#include "util/u_format.h"
-static unsigned si_array_mode(unsigned mode)
-{
- switch (mode) {
- case RADEON_SURF_MODE_LINEAR_ALIGNED:
- return V_009910_ARRAY_LINEAR_ALIGNED;
- case RADEON_SURF_MODE_1D:
- return V_009910_ARRAY_1D_TILED_THIN1;
- case RADEON_SURF_MODE_2D:
- return V_009910_ARRAY_2D_TILED_THIN1;
- default:
- case RADEON_SURF_MODE_LINEAR:
- return V_009910_ARRAY_LINEAR_GENERAL;
- }
-}
-
static uint32_t si_micro_tile_mode(struct si_screen *sscreen, unsigned tile_mode)
{
if (sscreen->b.info.si_tile_mode_array_valid) {
r600_need_dma_space(&ctx->b, ncopy * 5);
- r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, rsrc, RADEON_USAGE_READ,
+ radeon_add_to_buffer_list(&ctx->b, &ctx->b.rings.dma, rsrc, RADEON_USAGE_READ,
RADEON_PRIO_MIN);
- r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, rdst, RADEON_USAGE_WRITE,
+ radeon_add_to_buffer_list(&ctx->b, &ctx->b.rings.dma, rdst, RADEON_USAGE_WRITE,
RADEON_PRIO_MIN);
for (i = 0; i < ncopy; i++) {
csize = size < max_csize ? size : max_csize;
cs->buf[cs->cdw++] = SI_DMA_PACKET(SI_DMA_PACKET_COPY, sub_cmd, csize);
- cs->buf[cs->cdw++] = dst_offset & 0xffffffff;
- cs->buf[cs->cdw++] = src_offset & 0xffffffff;
+ cs->buf[cs->cdw++] = dst_offset;
+ cs->buf[cs->cdw++] = src_offset;
cs->buf[cs->cdw++] = (dst_offset >> 32UL) & 0xff;
cs->buf[cs->cdw++] = (src_offset >> 32UL) & 0xff;
dst_offset += csize << shift;
tiled_y = detile ? src_y : dst_y;
tiled_z = detile ? src_z : dst_z;
+ assert(!util_format_is_depth_and_stencil(rtiled->resource.b.b.format));
+
array_mode = si_array_mode(rtiled->surface.level[tiled_lvl].mode);
slice_tile_max = (rtiled->surface.level[tiled_lvl].nblk_x *
rtiled->surface.level[tiled_lvl].nblk_y) / (8*8) - 1;
* dma packet will be using the copy_height which is always smaller or equal
* to the linear height
*/
- height = rtiled->surface.level[tiled_lvl].npix_y;
+ height = rtiled->surface.level[tiled_lvl].nblk_y;
base = rtiled->surface.level[tiled_lvl].offset;
addr = rlinear->surface.level[linear_lvl].offset;
addr += rlinear->surface.level[linear_lvl].slice_size * linear_z;
bank_w = cik_bank_wh(rtiled->surface.bankw);
mt_aspect = cik_macro_tile_aspect(rtiled->surface.mtilea);
tile_split = cik_tile_split(rtiled->surface.tile_split);
- tile_mode_index = si_tile_mode_index(rtiled, tiled_lvl,
- util_format_has_stencil(util_format_description(rtiled->resource.b.b.format)));
+ tile_mode_index = si_tile_mode_index(rtiled, tiled_lvl, false);
nbanks = si_num_banks(sscreen, rtiled);
base += rtiled->resource.gpu_address;
addr += rlinear->resource.gpu_address;
ncopy = (size / SI_DMA_COPY_MAX_SIZE_DW) + !!(size % SI_DMA_COPY_MAX_SIZE_DW);
r600_need_dma_space(&ctx->b, ncopy * 9);
- r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, &rsrc->resource,
+ radeon_add_to_buffer_list(&ctx->b, &ctx->b.rings.dma, &rsrc->resource,
RADEON_USAGE_READ, RADEON_PRIO_MIN);
- r600_context_bo_reloc(&ctx->b, &ctx->b.rings.dma, &rdst->resource,
+ radeon_add_to_buffer_list(&ctx->b, &ctx->b.rings.dma, &rdst->resource,
RADEON_USAGE_WRITE, RADEON_PRIO_MIN);
for (i = 0; i < ncopy; i++) {
struct si_context *sctx = (struct si_context *)ctx;
struct r600_texture *rsrc = (struct r600_texture*)src;
struct r600_texture *rdst = (struct r600_texture*)dst;
- unsigned dst_pitch, src_pitch, bpp, dst_mode, src_mode, copy_height;
+ unsigned dst_pitch, src_pitch, bpp, dst_mode, src_mode;
unsigned src_w, dst_w;
unsigned src_x, src_y;
unsigned dst_x = dstx, dst_y = dsty, dst_z = dstz;
goto fallback;
}
- /* TODO: Implement DMA copy for CIK */
- if (sctx->b.chip_class >= CIK) {
- goto fallback;
- }
-
if (dst->target == PIPE_BUFFER && src->target == PIPE_BUFFER) {
si_dma_copy_buffer(sctx, dst, src, dst_x, src_box->x, src_box->width);
return;
}
+ /* XXX: Using the asynchronous DMA engine for multi-dimensional
+ * operations seems to cause random GPU lockups for various people.
