struct radeon_bo *rbo;
uint64_t fb_location;
uint32_t fb_format, fb_pitch_pixels, tiling_flags;
+ unsigned bankw, bankh, mtaspect, tile_split;
u32 fb_swap = EVERGREEN_GRPH_ENDIAN_SWAP(EVERGREEN_GRPH_ENDIAN_NONE);
u32 tmp, viewport_w, viewport_h;
int r;
break;
}
- switch ((tmp & 0xf000) >> 12) {
- case 0: /* 1KB rows */
- default:
- fb_format |= EVERGREEN_GRPH_TILE_SPLIT(EVERGREEN_ADDR_SURF_TILE_SPLIT_1KB);
- break;
- case 1: /* 2KB rows */
- fb_format |= EVERGREEN_GRPH_TILE_SPLIT(EVERGREEN_ADDR_SURF_TILE_SPLIT_2KB);
- break;
- case 2: /* 4KB rows */
- fb_format |= EVERGREEN_GRPH_TILE_SPLIT(EVERGREEN_ADDR_SURF_TILE_SPLIT_4KB);
- break;
- }
-
fb_format |= EVERGREEN_GRPH_ARRAY_MODE(EVERGREEN_GRPH_ARRAY_2D_TILED_THIN1);
+
+ evergreen_tiling_fields(tiling_flags, &bankw, &bankh, &mtaspect, &tile_split);
+ fb_format |= EVERGREEN_GRPH_TILE_SPLIT(tile_split);
+ fb_format |= EVERGREEN_GRPH_BANK_WIDTH(bankw);
+ fb_format |= EVERGREEN_GRPH_BANK_HEIGHT(bankh);
+ fb_format |= EVERGREEN_GRPH_MACRO_TILE_ASPECT(mtaspect);
} else if (tiling_flags & RADEON_TILING_MICRO)
fb_format |= EVERGREEN_GRPH_ARRAY_MODE(EVERGREEN_GRPH_ARRAY_1D_TILED_THIN1);
extern void cayman_cp_int_cntl_setup(struct radeon_device *rdev,
int ring, u32 cp_int_cntl);
+void evergreen_tiling_fields(unsigned tiling_flags, unsigned *bankw,
+ unsigned *bankh, unsigned *mtaspect,
+ unsigned *tile_split)
+{
+ *bankw = (tiling_flags >> RADEON_TILING_EG_BANKW_SHIFT) & RADEON_TILING_EG_BANKW_MASK;
+ *bankh = (tiling_flags >> RADEON_TILING_EG_BANKH_SHIFT) & RADEON_TILING_EG_BANKH_MASK;
+ *mtaspect = (tiling_flags >> RADEON_TILING_EG_MACRO_TILE_ASPECT_SHIFT) & RADEON_TILING_EG_MACRO_TILE_ASPECT_MASK;
+ *tile_split = (tiling_flags >> RADEON_TILING_EG_TILE_SPLIT_SHIFT) & RADEON_TILING_EG_TILE_SPLIT_MASK;
+ switch (*bankw) {
+ default:
+ case 1: *bankw = EVERGREEN_ADDR_SURF_BANK_WIDTH_1; break;
+ case 2: *bankw = EVERGREEN_ADDR_SURF_BANK_WIDTH_2; break;
+ case 4: *bankw = EVERGREEN_ADDR_SURF_BANK_WIDTH_4; break;
+ case 8: *bankw = EVERGREEN_ADDR_SURF_BANK_WIDTH_8; break;
+ }
+ switch (*bankh) {
+ default:
+ case 1: *bankh = EVERGREEN_ADDR_SURF_BANK_HEIGHT_1; break;
+ case 2: *bankh = EVERGREEN_ADDR_SURF_BANK_HEIGHT_2; break;
+ case 4: *bankh = EVERGREEN_ADDR_SURF_BANK_HEIGHT_4; break;
+ case 8: *bankh = EVERGREEN_ADDR_SURF_BANK_HEIGHT_8; break;
+ }
+ switch (*mtaspect) {
+ default:
+ case 1: *mtaspect = EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_1; break;
+ case 2: *mtaspect = EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_2; break;
+ case 4: *mtaspect = EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_4; break;
+ case 8: *mtaspect = EVERGREEN_ADDR_SURF_MACRO_TILE_ASPECT_8; break;
+ }
+}
+
void evergreen_fix_pci_max_read_req_size(struct radeon_device *rdev)
{
u16 ctl, v;
#include "evergreen_reg_safe.h"
#include "cayman_reg_safe.h"
+#define MAX(a,b) (((a)>(b))?(a):(b))
+#define MIN(a,b) (((a)<(b))?(a):(b))
+
static int evergreen_cs_packet_next_reloc(struct radeon_cs_parser *p,
struct radeon_cs_reloc **cs_reloc);
u32 cb_color_dim[12];
u32 cb_color_pitch[12];
u32 cb_color_slice[12];
+ u32 cb_color_attrib[12];
u32 cb_color_cmask_slice[8];
u32 cb_color_fmask_slice[8];
u32 cb_target_mask;
u32 vgt_strmout_size[4];
u32 db_depth_control;
u32 db_depth_view;
+ u32 db_depth_slice;
u32 db_depth_size;
u32 db_depth_size_idx;
u32 db_z_info;
}
}
-static u32 evergreen_cs_get_tile_split(u32 row_size)
-{
- switch (row_size) {
- case 1:
- default:
- return ADDR_SURF_TILE_SPLIT_1KB;
- case 2:
- return ADDR_SURF_TILE_SPLIT_2KB;
- case 4:
- return ADDR_SURF_TILE_SPLIT_4KB;
- }
-}
-
static void evergreen_cs_track_init(struct evergreen_cs_track *track)
{
int i;
track->cb_color_bo[i] = NULL;
track->cb_color_bo_offset[i] = 0xFFFFFFFF;
track->cb_color_info[i] = 0;
- track->cb_color_view[i] = 0;
+ track->cb_color_view[i] = 0xFFFFFFFF;
track->cb_color_pitch_idx[i] = 0;
track->cb_color_slice_idx[i] = 0;
track->cb_color_dim[i] = 0;
}
}
+struct eg_surface {
+ /* value gathered from cs */
+ unsigned nbx;
+ unsigned nby;
+ unsigned format;
+ unsigned mode;
+ unsigned nbanks;
+ unsigned bankw;
+ unsigned bankh;
+ unsigned tsplit;
+ unsigned mtilea;
+ unsigned nsamples;
+ /* output value */
+ unsigned bpe;
+ unsigned layer_size;
+ unsigned palign;
+ unsigned halign;
+ unsigned long base_align;
+};
+
+static int evergreen_surface_check_linear(struct radeon_cs_parser *p,
+ struct eg_surface *surf,
+ const char *prefix)
+{
+ surf->layer_size = surf->nbx * surf->nby * surf->bpe * surf->nsamples;
+ surf->base_align = surf->bpe;
+ surf->palign = 1;
+ surf->halign = 1;
+ return 0;
+}
+
+static int evergreen_surface_check_linear_aligned(struct radeon_cs_parser *p,
+ struct eg_surface *surf,
+ const char *prefix)
+{
+ struct evergreen_cs_track *track = p->track;
+ unsigned palign;
+
+ palign = MAX(64, track->group_size / surf->bpe);
+ surf->layer_size = surf->nbx * surf->nby * surf->bpe * surf->nsamples;
+ surf->base_align = track->group_size;
+ surf->palign = palign;
+ surf->halign = 1;
+ if (surf->nbx & (palign - 1)) {
+ if (prefix) {
+ dev_warn(p->dev, "%s:%d %s pitch %d invalid must be aligned with %d\n",
+ __func__, __LINE__, prefix, surf->nbx, palign);
+ }
+ return -EINVAL;
+ }
+ return 0;
+}
+
+static int evergreen_surface_check_1d(struct radeon_cs_parser *p,
+ struct eg_surface *surf,
+ const char *prefix)
+{
+ struct evergreen_cs_track *track = p->track;
+ unsigned palign;
+
+ palign = track->group_size / (8 * surf->bpe * surf->nsamples);
+ palign = MAX(8, palign);
+ surf->layer_size = surf->nbx * surf->nby * surf->bpe;
+ surf->base_align = track->group_size;
+ surf->palign = palign;
+ surf->halign = 8;
+ if ((surf->nbx & (palign - 1))) {
+ if (prefix) {
+ dev_warn(p->dev, "%s:%d %s pitch %d invalid must be aligned with %d (%d %d %d)\n",
+ __func__, __LINE__, prefix, surf->nbx, palign,
+ track->group_size, surf->bpe, surf->nsamples);
+ }
+ return -EINVAL;
+ }
+ if ((surf->nby & (8 - 1))) {
+ if (prefix) {
+ dev_warn(p->dev, "%s:%d %s height %d invalid must be aligned with 8\n",
+ __func__, __LINE__, prefix, surf->nby);
+ }
+ return -EINVAL;
+ }
+ return 0;
+}
+
+static int evergreen_surface_check_2d(struct radeon_cs_parser *p,
+ struct eg_surface *surf,
+ const char *prefix)
+{
+ struct evergreen_cs_track *track = p->track;
+ unsigned palign, halign, tileb, slice_pt;
+
+ tileb = 64 * surf->bpe * surf->nsamples;
+ palign = track->group_size / (8 * surf->bpe * surf->nsamples);
+ palign = MAX(8, palign);
+ slice_pt = 1;
+ if (tileb > surf->tsplit) {
+ slice_pt = tileb / surf->tsplit;
+ }
+ tileb = tileb / slice_pt;
+ /* macro tile width & height */
+ palign = (8 * surf->bankw * track->npipes) * surf->mtilea;
+ halign = (8 * surf->bankh * surf->nbanks) / surf->mtilea;
+ surf->layer_size = surf->nbx * surf->nby * surf->bpe * slice_pt;
+ surf->base_align = (palign / 8) * (halign / 8) * tileb;
+ surf->palign = palign;
+ surf->halign = halign;
+
+ if ((surf->nbx & (palign - 1))) {
+ if (prefix) {
+ dev_warn(p->dev, "%s:%d %s pitch %d invalid must be aligned with %d\n",
+ __func__, __LINE__, prefix, surf->nbx, palign);
+ }
+ return -EINVAL;
+ }
+ if ((surf->nby & (halign - 1))) {
+ if (prefix) {
+ dev_warn(p->dev, "%s:%d %s height %d invalid must be aligned with %d\n",
+ __func__, __LINE__, prefix, surf->nby, halign);
+ }
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int evergreen_surface_check(struct radeon_cs_parser *p,
+ struct eg_surface *surf,
+ const char *prefix)
+{
+ /* some common value computed here */
+ surf->bpe = r600_fmt_get_blocksize(surf->format);
+
+ switch (surf->mode) {
+ case ARRAY_LINEAR_GENERAL:
+ return evergreen_surface_check_linear(p, surf, prefix);
+ case ARRAY_LINEAR_ALIGNED:
+ return evergreen_surface_check_linear_aligned(p, surf, prefix);
+ case ARRAY_1D_TILED_THIN1:
+ return evergreen_surface_check_1d(p, surf, prefix);
+ case ARRAY_2D_TILED_THIN1:
+ return evergreen_surface_check_2d(p, surf, prefix);
+ default:
+ dev_warn(p->dev, "%s:%d invalid array mode %d\n",
+ __func__, __LINE__, surf->mode);
+ return -EINVAL;
+ }
+ return -EINVAL;
+}
+
+static int evergreen_surface_value_conv_check(struct radeon_cs_parser *p,
+ struct eg_surface *surf,
+ const char *prefix)
+{
+ switch (surf->mode) {
+ case ARRAY_2D_TILED_THIN1:
+ break;
+ case ARRAY_LINEAR_GENERAL:
+ case ARRAY_LINEAR_ALIGNED:
+ case ARRAY_1D_TILED_THIN1:
+ return 0;
+ default:
+ dev_warn(p->dev, "%s:%d invalid array mode %d\n",
+ __func__, __LINE__, surf->mode);
+ return -EINVAL;
+ }
+
+ switch (surf->nbanks) {
+ case 0: surf->nbanks = 2; break;
+ case 1: surf->nbanks = 4; break;
+ case 2: surf->nbanks = 8; break;
+ case 3: surf->nbanks = 16; break;
+ default:
+ dev_warn(p->dev, "%s:%d %s invalid number of banks %d\n",
+ __func__, __LINE__, prefix, surf->nbanks);
+ return -EINVAL;
+ }
+ switch (surf->bankw) {
+ case 0: surf->bankw = 1; break;
+ case 1: surf->bankw = 2; break;
+ case 2: surf->bankw = 4; break;
+ case 3: surf->bankw = 8; break;
+ default:
+ dev_warn(p->dev, "%s:%d %s invalid bankw %d\n",
+ __func__, __LINE__, prefix, surf->bankw);
+ return -EINVAL;
+ }
+ switch (surf->bankh) {
+ case 0: surf->bankh = 1; break;
+ case 1: surf->bankh = 2; break;
