The next commit will reduce the size even more.
v2: typecast to uint64_t manually
v3: add more typecasts, add asserts
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
surf_level = is_stencil ? &surf->u.legacy.stencil_level[level] : &surf->u.legacy.level[level];
surf_level->offset = align64(surf->surf_size, AddrSurfInfoOut->baseAlign);
- surf_level->slice_size = AddrSurfInfoOut->sliceSize;
+ surf_level->slice_size_dw = AddrSurfInfoOut->sliceSize / 4;
surf_level->nblk_x = AddrSurfInfoOut->pitch;
surf_level->nblk_y = AddrSurfInfoOut->height;
struct legacy_surf_level {
uint64_t offset;
- uint64_t slice_size;
+ uint32_t slice_size_dw; /* in dwords; max = 4GB / 4. */
uint32_t dcc_offset; /* relative offset within DCC mip tree */
uint32_t dcc_fast_clear_size;
uint16_t nblk_x;
if (image->type == VK_IMAGE_TYPE_3D)
pLayout->size *= u_minify(image->info.depth, level);
} else {
- pLayout->offset = surface->u.legacy.level[level].offset + surface->u.legacy.level[level].slice_size * layer;
+ pLayout->offset = surface->u.legacy.level[level].offset + (uint64_t)surface->u.legacy.level[level].slice_size_dw * 4 * layer;
pLayout->rowPitch = surface->u.legacy.level[level].nblk_x * surface->bpe;
- pLayout->arrayPitch = surface->u.legacy.level[level].slice_size;
- pLayout->depthPitch = surface->u.legacy.level[level].slice_size;
- pLayout->size = surface->u.legacy.level[level].slice_size;
+ pLayout->arrayPitch = (uint64_t)surface->u.legacy.level[level].slice_size_dw * 4;
+ pLayout->depthPitch = (uint64_t)surface->u.legacy.level[level].slice_size_dw * 4;
+ pLayout->size = (uint64_t)surface->u.legacy.level[level].slice_size_dw * 4;
if (image->type == VK_IMAGE_TYPE_3D)
pLayout->size *= u_minify(image->info.depth, level);
}
z = src_z;
base = rsrc->surface.u.legacy.level[src_level].offset;
addr = rdst->surface.u.legacy.level[dst_level].offset;
- addr += rdst->surface.u.legacy.level[dst_level].slice_size * dst_z;
+ addr += (uint64_t)rdst->surface.u.legacy.level[dst_level].slice_size_dw * 4 * dst_z;
addr += dst_y * pitch + dst_x * bpp;
bank_h = eg_bank_wh(rsrc->surface.u.legacy.bankh);
bank_w = eg_bank_wh(rsrc->surface.u.legacy.bankw);
z = dst_z;
base = rdst->surface.u.legacy.level[dst_level].offset;
addr = rsrc->surface.u.legacy.level[src_level].offset;
- addr += rsrc->surface.u.legacy.level[src_level].slice_size * src_z;
+ addr += (uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4 * src_z;
addr += src_y * pitch + src_x * bpp;
bank_h = eg_bank_wh(rdst->surface.u.legacy.bankh);
bank_w = eg_bank_wh(rdst->surface.u.legacy.bankw);
* dst_pitch == src_pitch
*/
src_offset= rsrc->surface.u.legacy.level[src_level].offset;
- src_offset += rsrc->surface.u.legacy.level[src_level].slice_size * src_box->z;
+ src_offset += (uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4 * src_box->z;
src_offset += src_y * src_pitch + src_x * bpp;
dst_offset = rdst->surface.u.legacy.level[dst_level].offset;
- dst_offset += rdst->surface.u.legacy.level[dst_level].slice_size * dst_z;
+ dst_offset += (uint64_t)rdst->surface.u.legacy.level[dst_level].slice_size_dw * 4 * dst_z;
dst_offset += dst_y * dst_pitch + dst_x * bpp;
evergreen_dma_copy_buffer(rctx, dst, src, dst_offset, src_offset,
src_box->height * src_pitch);
z = src_z;
base = rsrc->surface.u.legacy.level[src_level].offset;
addr = rdst->surface.u.legacy.level[dst_level].offset;
- addr += rdst->surface.u.legacy.level[dst_level].slice_size * dst_z;
+ addr += (uint64_t)rdst->surface.u.legacy.level[dst_level].slice_size_dw * 4 * dst_z;
addr += dst_y * pitch + dst_x * bpp;
} else {
/* L2T */
z = dst_z;
base = rdst->surface.u.legacy.level[dst_level].offset;
addr = rsrc->surface.u.legacy.level[src_level].offset;
- addr += rsrc->surface.u.legacy.level[src_level].slice_size * src_z;
+ addr += (uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4 * src_z;
addr += src_y * pitch + src_x * bpp;
}
/* check that we are in dw/base alignment constraint */
* dst_pitch == src_pitch
