drm/radeon/kms/evergreen: setup and enable the CP

The command processor (CP) fetches command buffers and
feeds the GPU.  This patch requires the evergreen
family me and pfp ucode files.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

drivers/gpu/drm/radeon/evergreen.c

@@ -32,6 +32,9 @@
 #include "avivod.h"
 #include "evergreen_reg.h"
 
+#define EVERGREEN_PFP_UCODE_SIZE 1120
+#define EVERGREEN_PM4_UCODE_SIZE 1376
+
 static void evergreen_gpu_init(struct radeon_device *rdev);
 void evergreen_fini(struct radeon_device *rdev);
 
@@ -418,23 +421,91 @@ static void evergreen_mc_program(struct radeon_device *rdev)
 	rv515_vga_render_disable(rdev);
 }
 
-#if 0
 /*
  * CP.
  */
-static void evergreen_cp_stop(struct radeon_device *rdev)
-{
-	/* XXX */
-}
-
 
 static int evergreen_cp_load_microcode(struct radeon_device *rdev)
 {
-	/* XXX */
+	const __be32 *fw_data;
+	int i;
 
+	if (!rdev->me_fw || !rdev->pfp_fw)
+		return -EINVAL;
+
+	r700_cp_stop(rdev);
+	WREG32(CP_RB_CNTL, RB_NO_UPDATE | (15 << 8) | (3 << 0));
+
+	fw_data = (const __be32 *)rdev->pfp_fw->data;
+	WREG32(CP_PFP_UCODE_ADDR, 0);
+	for (i = 0; i < EVERGREEN_PFP_UCODE_SIZE; i++)
+		WREG32(CP_PFP_UCODE_DATA, be32_to_cpup(fw_data++));
+	WREG32(CP_PFP_UCODE_ADDR, 0);
+
+	fw_data = (const __be32 *)rdev->me_fw->data;
+	WREG32(CP_ME_RAM_WADDR, 0);
+	for (i = 0; i < EVERGREEN_PM4_UCODE_SIZE; i++)
+		WREG32(CP_ME_RAM_DATA, be32_to_cpup(fw_data++));
+
+	WREG32(CP_PFP_UCODE_ADDR, 0);
+	WREG32(CP_ME_RAM_WADDR, 0);
+	WREG32(CP_ME_RAM_RADDR, 0);
 	return 0;
 }
+
+int evergreen_cp_resume(struct radeon_device *rdev)
+{
+	u32 tmp;
+	u32 rb_bufsz;
+	int r;
+
+	/* Reset cp; if cp is reset, then PA, SH, VGT also need to be reset */
+	WREG32(GRBM_SOFT_RESET, (SOFT_RESET_CP |
+				 SOFT_RESET_PA |
+				 SOFT_RESET_SH |
+				 SOFT_RESET_VGT |
+				 SOFT_RESET_SX));
+	RREG32(GRBM_SOFT_RESET);
+	mdelay(15);
+	WREG32(GRBM_SOFT_RESET, 0);
+	RREG32(GRBM_SOFT_RESET);
+
+	/* Set ring buffer size */
+	rb_bufsz = drm_order(rdev->cp.ring_size / 8);
+	tmp = RB_NO_UPDATE | (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz;
+#ifdef __BIG_ENDIAN
+	tmp |= BUF_SWAP_32BIT;
 #endif
+	WREG32(CP_RB_CNTL, tmp);
+	WREG32(CP_SEM_WAIT_TIMER, 0x4);
+
+	/* Set the write pointer delay */
+	WREG32(CP_RB_WPTR_DELAY, 0);
+
+	/* Initialize the ring buffer's read and write pointers */
+	WREG32(CP_RB_CNTL, tmp | RB_RPTR_WR_ENA);
+	WREG32(CP_RB_RPTR_WR, 0);
+	WREG32(CP_RB_WPTR, 0);
+	WREG32(CP_RB_RPTR_ADDR, rdev->cp.gpu_addr & 0xFFFFFFFF);
+	WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->cp.gpu_addr));
+	mdelay(1);
+	WREG32(CP_RB_CNTL, tmp);
+
+	WREG32(CP_RB_BASE, rdev->cp.gpu_addr >> 8);
+	WREG32(CP_DEBUG, (1 << 27) | (1 << 28));
+
+	rdev->cp.rptr = RREG32(CP_RB_RPTR);
+	rdev->cp.wptr = RREG32(CP_RB_WPTR);
+
+	r600_cp_start(rdev);
+	rdev->cp.ready = true;
+	r = radeon_ring_test(rdev);
+	if (r) {
+		rdev->cp.ready = false;
+		return r;
+	}
+	return 0;
+}
 