+ case 2: surf->bankh = 4; break;
+ case 3: surf->bankh = 8; break;
+ default:
+ dev_warn(p->dev, "%s:%d %s invalid bankh %d\n",
+ __func__, __LINE__, prefix, surf->bankh);
+ return -EINVAL;
+ }
+ switch (surf->mtilea) {
+ case 0: surf->mtilea = 1; break;
+ case 1: surf->mtilea = 2; break;
+ case 2: surf->mtilea = 4; break;
+ case 3: surf->mtilea = 8; break;
+ default:
+ dev_warn(p->dev, "%s:%d %s invalid macro tile aspect %d\n",
+ __func__, __LINE__, prefix, surf->mtilea);
+ return -EINVAL;
+ }
+ switch (surf->tsplit) {
+ case 0: surf->tsplit = 64; break;
+ case 1: surf->tsplit = 128; break;
+ case 2: surf->tsplit = 256; break;
+ case 3: surf->tsplit = 512; break;
+ case 4: surf->tsplit = 1024; break;
+ case 5: surf->tsplit = 2048; break;
+ case 6: surf->tsplit = 4096; break;
+ default:
+ dev_warn(p->dev, "%s:%d %s invalid tile split %d\n",
+ __func__, __LINE__, prefix, surf->tsplit);
+ return -EINVAL;
+ }
+ return 0;
+}
+
+static int evergreen_cs_track_validate_cb(struct radeon_cs_parser *p, unsigned id)
+{
+ struct evergreen_cs_track *track = p->track;
+ struct eg_surface surf;
+ unsigned pitch, slice, mslice;
+ unsigned long offset;
+ int r;
+
+ mslice = G_028C6C_SLICE_MAX(track->cb_color_view[id]) + 1;
+ pitch = track->cb_color_pitch[id];
+ slice = track->cb_color_slice[id];
+ surf.nbx = (pitch + 1) * 8;
+ surf.nby = ((slice + 1) * 64) / surf.nbx;
+ surf.mode = G_028C70_ARRAY_MODE(track->cb_color_info[id]);
+ surf.format = G_028C70_FORMAT(track->cb_color_info[id]);
+ surf.tsplit = G_028C74_TILE_SPLIT(track->cb_color_attrib[id]);
+ surf.nbanks = G_028C74_NUM_BANKS(track->cb_color_attrib[id]);
+ surf.bankw = G_028C74_BANK_WIDTH(track->cb_color_attrib[id]);
+ surf.bankh = G_028C74_BANK_HEIGHT(track->cb_color_attrib[id]);
+ surf.mtilea = G_028C74_MACRO_TILE_ASPECT(track->cb_color_attrib[id]);
+ surf.nsamples = 1;
+
+ if (!r600_fmt_is_valid_color(surf.format)) {
+ dev_warn(p->dev, "%s:%d cb invalid format %d for %d (0x%08x)\n",
+ __func__, __LINE__, surf.format,
+ id, track->cb_color_info[id]);
+ return -EINVAL;
+ }
+
+ r = evergreen_surface_value_conv_check(p, &surf, "cb");
+ if (r) {
+ return r;
+ }
+
+ r = evergreen_surface_check(p, &surf, "cb");
+ if (r) {
+ dev_warn(p->dev, "%s:%d cb[%d] invalid (0x%08x 0x%08x 0x%08x 0x%08x)\n",
+ __func__, __LINE__, id, track->cb_color_pitch[id],
+ track->cb_color_slice[id], track->cb_color_attrib[id],
+ track->cb_color_info[id]);
+ return r;
+ }
+
+ offset = track->cb_color_bo_offset[id] << 8;
+ if (offset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d cb[%d] bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, id, offset, surf.base_align);
+ return -EINVAL;
+ }
+
+ offset += surf.layer_size * mslice;
+ if (offset > radeon_bo_size(track->cb_color_bo[id])) {
+ dev_warn(p->dev, "%s:%d cb[%d] bo too small (layer size %d, "
+ "offset %d, max layer %d, bo size %ld, slice %d)\n",
+ __func__, __LINE__, id, surf.layer_size,
+ track->cb_color_bo_offset[id] << 8, mslice,
+ radeon_bo_size(track->cb_color_bo[id]), slice);
+ dev_warn(p->dev, "%s:%d problematic surf: (%d %d) (%d %d %d %d %d %d %d)\n",
+ __func__, __LINE__, surf.nbx, surf.nby,
+ surf.mode, surf.bpe, surf.nsamples,
+ surf.bankw, surf.bankh,
+ surf.tsplit, surf.mtilea);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int evergreen_cs_track_validate_stencil(struct radeon_cs_parser *p)
+{
+ struct evergreen_cs_track *track = p->track;
+ struct eg_surface surf;
+ unsigned pitch, slice, mslice;
+ unsigned long offset;
+ int r;
+
+ mslice = G_028008_SLICE_MAX(track->db_depth_view) + 1;
+ pitch = G_028058_PITCH_TILE_MAX(track->db_depth_size);
+ slice = track->db_depth_slice;
+ surf.nbx = (pitch + 1) * 8;
+ surf.nby = ((slice + 1) * 64) / surf.nbx;
+ surf.mode = G_028040_ARRAY_MODE(track->db_z_info);
+ surf.format = G_028044_FORMAT(track->db_s_info);
+ surf.tsplit = G_028044_TILE_SPLIT(track->db_s_info);
+ surf.nbanks = G_028040_NUM_BANKS(track->db_z_info);
+ surf.bankw = G_028040_BANK_WIDTH(track->db_z_info);
+ surf.bankh = G_028040_BANK_HEIGHT(track->db_z_info);
+ surf.mtilea = G_028040_MACRO_TILE_ASPECT(track->db_z_info);
+ surf.nsamples = 1;
+
+ if (surf.format != 1) {
+ dev_warn(p->dev, "%s:%d stencil invalid format %d\n",
+ __func__, __LINE__, surf.format);
+ return -EINVAL;
+ }
+ /* replace by color format so we can use same code */
+ surf.format = V_028C70_COLOR_8;
+
+ r = evergreen_surface_value_conv_check(p, &surf, "stencil");
+ if (r) {
+ return r;
+ }
+
+ r = evergreen_surface_check(p, &surf, NULL);
+ if (r) {
+ /* old userspace doesn't compute proper depth/stencil alignment
+ * check that alignment against a bigger byte per elements and
+ * only report if that alignment is wrong too.
+ */
+ surf.format = V_028C70_COLOR_8_8_8_8;
+ r = evergreen_surface_check(p, &surf, "stencil");
+ if (r) {
+ dev_warn(p->dev, "%s:%d stencil invalid (0x%08x 0x%08x 0x%08x 0x%08x)\n",
+ __func__, __LINE__, track->db_depth_size,
+ track->db_depth_slice, track->db_s_info, track->db_z_info);
+ }
+ return r;
+ }
+
+ offset = track->db_s_read_offset << 8;
+ if (offset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d stencil read bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, offset, surf.base_align);
+ return -EINVAL;
+ }
+ offset += surf.layer_size * mslice;
+ if (offset > radeon_bo_size(track->db_s_read_bo)) {
+ dev_warn(p->dev, "%s:%d stencil read bo too small (layer size %d, "
+ "offset %ld, max layer %d, bo size %ld)\n",
+ __func__, __LINE__, surf.layer_size,
+ (unsigned long)track->db_s_read_offset << 8, mslice,
+ radeon_bo_size(track->db_s_read_bo));
+ dev_warn(p->dev, "%s:%d stencil invalid (0x%08x 0x%08x 0x%08x 0x%08x)\n",
+ __func__, __LINE__, track->db_depth_size,
+ track->db_depth_slice, track->db_s_info, track->db_z_info);
+ return -EINVAL;
+ }
+
+ offset = track->db_s_write_offset << 8;
+ if (offset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d stencil write bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, offset, surf.base_align);
+ return -EINVAL;
+ }
+ offset += surf.layer_size * mslice;
+ if (offset > radeon_bo_size(track->db_s_write_bo)) {
+ dev_warn(p->dev, "%s:%d stencil write bo too small (layer size %d, "
+ "offset %ld, max layer %d, bo size %ld)\n",
+ __func__, __LINE__, surf.layer_size,
+ (unsigned long)track->db_s_write_offset << 8, mslice,
+ radeon_bo_size(track->db_s_write_bo));
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int evergreen_cs_track_validate_depth(struct radeon_cs_parser *p)
+{
+ struct evergreen_cs_track *track = p->track;
+ struct eg_surface surf;
+ unsigned pitch, slice, mslice;
+ unsigned long offset;
+ int r;
+
+ mslice = G_028008_SLICE_MAX(track->db_depth_view) + 1;
+ pitch = G_028058_PITCH_TILE_MAX(track->db_depth_size);
+ slice = track->db_depth_slice;
+ surf.nbx = (pitch + 1) * 8;
+ surf.nby = ((slice + 1) * 64) / surf.nbx;
+ surf.mode = G_028040_ARRAY_MODE(track->db_z_info);
+ surf.format = G_028040_FORMAT(track->db_z_info);
+ surf.tsplit = G_028040_TILE_SPLIT(track->db_z_info);
+ surf.nbanks = G_028040_NUM_BANKS(track->db_z_info);
+ surf.bankw = G_028040_BANK_WIDTH(track->db_z_info);
+ surf.bankh = G_028040_BANK_HEIGHT(track->db_z_info);
+ surf.mtilea = G_028040_MACRO_TILE_ASPECT(track->db_z_info);
+ surf.nsamples = 1;
+
+ switch (surf.format) {
+ case V_028040_Z_16:
+ surf.format = V_028C70_COLOR_16;
+ break;
+ case V_028040_Z_24:
+ case V_028040_Z_32_FLOAT:
+ surf.format = V_028C70_COLOR_8_8_8_8;
+ break;
+ default:
+ dev_warn(p->dev, "%s:%d depth invalid format %d\n",
+ __func__, __LINE__, surf.format);
+ return -EINVAL;
+ }
+
+ r = evergreen_surface_value_conv_check(p, &surf, "depth");
+ if (r) {
+ dev_warn(p->dev, "%s:%d depth invalid (0x%08x 0x%08x 0x%08x)\n",
+ __func__, __LINE__, track->db_depth_size,
+ track->db_depth_slice, track->db_z_info);
+ return r;
+ }
+
+ r = evergreen_surface_check(p, &surf, "depth");
+ if (r) {
+ dev_warn(p->dev, "%s:%d depth invalid (0x%08x 0x%08x 0x%08x)\n",
+ __func__, __LINE__, track->db_depth_size,
+ track->db_depth_slice, track->db_z_info);
+ return r;
+ }
+
+ offset = track->db_z_read_offset << 8;
+ if (offset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d stencil read bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, offset, surf.base_align);
+ return -EINVAL;
+ }
+ offset += surf.layer_size * mslice;
+ if (offset > radeon_bo_size(track->db_z_read_bo)) {
+ dev_warn(p->dev, "%s:%d depth read bo too small (layer size %d, "
+ "offset %ld, max layer %d, bo size %ld)\n",
+ __func__, __LINE__, surf.layer_size,
+ (unsigned long)track->db_z_read_offset << 8, mslice,
+ radeon_bo_size(track->db_z_read_bo));
+ return -EINVAL;
+ }
+
+ offset = track->db_z_write_offset << 8;
+ if (offset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d stencil write bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, offset, surf.base_align);
+ return -EINVAL;
+ }
+ offset += surf.layer_size * mslice;
+ if (offset > radeon_bo_size(track->db_z_write_bo)) {
+ dev_warn(p->dev, "%s:%d depth write bo too small (layer size %d, "
+ "offset %ld, max layer %d, bo size %ld)\n",
+ __func__, __LINE__, surf.layer_size,