+ * While the root cause for this might need to be fixed in the kernel,
+ * let's disable it for now.
+ *
+ * Before re-enabling this, please make sure you can hit all newly
+ * enabled paths in your testing, preferably with both piglit and real
+ * world apps, and get in touch with people on the bug reports below
+ * for stability testing.
+ *
+ * https://bugs.freedesktop.org/show_bug.cgi?id=85647
+ * https://bugs.freedesktop.org/show_bug.cgi?id=83500
+ */
+ goto fallback;
+
if (src->format != dst->format || src_box->depth > 1 ||
- rdst->dirty_level_mask != 0) {
+ rdst->dirty_level_mask != 0 ||
+ rdst->cmask.size || rdst->fmask.size ||
+ rsrc->cmask.size || rsrc->fmask.size) {
goto fallback;
}
src_pitch = rsrc->surface.level[src_level].pitch_bytes;
src_w = rsrc->surface.level[src_level].npix_x;
dst_w = rdst->surface.level[dst_level].npix_x;
- copy_height = src_box->height / rsrc->surface.blk_h;
dst_mode = rdst->surface.level[dst_level].mode;
src_mode = rsrc->surface.level[src_level].mode;
src_mode = src_mode == RADEON_SURF_MODE_LINEAR_ALIGNED ? RADEON_SURF_MODE_LINEAR : src_mode;
dst_mode = dst_mode == RADEON_SURF_MODE_LINEAR_ALIGNED ? RADEON_SURF_MODE_LINEAR : dst_mode;
- if (src_pitch != dst_pitch || src_box->x || dst_x || src_w != dst_w) {
+ if (src_pitch != dst_pitch || src_box->x || dst_x || src_w != dst_w ||
+ src_box->width != src_w ||
+ src_box->height != rsrc->surface.level[src_level].npix_y ||
+ src_box->height != rdst->surface.level[dst_level].npix_y ||
+ rsrc->surface.level[src_level].nblk_y !=
+ rdst->surface.level[dst_level].nblk_y) {
/* FIXME si can do partial blit */
goto fallback;
}
/* the x test here are currently useless (because we don't support partial blit)
* but keep them around so we don't forget about those
*/
- if ((src_pitch % 8) || (src_box->x % 8) || (dst_x % 8) || (src_box->y % 8) || (dst_y % 8)) {
+ if ((src_pitch % 8) || (src_box->x % 8) || (dst_x % 8) ||
+ (src_box->y % 8) || (dst_y % 8) || (src_box->height % 8)) {
goto fallback;
}
dst_offset += rdst->surface.level[dst_level].slice_size * dst_z;
dst_offset += dst_y * dst_pitch + dst_x * bpp;
si_dma_copy_buffer(sctx, dst, src, dst_offset, src_offset,
- src_box->height * src_pitch);
+ rsrc->surface.level[src_level].slice_size);
} 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,
- copy_height, dst_pitch, bpp);
+ src_box->height / rsrc->surface.blk_h,
+ dst_pitch, bpp);
}
return;
fallback:
- ctx->resource_copy_region(ctx, dst, dst_level, dstx, dsty, dstz,
- src, src_level, src_box);
+ si_resource_copy_region(ctx, dst, dst_level, dstx, dsty, dstz,
+ src, src_level, src_box);
}