 /*
  * Core functions
@@ -1138,15 +1209,15 @@ static int evergreen_startup(struct radeon_device *rdev)
 {
 	int r;
 
-#if 0
-	if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) {
+	/* XXX until interrupts are supported */
+	if (!rdev->me_fw || !rdev->pfp_fw /*|| !rdev->rlc_fw*/) {
 		r = r600_init_microcode(rdev);
 		if (r) {
 			DRM_ERROR("Failed to load firmware!\n");
 			return r;
 		}
 	}
-#endif
+
 	evergreen_mc_program(rdev);
 	if (rdev->flags & RADEON_IS_AGP) {
 		evergreen_agp_enable(rdev);
@@ -1184,6 +1255,7 @@ static int evergreen_startup(struct radeon_device *rdev)
 		return r;
 	}
 	r600_irq_set(rdev);
+#endif
 
 	r = radeon_ring_init(rdev, rdev->cp.ring_size);
 	if (r)
@@ -1191,12 +1263,12 @@ static int evergreen_startup(struct radeon_device *rdev)
 	r = evergreen_cp_load_microcode(rdev);
 	if (r)
 		return r;
-	r = r600_cp_resume(rdev);
+	r = evergreen_cp_resume(rdev);
 	if (r)
 		return r;
 	/* write back buffer are not vital so don't worry about failure */
 	r600_wb_enable(rdev);
-#endif
+
 	return 0;
 }
 
@@ -1221,13 +1293,13 @@ int evergreen_resume(struct radeon_device *rdev)
 		DRM_ERROR("r600 startup failed on resume\n");
 		return r;
 	}
-#if 0
+
 	r = r600_ib_test(rdev);
 	if (r) {
 		DRM_ERROR("radeon: failled testing IB (%d).\n", r);
 		return r;
 	}
-#endif
+
 	return r;
 
 }
@@ -1236,12 +1308,11 @@ int evergreen_suspend(struct radeon_device *rdev)
 {
 #if 0
 	int r;
-
+#endif
 	/* FIXME: we should wait for ring to be empty */
 	r700_cp_stop(rdev);
 	rdev->cp.ready = false;
 	r600_wb_disable(rdev);
-#endif
 
 	evergreen_pcie_gart_disable(rdev);
 #if 0
@@ -1348,10 +1419,10 @@ int evergreen_init(struct radeon_device *rdev)
 	r = radeon_irq_kms_init(rdev);
 	if (r)
 		return r;
-
+#endif
 	rdev->cp.ring_obj = NULL;
 	r600_ring_init(rdev, 1024 * 1024);
-
+#if 0
 	rdev->ih.ring_obj = NULL;
 	r600_ih_ring_init(rdev, 64 * 1024);
 #endif
@@ -1362,9 +1433,13 @@ int evergreen_init(struct radeon_device *rdev)
 	rdev->accel_working = false;
 	r = evergreen_startup(rdev);
 	if (r) {
-		evergreen_suspend(rdev);
-		/*r600_wb_fini(rdev);*/
-		/*radeon_ring_fini(rdev);*/
+		dev_err(rdev->dev, "disabling GPU acceleration\n");
+		r700_cp_fini(rdev);
+		r600_wb_fini(rdev);
+#if 0
+		r600_irq_fini(rdev);
+		radeon_irq_kms_fini(rdev);
+#endif
 		evergreen_pcie_gart_fini(rdev);
 		rdev->accel_working = false;
 	}

