aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/radeon
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/radeon')
-rw-r--r--drivers/gpu/drm/radeon/atom.c1
-rw-r--r--drivers/gpu/drm/radeon/evergreen.c39
-rw-r--r--drivers/gpu/drm/radeon/evergreen_blit_kms.c2
-rw-r--r--drivers/gpu/drm/radeon/r100.c4
-rw-r--r--drivers/gpu/drm/radeon/r300.c2
-rw-r--r--drivers/gpu/drm/radeon/r600.c31
-rw-r--r--drivers/gpu/drm/radeon/r600_blit_kms.c2
-rw-r--r--drivers/gpu/drm/radeon/r600_cs.c309
-rw-r--r--drivers/gpu/drm/radeon/r600_reg.h1
-rw-r--r--drivers/gpu/drm/radeon/r600d.h6
-rw-r--r--drivers/gpu/drm/radeon/radeon.h4
-rw-r--r--drivers/gpu/drm/radeon/radeon_atombios.c43
-rw-r--r--drivers/gpu/drm/radeon/radeon_benchmark.c4
-rw-r--r--drivers/gpu/drm/radeon/radeon_bios.c13
-rw-r--r--drivers/gpu/drm/radeon/radeon_combios.c15
-rw-r--r--drivers/gpu/drm/radeon/radeon_connectors.c68
-rw-r--r--drivers/gpu/drm/radeon/radeon_device.c6
-rw-r--r--drivers/gpu/drm/radeon/radeon_display.c18
-rw-r--r--drivers/gpu/drm/radeon/radeon_encoders.c382
-rw-r--r--drivers/gpu/drm/radeon/radeon_fence.c3
-rw-r--r--drivers/gpu/drm/radeon/radeon_gart.c4
-rw-r--r--drivers/gpu/drm/radeon/radeon_gem.c2
-rw-r--r--drivers/gpu/drm/radeon/radeon_i2c.c49
-rw-r--r--drivers/gpu/drm/radeon/radeon_irq.c4
-rw-r--r--drivers/gpu/drm/radeon/radeon_legacy_encoders.c2
-rw-r--r--drivers/gpu/drm/radeon/radeon_mode.h22
-rw-r--r--drivers/gpu/drm/radeon/radeon_object.c22
-rw-r--r--drivers/gpu/drm/radeon/radeon_object.h7
-rw-r--r--drivers/gpu/drm/radeon/radeon_ring.c6
-rw-r--r--drivers/gpu/drm/radeon/radeon_test.c4
-rw-r--r--drivers/gpu/drm/radeon/radeon_ttm.c5
-rw-r--r--drivers/gpu/drm/radeon/rs400.c2
-rw-r--r--drivers/gpu/drm/radeon/rs600.c4
-rw-r--r--drivers/gpu/drm/radeon/rv770.c4
34 files changed, 824 insertions, 266 deletions
diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c
index 8e421f644a5..05efb5b9f13 100644
--- a/drivers/gpu/drm/radeon/atom.c
+++ b/drivers/gpu/drm/radeon/atom.c
@@ -112,6 +112,7 @@ static uint32_t atom_iio_execute(struct atom_context *ctx, int base,
base += 3;
break;
case ATOM_IIO_WRITE:
+ (void)ctx->card->ioreg_read(ctx->card, CU16(base + 1));
ctx->card->ioreg_write(ctx->card, CU16(base + 1), temp);
base += 3;
break;
diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c
index f12a5b3ec05..4dc5b4714c5 100644
--- a/drivers/gpu/drm/radeon/evergreen.c
+++ b/drivers/gpu/drm/radeon/evergreen.c
@@ -1650,7 +1650,36 @@ static void evergreen_gpu_init(struct radeon_device *rdev)
}
}
- rdev->config.evergreen.tile_config = gb_addr_config;
+ /* setup tiling info dword. gb_addr_config is not adequate since it does
+ * not have bank info, so create a custom tiling dword.