+ (unsigned long)track->db_z_write_offset << 8, mslice,
+ radeon_bo_size(track->db_z_write_bo));
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int evergreen_cs_track_validate_texture(struct radeon_cs_parser *p,
+ struct radeon_bo *texture,
+ struct radeon_bo *mipmap,
+ unsigned idx)
+{
+ struct eg_surface surf;
+ unsigned long toffset, moffset;
+ unsigned dim, llevel, mslice, width, height, depth, i;
+ u32 texdw[7];
+ int r;
+
+ texdw[0] = radeon_get_ib_value(p, idx + 0);
+ texdw[1] = radeon_get_ib_value(p, idx + 1);
+ texdw[2] = radeon_get_ib_value(p, idx + 2);
+ texdw[3] = radeon_get_ib_value(p, idx + 3);
+ texdw[4] = radeon_get_ib_value(p, idx + 4);
+ texdw[5] = radeon_get_ib_value(p, idx + 5);
+ texdw[6] = radeon_get_ib_value(p, idx + 6);
+ texdw[7] = radeon_get_ib_value(p, idx + 7);
+ dim = G_030000_DIM(texdw[0]);
+ llevel = G_030014_LAST_LEVEL(texdw[5]);
+ mslice = G_030014_LAST_ARRAY(texdw[5]) + 1;
+ width = G_030000_TEX_WIDTH(texdw[0]) + 1;
+ height = G_030004_TEX_HEIGHT(texdw[1]) + 1;
+ depth = G_030004_TEX_DEPTH(texdw[1]) + 1;
+ surf.format = G_03001C_DATA_FORMAT(texdw[7]);
+ surf.nbx = (G_030000_PITCH(texdw[0]) + 1) * 8;
+ surf.nbx = r600_fmt_get_nblocksx(surf.format, surf.nbx);
+ surf.nby = r600_fmt_get_nblocksy(surf.format, height);
+ surf.mode = G_030004_ARRAY_MODE(texdw[1]);
+ surf.tsplit = G_030018_TILE_SPLIT(texdw[6]);
+ surf.nbanks = G_03001C_NUM_BANKS(texdw[7]);
+ surf.bankw = G_03001C_BANK_WIDTH(texdw[7]);
+ surf.bankh = G_03001C_BANK_HEIGHT(texdw[7]);
+ surf.mtilea = G_03001C_MACRO_TILE_ASPECT(texdw[7]);
+ surf.nsamples = 1;
+ toffset = texdw[2] << 8;
+ moffset = texdw[3] << 8;
+
+ if (!r600_fmt_is_valid_texture(surf.format, p->family)) {
+ dev_warn(p->dev, "%s:%d texture invalid format %d\n",
+ __func__, __LINE__, surf.format);
+ return -EINVAL;
+ }
+ switch (dim) {
+ case V_030000_SQ_TEX_DIM_1D:
+ case V_030000_SQ_TEX_DIM_2D:
+ case V_030000_SQ_TEX_DIM_CUBEMAP:
+ case V_030000_SQ_TEX_DIM_1D_ARRAY:
+ case V_030000_SQ_TEX_DIM_2D_ARRAY:
+ depth = 1;
+ case V_030000_SQ_TEX_DIM_3D:
+ break;
+ default:
+ dev_warn(p->dev, "%s:%d texture invalid dimension %d\n",
+ __func__, __LINE__, dim);
+ return -EINVAL;
+ }
+
+ r = evergreen_surface_value_conv_check(p, &surf, "texture");
+ if (r) {
+ return r;
+ }
+
+ /* align height */
+ evergreen_surface_check(p, &surf, NULL);
+ surf.nby = ALIGN(surf.nby, surf.halign);
+
+ r = evergreen_surface_check(p, &surf, "texture");
+ if (r) {
+ dev_warn(p->dev, "%s:%d texture invalid 0x%08x 0x%08x 0x%08x 0x%08x 0x%08x 0x%08x\n",
+ __func__, __LINE__, texdw[0], texdw[1], texdw[4],
+ texdw[5], texdw[6], texdw[7]);
+ return r;
+ }
+
+ /* check texture size */
+ if (toffset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d texture bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, toffset, surf.base_align);
+ return -EINVAL;
+ }
+ if (moffset & (surf.base_align - 1)) {
+ dev_warn(p->dev, "%s:%d mipmap bo base %ld not aligned with %ld\n",
+ __func__, __LINE__, moffset, surf.base_align);
+ return -EINVAL;
+ }
+ if (dim == SQ_TEX_DIM_3D) {
+ toffset += surf.layer_size * depth;
+ } else {
+ toffset += surf.layer_size * mslice;
+ }
+ if (toffset > radeon_bo_size(texture)) {
+ dev_warn(p->dev, "%s:%d texture bo too small (layer size %d, "
+ "offset %ld, max layer %d, depth %d, bo size %ld) (%d %d)\n",
+ __func__, __LINE__, surf.layer_size,
+ (unsigned long)texdw[2] << 8, mslice,
+ depth, radeon_bo_size(texture),
+ surf.nbx, surf.nby);
+ return -EINVAL;
+ }
+
+ /* check mipmap size */
+ for (i = 1; i <= llevel; i++) {
+ unsigned w, h, d;
+
+ w = r600_mip_minify(width, i);
+ h = r600_mip_minify(height, i);
+ d = r600_mip_minify(depth, i);
+ surf.nbx = r600_fmt_get_nblocksx(surf.format, w);
+ surf.nby = r600_fmt_get_nblocksy(surf.format, h);
+
+ switch (surf.mode) {
+ case ARRAY_2D_TILED_THIN1:
+ if (surf.nbx < surf.palign || surf.nby < surf.halign) {
+ surf.mode = ARRAY_1D_TILED_THIN1;
+ }
+ /* recompute alignment */
+ evergreen_surface_check(p, &surf, NULL);
+ break;
+ case ARRAY_LINEAR_GENERAL:
+ case ARRAY_LINEAR_ALIGNED:
+ case ARRAY_1D_TILED_THIN1:
+ break;
+ default:
+ dev_warn(p->dev, "%s:%d invalid array mode %d\n",
+ __func__, __LINE__, surf.mode);
+ return -EINVAL;
+ }
+ surf.nbx = ALIGN(surf.nbx, surf.palign);
+ surf.nby = ALIGN(surf.nby, surf.halign);
+
+ r = evergreen_surface_check(p, &surf, "mipmap");
+ if (r) {
+ return r;
+ }
+
+ if (dim == SQ_TEX_DIM_3D) {
+ moffset += surf.layer_size * d;
+ } else {
+ moffset += surf.layer_size * mslice;
+ }
+ if (moffset > radeon_bo_size(mipmap)) {
+ dev_warn(p->dev, "%s:%d mipmap [%d] bo too small (layer size %d, "
+ "offset %ld, coffset %ld, max layer %d, depth %d, "
+ "bo size %ld) level0 (%d %d %d)\n",
+ __func__, __LINE__, i, surf.layer_size,
+ (unsigned long)texdw[3] << 8, moffset, mslice,
+ d, radeon_bo_size(mipmap),
+ width, height, depth);
+ dev_warn(p->dev, "%s:%d problematic surf: (%d %d) (%d %d %d %d %d %d %d)\n",
+ __func__, __LINE__, surf.nbx, surf.nby,
+ surf.mode, surf.bpe, surf.nsamples,
+ surf.bankw, surf.bankh,
+ surf.tsplit, surf.mtilea);
+ return -EINVAL;
+ }
+ }
+
+ return 0;
+}
+
static int evergreen_cs_track_check(struct radeon_cs_parser *p)
{
struct evergreen_cs_track *track = p->track;
- int i, j;
+ unsigned tmp, i, j;
+ int r;
/* check streamout */
for (i = 0; i < 4; i++) {
}
}
- /* XXX fill in */
+ /* check that we have a cb for each enabled target
+ */
+ tmp = track->cb_target_mask;
+ for (i = 0; i < 8; i++) {
+ if ((tmp >> (i * 4)) & 0xF) {
+ /* at least one component is enabled */
+ if (track->cb_color_bo[i] == NULL) {
+ dev_warn(p->dev, "%s:%d mask 0x%08X | 0x%08X no cb for %d\n",
+ __func__, __LINE__, track->cb_target_mask, track->cb_shader_mask, i);
+ return -EINVAL;
+ }
+ /* check cb */
+ r = evergreen_cs_track_validate_cb(p, i);
+ if (r) {
+ return r;
+ }
+ }
+ }
+
+ /* Check stencil buffer */
+ if (G_028800_STENCIL_ENABLE(track->db_depth_control)) {
+ r = evergreen_cs_track_validate_stencil(p);
+ if (r)
+ return r;
+ }
+ /* Check depth buffer */
+ if (G_028800_Z_WRITE_ENABLE(track->db_depth_control)) {
+ r = evergreen_cs_track_validate_depth(p);
+ if (r)
+ return r;
+ }
+
return 0;
}
ib[idx] |= Z_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags));
track->db_z_info |= Z_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags));
if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) {
+ unsigned bankw, bankh, mtaspect, tile_split;
+
+ evergreen_tiling_fields(reloc->lobj.tiling_flags,
+ &bankw, &bankh, &mtaspect,
+ &tile_split);
ib[idx] |= DB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks));
- ib[idx] |= DB_TILE_SPLIT(evergreen_cs_get_tile_split(track->row_size));
+ ib[idx] |= DB_TILE_SPLIT(tile_split) |
+ DB_BANK_WIDTH(bankw) |
+ DB_BANK_HEIGHT(bankh) |
+ DB_MACRO_TILE_ASPECT(mtaspect);
}
}
break;
track->db_depth_size = radeon_get_ib_value(p, idx);
track->db_depth_size_idx = idx;
break;
+ case R_02805C_DB_DEPTH_SLICE:
+ track->db_depth_slice = radeon_get_ib_value(p, idx);
+ break;
case DB_Z_READ_BASE:
r = evergreen_cs_packet_next_reloc(p, &reloc);
if (r) {
case CB_COLOR5_ATTRIB:
case CB_COLOR6_ATTRIB:
case CB_COLOR7_ATTRIB:
+ r = evergreen_cs_packet_next_reloc(p, &reloc);
+ if (r) {
+ dev_warn(p->dev, "bad SET_CONTEXT_REG "
+ "0x%04X\n", reg);
+ return -EINVAL;
+ }
+ if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) {
+ if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) {
+ unsigned bankw, bankh, mtaspect, tile_split;
+
+ evergreen_tiling_fields(reloc->lobj.tiling_flags,
+ &bankw, &bankh, &mtaspect,
+ &tile_split);
+ ib[idx] |= CB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks));
+ ib[idx] |= CB_TILE_SPLIT(tile_split) |
+ CB_BANK_WIDTH(bankw) |
+ CB_BANK_HEIGHT(bankh) |
+ CB_MACRO_TILE_ASPECT(mtaspect);
+ }
+ }
+ tmp = ((reg - CB_COLOR0_ATTRIB) / 0x3c);
+ track->cb_color_attrib[tmp] = ib[idx];
+ break;
case CB_COLOR8_ATTRIB:
case CB_COLOR9_ATTRIB:
case CB_COLOR10_ATTRIB:
"0x%04X\n", reg);
return -EINVAL;
}
- if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) {
- ib[idx] |= CB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks));
- ib[idx] |= CB_TILE_SPLIT(evergreen_cs_get_tile_split(track->row_size));
+ if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) {
+ if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) {
+ unsigned bankw, bankh, mtaspect, tile_split;
+
+ evergreen_tiling_fields(reloc->lobj.tiling_flags,
+ &bankw, &bankh, &mtaspect,
+ &tile_split);
+ ib[idx] |= CB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks));
+ ib[idx] |= CB_TILE_SPLIT(tile_split) |
+ CB_BANK_WIDTH(bankw) |
+ CB_BANK_HEIGHT(bankh) |
+ CB_MACRO_TILE_ASPECT(mtaspect);
+ }
}
+ tmp = ((reg - CB_COLOR8_ATTRIB) / 0x1c) + 8;
+ track->cb_color_attrib[tmp] = ib[idx];
break;
case CB_COLOR0_DIM:
case CB_COLOR1_DIM:
return 0;
}
-/**
- * evergreen_check_texture_resource() - check if register is authorized or not
- * @p: parser structure holding parsing context
- * @idx: index into the cs buffer
- * @texture: texture's bo structure
- * @mipmap: mipmap's bo structure
- *
- * This function will check that the resource has valid field and that
- * the texture and mipmap bo object are big enough to cover this resource.