drivers/gpu/drm/radeon/evergreend.h

@@ -89,6 +89,7 @@
 #define	CP_QUEUE_THRESHOLDS				0x8760
 #define		ROQ_IB1_START(x)				((x) << 0)
 #define		ROQ_IB2_START(x)				((x) << 8)
+#define	CP_RB_BASE					0xC100
 #define	CP_RB_CNTL					0xC104
 #define		RB_BUFSZ(x)					((x) << 0)
 #define		RB_BLKSZ(x)					((x) << 8)
@@ -104,6 +105,7 @@
 #define	CP_RB_WPTR_ADDR_HI				0xC11C
 #define	CP_RB_WPTR_DELAY				0x8704
 #define	CP_SEM_WAIT_TIMER				0x85BC
+#define	CP_DEBUG					0xC1FC
 
 
 #define	GC_USER_SHADER_PIPE_CONFIG			0x8954

drivers/gpu/drm/radeon/r600.c

@@ -43,6 +43,8 @@
 #define R700_PFP_UCODE_SIZE 848
 #define R700_PM4_UCODE_SIZE 1360
 #define R700_RLC_UCODE_SIZE 1024
+#define EVERGREEN_PFP_UCODE_SIZE 1120
+#define EVERGREEN_PM4_UCODE_SIZE 1376
 
 /* Firmware Names */
 MODULE_FIRMWARE("radeon/R600_pfp.bin");
@@ -67,6 +69,14 @@ MODULE_FIRMWARE("radeon/RV710_pfp.bin");
 MODULE_FIRMWARE("radeon/RV710_me.bin");
 MODULE_FIRMWARE("radeon/R600_rlc.bin");
 MODULE_FIRMWARE("radeon/R700_rlc.bin");
+MODULE_FIRMWARE("radeon/CEDAR_pfp.bin");
+MODULE_FIRMWARE("radeon/CEDAR_me.bin");
+MODULE_FIRMWARE("radeon/REDWOOD_pfp.bin");
+MODULE_FIRMWARE("radeon/REDWOOD_me.bin");
+MODULE_FIRMWARE("radeon/JUNIPER_pfp.bin");
+MODULE_FIRMWARE("radeon/JUNIPER_me.bin");
+MODULE_FIRMWARE("radeon/CYRPESS_pfp.bin");
+MODULE_FIRMWARE("radeon/CYPRESS_me.bin");
 
 int r600_debugfs_mc_info_init(struct radeon_device *rdev);
 
@@ -1449,10 +1459,31 @@ int r600_init_microcode(struct radeon_device *rdev)
 		chip_name = "RV710";
 		rlc_chip_name = "R700";
 		break;
+	case CHIP_CEDAR:
+		chip_name = "CEDAR";
+		rlc_chip_name = "";
+		break;
+	case CHIP_REDWOOD:
+		chip_name = "REDWOOD";
+		rlc_chip_name = "";
+		break;
+	case CHIP_JUNIPER:
+		chip_name = "JUNIPER";
+		rlc_chip_name = "";
+		break;
+	case CHIP_CYPRESS:
+	case CHIP_HEMLOCK:
+		chip_name = "CYPRESS";
+		rlc_chip_name = "";
+		break;
 	default: BUG();
 	}
 
-	if (rdev->family >= CHIP_RV770) {
+	if (rdev->family >= CHIP_CEDAR) {
+		pfp_req_size = EVERGREEN_PFP_UCODE_SIZE * 4;
+		me_req_size = EVERGREEN_PM4_UCODE_SIZE * 4;
+		rlc_req_size = 0;
+	} else if (rdev->family >= CHIP_RV770) {
 		pfp_req_size = R700_PFP_UCODE_SIZE * 4;
 		me_req_size = R700_PM4_UCODE_SIZE * 4;
 		rlc_req_size = R700_RLC_UCODE_SIZE * 4;
@@ -1487,6 +1518,8 @@ int r600_init_microcode(struct radeon_device *rdev)
 		err = -EINVAL;
 	}
 