+ * bits 3:0 num_pipes
+ * bits 7:4 num_banks
+ * bits 11:8 group_size
+ * bits 15:12 row_size
+ */
+ rdev->config.evergreen.tile_config = 0;
+ switch (rdev->config.evergreen.max_tile_pipes) {
+ case 1:
+ default:
+ rdev->config.evergreen.tile_config |= (0 << 0);
+ break;
+ case 2:
+ rdev->config.evergreen.tile_config |= (1 << 0);
+ break;
+ case 4:
+ rdev->config.evergreen.tile_config |= (2 << 0);
+ break;
+ case 8:
+ rdev->config.evergreen.tile_config |= (3 << 0);
+ break;
+ }
+ rdev->config.evergreen.tile_config |=
+ ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4;
+ rdev->config.evergreen.tile_config |=
+ ((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT) << 8;
+ rdev->config.evergreen.tile_config |=
+ ((gb_addr_config & 0x30000000) >> 28) << 12;
+
WREG32(GB_BACKEND_MAP, gb_backend_map);
WREG32(GB_ADDR_CONFIG, gb_addr_config);
WREG32(DMIF_ADDR_CONFIG, gb_addr_config);
@@ -2033,7 +2062,7 @@ int evergreen_irq_set(struct radeon_device *rdev)
u32 grbm_int_cntl = 0;
if (!rdev->irq.installed) {
- WARN(1, "Can't enable IRQ/MSI because no handler is installed.\n");
+ WARN(1, "Can't enable IRQ/MSI because no handler is installed\n");
return -EINVAL;
}
/* don't enable anything if the ih is disabled */
@@ -2295,6 +2324,7 @@ restart_ih:
case 0: /* D1 vblank */
if (disp_int & LB_D1_VBLANK_INTERRUPT) {
drm_handle_vblank(rdev->ddev, 0);
+ rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
disp_int &= ~LB_D1_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D1 vblank\n");
@@ -2316,6 +2346,7 @@ restart_ih:
case 0: /* D2 vblank */
if (disp_int_cont & LB_D2_VBLANK_INTERRUPT) {
drm_handle_vblank(rdev->ddev, 1);
+ rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
disp_int_cont &= ~LB_D2_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D2 vblank\n");
@@ -2337,6 +2368,7 @@ restart_ih:
case 0: /* D3 vblank */
if (disp_int_cont2 & LB_D3_VBLANK_INTERRUPT) {
drm_handle_vblank(rdev->ddev, 2);
+ rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
disp_int_cont2 &= ~LB_D3_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D3 vblank\n");
@@ -2358,6 +2390,7 @@ restart_ih:
case 0: /* D4 vblank */
if (disp_int_cont3 & LB_D4_VBLANK_INTERRUPT) {
drm_handle_vblank(rdev->ddev, 3);
+ rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
disp_int_cont3 &= ~LB_D4_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D4 vblank\n");
@@ -2379,6 +2412,7 @@ restart_ih:
case 0: /* D5 vblank */
if (disp_int_cont4 & LB_D5_VBLANK_INTERRUPT) {
drm_handle_vblank(rdev->ddev, 4);
+ rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
disp_int_cont4 &= ~LB_D5_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D5 vblank\n");
@@ -2400,6 +2434,7 @@ restart_ih:
case 0: /* D6 vblank */
if (disp_int_cont5 & LB_D6_VBLANK_INTERRUPT) {
drm_handle_vblank(rdev->ddev, 5);
+ rdev->pm.vblank_sync = true;
wake_up(&rdev->irq.vblank_queue);
disp_int_cont5 &= ~LB_D6_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D6 vblank\n");
diff --git a/drivers/gpu/drm/radeon/evergreen_blit_kms.c b/drivers/gpu/drm/radeon/evergreen_blit_kms.c
index ac3b6dde23d..e0e590110dd 100644
--- a/drivers/gpu/drm/radeon/evergreen_blit_kms.c
+++ b/drivers/gpu/drm/radeon/evergreen_blit_kms.c
@@ -459,7 +459,7 @@ int evergreen_blit_init(struct radeon_device *rdev)
obj_size += evergreen_ps_size * 4;
obj_size = ALIGN(obj_size, 256);
- r = radeon_bo_create(rdev, NULL, obj_size, true, RADEON_GEM_DOMAIN_VRAM,
+ r = radeon_bo_create(rdev, NULL, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
&rdev->r600_blit.shader_obj);
if (r) {
DRM_ERROR("evergreen failed to allocate shader\n");
diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c
index 0e8f28a6892..8e10aa9f74b 100644
--- a/drivers/gpu/drm/radeon/r100.c
+++ b/drivers/gpu/drm/radeon/r100.c