- */
-static int evergreen_check_texture_resource(struct radeon_cs_parser *p, u32 idx,
- struct radeon_bo *texture,
- struct radeon_bo *mipmap)
-{
- /* XXX fill in */
- return 0;
-}
-
static bool evergreen_is_safe_reg(struct radeon_cs_parser *p, u32 reg, u32 idx)
{
u32 last_reg, m, i;
}
for (i = 0; i < (pkt->count / 8); i++) {
struct radeon_bo *texture, *mipmap;
+ u32 toffset, moffset;
u32 size, offset;
switch (G__SQ_CONSTANT_TYPE(radeon_get_ib_value(p, idx+1+(i*8)+7))) {
DRM_ERROR("bad SET_RESOURCE (tex)\n");
return -EINVAL;
}
- ib[idx+1+(i*8)+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) {
ib[idx+1+(i*8)+1] |=
TEX_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags));
if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) {
- ib[idx+1+(i*8)+6] |=
- TEX_TILE_SPLIT(evergreen_cs_get_tile_split(track->row_size));
+ unsigned bankw, bankh, mtaspect, tile_split;
+
+ evergreen_tiling_fields(reloc->lobj.tiling_flags,
+ &bankw, &bankh, &mtaspect,
+ &tile_split);
+ ib[idx+1+(i*8)+6] |= TEX_TILE_SPLIT(tile_split);
ib[idx+1+(i*8)+7] |=
+ TEX_BANK_WIDTH(bankw) |
+ TEX_BANK_HEIGHT(bankh) |
+ MACRO_TILE_ASPECT(mtaspect) |
TEX_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks));
}
}
texture = reloc->robj;
+ toffset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
/* tex mip base */
r = evergreen_cs_packet_next_reloc(p, &reloc);
if (r) {
DRM_ERROR("bad SET_RESOURCE (tex)\n");
return -EINVAL;
}
- ib[idx+1+(i*8)+3] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
+ moffset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
mipmap = reloc->robj;
- r = evergreen_check_texture_resource(p, idx+1+(i*8),
- texture, mipmap);
+ r = evergreen_cs_track_validate_texture(p, texture, mipmap, idx+1+(i*8));
if (r)
return r;
+ ib[idx+1+(i*8)+2] += toffset;
+ ib[idx+1+(i*8)+3] += moffset;
break;
case SQ_TEX_VTX_VALID_BUFFER:
/* vtx base */
#define DB_DEBUG4 0x983C
#define DB_WATERMARKS 0x9854
#define DB_DEPTH_CONTROL 0x28800
+#define R_028800_DB_DEPTH_CONTROL 0x028800
+#define S_028800_STENCIL_ENABLE(x) (((x) & 0x1) << 0)
+#define G_028800_STENCIL_ENABLE(x) (((x) >> 0) & 0x1)
+#define C_028800_STENCIL_ENABLE 0xFFFFFFFE
+#define S_028800_Z_ENABLE(x) (((x) & 0x1) << 1)
+#define G_028800_Z_ENABLE(x) (((x) >> 1) & 0x1)
+#define C_028800_Z_ENABLE 0xFFFFFFFD
+#define S_028800_Z_WRITE_ENABLE(x) (((x) & 0x1) << 2)
+#define G_028800_Z_WRITE_ENABLE(x) (((x) >> 2) & 0x1)
+#define C_028800_Z_WRITE_ENABLE 0xFFFFFFFB
+#define S_028800_ZFUNC(x) (((x) & 0x7) << 4)
+#define G_028800_ZFUNC(x) (((x) >> 4) & 0x7)
+#define C_028800_ZFUNC 0xFFFFFF8F
+#define S_028800_BACKFACE_ENABLE(x) (((x) & 0x1) << 7)
+#define G_028800_BACKFACE_ENABLE(x) (((x) >> 7) & 0x1)
+#define C_028800_BACKFACE_ENABLE 0xFFFFFF7F
+#define S_028800_STENCILFUNC(x) (((x) & 0x7) << 8)
+#define G_028800_STENCILFUNC(x) (((x) >> 8) & 0x7)
+#define C_028800_STENCILFUNC 0xFFFFF8FF
+#define V_028800_STENCILFUNC_NEVER 0x00000000
+#define V_028800_STENCILFUNC_LESS 0x00000001
+#define V_028800_STENCILFUNC_EQUAL 0x00000002
+#define V_028800_STENCILFUNC_LEQUAL 0x00000003
+#define V_028800_STENCILFUNC_GREATER 0x00000004
+#define V_028800_STENCILFUNC_NOTEQUAL 0x00000005
+#define V_028800_STENCILFUNC_GEQUAL 0x00000006
+#define V_028800_STENCILFUNC_ALWAYS 0x00000007
+#define S_028800_STENCILFAIL(x) (((x) & 0x7) << 11)
+#define G_028800_STENCILFAIL(x) (((x) >> 11) & 0x7)
+#define C_028800_STENCILFAIL 0xFFFFC7FF
+#define V_028800_STENCIL_KEEP 0x00000000
+#define V_028800_STENCIL_ZERO 0x00000001
+#define V_028800_STENCIL_REPLACE 0x00000002
+#define V_028800_STENCIL_INCR 0x00000003
+#define V_028800_STENCIL_DECR 0x00000004
+#define V_028800_STENCIL_INVERT 0x00000005
+#define V_028800_STENCIL_INCR_WRAP 0x00000006
+#define V_028800_STENCIL_DECR_WRAP 0x00000007
+#define S_028800_STENCILZPASS(x) (((x) & 0x7) << 14)
+#define G_028800_STENCILZPASS(x) (((x) >> 14) & 0x7)
+#define C_028800_STENCILZPASS 0xFFFE3FFF
+#define S_028800_STENCILZFAIL(x) (((x) & 0x7) << 17)
+#define G_028800_STENCILZFAIL(x) (((x) >> 17) & 0x7)
+#define C_028800_STENCILZFAIL 0xFFF1FFFF
+#define S_028800_STENCILFUNC_BF(x) (((x) & 0x7) << 20)
+#define G_028800_STENCILFUNC_BF(x) (((x) >> 20) & 0x7)
+#define C_028800_STENCILFUNC_BF 0xFF8FFFFF
+#define S_028800_STENCILFAIL_BF(x) (((x) & 0x7) << 23)
+#define G_028800_STENCILFAIL_BF(x) (((x) >> 23) & 0x7)
+#define C_028800_STENCILFAIL_BF 0xFC7FFFFF
+#define S_028800_STENCILZPASS_BF(x) (((x) & 0x7) << 26)
+#define G_028800_STENCILZPASS_BF(x) (((x) >> 26) & 0x7)
+#define C_028800_STENCILZPASS_BF 0xE3FFFFFF
+#define S_028800_STENCILZFAIL_BF(x) (((x) & 0x7) << 29)
+#define G_028800_STENCILZFAIL_BF(x) (((x) >> 29) & 0x7)
+#define C_028800_STENCILZFAIL_BF 0x1FFFFFFF
#define DB_DEPTH_VIEW 0x28008
+#define R_028008_DB_DEPTH_VIEW 0x00028008
+#define S_028008_SLICE_START(x) (((x) & 0x7FF) << 0)
+#define G_028008_SLICE_START(x) (((x) >> 0) & 0x7FF)
+#define C_028008_SLICE_START 0xFFFFF800
+#define S_028008_SLICE_MAX(x) (((x) & 0x7FF) << 13)
+#define G_028008_SLICE_MAX(x) (((x) >> 13) & 0x7FF)
+#define C_028008_SLICE_MAX 0xFF001FFF
#define DB_HTILE_DATA_BASE 0x28014
#define DB_Z_INFO 0x28040
# define Z_ARRAY_MODE(x) ((x) << 4)
# define DB_NUM_BANKS(x) (((x) & 0x3) << 12)
# define DB_BANK_WIDTH(x) (((x) & 0x3) << 16)
# define DB_BANK_HEIGHT(x) (((x) & 0x3) << 20)
+# define DB_MACRO_TILE_ASPECT(x) (((x) & 0x3) << 24)
+#define R_028040_DB_Z_INFO 0x028040
+#define S_028040_FORMAT(x) (((x) & 0x3) << 0)
+#define G_028040_FORMAT(x) (((x) >> 0) & 0x3)
+#define C_028040_FORMAT 0xFFFFFFFC
+#define V_028040_Z_INVALID 0x00000000
+#define V_028040_Z_16 0x00000001
+#define V_028040_Z_24 0x00000002
+#define V_028040_Z_32_FLOAT 0x00000003
+#define S_028040_ARRAY_MODE(x) (((x) & 0xF) << 4)
+#define G_028040_ARRAY_MODE(x) (((x) >> 4) & 0xF)
+#define C_028040_ARRAY_MODE 0xFFFFFF0F
+#define S_028040_READ_SIZE(x) (((x) & 0x1) << 28)
+#define G_028040_READ_SIZE(x) (((x) >> 28) & 0x1)
+#define C_028040_READ_SIZE 0xEFFFFFFF
+#define S_028040_TILE_SURFACE_ENABLE(x) (((x) & 0x1) << 29)
+#define G_028040_TILE_SURFACE_ENABLE(x) (((x) >> 29) & 0x1)
+#define C_028040_TILE_SURFACE_ENABLE 0xDFFFFFFF
+#define S_028040_ZRANGE_PRECISION(x) (((x) & 0x1) << 31)
+#define G_028040_ZRANGE_PRECISION(x) (((x) >> 31) & 0x1)
+#define C_028040_ZRANGE_PRECISION 0x7FFFFFFF
+#define S_028040_TILE_SPLIT(x) (((x) & 0x7) << 8)
+#define G_028040_TILE_SPLIT(x) (((x) >> 8) & 0x7)
+#define S_028040_NUM_BANKS(x) (((x) & 0x3) << 12)
+#define G_028040_NUM_BANKS(x) (((x) >> 12) & 0x3)
