summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/drm/radeon/si.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/radeon/si.c')
-rw-r--r--drivers/gpu/drm/radeon/si.c67
1 files changed, 48 insertions, 19 deletions
diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c
index ac708e00618..eba0225259a 100644
--- a/drivers/gpu/drm/radeon/si.c
+++ b/drivers/gpu/drm/radeon/si.c
@@ -71,6 +71,7 @@ MODULE_FIRMWARE("radeon/HAINAN_mc2.bin");
MODULE_FIRMWARE("radeon/HAINAN_rlc.bin");
MODULE_FIRMWARE("radeon/HAINAN_smc.bin");
+static u32 si_get_cu_active_bitmap(struct radeon_device *rdev, u32 se, u32 sh);
static void si_pcie_gen3_enable(struct radeon_device *rdev);
static void si_program_aspm(struct radeon_device *rdev);
extern void sumo_rlc_fini(struct radeon_device *rdev);
@@ -2900,7 +2901,7 @@ static void si_gpu_init(struct radeon_device *rdev)
u32 sx_debug_1;
u32 hdp_host_path_cntl;
u32 tmp;
- int i, j;
+ int i, j, k;
switch (rdev->family) {
case CHIP_TAHITI:
@@ -3098,6 +3099,14 @@ static void si_gpu_init(struct radeon_device *rdev)
rdev->config.si.max_sh_per_se,
rdev->config.si.max_cu_per_sh);
+ for (i = 0; i < rdev->config.si.max_shader_engines; i++) {
+ for (j = 0; j < rdev->config.si.max_sh_per_se; j++) {
+ for (k = 0; k < rdev->config.si.max_cu_per_sh; k++) {
+ rdev->config.si.active_cus +=
+ hweight32(si_get_cu_active_bitmap(rdev, i, j));
+ }
+ }
+ }
/* set HW defaults for 3D engine */
WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) |
@@ -3186,7 +3195,7 @@ void si_fence_ring_emit(struct radeon_device *rdev,
/* EVENT_WRITE_EOP - flush caches, send int */
radeon_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4));
radeon_ring_write(ring, EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) | EVENT_INDEX(5));
- radeon_ring_write(ring, addr & 0xffffffff);
+ radeon_ring_write(ring, lower_32_bits(addr));
radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | DATA_SEL(1) | INT_SEL(2));
radeon_ring_write(ring, fence->seq);
radeon_ring_write(ring, 0);
@@ -3219,7 +3228,7 @@ void si_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib)
radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
radeon_ring_write(ring, (1 << 8));
radeon_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
- radeon_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
+ radeon_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr));
radeon_ring_write(ring, next_rptr);
}
@@ -4044,18 +4053,21 @@ static int si_pcie_gart_enable(struct radeon_device *rdev)
WREG32(MC_VM_MX_L1_TLB_CNTL,
(0xA << 7) |
ENABLE_L1_TLB |
+ ENABLE_L1_FRAGMENT_PROCESSING |
SYSTEM_ACCESS_MODE_NOT_IN_SYS |
ENABLE_ADVANCED_DRIVER_MODEL |
SYSTEM_APERTURE_UNMAPPED_ACCESS_PASS_THRU);
/* Setup L2 cache */
WREG32(VM_L2_CNTL, ENABLE_L2_CACHE |
+ ENABLE_L2_FRAGMENT_PROCESSING |
ENABLE_L2_PTE_CACHE_LRU_UPDATE_BY_WRITE |
ENABLE_L2_PDE0_CACHE_LRU_UPDATE_BY_WRITE |
EFFECTIVE_L2_QUEUE_SIZE(7) |
CONTEXT1_IDENTITY_ACCESS_MODE(1));
WREG32(VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS | INVALIDATE_L2_CACHE);
WREG32(VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY |
- L2_CACHE_BIGK_FRAGMENT_SIZE(0));
+ BANK_SELECT(4) |
+ L2_CACHE_BIGK_FRAGMENT_SIZE(4));
/* setup context0 */
WREG32(VM_CONTEXT0_PAGE_TABLE_START_ADDR, rdev->mc.gtt_start >> 12);
WREG32(VM_CONTEXT0_PAGE_TABLE_END_ADDR, rdev->mc.gtt_end >> 12);
@@ -4092,6 +4104,7 @@ static int si_pcie_gart_enable(struct radeon_device *rdev)
(u32)(rdev->dummy_page.addr >> 12));
WREG32(VM_CONTEXT1_CNTL2, 4);
WREG32(VM_CONTEXT1_CNTL, ENABLE_CONTEXT | PAGE_TABLE_DEPTH(1) |
+ PAGE_TABLE_BLOCK_SIZE(radeon_vm_block_size - 9) |
RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT |
RANGE_PROTECTION_FAULT_ENABLE_DEFAULT |
DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT |
@@ -5780,7 +5793,6 @@ int si_irq_set(struct radeon_device *rdev)
u32 crtc1 = 0, crtc2 = 0, crtc3 = 0, crtc4 = 0, crtc5 = 0, crtc6 = 0;
u32 hpd1 = 0, hpd2 = 0, hpd3 = 0, hpd4 = 0, hpd5 = 0, hpd6 = 0;
u32 grbm_int_cntl = 0;
- u32 grph1 = 0, grph2 = 0, grph3 = 0, grph4 = 0, grph5 = 0, grph6 = 0;
u32 dma_cntl, dma_cntl1;
u32 thermal_int = 0;