@@ -442,7 +442,7 @@ int r100_pci_gart_init(struct radeon_device *rdev)
int r;
if (rdev->gart.table.ram.ptr) {
- WARN(1, "R100 PCI GART already initialized.\n");
+ WARN(1, "R100 PCI GART already initialized\n");
return 0;
}
/* Initialize common gart structure */
@@ -516,7 +516,7 @@ int r100_irq_set(struct radeon_device *rdev)
uint32_t tmp = 0;
if (!rdev->irq.installed) {
- WARN(1, "Can't enable IRQ/MSI because no handler is installed.\n");
+ WARN(1, "Can't enable IRQ/MSI because no handler is installed\n");
WREG32(R_000040_GEN_INT_CNTL, 0);
return -EINVAL;
}
diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c
index 34527e600fe..cde1d3480d9 100644
--- a/drivers/gpu/drm/radeon/r300.c
+++ b/drivers/gpu/drm/radeon/r300.c
@@ -91,7 +91,7 @@ int rv370_pcie_gart_init(struct radeon_device *rdev)
int r;
if (rdev->gart.table.vram.robj) {
- WARN(1, "RV370 PCIE GART already initialized.\n");
+ WARN(1, "RV370 PCIE GART already initialized\n");
return 0;
}
/* Initialize common gart structure */
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c
index 33952a12f0a..4d7a2e1bdb9 100644
--- a/drivers/gpu/drm/radeon/r600.c
+++ b/drivers/gpu/drm/radeon/r600.c
@@ -97,14 +97,8 @@ u32 rv6xx_get_temp(struct radeon_device *rdev)
{
u32 temp = (RREG32(CG_THERMAL_STATUS) & ASIC_T_MASK) >>
ASIC_T_SHIFT;
- u32 actual_temp = 0;
- if ((temp >> 7) & 1)
- actual_temp = 0;
- else
- actual_temp = (temp >> 1) & 0xff;
-
- return actual_temp * 1000;
+ return temp * 1000;
}
void r600_pm_get_dynpm_state(struct radeon_device *rdev)
@@ -884,12 +878,15 @@ void r600_pcie_gart_tlb_flush(struct radeon_device *rdev)
u32 tmp;
/* flush hdp cache so updates hit vram */
- if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740)) {
+ if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) &&
+ !(rdev->flags & RADEON_IS_AGP)) {
void __iomem *ptr = (void *)rdev->gart.table.vram.ptr;
u32 tmp;
/* r7xx hw bug. write to HDP_DEBUG1 followed by fb read
* rather than write to HDP_REG_COHERENCY_FLUSH_CNTL
+ * This seems to cause problems on some AGP cards. Just use the old
+ * method for them.
*/
WREG32(HDP_DEBUG1, 0);
tmp = readl((void __iomem *)ptr);
@@ -919,7 +916,7 @@ int r600_pcie_gart_init(struct radeon_device *rdev)
int r;
if (rdev->gart.table.vram.robj) {
- WARN(1, "R600 PCIE GART already initialized.\n");
+ WARN(1, "R600 PCIE GART already initialized\n");
return 0;
}
/* Initialize common gart structure */
@@ -1201,8 +1198,10 @@ void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc)
mc->vram_end, mc->real_vram_size >> 20);
} else {
u64 base = 0;
- if (rdev->flags & RADEON_IS_IGP)
- base = (RREG32(MC_VM_FB_LOCATION) & 0xFFFF) << 24;
+ if (rdev->flags & RADEON_IS_IGP) {
+ base = RREG32(MC_VM_FB_LOCATION) & 0xFFFF;
+ base <<= 24;
+ }
radeon_vram_location(rdev, &rdev->mc, base);
rdev->mc.gtt_base_align = 0;
radeon_gtt_location(rdev, mc);
@@ -2724,7 +2723,7 @@ static int r600_ih_ring_alloc(struct radeon_device *rdev)
/* Allocate ring buffer */
if (rdev->ih.ring_obj == NULL) {
r = radeon_bo_create(rdev, NULL, rdev->ih.ring_size,
- true,
+ PAGE_SIZE, true,
RADEON_GEM_DOMAIN_GTT,
&rdev->ih.ring_obj);
if (r) {
@@ -2995,7 +2994,7 @@ int r600_irq_set(struct radeon_device *rdev)
u32 hdmi1, hdmi2;
if (!rdev->irq.installed) {
- WARN(1, "Can't enable IRQ/MSI because no handler is installed.\n");
+ WARN(1, "Can't enable IRQ/MSI because no handler is installed\n");
return -EINVAL;
}
/* don't enable anything if the ih is disabled */
@@ -3489,10 +3488,12 @@ int r600_debugfs_mc_info_init(struct radeon_device *rdev)
void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo)
{
/* r7xx hw bug. write to HDP_DEBUG1 followed by fb read
- * rather than write to HDP_REG_COHERENCY_FLUSH_CNTL
+ * rather than write to HDP_REG_COHERENCY_FLUSH_CNTL.