+#define S_028040_BANK_WIDTH(x) (((x) & 0x3) << 16)
+#define G_028040_BANK_WIDTH(x) (((x) >> 16) & 0x3)
+#define S_028040_BANK_HEIGHT(x) (((x) & 0x3) << 20)
+#define G_028040_BANK_HEIGHT(x) (((x) >> 20) & 0x3)
+#define S_028040_MACRO_TILE_ASPECT(x) (((x) & 0x3) << 24)
+#define G_028040_MACRO_TILE_ASPECT(x) (((x) >> 24) & 0x3)
#define DB_STENCIL_INFO 0x28044
+#define R_028044_DB_STENCIL_INFO 0x028044
+#define S_028044_FORMAT(x) (((x) & 0x1) << 0)
+#define G_028044_FORMAT(x) (((x) >> 0) & 0x1)
+#define C_028044_FORMAT 0xFFFFFFFE
+#define G_028044_TILE_SPLIT(x) (((x) >> 8) & 0x7)
#define DB_Z_READ_BASE 0x28048
#define DB_STENCIL_READ_BASE 0x2804c
#define DB_Z_WRITE_BASE 0x28050
#define DB_STENCIL_WRITE_BASE 0x28054
#define DB_DEPTH_SIZE 0x28058
+#define R_028058_DB_DEPTH_SIZE 0x028058
+#define S_028058_PITCH_TILE_MAX(x) (((x) & 0x7FF) << 0)
+#define G_028058_PITCH_TILE_MAX(x) (((x) >> 0) & 0x7FF)
+#define C_028058_PITCH_TILE_MAX 0xFFFFF800
+#define S_028058_HEIGHT_TILE_MAX(x) (((x) & 0x7FF) << 11)
+#define G_028058_HEIGHT_TILE_MAX(x) (((x) >> 11) & 0x7FF)
+#define C_028058_HEIGHT_TILE_MAX 0xFFC007FF
+#define R_02805C_DB_DEPTH_SLICE 0x02805C
+#define S_02805C_SLICE_TILE_MAX(x) (((x) & 0x3FFFFF) << 0)
+#define G_02805C_SLICE_TILE_MAX(x) (((x) >> 0) & 0x3FFFFF)
+#define C_02805C_SLICE_TILE_MAX 0xFFC00000
#define SQ_PGM_START_PS 0x28840
#define SQ_PGM_START_VS 0x2885c
#define CB_COLOR0_PITCH 0x28c64
#define CB_COLOR0_SLICE 0x28c68
#define CB_COLOR0_VIEW 0x28c6c
+#define R_028C6C_CB_COLOR0_VIEW 0x00028C6C
+#define S_028C6C_SLICE_START(x) (((x) & 0x7FF) << 0)
+#define G_028C6C_SLICE_START(x) (((x) >> 0) & 0x7FF)
+#define C_028C6C_SLICE_START 0xFFFFF800
+#define S_028C6C_SLICE_MAX(x) (((x) & 0x7FF) << 13)
+#define G_028C6C_SLICE_MAX(x) (((x) >> 13) & 0x7FF)
+#define C_028C6C_SLICE_MAX 0xFF001FFF
+#define R_028C70_CB_COLOR0_INFO 0x028C70
+#define S_028C70_ENDIAN(x) (((x) & 0x3) << 0)
+#define G_028C70_ENDIAN(x) (((x) >> 0) & 0x3)
+#define C_028C70_ENDIAN 0xFFFFFFFC
+#define S_028C70_FORMAT(x) (((x) & 0x3F) << 2)
+#define G_028C70_FORMAT(x) (((x) >> 2) & 0x3F)
+#define C_028C70_FORMAT 0xFFFFFF03
+#define V_028C70_COLOR_INVALID 0x00000000
+#define V_028C70_COLOR_8 0x00000001
+#define V_028C70_COLOR_4_4 0x00000002
+#define V_028C70_COLOR_3_3_2 0x00000003
+#define V_028C70_COLOR_16 0x00000005
+#define V_028C70_COLOR_16_FLOAT 0x00000006
+#define V_028C70_COLOR_8_8 0x00000007
+#define V_028C70_COLOR_5_6_5 0x00000008
+#define V_028C70_COLOR_6_5_5 0x00000009
+#define V_028C70_COLOR_1_5_5_5 0x0000000A
+#define V_028C70_COLOR_4_4_4_4 0x0000000B
+#define V_028C70_COLOR_5_5_5_1 0x0000000C
+#define V_028C70_COLOR_32 0x0000000D
+#define V_028C70_COLOR_32_FLOAT 0x0000000E
+#define V_028C70_COLOR_16_16 0x0000000F
+#define V_028C70_COLOR_16_16_FLOAT 0x00000010
+#define V_028C70_COLOR_8_24 0x00000011
+#define V_028C70_COLOR_8_24_FLOAT 0x00000012
+#define V_028C70_COLOR_24_8 0x00000013
+#define V_028C70_COLOR_24_8_FLOAT 0x00000014
+#define V_028C70_COLOR_10_11_11 0x00000015
+#define V_028C70_COLOR_10_11_11_FLOAT 0x00000016
+#define V_028C70_COLOR_11_11_10 0x00000017
+#define V_028C70_COLOR_11_11_10_FLOAT 0x00000018
+#define V_028C70_COLOR_2_10_10_10 0x00000019
+#define V_028C70_COLOR_8_8_8_8 0x0000001A
+#define V_028C70_COLOR_10_10_10_2 0x0000001B
+#define V_028C70_COLOR_X24_8_32_FLOAT 0x0000001C
+#define V_028C70_COLOR_32_32 0x0000001D
+#define V_028C70_COLOR_32_32_FLOAT 0x0000001E
+#define V_028C70_COLOR_16_16_16_16 0x0000001F
+#define V_028C70_COLOR_16_16_16_16_FLOAT 0x00000020
+#define V_028C70_COLOR_32_32_32_32 0x00000022
+#define V_028C70_COLOR_32_32_32_32_FLOAT 0x00000023
+#define V_028C70_COLOR_32_32_32_FLOAT 0x00000030
+#define S_028C70_ARRAY_MODE(x) (((x) & 0xF) << 8)
+#define G_028C70_ARRAY_MODE(x) (((x) >> 8) & 0xF)
+#define C_028C70_ARRAY_MODE 0xFFFFF0FF
+#define V_028C70_ARRAY_LINEAR_GENERAL 0x00000000
+#define V_028C70_ARRAY_LINEAR_ALIGNED 0x00000001
+#define V_028C70_ARRAY_1D_TILED_THIN1 0x00000002
+#define V_028C70_ARRAY_2D_TILED_THIN1 0x00000004
+#define S_028C70_NUMBER_TYPE(x) (((x) & 0x7) << 12)
+#define G_028C70_NUMBER_TYPE(x) (((x) >> 12) & 0x7)
+#define C_028C70_NUMBER_TYPE 0xFFFF8FFF
+#define V_028C70_NUMBER_UNORM 0x00000000
+#define V_028C70_NUMBER_SNORM 0x00000001
+#define V_028C70_NUMBER_USCALED 0x00000002
+#define V_028C70_NUMBER_SSCALED 0x00000003
+#define V_028C70_NUMBER_UINT 0x00000004
+#define V_028C70_NUMBER_SINT 0x00000005
+#define V_028C70_NUMBER_SRGB 0x00000006
+#define V_028C70_NUMBER_FLOAT 0x00000007
+#define S_028C70_COMP_SWAP(x) (((x) & 0x3) << 15)
+#define G_028C70_COMP_SWAP(x) (((x) >> 15) & 0x3)
+#define C_028C70_COMP_SWAP 0xFFFE7FFF
+#define V_028C70_SWAP_STD 0x00000000
+#define V_028C70_SWAP_ALT 0x00000001
+#define V_028C70_SWAP_STD_REV 0x00000002
+#define V_028C70_SWAP_ALT_REV 0x00000003
+#define S_028C70_FAST_CLEAR(x) (((x) & 0x1) << 17)
+#define G_028C70_FAST_CLEAR(x) (((x) >> 17) & 0x1)
+#define C_028C70_FAST_CLEAR 0xFFFDFFFF
+#define S_028C70_COMPRESSION(x) (((x) & 0x3) << 18)
+#define G_028C70_COMPRESSION(x) (((x) >> 18) & 0x3)
+#define C_028C70_COMPRESSION 0xFFF3FFFF
+#define S_028C70_BLEND_CLAMP(x) (((x) & 0x1) << 19)
+#define G_028C70_BLEND_CLAMP(x) (((x) >> 19) & 0x1)
+#define C_028C70_BLEND_CLAMP 0xFFF7FFFF
+#define S_028C70_BLEND_BYPASS(x) (((x) & 0x1) << 20)
+#define G_028C70_BLEND_BYPASS(x) (((x) >> 20) & 0x1)
+#define C_028C70_BLEND_BYPASS 0xFFEFFFFF
+#define S_028C70_SIMPLE_FLOAT(x) (((x) & 0x1) << 21)
+#define G_028C70_SIMPLE_FLOAT(x) (((x) >> 21) & 0x1)
+#define C_028C70_SIMPLE_FLOAT 0xFFDFFFFF
+#define S_028C70_ROUND_MODE(x) (((x) & 0x1) << 22)
+#define G_028C70_ROUND_MODE(x) (((x) >> 22) & 0x1)
+#define C_028C70_ROUND_MODE 0xFFBFFFFF
+#define S_028C70_TILE_COMPACT(x) (((x) & 0x1) << 23)
+#define G_028C70_TILE_COMPACT(x) (((x) >> 23) & 0x1)
+#define C_028C70_TILE_COMPACT 0xFF7FFFFF
+#define S_028C70_SOURCE_FORMAT(x) (((x) & 0x3) << 24)
+#define G_028C70_SOURCE_FORMAT(x) (((x) >> 24) & 0x3)
+#define C_028C70_SOURCE_FORMAT 0xFCFFFFFF
+#define V_028C70_EXPORT_4C_32BPC 0x0
+#define V_028C70_EXPORT_4C_16BPC 0x1
+#define V_028C70_EXPORT_2C_32BPC 0x2 /* Do not use */
+#define S_028C70_RAT(x) (((x) & 0x1) << 26)
+#define G_028C70_RAT(x) (((x) >> 26) & 0x1)
+#define C_028C70_RAT 0xFBFFFFFF
+#define S_028C70_RESOURCE_TYPE(x) (((x) & 0x7) << 27)
+#define G_028C70_RESOURCE_TYPE(x) (((x) >> 27) & 0x7)
+#define C_028C70_RESOURCE_TYPE 0xC7FFFFFF
+
#define CB_COLOR0_INFO 0x28c70