*/
src_offset= rsrc->surface.u.legacy.level[src_level].offset;
- src_offset += rsrc->surface.u.legacy.level[src_level].slice_size * src_box->z;
+ src_offset += (uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4 * src_box->z;
src_offset += src_y * src_pitch + src_x * bpp;
dst_offset = rdst->surface.u.legacy.level[dst_level].offset;
- dst_offset += rdst->surface.u.legacy.level[dst_level].slice_size * dst_z;
+ dst_offset += (uint64_t)rdst->surface.u.legacy.level[dst_level].slice_size_dw * 4 * dst_z;
dst_offset += dst_y * dst_pitch + dst_x * bpp;
size = src_box->height * src_pitch;
/* must be dw aligned */
{
*stride = rtex->surface.u.legacy.level[level].nblk_x *
rtex->surface.bpe;
- *layer_stride = rtex->surface.u.legacy.level[level].slice_size;
+ assert((uint64_t)rtex->surface.u.legacy.level[level].slice_size_dw * 4 <= UINT_MAX);
+ *layer_stride = (uint64_t)rtex->surface.u.legacy.level[level].slice_size_dw * 4;
if (!box)
return rtex->surface.u.legacy.level[level].offset;
/* Each texture is an array of mipmap levels. Each level is
* an array of slices. */
return rtex->surface.u.legacy.level[level].offset +
- box->z * rtex->surface.u.legacy.level[level].slice_size +
+ box->z * (uint64_t)rtex->surface.u.legacy.level[level].slice_size_dw * 4 +
(box->y / rtex->surface.blk_h *
rtex->surface.u.legacy.level[level].nblk_x +
box->x / rtex->surface.blk_w) * rtex->surface.bpe;
* for those
*/
surface->u.legacy.level[0].nblk_x = pitch_in_bytes_override / bpe;
- surface->u.legacy.level[0].slice_size = pitch_in_bytes_override *
- surface->u.legacy.level[0].nblk_y;
+ surface->u.legacy.level[0].slice_size_dw =
+ ((uint64_t)pitch_in_bytes_override * surface->u.legacy.level[0].nblk_y) / 4;
}
if (offset) {
offset = rtex->surface.u.legacy.level[0].offset;
stride = rtex->surface.u.legacy.level[0].nblk_x *
rtex->surface.bpe;
- slice_size = rtex->surface.u.legacy.level[0].slice_size;
+ slice_size = (uint64_t)rtex->surface.u.legacy.level[0].slice_size_dw * 4;
} else {
/* Move a suballocated buffer into a non-suballocated allocation. */
if (rscreen->ws->buffer_is_suballocated(res->buf)) {
"npix_x=%u, npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
"mode=%u, tiling_index = %u\n",
i, rtex->surface.u.legacy.level[i].offset,
- rtex->surface.u.legacy.level[i].slice_size,
+ (uint64_t)rtex->surface.u.legacy.level[i].slice_size_dw * 4,
u_minify(rtex->resource.b.b.width0, i),
u_minify(rtex->resource.b.b.height0, i),
u_minify(rtex->resource.b.b.depth0, i),
"npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
"mode=%u, tiling_index = %u\n",
i, rtex->surface.u.legacy.stencil_level[i].offset,
- rtex->surface.u.legacy.stencil_level[i].slice_size,
+ (uint64_t)rtex->surface.u.legacy.stencil_level[i].slice_size_dw * 4,
u_minify(rtex->resource.b.b.width0, i),
u_minify(rtex->resource.b.b.height0, i),
u_minify(rtex->resource.b.b.depth0, i),
static unsigned texture_offset(struct radeon_surf *surface, unsigned layer)
{
return surface->u.legacy.level[0].offset +
- layer * surface->u.legacy.level[0].slice_size;
+ layer * (uint64_t)surface->u.legacy.level[0].slice_size_dw * 4;
}
/* hw encode the aspect of macro tiles */
} else {
*stride = rtex->surface.u.legacy.level[level].nblk_x *
rtex->surface.bpe;
- *layer_stride = rtex->surface.u.legacy.level[level].slice_size;
+ assert((uint64_t)rtex->surface.u.legacy.level[level].slice_size_dw * 4 <= UINT_MAX);
+ *layer_stride = (uint64_t)rtex->surface.u.legacy.level[level].slice_size_dw * 4;
if (!box)
return rtex->surface.u.legacy.level[level].offset;
/* Each texture is an array of mipmap levels. Each level is
* an array of slices. */
return rtex->surface.u.legacy.level[level].offset +
- box->z * rtex->surface.u.legacy.level[level].slice_size +
+ box->z * (uint64_t)rtex->surface.u.legacy.level[level].slice_size_dw * 4 +
(box->y / rtex->surface.blk_h *
rtex->surface.u.legacy.level[level].nblk_x +
box->x / rtex->surface.blk_w) * rtex->surface.bpe;