+ * This seems to cause problems on some AGP cards. Just use the old
+ * method for them.
*/
if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_RV740) &&
- rdev->vram_scratch.ptr) {
+ rdev->vram_scratch.ptr && !(rdev->flags & RADEON_IS_AGP)) {
void __iomem *ptr = (void *)rdev->vram_scratch.ptr;
u32 tmp;
diff --git a/drivers/gpu/drm/radeon/r600_blit_kms.c b/drivers/gpu/drm/radeon/r600_blit_kms.c
index 8362974ef41..86e5aa07f0d 100644
--- a/drivers/gpu/drm/radeon/r600_blit_kms.c
+++ b/drivers/gpu/drm/radeon/r600_blit_kms.c
@@ -501,7 +501,7 @@ int r600_blit_init(struct radeon_device *rdev)
obj_size += r6xx_ps_size * 4;
obj_size = ALIGN(obj_size, 256);
- r = radeon_bo_create(rdev, NULL, obj_size, true, RADEON_GEM_DOMAIN_VRAM,
+ r = radeon_bo_create(rdev, NULL, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM,
&rdev->r600_blit.shader_obj);
if (r) {
DRM_ERROR("r600 failed to allocate shader\n");
diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c
index 37cc2aa9f92..0f90fc3482c 100644
--- a/drivers/gpu/drm/radeon/r600_cs.c
+++ b/drivers/gpu/drm/radeon/r600_cs.c
@@ -50,6 +50,7 @@ struct r600_cs_track {
u32 nsamples;
u32 cb_color_base_last[8];
struct radeon_bo *cb_color_bo[8];
+ u64 cb_color_bo_mc[8];
u32 cb_color_bo_offset[8];
struct radeon_bo *cb_color_frag_bo[8];
struct radeon_bo *cb_color_tile_bo[8];
@@ -67,6 +68,7 @@ struct r600_cs_track {
u32 db_depth_size;
u32 db_offset;
struct radeon_bo *db_bo;
+ u64 db_bo_mc;
};
static inline int r600_bpe_from_format(u32 *bpe, u32 format)
@@ -140,6 +142,68 @@ static inline int r600_bpe_from_format(u32 *bpe, u32 format)
return 0;
}
+struct array_mode_checker {
+ int array_mode;
+ u32 group_size;
+ u32 nbanks;
+ u32 npipes;
+ u32 nsamples;
+ u32 bpe;
+};
+
+/* returns alignment in pixels for pitch/height/depth and bytes for base */
+static inline int r600_get_array_mode_alignment(struct array_mode_checker *values,
+ u32 *pitch_align,
+ u32 *height_align,
+ u32 *depth_align,
+ u64 *base_align)
+{
+ u32 tile_width = 8;
+ u32 tile_height = 8;
+ u32 macro_tile_width = values->nbanks;
+ u32 macro_tile_height = values->npipes;
+ u32 tile_bytes = tile_width * tile_height * values->bpe * values->nsamples;
+ u32 macro_tile_bytes = macro_tile_width * macro_tile_height * tile_bytes;
+
+ switch (values->array_mode) {
+ case ARRAY_LINEAR_GENERAL:
+ /* technically tile_width/_height for pitch/height */
+ *pitch_align = 1; /* tile_width */
+ *height_align = 1; /* tile_height */
+ *depth_align = 1;
+ *base_align = 1;
+ break;
+ case ARRAY_LINEAR_ALIGNED:
+ *pitch_align = max((u32)64, (u32)(values->group_size / values->bpe));
+ *height_align = tile_height;
+ *depth_align = 1;
+ *base_align = values->group_size;
+ break;
+ case ARRAY_1D_TILED_THIN1:
+ *pitch_align = max((u32)tile_width,
+ (u32)(values->group_size /
+ (tile_height * values->bpe * values->nsamples)));
+ *height_align = tile_height;
+ *depth_align = 1;
+ *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->bpe * values->nsamples)) *
+ values->nbanks)) * tile_width;
+ *height_align = macro_tile_height * tile_height;
+ *depth_align = 1;
+ *base_align = max(macro_tile_bytes,
+ (*pitch_align) * values->bpe * (*height_align) * values->nsamples);