# define CB_FORMAT(x) ((x) << 2)
# define CB_ARRAY_MODE(x) ((x) << 8)
# define CB_SOURCE_FORMAT(x) ((x) << 24)
# define CB_SF_EXPORT_FULL 0
# define CB_SF_EXPORT_NORM 1
+#define R_028C74_CB_COLOR0_ATTRIB 0x028C74
+#define S_028C74_NON_DISP_TILING_ORDER(x) (((x) & 0x1) << 4)
+#define G_028C74_NON_DISP_TILING_ORDER(x) (((x) >> 4) & 0x1)
+#define C_028C74_NON_DISP_TILING_ORDER 0xFFFFFFEF
+#define S_028C74_TILE_SPLIT(x) (((x) & 0xf) << 5)
+#define G_028C74_TILE_SPLIT(x) (((x) >> 5) & 0xf)
+#define S_028C74_NUM_BANKS(x) (((x) & 0x3) << 10)
+#define G_028C74_NUM_BANKS(x) (((x) >> 10) & 0x3)
+#define S_028C74_BANK_WIDTH(x) (((x) & 0x3) << 13)
+#define G_028C74_BANK_WIDTH(x) (((x) >> 13) & 0x3)
+#define S_028C74_BANK_HEIGHT(x) (((x) & 0x3) << 16)
+#define G_028C74_BANK_HEIGHT(x) (((x) >> 16) & 0x3)
+#define S_028C74_MACRO_TILE_ASPECT(x) (((x) & 0x3) << 19)
+#define G_028C74_MACRO_TILE_ASPECT(x) (((x) >> 19) & 0x3)
#define CB_COLOR0_ATTRIB 0x28c74
# define CB_TILE_SPLIT(x) (((x) & 0x7) << 5)
# define ADDR_SURF_TILE_SPLIT_64B 0
# define ADDR_SURF_BANK_HEIGHT_2 1
# define ADDR_SURF_BANK_HEIGHT_4 2
# define ADDR_SURF_BANK_HEIGHT_8 3
+# define CB_MACRO_TILE_ASPECT(x) (((x) & 0x3) << 19)
#define CB_COLOR0_DIM 0x28c78
/* only CB0-7 blocks have these regs */
#define CB_COLOR0_CMASK 0x28c7c
#define SQ_TEX_RESOURCE_WORD6_0 0x30018
# define TEX_TILE_SPLIT(x) (((x) & 0x7) << 29)
#define SQ_TEX_RESOURCE_WORD7_0 0x3001c
+# define MACRO_TILE_ASPECT(x) (((x) & 0x3) << 6)
# define TEX_BANK_WIDTH(x) (((x) & 0x3) << 8)
# define TEX_BANK_HEIGHT(x) (((x) & 0x3) << 10)
# define TEX_NUM_BANKS(x) (((x) & 0x3) << 16)
+#define R_030000_SQ_TEX_RESOURCE_WORD0_0 0x030000
+#define S_030000_DIM(x) (((x) & 0x7) << 0)
+#define G_030000_DIM(x) (((x) >> 0) & 0x7)
+#define C_030000_DIM 0xFFFFFFF8
+#define V_030000_SQ_TEX_DIM_1D 0x00000000
+#define V_030000_SQ_TEX_DIM_2D 0x00000001
+#define V_030000_SQ_TEX_DIM_3D 0x00000002
+#define V_030000_SQ_TEX_DIM_CUBEMAP 0x00000003
+#define V_030000_SQ_TEX_DIM_1D_ARRAY 0x00000004
+#define V_030000_SQ_TEX_DIM_2D_ARRAY 0x00000005
+#define V_030000_SQ_TEX_DIM_2D_MSAA 0x00000006
+#define V_030000_SQ_TEX_DIM_2D_ARRAY_MSAA 0x00000007
+#define S_030000_NON_DISP_TILING_ORDER(x) (((x) & 0x1) << 5)
+#define G_030000_NON_DISP_TILING_ORDER(x) (((x) >> 5) & 0x1)
+#define C_030000_NON_DISP_TILING_ORDER 0xFFFFFFDF
+#define S_030000_PITCH(x) (((x) & 0xFFF) << 6)
+#define G_030000_PITCH(x) (((x) >> 6) & 0xFFF)
+#define C_030000_PITCH 0xFFFC003F
+#define S_030000_TEX_WIDTH(x) (((x) & 0x3FFF) << 18)
+#define G_030000_TEX_WIDTH(x) (((x) >> 18) & 0x3FFF)
+#define C_030000_TEX_WIDTH 0x0003FFFF
+#define R_030004_SQ_TEX_RESOURCE_WORD1_0 0x030004
+#define S_030004_TEX_HEIGHT(x) (((x) & 0x3FFF) << 0)
+#define G_030004_TEX_HEIGHT(x) (((x) >> 0) & 0x3FFF)
+#define C_030004_TEX_HEIGHT 0xFFFFC000
+#define S_030004_TEX_DEPTH(x) (((x) & 0x1FFF) << 14)
+#define G_030004_TEX_DEPTH(x) (((x) >> 14) & 0x1FFF)
+#define C_030004_TEX_DEPTH 0xF8003FFF
+#define S_030004_ARRAY_MODE(x) (((x) & 0xF) << 28)
+#define G_030004_ARRAY_MODE(x) (((x) >> 28) & 0xF)
+#define C_030004_ARRAY_MODE 0x0FFFFFFF
+#define R_030008_SQ_TEX_RESOURCE_WORD2_0 0x030008
+#define S_030008_BASE_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0)
+#define G_030008_BASE_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF)
+#define C_030008_BASE_ADDRESS 0x00000000
+#define R_03000C_SQ_TEX_RESOURCE_WORD3_0 0x03000C
+#define S_03000C_MIP_ADDRESS(x) (((x) & 0xFFFFFFFF) << 0)
+#define G_03000C_MIP_ADDRESS(x) (((x) >> 0) & 0xFFFFFFFF)
+#define C_03000C_MIP_ADDRESS 0x00000000
+#define R_030010_SQ_TEX_RESOURCE_WORD4_0 0x030010
+#define S_030010_FORMAT_COMP_X(x) (((x) & 0x3) << 0)
+#define G_030010_FORMAT_COMP_X(x) (((x) >> 0) & 0x3)
+#define C_030010_FORMAT_COMP_X 0xFFFFFFFC
+#define V_030010_SQ_FORMAT_COMP_UNSIGNED 0x00000000
+#define V_030010_SQ_FORMAT_COMP_SIGNED 0x00000001
+#define V_030010_SQ_FORMAT_COMP_UNSIGNED_BIASED 0x00000002
+#define S_030010_FORMAT_COMP_Y(x) (((x) & 0x3) << 2)
+#define G_030010_FORMAT_COMP_Y(x) (((x) >> 2) & 0x3)
+#define C_030010_FORMAT_COMP_Y 0xFFFFFFF3
+#define S_030010_FORMAT_COMP_Z(x) (((x) & 0x3) << 4)
+#define G_030010_FORMAT_COMP_Z(x) (((x) >> 4) & 0x3)
+#define C_030010_FORMAT_COMP_Z 0xFFFFFFCF
+#define S_030010_FORMAT_COMP_W(x) (((x) & 0x3) << 6)
+#define G_030010_FORMAT_COMP_W(x) (((x) >> 6) & 0x3)
+#define C_030010_FORMAT_COMP_W 0xFFFFFF3F
+#define S_030010_NUM_FORMAT_ALL(x) (((x) & 0x3) << 8)
+#define G_030010_NUM_FORMAT_ALL(x) (((x) >> 8) & 0x3)
+#define C_030010_NUM_FORMAT_ALL 0xFFFFFCFF
+#define V_030010_SQ_NUM_FORMAT_NORM 0x00000000
+#define V_030010_SQ_NUM_FORMAT_INT 0x00000001
+#define V_030010_SQ_NUM_FORMAT_SCALED 0x00000002
+#define S_030010_SRF_MODE_ALL(x) (((x) & 0x1) << 10)
+#define G_030010_SRF_MODE_ALL(x) (((x) >> 10) & 0x1)
+#define C_030010_SRF_MODE_ALL 0xFFFFFBFF
+#define V_030010_SRF_MODE_ZERO_CLAMP_MINUS_ONE 0x00000000
+#define V_030010_SRF_MODE_NO_ZERO 0x00000001
+#define S_030010_FORCE_DEGAMMA(x) (((x) & 0x1) << 11)
+#define G_030010_FORCE_DEGAMMA(x) (((x) >> 11) & 0x1)
+#define C_030010_FORCE_DEGAMMA 0xFFFFF7FF
+#define S_030010_ENDIAN_SWAP(x) (((x) & 0x3) << 12)
+#define G_030010_ENDIAN_SWAP(x) (((x) >> 12) & 0x3)
+#define C_030010_ENDIAN_SWAP 0xFFFFCFFF
+#define S_030010_DST_SEL_X(x) (((x) & 0x7) << 16)
+#define G_030010_DST_SEL_X(x) (((x) >> 16) & 0x7)
+#define C_030010_DST_SEL_X 0xFFF8FFFF
+#define V_030010_SQ_SEL_X 0x00000000
+#define V_030010_SQ_SEL_Y 0x00000001
+#define V_030010_SQ_SEL_Z 0x00000002
+#define V_030010_SQ_SEL_W 0x00000003
+#define V_030010_SQ_SEL_0 0x00000004
+#define V_030010_SQ_SEL_1 0x00000005
+#define S_030010_DST_SEL_Y(x) (((x) & 0x7) << 19)
+#define G_030010_DST_SEL_Y(x) (((x) >> 19) & 0x7)
+#define C_030010_DST_SEL_Y 0xFFC7FFFF
+#define S_030010_DST_SEL_Z(x) (((x) & 0x7) << 22)
+#define G_030010_DST_SEL_Z(x) (((x) >> 22) & 0x7)
+#define C_030010_DST_SEL_Z 0xFE3FFFFF
+#define S_030010_DST_SEL_W(x) (((x) & 0x7) << 25)
+#define G_030010_DST_SEL_W(x) (((x) >> 25) & 0x7)
+#define C_030010_DST_SEL_W 0xF1FFFFFF
+#define S_030010_BASE_LEVEL(x) (((x) & 0xF) << 28)
+#define G_030010_BASE_LEVEL(x) (((x) >> 28) & 0xF)
+#define C_030010_BASE_LEVEL 0x0FFFFFFF
+#define R_030014_SQ_TEX_RESOURCE_WORD5_0 0x030014
+#define S_030014_LAST_LEVEL(x) (((x) & 0xF) << 0)
+#define G_030014_LAST_LEVEL(x) (((x) >> 0) & 0xF)
+#define C_030014_LAST_LEVEL 0xFFFFFFF0
+#define S_030014_BASE_ARRAY(x) (((x) & 0x1FFF) << 4)
+#define G_030014_BASE_ARRAY(x) (((x) >> 4) & 0x1FFF)