offset = rtex->surface.u.legacy.level[0].offset;
stride = rtex->surface.u.legacy.level[0].nblk_x *
rtex->surface.bpe;
- slice_size = rtex->surface.u.legacy.level[0].slice_size;
+ slice_size = (uint64_t)rtex->surface.u.legacy.level[0].slice_size_dw * 4;
}
} else {
/* Move a suballocated buffer into a non-suballocated allocation. */
"npix_x=%u, npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
"mode=%u, tiling_index = %u\n",
i, rtex->surface.u.legacy.level[i].offset,
- rtex->surface.u.legacy.level[i].slice_size,
+ (uint64_t)rtex->surface.u.legacy.level[i].slice_size_dw * 4,
u_minify(rtex->resource.b.b.width0, i),
u_minify(rtex->resource.b.b.height0, i),
u_minify(rtex->resource.b.b.depth0, i),
"npix_y=%u, npix_z=%u, nblk_x=%u, nblk_y=%u, "
"mode=%u, tiling_index = %u\n",
i, rtex->surface.u.legacy.stencil_level[i].offset,
- rtex->surface.u.legacy.stencil_level[i].slice_size,
+ (uint64_t)rtex->surface.u.legacy.stencil_level[i].slice_size_dw * 4,
u_minify(rtex->resource.b.b.width0, i),
u_minify(rtex->resource.b.b.height0, i),
u_minify(rtex->resource.b.b.depth0, i),
default:
case RUVD_SURFACE_TYPE_LEGACY:
return surface->u.legacy.level[0].offset +
- layer * surface->u.legacy.level[0].slice_size;
+ layer * (uint64_t)surface->u.legacy.level[0].slice_size_dw * 4;
break;
case RUVD_SURFACE_TYPE_GFX9:
return surface->u.gfx9.surf_offset +
rsrc->surface.tile_swizzle : 0;
unsigned dst_pitch = rdst->surface.u.legacy.level[dst_level].nblk_x;
unsigned src_pitch = rsrc->surface.u.legacy.level[src_level].nblk_x;
- uint64_t dst_slice_pitch = rdst->surface.u.legacy.level[dst_level].slice_size / bpp;
- uint64_t src_slice_pitch = rsrc->surface.u.legacy.level[src_level].slice_size / bpp;
+ uint64_t dst_slice_pitch = ((uint64_t)rdst->surface.u.legacy.level[dst_level].slice_size_dw * 4) / bpp;
+ uint64_t src_slice_pitch = ((uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4) / bpp;
unsigned dst_width = minify_as_blocks(rdst->resource.b.b.width0,
dst_level, rdst->surface.blk_w);
unsigned src_width = minify_as_blocks(rsrc->resource.b.b.width0,
height = rtiled->surface.u.legacy.level[tiled_lvl].nblk_y;
base = rtiled->surface.u.legacy.level[tiled_lvl].offset;
addr = rlinear->surface.u.legacy.level[linear_lvl].offset;
- addr += rlinear->surface.u.legacy.level[linear_lvl].slice_size * linear_z;
+ addr += (uint64_t)rlinear->surface.u.legacy.level[linear_lvl].slice_size_dw * 4 * linear_z;
addr += linear_y * pitch + linear_x * bpp;
bank_h = G_009910_BANK_HEIGHT(tile_mode);
bank_w = G_009910_BANK_WIDTH(tile_mode);
* dst_pitch == src_pitch
*/
src_offset= rsrc->surface.u.legacy.level[src_level].offset;
- src_offset += rsrc->surface.u.legacy.level[src_level].slice_size * src_box->z;
+ src_offset += (uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4 * src_box->z;
src_offset += src_y * src_pitch + src_x * bpp;
dst_offset = rdst->surface.u.legacy.level[dst_level].offset;
- dst_offset += rdst->surface.u.legacy.level[dst_level].slice_size * dst_z;
+ dst_offset += (uint64_t)rdst->surface.u.legacy.level[dst_level].slice_size_dw * 4 * dst_z;
dst_offset += dst_y * dst_pitch + dst_x * bpp;
si_dma_copy_buffer(sctx, dst, src, dst_offset, src_offset,
- rsrc->surface.u.legacy.level[src_level].slice_size);
+ (uint64_t)rsrc->surface.u.legacy.level[src_level].slice_size_dw * 4);
} else {
si_dma_copy_tile(sctx, dst, dst_level, dst_x, dst_y, dst_z,
src, src_level, src_x, src_y, src_box->z,
unsigned bpe)
{
level_drm->offset = level_ws->offset;
- level_drm->slice_size = level_ws->slice_size;
+ level_drm->slice_size = (uint64_t)level_ws->slice_size_dw * 4;
level_drm->nblk_x = level_ws->nblk_x;
level_drm->nblk_y = level_ws->nblk_y;
level_drm->pitch_bytes = level_ws->nblk_x * bpe;
unsigned bpe)
{
level_ws->offset = level_drm->offset;
- level_ws->slice_size = level_drm->slice_size;
+ level_ws->slice_size_dw = level_drm->slice_size / 4;
level_ws->nblk_x = level_drm->nblk_x;
level_ws->nblk_y = level_drm->nblk_y;
level_ws->mode = level_drm->mode;