format_is_pure_uint(surface_format)) {
+ values[i] = color->ui[i] != 0U;
+ if (color->ui[i] != 0U && color->ui[i] != UINT32_MAX)
+ return;
+ } else {
+ values[i] = color->f[i] != 0.0F;
+ if (color->f[i] != 0.0F && color->f[i] != 1.0F)
+ return;
+ }
+
+ if (index == extra_channel)
+ extra_value = values[i];
+ else
+ main_value = values[i];
+ }
+
+ for (int i = 0; i < 4; ++i)
+ if (values[i] != main_value &&
+ desc->swizzle[i] - UTIL_FORMAT_SWIZZLE_X != extra_channel &&
+ desc->swizzle[i] >= UTIL_FORMAT_SWIZZLE_X &&
+ desc->swizzle[i] <= UTIL_FORMAT_SWIZZLE_W)
+ return;
+
+ *clear_words_needed = false;
+ if (main_value)
+ *reset_value |= 0x80808080U;
+
+ if (extra_value)
+ *reset_value |= 0x40404040U;
+}
+
+void evergreen_do_fast_color_clear(struct r600_common_context *rctx,
+ struct pipe_framebuffer_state *fb,
+ struct r600_atom *fb_state,
+ unsigned *buffers, unsigned *dirty_cbufs,
+ const union pipe_color_union *color)
+{
+ int i;
+
+ if (rctx->render_cond)
+ return;
+
+ for (i = 0; i < fb->nr_cbufs; i++) {
+ struct r600_texture *tex;
+ unsigned clear_bit = PIPE_CLEAR_COLOR0 << i;
+
+ if (!fb->cbufs[i])
+ continue;
+
+ /* if this colorbuffer is not being cleared */
+ if (!(*buffers & clear_bit))
+ continue;
+
+ tex = (struct r600_texture *)fb->cbufs[i]->texture;
+
+ /* 128-bit formats are unusupported */
+ if (util_format_get_blocksizebits(fb->cbufs[i]->format) > 64) {
+ continue;
+ }
+
+ /* the clear is allowed if all layers are bound */
+ if (fb->cbufs[i]->u.tex.first_layer != 0 ||
+ fb->cbufs[i]->u.tex.last_layer != util_max_layer(&tex->resource.b.b, 0)) {
+ continue;
+ }
+
+ /* cannot clear mipmapped textures */
+ if (fb->cbufs[i]->texture->last_level != 0) {
+ continue;
+ }
+
+ /* only supported on tiled surfaces */
+ if (tex->surface.level[0].mode < RADEON_SURF_MODE_1D) {
+ continue;
+ }
+
+ /* fast color clear with 1D tiling doesn't work on old kernels and CIK */
+ if (tex->surface.level[0].mode == RADEON_SURF_MODE_1D &&
+ rctx->chip_class >= CIK &&
+ rctx->screen->info.drm_major == 2 &&
+ rctx->screen->info.drm_minor < 38) {
+ continue;
+ }
+
+ if (tex->dcc_buffer) {
+ uint32_t reset_value;
+ bool clear_words_needed;
+
+ if (rctx->screen->debug_flags & DBG_NO_DCC_CLEAR)
+ continue;
+
+ vi_get_fast_clear_parameters(fb->cbufs[i]->format, color, &reset_value, &clear_words_needed);
+
+ rctx->clear_buffer(&rctx->b, &tex->dcc_buffer->b.b,
+ 0, tex->surface.dcc_size, reset_value, true);
+
+ if (clear_words_needed)
+ tex->dirty_level_mask |= 1 << fb->cbufs[i]->u.tex.level;
+ } else {
+ /* Stoney/RB+ doesn't work with CMASK fast clear. */
+ if (rctx->family == CHIP_STONEY)
+ continue;
+
+ /* ensure CMASK is enabled */
+ r600_texture_alloc_cmask_separate(rctx->screen, tex);
+ if (tex->cmask.size == 0) {
+ continue;
+ }
+
+ /* Do the fast clear. */
+ rctx->clear_buffer(&rctx->b, &tex->cmask_buffer->b.b,
+ tex->cmask.offset, tex->cmask.size, 0, true);
+
+ tex->dirty_level_mask |= 1 << fb->cbufs[i]->u.tex.level;
+ }
+
+ evergreen_set_clear_color(tex, fb->cbufs[i]->format, color);
+
+ if (dirty_cbufs)
+ *dirty_cbufs |= 1 << i;
+ rctx->set_atom_dirty(rctx, fb_state, true);
+ *buffers &= ~clear_bit;
+ }
+}
+
+void r600_init_screen_texture_functions(struct r600_common_screen *rscreen)
+{
+ rscreen->b.resource_from_handle = r600_texture_from_handle;
+ rscreen->b.resource_get_handle = r600_texture_get_handle;
+}
+
+void r600_init_context_texture_functions(struct r600_common_context *rctx)
+{
+ rctx->b.create_surface = r600_create_surface;
+ rctx->b.surface_destroy = r600_surface_destroy;
+}