+#define C_030014_BASE_ARRAY 0xFFFE000F
+#define S_030014_LAST_ARRAY(x) (((x) & 0x1FFF) << 17)
+#define G_030014_LAST_ARRAY(x) (((x) >> 17) & 0x1FFF)
+#define C_030014_LAST_ARRAY 0xC001FFFF
+#define R_030018_SQ_TEX_RESOURCE_WORD6_0 0x030018
+#define S_030018_MAX_ANISO(x) (((x) & 0x7) << 0)
+#define G_030018_MAX_ANISO(x) (((x) >> 0) & 0x7)
+#define C_030018_MAX_ANISO 0xFFFFFFF8
+#define S_030018_PERF_MODULATION(x) (((x) & 0x7) << 3)
+#define G_030018_PERF_MODULATION(x) (((x) >> 3) & 0x7)
+#define C_030018_PERF_MODULATION 0xFFFFFFC7
+#define S_030018_INTERLACED(x) (((x) & 0x1) << 6)
+#define G_030018_INTERLACED(x) (((x) >> 6) & 0x1)
+#define C_030018_INTERLACED 0xFFFFFFBF
+#define S_030018_TILE_SPLIT(x) (((x) & 0x7) << 29)
+#define G_030018_TILE_SPLIT(x) (((x) >> 29) & 0x7)
+#define R_03001C_SQ_TEX_RESOURCE_WORD7_0 0x03001C
+#define S_03001C_MACRO_TILE_ASPECT(x) (((x) & 0x3) << 6)
+#define G_03001C_MACRO_TILE_ASPECT(x) (((x) >> 6) & 0x3)
+#define S_03001C_BANK_WIDTH(x) (((x) & 0x3) << 8)
+#define G_03001C_BANK_WIDTH(x) (((x) >> 8) & 0x3)
+#define S_03001C_BANK_HEIGHT(x) (((x) & 0x3) << 10)
+#define G_03001C_BANK_HEIGHT(x) (((x) >> 10) & 0x3)
+#define S_03001C_NUM_BANKS(x) (((x) & 0x3) << 16)
+#define G_03001C_NUM_BANKS(x) (((x) >> 16) & 0x3)
+#define S_03001C_TYPE(x) (((x) & 0x3) << 30)
+#define G_03001C_TYPE(x) (((x) >> 30) & 0x3)
+#define C_03001C_TYPE 0x3FFFFFFF
+#define V_03001C_SQ_TEX_VTX_INVALID_TEXTURE 0x00000000
+#define V_03001C_SQ_TEX_VTX_INVALID_BUFFER 0x00000001
+#define V_03001C_SQ_TEX_VTX_VALID_TEXTURE 0x00000002
+#define V_03001C_SQ_TEX_VTX_VALID_BUFFER 0x00000003
+#define S_03001C_DATA_FORMAT(x) (((x) & 0x3F) << 0)
+#define G_03001C_DATA_FORMAT(x) (((x) >> 0) & 0x3F)
+#define C_03001C_DATA_FORMAT 0xFFFFFFC0
#define SQ_VTX_CONSTANT_WORD0_0 0x30000
#define SQ_VTX_CONSTANT_WORD1_0 0x30004
struct radeon_bo *cb_color_frag_bo[8];
struct radeon_bo *cb_color_tile_bo[8];
u32 cb_color_info[8];
+ u32 cb_color_view[8];
u32 cb_color_size_idx[8];
u32 cb_target_mask;
u32 cb_shader_mask;
#define FMT_8_BIT(fmt, vc) [fmt] = { 1, 1, 1, vc, CHIP_R600 }
#define FMT_16_BIT(fmt, vc) [fmt] = { 1, 1, 2, vc, CHIP_R600 }
-#define FMT_24_BIT(fmt) [fmt] = { 1, 1, 3, 0, CHIP_R600 }
+#define FMT_24_BIT(fmt) [fmt] = { 1, 1, 4, 0, CHIP_R600 }
#define FMT_32_BIT(fmt, vc) [fmt] = { 1, 1, 4, vc, CHIP_R600 }
-#define FMT_48_BIT(fmt) [fmt] = { 1, 1, 6, 0, CHIP_R600 }
+#define FMT_48_BIT(fmt) [fmt] = { 1, 1, 8, 0, CHIP_R600 }
#define FMT_64_BIT(fmt, vc) [fmt] = { 1, 1, 8, vc, CHIP_R600 }
#define FMT_96_BIT(fmt) [fmt] = { 1, 1, 12, 0, CHIP_R600 }
#define FMT_128_BIT(fmt, vc) [fmt] = { 1, 1, 16,vc, CHIP_R600 }
/* 24-bit */
FMT_24_BIT(V_038004_FMT_8_8_8),
-
+
/* 32-bit */
FMT_32_BIT(V_038004_COLOR_32, 1),
FMT_32_BIT(V_038004_COLOR_32_FLOAT, 1),
[V_038004_FMT_32_AS_32_32_32_32] = { 1, 1, 4, 0, CHIP_CEDAR},
};
-static bool fmt_is_valid_color(u32 format)
+bool r600_fmt_is_valid_color(u32 format)
{
if (format >= ARRAY_SIZE(color_formats_table))
return false;
-
+
if (color_formats_table[format].valid_color)
return true;
return false;
}
-static bool fmt_is_valid_texture(u32 format, enum radeon_family family)
+bool r600_fmt_is_valid_texture(u32 format, enum radeon_family family)
{
if (format >= ARRAY_SIZE(color_formats_table))
return false;
-
+
if (family < color_formats_table[format].min_family)
return false;
return false;
}
-static int fmt_get_blocksize(u32 format)
+int r600_fmt_get_blocksize(u32 format)
{
if (format >= ARRAY_SIZE(color_formats_table))
return 0;
return color_formats_table[format].blocksize;
}
-static int fmt_get_nblocksx(u32 format, u32 w)
+int r600_fmt_get_nblocksx(u32 format, u32 w)
{
unsigned bw;
return (w + bw - 1) / bw;
}
-static int fmt_get_nblocksy(u32 format, u32 h)
+int r600_fmt_get_nblocksy(u32 format, u32 h)
{
unsigned bh;
break;
case ARRAY_LINEAR_ALIGNED:
*pitch_align = max((u32)64, (u32)(values->group_size / values->blocksize));
- *height_align = tile_height;
+ *height_align = 1;
*depth_align = 1;
*base_align = values->group_size;
break;
*base_align = values->group_size;
break;
case ARRAY_2D_TILED_THIN1:
- *pitch_align = max((u32)macro_tile_width,
- (u32)(((values->group_size / tile_height) /
- (values->blocksize * values->nsamples)) *
- values->nbanks)) * tile_width;
+ *pitch_align = max((u32)macro_tile_width * tile_width,
+ (u32)((values->group_size * values->nbanks) /
+ (values->blocksize * values->nsamples * tile_width)));
*height_align = macro_tile_height * tile_height;
*depth_align = 1;
*base_align = max(macro_tile_bytes,
track->cb_color_size[i] = 0;
track->cb_color_size_idx[i] = 0;
track->cb_color_info[i] = 0;
+ track->cb_color_view[i] = 0xFFFFFFFF;
track->cb_color_bo[i] = NULL;
track->cb_color_bo_offset[i] = 0xFFFFFFFF;
track->cb_color_bo_mc[i] = 0xFFFFFFFF;
volatile u32 *ib = p->ib->ptr;
unsigned array_mode;
u32 format;
+
if (G_0280A0_TILE_MODE(track->cb_color_info[i])) {
dev_warn(p->dev, "FMASK or CMASK buffer are not supported by this kernel\n");
return -EINVAL;
}
size = radeon_bo_size(track->cb_color_bo[i]) - track->cb_color_bo_offset[i];
format = G_0280A0_FORMAT(track->cb_color_info[i]);
- if (!fmt_is_valid_color(format)) {
+ if (!r600_fmt_is_valid_color(format)) {
dev_warn(p->dev, "%s:%d cb invalid format %d for %d (0x%08X)\n",
__func__, __LINE__, format,
i, track->cb_color_info[i]);
array_check.nbanks = track->nbanks;
array_check.npipes = track->npipes;
array_check.nsamples = track->nsamples;
- array_check.blocksize = fmt_get_blocksize(format);
+ array_check.blocksize = r600_fmt_get_blocksize(format);
if (r600_get_array_mode_alignment(&array_check,
&pitch_align, &height_align, &depth_align, &base_align)) {
dev_warn(p->dev, "%s invalid tiling %d for %d (0x%08X)\n", __func__,
}
/* check offset */
- tmp = fmt_get_nblocksy(format, height) * fmt_get_nblocksx(format, pitch) * fmt_get_blocksize(format);
+ tmp = r600_fmt_get_nblocksy(format, height) * r600_fmt_get_nblocksx(format, pitch) * r600_fmt_get_blocksize(format);
+ switch (array_mode) {
+ default:
+ case V_0280A0_ARRAY_LINEAR_GENERAL:
+ case V_0280A0_ARRAY_LINEAR_ALIGNED:
+ tmp += track->cb_color_view[i] & 0xFF;
+ break;
+ case V_0280A0_ARRAY_1D_TILED_THIN1:
+ case V_0280A0_ARRAY_2D_TILED_THIN1:
+ tmp += G_028080_SLICE_MAX(track->cb_color_view[i]) * tmp;
+ break;
+ }
if ((tmp + track->cb_color_bo_offset[i]) > radeon_bo_size(track->cb_color_bo[i])) {
if (array_mode == V_0280A0_ARRAY_LINEAR_GENERAL) {
/* the initial DDX does bad things with the CB size occasionally */
* broken userspace.
