drm/radeon: use status regs to determine what to reset (cayman)

When we attempt the reset the GPU, look at the status registers
to determine what blocks need to be reset.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

3 files changed, 136 insertions, 32 deletions

diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c
index 8b73743b0b3f..fd5e0025b656 100644
--- a/drivers/gpu/drm/radeon/evergreen.c
+++ b/drivers/gpu/drm/radeon/evergreen.c
@@ -2349,9 +2349,13 @@ void evergreen_print_gpu_status_regs(struct radeon_device *rdev)
 		RREG32(CP_STAT));
 	dev_info(rdev->dev, "  R_00D034_DMA_STATUS_REG   = 0x%08X\n",
 		RREG32(DMA_STATUS_REG));
+	if (rdev->family >= CHIP_CAYMAN) {
+		dev_info(rdev->dev, "  R_00D834_DMA_STATUS_REG   = 0x%08X\n",
+			 RREG32(DMA_STATUS_REG + 0x800));
+	}
 }
 
-static bool evergreen_is_display_hung(struct radeon_device *rdev)
+bool evergreen_is_display_hung(struct radeon_device *rdev)
 {
 	u32 crtc_hung = 0;
 	u32 crtc_status[6];
diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c
index 88f6fbbe4be6..eee013f5699a 100644
--- a/drivers/gpu/drm/radeon/ni.c
+++ b/drivers/gpu/drm/radeon/ni.c
@@ -34,6 +34,7 @@
 #include "ni_reg.h"
 #include "cayman_blit_shaders.h"
 
+extern bool evergreen_is_display_hung(struct radeon_device *rdev);
 extern void evergreen_print_gpu_status_regs(struct radeon_device *rdev);
 extern void evergreen_mc_stop(struct radeon_device *rdev, struct evergreen_mc_save *save);
 extern void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *save);
@@ -1311,21 +1312,86 @@ void cayman_dma_fini(struct radeon_device *rdev)
 	radeon_ring_fini(rdev, &rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX]);
 }
 
-static int cayman_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask)
+static u32 cayman_gpu_check_soft_reset(struct radeon_device *rdev)
 {
-	struct evergreen_mc_save save;
-	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 reset_mask = 0;
 	u32 tmp;
-	int ret = 0;
 
-	if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE))
-		reset_mask &= ~(RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP);
+	/* GRBM_STATUS */
+	tmp = RREG32(GRBM_STATUS);
+	if (tmp & (PA_BUSY | SC_BUSY |
+		   SH_BUSY | SX_BUSY |
+		   TA_BUSY | VGT_BUSY |
+		   DB_BUSY | CB_BUSY |
+		   GDS_BUSY | SPI_BUSY |
+		   IA_BUSY | IA_BUSY_NO_DMA))
+		reset_mask |= RADEON_RESET_GFX;
+
+	if (tmp & (CF_RQ_PENDING | PF_RQ_PENDING |
+		   CP_BUSY | CP_COHERENCY_BUSY))
+		reset_mask |= RADEON_RESET_CP;
+
+	if (tmp & GRBM_EE_BUSY)
+		reset_mask |= RADEON_RESET_GRBM | RADEON_RESET_GFX | RADEON_RESET_CP;
+
+	/* DMA_STATUS_REG 0 */
+	tmp = RREG32(DMA_STATUS_REG + DMA0_REGISTER_OFFSET);
+	if (!(tmp & DMA_IDLE))
+		reset_mask |= RADEON_RESET_DMA;
+
+	/* DMA_STATUS_REG 1 */
+	tmp = RREG32(DMA_STATUS_REG + DMA1_REGISTER_OFFSET);
+	if (!(tmp & DMA_IDLE))
+		reset_mask |= RADEON_RESET_DMA1;
+
+	/* SRBM_STATUS2 */
+	tmp = RREG32(SRBM_STATUS2);
+	if (tmp & DMA_BUSY)
+		reset_mask |= RADEON_RESET_DMA;
+
+	if (tmp & DMA1_BUSY)
+		reset_mask |= RADEON_RESET_DMA1;
+
+	/* SRBM_STATUS */
+	tmp = RREG32(SRBM_STATUS);
+	if (tmp & (RLC_RQ_PENDING | RLC_BUSY))
+		reset_mask |= RADEON_RESET_RLC;
+
+	if (tmp & IH_BUSY)
+		reset_mask |= RADEON_RESET_IH;
+
+	if (tmp & SEM_BUSY)
+		reset_mask |= RADEON_RESET_SEM;
+
+	if (tmp & GRBM_RQ_PENDING)
+		reset_mask |= RADEON_RESET_GRBM;
+
+	if (tmp & VMC_BUSY)
+		reset_mask |= RADEON_RESET_VMC;
 
-	if (RREG32(DMA_STATUS_REG) & DMA_IDLE)
-		reset_mask &= ~RADEON_RESET_DMA;
+	if (tmp & (MCB_BUSY | MCB_NON_DISPLAY_BUSY |
+		   MCC_BUSY | MCD_BUSY))
+		reset_mask |= RADEON_RESET_MC;
+
+	if (evergreen_is_display_hung(rdev))
+		reset_mask |= RADEON_RESET_DISPLAY;
+
+	/* VM_L2_STATUS */
+	tmp = RREG32(VM_L2_STATUS);
+	if (tmp & L2_BUSY)
+		reset_mask |= RADEON_RESET_VMC;
+
+	return reset_mask;
+}
+
+static void cayman_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask)
+{
+	struct evergreen_mc_save save;
+	u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+	u32 tmp;
 
 	if (reset_mask == 0)
-		return 0;
+		return;
 
 	dev_info(rdev->dev, "GPU softreset: 0x%08X\n", reset_mask);
 
