summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/radeon/r200.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/radeon/r200.c')
-rw-r--r--drivers/gpu/drm/radeon/r200.c145
1 files changed, 99 insertions, 46 deletions
diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c
index 568c74bfba3d..1146c9909c2c 100644
--- a/drivers/gpu/drm/radeon/r200.c
+++ b/drivers/gpu/drm/radeon/r200.c
@@ -31,6 +31,7 @@
#include "radeon_reg.h"
#include "radeon.h"
+#include "r100d.h"
#include "r200_reg_safe.h"
#include "r100_track.h"
@@ -79,6 +80,51 @@ static int r200_get_vtx_size_0(uint32_t vtx_fmt_0)
return vtx_size;
}
+int r200_copy_dma(struct radeon_device *rdev,
+ uint64_t src_offset,
+ uint64_t dst_offset,
+ unsigned num_pages,
+ struct radeon_fence *fence)
+{
+ uint32_t size;
+ uint32_t cur_size;
+ int i, num_loops;
+ int r = 0;
+
+ /* radeon pitch is /64 */
+ size = num_pages << PAGE_SHIFT;
+ num_loops = DIV_ROUND_UP(size, 0x1FFFFF);
+ r = radeon_ring_lock(rdev, num_loops * 4 + 64);
+ if (r) {
+ DRM_ERROR("radeon: moving bo (%d).\n", r);
+ return r;
+ }
+ /* Must wait for 2D idle & clean before DMA or hangs might happen */
+ radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
+ radeon_ring_write(rdev, (1 << 16));
+ for (i = 0; i < num_loops; i++) {
+ cur_size = size;
+ if (cur_size > 0x1FFFFF) {
+ cur_size = 0x1FFFFF;
+ }
+ size -= cur_size;
+ radeon_ring_write(rdev, PACKET0(0x720, 2));
+ radeon_ring_write(rdev, src_offset);
+ radeon_ring_write(rdev, dst_offset);
+ radeon_ring_write(rdev, cur_size | (1 << 31) | (1 << 30));
+ src_offset += cur_size;
+ dst_offset += cur_size;
+ }
+ radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0));
+ radeon_ring_write(rdev, RADEON_WAIT_DMA_GUI_IDLE);
+ if (fence) {
+ r = radeon_fence_emit(rdev, fence);
+ }
+ radeon_ring_unlock_commit(rdev);
+ return r;
+}
+
+
static int r200_get_vtx_size_1(uint32_t vtx_fmt_1)
{
int vtx_size, i, tex_size;
@@ -96,7 +142,6 @@ int r200_packet0_check(struct radeon_cs_parser *p,
struct radeon_cs_packet *pkt,
unsigned idx, unsigned reg)
{
- struct radeon_cs_chunk *ib_chunk;
struct radeon_cs_reloc *reloc;
struct r100_cs_track *track;
volatile uint32_t *ib;
@@ -105,11 +150,11 @@ int r200_packet0_check(struct radeon_cs_parser *p,
int i;
int face;
u32 tile_flags = 0;
+ u32 idx_value;
ib = p->ib->ptr;
- ib_chunk = &p->chunks[p->chunk_ib_idx];
track = (struct r100_cs_track *)p->track;
-
+ idx_value = radeon_get_ib_value(p, idx);
switch (reg) {
case RADEON_CRTC_GUI_TRIG_VLINE:
r = r100_cs_packet_parse_vline(p);
@@ -137,8 +182,8 @@ int r200_packet0_check(struct radeon_cs_parser *p,
return r;
}
track->zb.robj = reloc->robj;
- track->zb.offset = ib_chunk->kdata[idx];
- ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset);
+ track->zb.offset = idx_value;
+ ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
break;
case RADEON_RB3D_COLOROFFSET:
r = r100_cs_packet_next_reloc(p, &reloc);
@@ -149,8 +194,8 @@ int r200_packet0_check(struct radeon_cs_parser *p,
return r;
}
track->cb[0].robj = reloc->robj;
- track->cb[0].offset = ib_chunk->kdata[idx];
- ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset);
+ track->cb[0].offset = idx_value;
+ ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
break;
case R200_PP_TXOFFSET_0:
case R200_PP_TXOFFSET_1:
@@ -166,7 +211,7 @@ int r200_packet0_check(struct radeon_cs_parser *p,
r100_cs_dump_packet(p, pkt);
return r;
}
- ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset);
+ ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
track->textures[i].robj = reloc->robj;
break;
case R200_PP_CUBIC_OFFSET_F1_0:
@@ -208,12 +253,12 @@ int r200_packet0_check(struct radeon_cs_parser *p,
r100_cs_dump_packet(p, pkt);
return r;
}
- track->textures[i].cube_info[face - 1].offset = ib_chunk->kdata[idx];
- ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset);
+ track->textures[i].cube_info[face - 1].offset = idx_value;
+ ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
track->textures[i].cube_info[face - 1].robj = reloc->robj;
break;
case RADEON_RE_WIDTH_HEIGHT:
- track->maxy = ((ib_chunk->kdata[idx] >> 16) & 0x7FF);
+ track->maxy = ((idx_value >> 16) & 0x7FF);
break;
case RADEON_RB3D_COLORPITCH:
r = r100_cs_packet_next_reloc(p, &reloc);
@@ -229,17 +274,17 @@ int r200_packet0_check(struct radeon_cs_parser *p,
if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO)
tile_flags |= RADEON_COLOR_MICROTILE_ENABLE;
- tmp = ib_chunk->kdata[idx] & ~(0x7 << 16);
+ tmp = idx_value & ~(0x7 << 16);
tmp |= tile_flags;
ib[idx] = tmp;
- track->cb[0].pitch = ib_chunk->kdata[idx] & RADEON_COLORPITCH_MASK;
+ track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK;
break;
case RADEON_RB3D_DEPTHPITCH:
- track->zb.pitch = ib_chunk->kdata[idx] & RADEON_DEPTHPITCH_MASK;
+ track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK;
break;
case RADEON_RB3D_CNTL:
- switch ((ib_chunk->kdata[idx] >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) {
+ switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) {
case 7:
case 8:
case 9:
@@ -257,18 +302,18 @@ int r200_packet0_check(struct radeon_cs_parser *p,
break;
default:
DRM_ERROR("Invalid color buffer format (%d) !\n",
- ((ib_chunk->kdata[idx] >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f));
+ ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f));
return -EINVAL;
}
- if (ib_chunk->kdata[idx] & RADEON_DEPTHXY_OFFSET_ENABLE) {
+ if (idx_value & RADEON_DEPTHXY_OFFSET_ENABLE) {
DRM_ERROR("No support for depth xy offset in kms\n");
return -EINVAL;
}
- track->z_enabled = !!(ib_chunk->kdata[idx] & RADEON_Z_ENABLE);
+ track->z_enabled = !!(idx_value & RADEON_Z_ENABLE);
break;
case RADEON_RB3D_ZSTENCILCNTL:
- switch (ib_chunk->kdata[idx] & 0xf) {
+ switch (idx_value & 0xf) {
case 0:
track->zb.cpp = 2;
break;
@@ -292,27 +337,27 @@ int r200_packet0_check(struct radeon_cs_parser *p,
r100_cs_dump_packet(p, pkt);
return r;
}
- ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset);
+ ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset);
break;
case RADEON_PP_CNTL:
{
- uint32_t temp = ib_chunk->kdata[idx] >> 4;
+ uint32_t temp = idx_value >> 4;
for (i = 0; i < track->num_texture; i++)
track->textures[i].enabled = !!(temp & (1 << i));
}
break;
case RADEON_SE_VF_CNTL:
- track->vap_vf_cntl = ib_chunk->kdata[idx];
+ track->vap_vf_cntl = idx_value;
break;
case 0x210c:
/* VAP_VF_MAX_VTX_INDX */
- track->max_indx = ib_chunk->kdata[idx] & 0x00FFFFFFUL;
+ track->max_indx = idx_value & 0x00FFFFFFUL;
break;
case R200_SE_VTX_FMT_0:
- track->vtx_size = r200_get_vtx_size_0(ib_chunk->kdata[idx]);
+ track->vtx_size = r200_get_vtx_size_0(idx_value);
break;
case R200_SE_VTX_FMT_1:
- track->vtx_size += r200_get_vtx_size_1(ib_chunk->kdata[idx]);
+ track->vtx_size += r200_get_vtx_size_1(idx_value);
break;
case R200_PP_TXSIZE_0:
case R200_PP_TXSIZE_1:
@@ -321,8 +366,8 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_PP_TXSIZE_4:
case R200_PP_TXSIZE_5:
i = (reg - R200_PP_TXSIZE_0) / 32;
- track->textures[i].width = (ib_chunk->kdata[idx] & RADEON_TEX_USIZE_MASK) + 1;
- track->textures[i].height = ((ib_chunk->kdata[idx] & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1;
+ track->textures[i].width = (idx_value & RADEON_TEX_USIZE_MASK) + 1;
+ track->textures[i].height = ((idx_value & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1;
break;
case R200_PP_TXPITCH_0:
case R200_PP_TXPITCH_1:
@@ -331,7 +376,7 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_PP_TXPITCH_4:
case R200_PP_TXPITCH_5:
i = (reg - R200_PP_TXPITCH_0) / 32;
- track->textures[i].pitch = ib_chunk->kdata[idx] + 32;
+ track->textures[i].pitch = idx_value + 32;
break;
case R200_PP_TXFILTER_0:
case R200_PP_TXFILTER_1:
@@ -340,12 +385,12 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_PP_TXFILTER_4:
case R200_PP_TXFILTER_5:
i = (reg - R200_PP_TXFILTER_0) / 32;