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
static void r600_cs_track_init(struct r600_cs_track *track)
{
int i;
@@ -153,10 +217,12 @@ static void r600_cs_track_init(struct r600_cs_track *track)
track->cb_color_info[i] = 0;
track->cb_color_bo[i] = NULL;
track->cb_color_bo_offset[i] = 0xFFFFFFFF;
+ track->cb_color_bo_mc[i] = 0xFFFFFFFF;
}
track->cb_target_mask = 0xFFFFFFFF;
track->cb_shader_mask = 0xFFFFFFFF;
track->db_bo = NULL;
+ track->db_bo_mc = 0xFFFFFFFF;
/* assume the biggest format and that htile is enabled */
track->db_depth_info = 7 | (1 << 25);
track->db_depth_view = 0xFFFFC000;
@@ -168,7 +234,10 @@ static void r600_cs_track_init(struct r600_cs_track *track)
static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
{
struct r600_cs_track *track = p->track;
- u32 bpe = 0, pitch, slice_tile_max, size, tmp, height, pitch_align;
+ u32 bpe = 0, slice_tile_max, size, tmp;
+ u32 height, height_align, pitch, pitch_align, depth_align;
+ u64 base_offset, base_align;
+ struct array_mode_checker array_check;
volatile u32 *ib = p->ib->ptr;
unsigned array_mode;
@@ -183,60 +252,40 @@ static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
i, track->cb_color_info[i]);
return -EINVAL;
}
- /* pitch is the number of 8x8 tiles per row */
- pitch = G_028060_PITCH_TILE_MAX(track->cb_color_size[i]) + 1;
+ /* pitch in pixels */
+ pitch = (G_028060_PITCH_TILE_MAX(track->cb_color_size[i]) + 1) * 8;
slice_tile_max = G_028060_SLICE_TILE_MAX(track->cb_color_size[i]) + 1;
slice_tile_max *= 64;
- height = slice_tile_max / (pitch * 8);
+ height = slice_tile_max / pitch;
if (height > 8192)
height = 8192;
array_mode = G_0280A0_ARRAY_MODE(track->cb_color_info[i]);
+
+ base_offset = track->cb_color_bo_mc[i] + track->cb_color_bo_offset[i];
+ array_check.array_mode = array_mode;
+ array_check.group_size = track->group_size;
+ array_check.nbanks = track->nbanks;
+ array_check.npipes = track->npipes;
+ array_check.nsamples = track->nsamples;
+ array_check.bpe = bpe;
+ 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__,
+ G_0280A0_ARRAY_MODE(track->cb_color_info[i]), i,
+ track->cb_color_info[i]);
+ return -EINVAL;
+ }
switch (array_mode) {
case V_0280A0_ARRAY_LINEAR_GENERAL:
- /* technically height & 0x7 */
break;
case V_0280A0_ARRAY_LINEAR_ALIGNED:
- pitch_align = max((u32)64, (u32)(track->group_size / bpe)) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
- if (!IS_ALIGNED(height, 8)) {
- dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
- __func__, __LINE__, height);
- return -EINVAL;
- }
break;
case V_0280A0_ARRAY_1D_TILED_THIN1:
- pitch_align = max((u32)8, (u32)(track->group_size / (8 * bpe * track->nsamples))) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
/* avoid breaking userspace */
if (height > 7)
height &= ~0x7;
- if (!IS_ALIGNED(height, 8)) {
- dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
- __func__, __LINE__, height);
- return -EINVAL;
- }
break;
case V_0280A0_ARRAY_2D_TILED_THIN1:
- pitch_align = max((u32)track->nbanks,
- (u32)(((track->group_size / 8) / (bpe * track->nsamples)) * track->nbanks)) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