@@ -5919,16 +5931,22 @@ int si_irq_set(struct radeon_device *rdev)
}
if (rdev->num_crtc >= 2) {
- WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, grph1);
- WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, grph2);
+ WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET,
+ GRPH_PFLIP_INT_MASK);
+ WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET,
+ GRPH_PFLIP_INT_MASK);
}
if (rdev->num_crtc >= 4) {
- WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, grph3);
- WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, grph4);
+ WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET,
+ GRPH_PFLIP_INT_MASK);
+ WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET,
+ GRPH_PFLIP_INT_MASK);
}
if (rdev->num_crtc >= 6) {
- WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, grph5);
- WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, grph6);
+ WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET,
+ GRPH_PFLIP_INT_MASK);
+ WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET,
+ GRPH_PFLIP_INT_MASK);
}
if (!ASIC_IS_NODCE(rdev)) {
@@ -6146,7 +6164,7 @@ restart_ih:
wake_up(&rdev->irq.vblank_queue);
}
if (atomic_read(&rdev->irq.pflip[0]))
- radeon_crtc_handle_flip(rdev, 0);
+ radeon_crtc_handle_vblank(rdev, 0);
rdev->irq.stat_regs.evergreen.disp_int &= ~LB_D1_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D1 vblank\n");
}
@@ -6172,7 +6190,7 @@ restart_ih:
wake_up(&rdev->irq.vblank_queue);
}
if (atomic_read(&rdev->irq.pflip[1]))
- radeon_crtc_handle_flip(rdev, 1);
+ radeon_crtc_handle_vblank(rdev, 1);
rdev->irq.stat_regs.evergreen.disp_int_cont &= ~LB_D2_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D2 vblank\n");
}
@@ -6198,7 +6216,7 @@ restart_ih:
wake_up(&rdev->irq.vblank_queue);
}
if (atomic_read(&rdev->irq.pflip[2]))
- radeon_crtc_handle_flip(rdev, 2);
+ radeon_crtc_handle_vblank(rdev, 2);
rdev->irq.stat_regs.evergreen.disp_int_cont2 &= ~LB_D3_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D3 vblank\n");
}
@@ -6224,7 +6242,7 @@ restart_ih:
wake_up(&rdev->irq.vblank_queue);
}
if (atomic_read(&rdev->irq.pflip[3]))
- radeon_crtc_handle_flip(rdev, 3);
+ radeon_crtc_handle_vblank(rdev, 3);
rdev->irq.stat_regs.evergreen.disp_int_cont3 &= ~LB_D4_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D4 vblank\n");
}
@@ -6250,7 +6268,7 @@ restart_ih:
wake_up(&rdev->irq.vblank_queue);
}
if (atomic_read(&rdev->irq.pflip[4]))
- radeon_crtc_handle_flip(rdev, 4);
+ radeon_crtc_handle_vblank(rdev, 4);
rdev->irq.stat_regs.evergreen.disp_int_cont4 &= ~LB_D5_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D5 vblank\n");
}
@@ -6276,7 +6294,7 @@ restart_ih:
wake_up(&rdev->irq.vblank_queue);
}
if (atomic_read(&rdev->irq.pflip[5]))
- radeon_crtc_handle_flip(rdev, 5);
+ radeon_crtc_handle_vblank(rdev, 5);
rdev->irq.stat_regs.evergreen.disp_int_cont5 &= ~LB_D6_VBLANK_INTERRUPT;
DRM_DEBUG("IH: D6 vblank\n");
}
@@ -6292,6 +6310,15 @@ restart_ih:
break;
}
break;
+ case 8: /* D1 page flip */
+ case 10: /* D2 page flip */
+ case 12: /* D3 page flip */
+ case 14: /* D4 page flip */
+ case 16: /* D5 page flip */
+ case 18: /* D6 page flip */
+ DRM_DEBUG("IH: D%d flip\n", ((src_id - 8) >> 1) + 1);
+ radeon_crtc_handle_flip(rdev, (src_id - 8) >> 1);
+ break;
case 42: /* HPD hotplug */
switch (src_data) {
case 0:
@@ -6349,14 +6376,16 @@ restart_ih:
case 147:
addr = RREG32(VM_CONTEXT1_PROTECTION_FAULT_ADDR);
status = RREG32(VM_CONTEXT1_PROTECTION_FAULT_STATUS);
+ /* reset addr and status */
+ WREG32_P(VM_CONTEXT1_CNTL2, 1, ~1);
+ if (addr == 0x0 && status == 0x0)
+ break;
dev_err(rdev->dev, "GPU fault detected: %d 0x%08x\n", src_id, src_data);
dev_err(rdev->dev, " VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x%08X\n",
addr);
dev_err(rdev->dev, " VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n",
status);
si_vm_decode_fault(rdev, status, addr);
- /* reset addr and status */
- WREG32_P(VM_CONTEXT1_CNTL2, 1, ~1);
break;
case 176: /* RINGID0 CP_INT */
radeon_fence_process(rdev, RADEON_RING_TYPE_GFX_INDEX);
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-04-01 00:55:47 +0000