@@ -1354,7 +1420,9 @@ static int cayman_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask)
 		tmp = RREG32(DMA_RB_CNTL + DMA0_REGISTER_OFFSET);
 		tmp &= ~DMA_RB_ENABLE;
 		WREG32(DMA_RB_CNTL + DMA0_REGISTER_OFFSET, tmp);
+	}
 
+	if (reset_mask & RADEON_RESET_DMA1) {
 		/* dma1 */
 		tmp = RREG32(DMA_RB_CNTL + DMA1_REGISTER_OFFSET);
 		tmp &= ~DMA_RB_ENABLE;
@@ -1383,7 +1451,31 @@ static int cayman_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask)
 	}
 
 	if (reset_mask & RADEON_RESET_DMA)
-		srbm_soft_reset |= SOFT_RESET_DMA | SOFT_RESET_DMA1;
+		srbm_soft_reset |= SOFT_RESET_DMA;
+
+	if (reset_mask & RADEON_RESET_DMA1)
+		srbm_soft_reset |= SOFT_RESET_DMA1;
+
+	if (reset_mask & RADEON_RESET_DISPLAY)
+		srbm_soft_reset |= SOFT_RESET_DC;
+
+	if (reset_mask & RADEON_RESET_RLC)
+		srbm_soft_reset |= SOFT_RESET_RLC;
+
+	if (reset_mask & RADEON_RESET_SEM)
+		srbm_soft_reset |= SOFT_RESET_SEM;
+
+	if (reset_mask & RADEON_RESET_IH)
+		srbm_soft_reset |= SOFT_RESET_IH;
+
+	if (reset_mask & RADEON_RESET_GRBM)
+		srbm_soft_reset |= SOFT_RESET_GRBM;
+
+	if (reset_mask & RADEON_RESET_VMC)
+		srbm_soft_reset |= SOFT_RESET_VMC;
+
+	if (reset_mask & RADEON_RESET_MC)
+		srbm_soft_reset |= SOFT_RESET_MC;
 
 	if (grbm_soft_reset) {
 		tmp = RREG32(GRBM_SOFT_RESET);
@@ -1419,32 +1511,26 @@ static int cayman_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask)
 	evergreen_mc_resume(rdev, &save);
 	udelay(50);
 
-#if 0
-	if (reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP)) {
-		if (RREG32(GRBM_STATUS) & GUI_ACTIVE)
-			ret = -EAGAIN;
-	}
-
-	if (reset_mask & RADEON_RESET_DMA) {
-		if (!(RREG32(DMA_STATUS_REG) & DMA_IDLE))
-			ret = -EAGAIN;
-	}
-#endif
-
-	if (!ret)
-		r600_set_bios_scratch_engine_hung(rdev, false);
-
 	evergreen_print_gpu_status_regs(rdev);
-
-	return 0;
 }
 
 int cayman_asic_reset(struct radeon_device *rdev)
 {
-	return cayman_gpu_soft_reset(rdev, (RADEON_RESET_GFX |
-					    RADEON_RESET_COMPUTE |
-					    RADEON_RESET_DMA |
-					    RADEON_RESET_CP));
+	u32 reset_mask;
+
+	reset_mask = cayman_gpu_check_soft_reset(rdev);
+
+	if (reset_mask)
+		r600_set_bios_scratch_engine_hung(rdev, true);
+
+	cayman_gpu_soft_reset(rdev, reset_mask);
+
+	reset_mask = cayman_gpu_check_soft_reset(rdev);
+
+	if (!reset_mask)
+		r600_set_bios_scratch_engine_hung(rdev, false);
+
+	return 0;
 }
 
 /**
diff --git a/drivers/gpu/drm/radeon/nid.h b/drivers/gpu/drm/radeon/nid.h
index c4b84c5a9ab7..079dee202a9e 100644
--- a/drivers/gpu/drm/radeon/nid.h
+++ b/drivers/gpu/drm/radeon/nid.h
@@ -49,6 +49,16 @@
 #define		RINGID(x)					(((x) & 0x3) << 0)
 #define		VMID(x)						(((x) & 0x7) << 0)
 #define	SRBM_STATUS				        0x0E50
+#define		RLC_RQ_PENDING 				(1 << 3)
+#define		GRBM_RQ_PENDING 			(1 << 5)
+#define		VMC_BUSY 				(1 << 8)
+#define		MCB_BUSY 				(1 << 9)
+#define		MCB_NON_DISPLAY_BUSY 			(1 << 10)
+#define		MCC_BUSY 				(1 << 11)
+#define		MCD_BUSY 				(1 << 12)
+#define		SEM_BUSY 				(1 << 14)
+#define		RLC_BUSY 				(1 << 15)
+#define		IH_BUSY 				(1 << 17)
 
 #define	SRBM_SOFT_RESET				        0x0E60
 #define		SOFT_RESET_BIF				(1 << 1)
@@ -68,6 +78,10 @@
 #define		SOFT_RESET_REGBB			(1 << 22)
 #define		SOFT_RESET_ORB				(1 << 23)
 
+#define	SRBM_STATUS2				        0x0EC4
+#define		DMA_BUSY 				(1 << 5)
+#define		DMA1_BUSY 				(1 << 6)
+
 #define VM_CONTEXT0_REQUEST_RESPONSE			0x1470
 #define		REQUEST_TYPE(x)					(((x) & 0xf) << 0)
 #define		RESPONSE_TYPE_MASK				0x000000F0