- if (!IS_ALIGNED((height / 8), track->npipes)) {
- dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
- __func__, __LINE__, height);
- return -EINVAL;
- }
break;
default:
dev_warn(p->dev, "%s invalid tiling %d for %d (0x%08X)\n", __func__,
@@ -244,8 +293,24 @@ static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
track->cb_color_info[i]);
return -EINVAL;
}
+
+ if (!IS_ALIGNED(pitch, pitch_align)) {
+ dev_warn(p->dev, "%s:%d cb pitch (%d) invalid\n",
+ __func__, __LINE__, pitch);
+ return -EINVAL;
+ }
+ if (!IS_ALIGNED(height, height_align)) {
+ dev_warn(p->dev, "%s:%d cb height (%d) invalid\n",
+ __func__, __LINE__, height);
+ return -EINVAL;
+ }
+ if (!IS_ALIGNED(base_offset, base_align)) {
+ dev_warn(p->dev, "%s offset[%d] 0x%llx not aligned\n", __func__, i, base_offset);
+ return -EINVAL;
+ }
+
/* check offset */
- tmp = height * pitch * 8 * bpe;
+ tmp = height * pitch * bpe;
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 */
@@ -260,15 +325,11 @@ static inline int r600_cs_track_validate_cb(struct radeon_cs_parser *p, int i)
return -EINVAL;
}
}
- if (!IS_ALIGNED(track->cb_color_bo_offset[i], track->group_size)) {
- dev_warn(p->dev, "%s offset[%d] %d not aligned\n", __func__, i, track->cb_color_bo_offset[i]);
- return -EINVAL;
- }
/* limit max tile */
- tmp = (height * pitch * 8) >> 6;
+ tmp = (height * pitch) >> 6;
if (tmp < slice_tile_max)
slice_tile_max = tmp;
- tmp = S_028060_PITCH_TILE_MAX(pitch - 1) |
+ tmp = S_028060_PITCH_TILE_MAX((pitch / 8) - 1) |
S_028060_SLICE_TILE_MAX(slice_tile_max - 1);
ib[track->cb_color_size_idx[i]] = tmp;
return 0;
@@ -310,7 +371,12 @@ static int r600_cs_track_check(struct radeon_cs_parser *p)
/* Check depth buffer */
if (G_028800_STENCIL_ENABLE(track->db_depth_control) ||
G_028800_Z_ENABLE(track->db_depth_control)) {
- u32 nviews, bpe, ntiles, pitch, pitch_align, height, size, slice_tile_max;
+ u32 nviews, bpe, ntiles, size, slice_tile_max;
+ u32 height, height_align, pitch, pitch_align, depth_align;
+ u64 base_offset, base_align;
+ struct array_mode_checker array_check;
+ int array_mode;
+
if (track->db_bo == NULL) {
dev_warn(p->dev, "z/stencil with no depth buffer\n");
return -EINVAL;
@@ -353,41 +419,34 @@ static int r600_cs_track_check(struct radeon_cs_parser *p)
ib[track->db_depth_size_idx] = S_028000_SLICE_TILE_MAX(tmp - 1) | (track->db_depth_size & 0x3FF);
} else {
size = radeon_bo_size(track->db_bo);
- pitch = G_028000_PITCH_TILE_MAX(track->db_depth_size) + 1;
+ /* pitch in pixels */
+ pitch = (G_028000_PITCH_TILE_MAX(track->db_depth_size) + 1) * 8;
slice_tile_max = G_028000_SLICE_TILE_MAX(track->db_depth_size) + 1;
slice_tile_max *= 64;
- height = slice_tile_max / (pitch * 8);
+ height = slice_tile_max / pitch;
if (height > 8192)
height = 8192;
- switch (G_028010_ARRAY_MODE(track->db_depth_info)) {
+ base_offset = track->db_bo_mc + track->db_offset;
+ array_mode = G_028010_ARRAY_MODE(track->db_depth_info);
+ array_check.array_mode = array_mode;
+ array_check.group_size = track->group_size;
+ array_check.nbanks = track->nbanks;
+ array_check.npipes = track->npipes;
+ array_check.nsamples = track->nsamples;
+ array_check.bpe = bpe;