- track->textures[i].num_levels = ((ib_chunk->kdata[idx] & R200_MAX_MIP_LEVEL_MASK)
+ track->textures[i].num_levels = ((idx_value & R200_MAX_MIP_LEVEL_MASK)
>> R200_MAX_MIP_LEVEL_SHIFT);
- tmp = (ib_chunk->kdata[idx] >> 23) & 0x7;
+ tmp = (idx_value >> 23) & 0x7;
if (tmp == 2 || tmp == 6)
track->textures[i].roundup_w = false;
- tmp = (ib_chunk->kdata[idx] >> 27) & 0x7;
+ tmp = (idx_value >> 27) & 0x7;
if (tmp == 2 || tmp == 6)
track->textures[i].roundup_h = false;
break;
@@ -364,21 +409,24 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_PP_TXFORMAT_X_4:
case R200_PP_TXFORMAT_X_5:
i = (reg - R200_PP_TXFORMAT_X_0) / 32;
- track->textures[i].txdepth = ib_chunk->kdata[idx] & 0x7;
- tmp = (ib_chunk->kdata[idx] >> 16) & 0x3;
+ track->textures[i].txdepth = idx_value & 0x7;
+ tmp = (idx_value >> 16) & 0x3;
/* 2D, 3D, CUBE */
switch (tmp) {
case 0:
case 5:
case 6:
case 7:
+ /* 1D/2D */
track->textures[i].tex_coord_type = 0;
break;
case 1:
- track->textures[i].tex_coord_type = 1;
+ /* CUBE */
+ track->textures[i].tex_coord_type = 2;
break;
case 2:
- track->textures[i].tex_coord_type = 2;
+ /* 3D */
+ track->textures[i].tex_coord_type = 1;
break;
}
break;
@@ -389,20 +437,19 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_PP_TXFORMAT_4:
case R200_PP_TXFORMAT_5:
i = (reg - R200_PP_TXFORMAT_0) / 32;
- if (ib_chunk->kdata[idx] & R200_TXFORMAT_NON_POWER2) {
+ if (idx_value & R200_TXFORMAT_NON_POWER2) {
track->textures[i].use_pitch = 1;
} else {
track->textures[i].use_pitch = 0;
- track->textures[i].width = 1 << ((ib_chunk->kdata[idx] >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK);
- track->textures[i].height = 1 << ((ib_chunk->kdata[idx] >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK);
+ track->textures[i].width = 1 << ((idx_value >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK);
+ track->textures[i].height = 1 << ((idx_value >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK);
}
- switch ((ib_chunk->kdata[idx] & RADEON_TXFORMAT_FORMAT_MASK)) {
+ switch ((idx_value & RADEON_TXFORMAT_FORMAT_MASK)) {
case R200_TXFORMAT_I8:
case R200_TXFORMAT_RGB332:
case R200_TXFORMAT_Y8:
track->textures[i].cpp = 1;
break;
- case R200_TXFORMAT_DXT1:
case R200_TXFORMAT_AI88:
case R200_TXFORMAT_ARGB1555:
case R200_TXFORMAT_RGB565:
@@ -419,13 +466,20 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_TXFORMAT_ABGR8888:
case R200_TXFORMAT_BGR111110:
case R200_TXFORMAT_LDVDU8888:
+ track->textures[i].cpp = 4;
+ break;
+ case R200_TXFORMAT_DXT1:
+ track->textures[i].cpp = 1;
+ track->textures[i].compress_format = R100_TRACK_COMP_DXT1;
+ break;
case R200_TXFORMAT_DXT23:
case R200_TXFORMAT_DXT45:
- track->textures[i].cpp = 4;
+ track->textures[i].cpp = 1;
+ track->textures[i].compress_format = R100_TRACK_COMP_DXT1;
break;
}
- track->textures[i].cube_info[4].width = 1 << ((ib_chunk->kdata[idx] >> 16) & 0xf);
- track->textures[i].cube_info[4].height = 1 << ((ib_chunk->kdata[idx] >> 20) & 0xf);
+ track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf);
+ track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf);
break;
case R200_PP_CUBIC_FACES_0:
case R200_PP_CUBIC_FACES_1:
@@ -433,7 +487,7 @@ int r200_packet0_check(struct radeon_cs_parser *p,
case R200_PP_CUBIC_FACES_3:
case R200_PP_CUBIC_FACES_4:
case R200_PP_CUBIC_FACES_5:
- tmp = ib_chunk->kdata[idx];
+ tmp = idx_value;
i = (reg - R200_PP_CUBIC_FACES_0) / 32;
for (face = 0; face < 4; face++) {
track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf);
@@ -448,9 +502,8 @@ int r200_packet0_check(struct radeon_cs_parser *p,
return 0;
}
-int r200_init(struct radeon_device *rdev)
+void r200_set_safe_registers(struct radeon_device *rdev)
{
rdev->config.r100.reg_safe_bm = r200_reg_safe_bm;
rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(r200_reg_safe_bm);
- return 0;
}
generated by cgit v1.2.3-55-g7522 (git 2.39.0) at 2025-04-27 12:22:36 +0200