*/
} else {
- dev_warn(p->dev, "%s offset[%d] %d %d %d %lu too big\n", __func__, i,
- array_mode,
+ dev_warn(p->dev, "%s offset[%d] %d %d %d %lu too big (%d %d) (%d %d %d)\n",
+ __func__, i, array_mode,
track->cb_color_bo_offset[i], tmp,
- radeon_bo_size(track->cb_color_bo[i]));
+ radeon_bo_size(track->cb_color_bo[i]),
+ pitch, height, r600_fmt_get_nblocksx(format, pitch),
+ r600_fmt_get_nblocksy(format, height),
+ r600_fmt_get_blocksize(format));
return -EINVAL;
}
}
track->cb_color_info[tmp] = radeon_get_ib_value(p, idx);
}
break;
+ case R_028080_CB_COLOR0_VIEW:
+ case R_028084_CB_COLOR1_VIEW:
+ case R_028088_CB_COLOR2_VIEW:
+ case R_02808C_CB_COLOR3_VIEW:
+ case R_028090_CB_COLOR4_VIEW:
+ case R_028094_CB_COLOR5_VIEW:
+ case R_028098_CB_COLOR6_VIEW:
+ case R_02809C_CB_COLOR7_VIEW:
+ tmp = (reg - R_028080_CB_COLOR0_VIEW) / 4;
+ track->cb_color_view[tmp] = radeon_get_ib_value(p, idx);
+ break;
case R_028060_CB_COLOR0_SIZE:
case R_028064_CB_COLOR1_SIZE:
case R_028068_CB_COLOR2_SIZE:
return 0;
}
-static unsigned mip_minify(unsigned size, unsigned level)
+unsigned r600_mip_minify(unsigned size, unsigned level)
{
unsigned val;
unsigned nlevels = llevel - blevel + 1;
*l0_size = -1;
- blocksize = fmt_get_blocksize(format);
+ blocksize = r600_fmt_get_blocksize(format);
- w0 = mip_minify(w0, 0);
- h0 = mip_minify(h0, 0);
- d0 = mip_minify(d0, 0);
+ w0 = r600_mip_minify(w0, 0);
+ h0 = r600_mip_minify(h0, 0);
+ d0 = r600_mip_minify(d0, 0);
for(i = 0, offset = 0, level = blevel; i < nlevels; i++, level++) {
- width = mip_minify(w0, i);
- nbx = fmt_get_nblocksx(format, width);
+ width = r600_mip_minify(w0, i);
+ nbx = r600_fmt_get_nblocksx(format, width);
nbx = round_up(nbx, block_align);
- height = mip_minify(h0, i);
- nby = fmt_get_nblocksy(format, height);
+ height = r600_mip_minify(h0, i);
+ nby = r600_fmt_get_nblocksy(format, height);
nby = round_up(nby, height_align);
- depth = mip_minify(d0, i);
+ depth = r600_mip_minify(d0, i);
size = nbx * nby * blocksize;
if (nfaces)
return -EINVAL;
}
format = G_038004_DATA_FORMAT(word1);
- if (!fmt_is_valid_texture(format, p->family)) {
+ if (!r600_fmt_is_valid_texture(format, p->family)) {
dev_warn(p->dev, "%s:%d texture invalid format %d\n",
__func__, __LINE__, format);
return -EINVAL;
array_check.nbanks = track->nbanks;
array_check.npipes = track->npipes;
array_check.nsamples = 1;
- array_check.blocksize = fmt_get_blocksize(format);
+ array_check.blocksize = r600_fmt_get_blocksize(format);
if (r600_get_array_mode_alignment(&array_check,
&pitch_align, &height_align, &depth_align, &base_align)) {
dev_warn(p->dev, "%s:%d tex array mode (%d) invalid\n",
word1 = radeon_get_ib_value(p, idx + 5);
blevel = G_038010_BASE_LEVEL(word0);
llevel = G_038014_LAST_LEVEL(word1);
+ if (blevel > llevel) {
+ dev_warn(p->dev, "texture blevel %d > llevel %d\n",
+ blevel, llevel);
+ }
if (array == 1) {
barray = G_038014_BASE_ARRAY(word1);
larray = G_038014_LAST_ARRAY(word1);
&l0_size, &mipmap_size);
/* using get ib will give us the offset into the texture bo */
if ((l0_size + word2) > radeon_bo_size(texture)) {
- dev_warn(p->dev, "texture bo too small (%d %d %d %d -> %d have %ld)\n",
- w0, h0, format, word2, l0_size, radeon_bo_size(texture));
+ dev_warn(p->dev, "texture bo too small ((%d %d) (%d %d) %d %d %d -> %d have %ld)\n",
+ w0, h0, pitch_align, height_align,
+ array_check.array_mode, format, word2,
+ l0_size, radeon_bo_size(texture));
dev_warn(p->dev, "alignments %d %d %d %lld\n", pitch, pitch_align, height_align, base_align);
return -EINVAL;
}
#define CB_COLOR0_SIZE 0x28060
#define CB_COLOR0_VIEW 0x28080
+#define R_028080_CB_COLOR0_VIEW 0x028080
+#define S_028080_SLICE_START(x) (((x) & 0x7FF) << 0)
+#define G_028080_SLICE_START(x) (((x) >> 0) & 0x7FF)
+#define C_028080_SLICE_START 0xFFFFF800
+#define S_028080_SLICE_MAX(x) (((x) & 0x7FF) << 13)
+#define G_028080_SLICE_MAX(x) (((x) >> 13) & 0x7FF)
+#define C_028080_SLICE_MAX 0xFF001FFF
+#define R_028084_CB_COLOR1_VIEW 0x028084
+#define R_028088_CB_COLOR2_VIEW 0x028088
+#define R_02808C_CB_COLOR3_VIEW 0x02808C
+#define R_028090_CB_COLOR4_VIEW 0x028090
+#define R_028094_CB_COLOR5_VIEW 0x028094
+#define R_028098_CB_COLOR6_VIEW 0x028098
+#define R_02809C_CB_COLOR7_VIEW 0x02809C
#define CB_COLOR0_INFO 0x280a0
# define CB_FORMAT(x) ((x) << 2)
# define CB_ARRAY_MODE(x) ((x) << 8)
extern int rv770_get_temp(struct radeon_device *rdev);
extern int evergreen_get_temp(struct radeon_device *rdev);
extern int sumo_get_temp(struct radeon_device *rdev);
+extern void evergreen_tiling_fields(unsigned tiling_flags, unsigned *bankw,
+ unsigned *bankh, unsigned *mtaspect,
+ unsigned *tile_split);
/*
* Fences.
int r600_vram_scratch_init(struct radeon_device *rdev);
void r600_vram_scratch_fini(struct radeon_device *rdev);
+/*
+ * r600 cs checking helper
+ */
+unsigned r600_mip_minify(unsigned size, unsigned level);
+bool r600_fmt_is_valid_color(u32 format);
+bool r600_fmt_is_valid_texture(u32 format, enum radeon_family family);
+int r600_fmt_get_blocksize(u32 format);
+int r600_fmt_get_nblocksx(u32 format, u32 w);
+int r600_fmt_get_nblocksy(u32 format, u32 h);
+
/*
* r600 functions used by radeon_encoder.c
*/
* 2.11.0 - backend map, initial compute support for the CS checker
* 2.12.0 - RADEON_CS_KEEP_TILING_FLAGS
* 2.13.0 - virtual memory support, streamout
+ * 2.14.0 - add evergreen tiling informations
*/
#define KMS_DRIVER_MAJOR 2
-#define KMS_DRIVER_MINOR 13
+#define KMS_DRIVER_MINOR 14
#define KMS_DRIVER_PATCHLEVEL 0
int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags);
int radeon_driver_unload_kms(struct drm_device *dev);
int radeon_bo_set_tiling_flags(struct radeon_bo *bo,
uint32_t tiling_flags, uint32_t pitch)
{
+ struct radeon_device *rdev = bo->rdev;
int r;
+ if (rdev->family >= CHIP_CEDAR) {
+ unsigned bankw, bankh, mtaspect, tilesplit, stilesplit;
+
+ bankw = (tiling_flags >> RADEON_TILING_EG_BANKW_SHIFT) & RADEON_TILING_EG_BANKW_MASK;
+ bankh = (tiling_flags >> RADEON_TILING_EG_BANKH_SHIFT) & RADEON_TILING_EG_BANKH_MASK;
+ mtaspect = (tiling_flags >> RADEON_TILING_EG_MACRO_TILE_ASPECT_SHIFT) & RADEON_TILING_EG_MACRO_TILE_ASPECT_MASK;
+ tilesplit = (tiling_flags >> RADEON_TILING_EG_TILE_SPLIT_SHIFT) & RADEON_TILING_EG_TILE_SPLIT_MASK;
+ stilesplit = (tiling_flags >> RADEON_TILING_EG_STENCIL_TILE_SPLIT_SHIFT) & RADEON_TILING_EG_STENCIL_TILE_SPLIT_MASK;
+ switch (bankw) {
+ case 0:
+ case 1:
+ case 2:
+ case 4:
+ case 8:
+ break;
+ default:
+ return -EINVAL;
+ }
+ switch (bankh) {
+ case 0:
+ case 1:
+ case 2:
+ case 4:
+ case 8:
+ break;
+ default:
+ return -EINVAL;
+ }
+ switch (mtaspect) {
+ case 0:
+ case 1:
+ case 2:
+ case 4:
+ case 8:
+ break;
+ default:
+ return -EINVAL;
+ }
+ if (tilesplit > 6) {
+ return -EINVAL;
+ }
+ if (stilesplit > 6) {
+ return -EINVAL;
+ }
+ }
r = radeon_bo_reserve(bo, false);
if (unlikely(r != 0))
return r;
0x0002802C DB_DEPTH_CLEAR
0x00028030 PA_SC_SCREEN_SCISSOR_TL
0x00028034 PA_SC_SCREEN_SCISSOR_BR
-0x0002805C DB_DEPTH_SLICE
0x00028140 SQ_ALU_CONST_BUFFER_SIZE_PS_0
0x00028144 SQ_ALU_CONST_BUFFER_SIZE_PS_1
0x00028148 SQ_ALU_CONST_BUFFER_SIZE_PS_2
0x0002802C DB_DEPTH_CLEAR
0x00028030 PA_SC_SCREEN_SCISSOR_TL
0x00028034 PA_SC_SCREEN_SCISSOR_BR
-0x0002805C DB_DEPTH_SLICE
0x00028140 SQ_ALU_CONST_BUFFER_SIZE_PS_0
0x00028144 SQ_ALU_CONST_BUFFER_SIZE_PS_1
0x00028148 SQ_ALU_CONST_BUFFER_SIZE_PS_2
0x00028114 CB_COLOR5_MASK
0x00028118 CB_COLOR6_MASK
0x0002811C CB_COLOR7_MASK
-0x00028080 CB_COLOR0_VIEW
-0x00028084 CB_COLOR1_VIEW
-0x00028088 CB_COLOR2_VIEW
-0x0002808C CB_COLOR3_VIEW
-0x00028090 CB_COLOR4_VIEW
-0x00028094 CB_COLOR5_VIEW
-0x00028098 CB_COLOR6_VIEW
-0x0002809C CB_COLOR7_VIEW
0x00028808 CB_COLOR_CONTROL
0x0002842C CB_FOG_BLUE
0x00028428 CB_FOG_GREEN
uint32_t flags;
};
-#define RADEON_TILING_MACRO 0x1
-#define RADEON_TILING_MICRO 0x2
-#define RADEON_TILING_SWAP_16BIT 0x4
-#define RADEON_TILING_SWAP_32BIT 0x8
-#define RADEON_TILING_SURFACE 0x10 /* this object requires a surface
- * when mapped - i.e. front buffer */
-#define RADEON_TILING_MICRO_SQUARE 0x20
+#define RADEON_TILING_MACRO 0x1
+#define RADEON_TILING_MICRO 0x2
+#define RADEON_TILING_SWAP_16BIT 0x4
+#define RADEON_TILING_SWAP_32BIT 0x8
+/* this object requires a surface when mapped - i.e. front buffer */
+#define RADEON_TILING_SURFACE 0x10
+#define RADEON_TILING_MICRO_SQUARE 0x20
+#define RADEON_TILING_EG_BANKW_SHIFT 8
+#define RADEON_TILING_EG_BANKW_MASK 0xf
+#define RADEON_TILING_EG_BANKH_SHIFT 12
+#define RADEON_TILING_EG_BANKH_MASK 0xf
+#define RADEON_TILING_EG_MACRO_TILE_ASPECT_SHIFT 16
+#define RADEON_TILING_EG_MACRO_TILE_ASPECT_MASK 0xf
+#define RADEON_TILING_EG_TILE_SPLIT_SHIFT 24
+#define RADEON_TILING_EG_TILE_SPLIT_MASK 0xf
+#define RADEON_TILING_EG_STENCIL_TILE_SPLIT_SHIFT 28
+#define RADEON_TILING_EG_STENCIL_TILE_SPLIT_MASK 0xf
struct drm_radeon_gem_set_tiling {
uint32_t handle;
projects / GitHub / exynos8895 / android_kernel_samsung_universal8895.git / commitdiff