+ if (r600_get_array_mode_alignment(&array_check,
+ &pitch_align, &height_align, &depth_align, &base_align)) {
+ dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__,
+ G_028010_ARRAY_MODE(track->db_depth_info),
+ track->db_depth_info);
+ return -EINVAL;
+ }
+ switch (array_mode) {
case V_028010_ARRAY_1D_TILED_THIN1:
- pitch_align = (max((u32)8, (u32)(track->group_size / (8 * bpe))) / 8);
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d db pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
/* don't break userspace */
height &= ~0x7;
- if (!IS_ALIGNED(height, 8)) {
- dev_warn(p->dev, "%s:%d db height (%d) invalid\n",
- __func__, __LINE__, height);
- return -EINVAL;
- }
break;
case V_028010_ARRAY_2D_TILED_THIN1:
- pitch_align = max((u32)track->nbanks,
- (u32)(((track->group_size / 8) / bpe) * track->nbanks)) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d db pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
- if (!IS_ALIGNED((height / 8), track->npipes)) {
- dev_warn(p->dev, "%s:%d db height (%d) invalid\n",
- __func__, __LINE__, height);
- return -EINVAL;
- }
break;
default:
dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__,
@@ -395,15 +454,27 @@ static int r600_cs_track_check(struct radeon_cs_parser *p)
track->db_depth_info);
return -EINVAL;
}
- if (!IS_ALIGNED(track->db_offset, track->group_size)) {
- dev_warn(p->dev, "%s offset[%d] %d not aligned\n", __func__, i, track->db_offset);
+
+ if (!IS_ALIGNED(pitch, pitch_align)) {
+ dev_warn(p->dev, "%s:%d db pitch (%d) invalid\n",
+ __func__, __LINE__, pitch);
+ return -EINVAL;
+ }
+ if (!IS_ALIGNED(height, height_align)) {
+ dev_warn(p->dev, "%s:%d db height (%d) invalid\n",
+ __func__, __LINE__, height);
return -EINVAL;
}
+ if (!IS_ALIGNED(base_offset, base_align)) {
+ dev_warn(p->dev, "%s offset[%d] 0x%llx not aligned\n", __func__, i, base_offset);
+ return -EINVAL;
+ }
+
ntiles = G_028000_SLICE_TILE_MAX(track->db_depth_size) + 1;
nviews = G_028004_SLICE_MAX(track->db_depth_view) + 1;
tmp = ntiles * bpe * 64 * nviews;
if ((tmp + track->db_offset) > radeon_bo_size(track->db_bo)) {
- dev_warn(p->dev, "z/stencil buffer too small (0x%08X %d %d %d -> %d have %ld)\n",
+ dev_warn(p->dev, "z/stencil buffer too small (0x%08X %d %d %d -> %u have %lu)\n",
track->db_depth_size, ntiles, nviews, bpe, tmp + track->db_offset,
radeon_bo_size(track->db_bo));
return -EINVAL;
@@ -954,6 +1025,7 @@ static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx
ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
track->cb_color_base_last[tmp] = ib[idx];
track->cb_color_bo[tmp] = reloc->robj;
+ track->cb_color_bo_mc[tmp] = reloc->lobj.gpu_offset;
break;
case DB_DEPTH_BASE:
r = r600_cs_packet_next_reloc(p, &reloc);
@@ -965,6 +1037,7 @@ static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx
track->db_offset = radeon_get_ib_value(p, idx) << 8;
ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
track->db_bo = reloc->robj;
+ track->db_bo_mc = reloc->lobj.gpu_offset;
break;
case DB_HTILE_DATA_BASE:
case SQ_PGM_START_FS:
@@ -1086,16 +1159,25 @@ static void r600_texture_size(unsigned nfaces, unsigned blevel, unsigned nlevels
static inline int r600_check_texture_resource(struct radeon_cs_parser *p, u32 idx,
struct radeon_bo *texture,
struct radeon_bo *mipmap,
+ u64 base_offset,
+ u64 mip_offset,
u32 tiling_flags)
{
struct r600_cs_track *track = p->track;
u32 nfaces, nlevels, blevel, w0, h0, d0, bpe = 0;
- u32 word0, word1, l0_size, mipmap_size, pitch, pitch_align;
+ u32 word0, word1, l0_size, mipmap_size;
+ u32 height_align, pitch, pitch_align, depth_align;
+ u64 base_align;
+ struct array_mode_checker array_check;
/* on legacy kernel we don't perform advanced check */
if (p->rdev == NULL)
return 0;
+ /* convert to bytes */
+ base_offset <<= 8;
+ mip_offset <<= 8;
+
word0 = radeon_get_ib_value(p, idx + 0);
if (tiling_flags & RADEON_TILING_MACRO)
word0 |= S_038000_TILE_MODE(V_038000_ARRAY_2D_TILED_THIN1);
@@ -1128,46 +1210,38 @@ static inline int r600_check_texture_resource(struct radeon_cs_parser *p, u32 i
return -EINVAL;
}
- pitch = G_038000_PITCH(word0) + 1;
- switch (G_038000_TILE_MODE(word0)) {
- case V_038000_ARRAY_LINEAR_GENERAL:
- pitch_align = 1;
- /* XXX check height align */
- break;
- case V_038000_ARRAY_LINEAR_ALIGNED:
- pitch_align = max((u32)64, (u32)(track->group_size / bpe)) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
- /* XXX check height align */
- break;
- case V_038000_ARRAY_1D_TILED_THIN1:
- pitch_align = max((u32)8, (u32)(track->group_size / (8 * bpe))) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
- /* XXX check height align */
- break;
- case V_038000_ARRAY_2D_TILED_THIN1:
- pitch_align = max((u32)track->nbanks,
- (u32)(((track->group_size / 8) / bpe) * track->nbanks)) / 8;
- if (!IS_ALIGNED(pitch, pitch_align)) {
- dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
- __func__, __LINE__, pitch);
- return -EINVAL;
- }
- /* XXX check height align */
- break;
- default:
- dev_warn(p->dev, "%s invalid tiling %d (0x%08X)\n", __func__,
- G_038000_TILE_MODE(word0), word0);
+ /* pitch in texels */
+ pitch = (G_038000_PITCH(word0) + 1) * 8;
+ array_check.array_mode = G_038000_TILE_MODE(word0);
+ array_check.group_size = track->group_size;
+ array_check.nbanks = track->nbanks;
+ array_check.npipes = track->npipes;
+ array_check.nsamples = 1;
+ array_check.bpe = bpe;
+ 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",
+ __func__, __LINE__, G_038000_TILE_MODE(word0));
+ return -EINVAL;
+ }
+
+ /* XXX check height as well... */
+
+ if (!IS_ALIGNED(pitch, pitch_align)) {
+ dev_warn(p->dev, "%s:%d tex pitch (%d) invalid\n",
+ __func__, __LINE__, pitch);
+ return -EINVAL;
+ }
+ if (!IS_ALIGNED(base_offset, base_align)) {
+ dev_warn(p->dev, "%s:%d tex base offset (0x%llx) invalid\n",
+ __func__, __LINE__, base_offset);
+ return -EINVAL;
+ }
+ if (!IS_ALIGNED(mip_offset, base_align)) {
+ dev_warn(p->dev, "%s:%d tex mip offset (0x%llx) invalid\n",
+ __func__, __LINE__, mip_offset);
return -EINVAL;
}
- /* XXX check offset align */
word0 = radeon_get_ib_value(p, idx + 4);
word1 = radeon_get_ib_value(p, idx + 5);
@@ -1402,7 +1476,10 @@ static int r600_packet3_check(struct radeon_cs_parser *p,
mip_offset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff);
mipmap = reloc->robj;
r = r600_check_texture_resource(p