+	/* XXX until evergreen interrupts are supported */
+	if (rdev->family < CHIP_CEDAR) {
 	snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", rlc_chip_name);
 	err = request_firmware(&rdev->rlc_fw, fw_name, &pdev->dev);
 	if (err)
@@ -1497,6 +1530,7 @@ int r600_init_microcode(struct radeon_device *rdev)
 		       rdev->rlc_fw->size, fw_name);
 		err = -EINVAL;
 	}
+	}
 
 out:
 	platform_device_unregister(pdev);
@@ -1566,12 +1600,15 @@ int r600_cp_start(struct radeon_device *rdev)
 	}
 	radeon_ring_write(rdev, PACKET3(PACKET3_ME_INITIALIZE, 5));
 	radeon_ring_write(rdev, 0x1);
-	if (rdev->family < CHIP_RV770) {
-		radeon_ring_write(rdev, 0x3);
-		radeon_ring_write(rdev, rdev->config.r600.max_hw_contexts - 1);
-	} else {
+	if (rdev->family >= CHIP_CEDAR) {
+		radeon_ring_write(rdev, 0x0);
+		radeon_ring_write(rdev, rdev->config.evergreen.max_hw_contexts - 1);
+	} else if (rdev->family >= CHIP_RV770) {
 		radeon_ring_write(rdev, 0x0);
 		radeon_ring_write(rdev, rdev->config.rv770.max_hw_contexts - 1);
+	} else {
+		radeon_ring_write(rdev, 0x3);
+		radeon_ring_write(rdev, rdev->config.r600.max_hw_contexts - 1);
 	}
 	radeon_ring_write(rdev, PACKET3_ME_INITIALIZE_DEVICE_ID(1));
 	radeon_ring_write(rdev, 0);

drivers/gpu/drm/radeon/radeon.h

@@ -1300,6 +1300,7 @@ extern void rs690_line_buffer_adjust(struct radeon_device *rdev,
 extern void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc);
 extern bool r600_card_posted(struct radeon_device *rdev);
 extern void r600_cp_stop(struct radeon_device *rdev);
+extern int r600_cp_start(struct radeon_device *rdev);
 extern void r600_ring_init(struct radeon_device *rdev, unsigned ring_size);
 extern int r600_cp_resume(struct radeon_device *rdev);
 extern void r600_cp_fini(struct radeon_device *rdev);
@@ -1340,6 +1341,9 @@ extern void r600_hdmi_update_audio_settings(struct drm_encoder *encoder,
 					    uint8_t status_bits,
 					    uint8_t category_code);
 
+extern void r700_cp_stop(struct radeon_device *rdev);
+extern void r700_cp_fini(struct radeon_device *rdev);
+
 /* evergreen */
 struct evergreen_mc_save {
 	u32 vga_control[6];

drivers/gpu/drm/radeon/radeon_asic.c

@@ -635,14 +635,14 @@ static struct radeon_asic evergreen_asic = {
 	.fini = &evergreen_fini,
 	.suspend = &evergreen_suspend,
 	.resume = &evergreen_resume,
-	.cp_commit = NULL,
+	.cp_commit = &r600_cp_commit,
 	.gpu_is_lockup = &evergreen_gpu_is_lockup,
 	.asic_reset = &evergreen_asic_reset,
 	.vga_set_state = &r600_vga_set_state,
 	.gart_tlb_flush = &evergreen_pcie_gart_tlb_flush,
 	.gart_set_page = &rs600_gart_set_page,
-	.ring_test = NULL,
-	.ring_ib_execute = NULL,
+	.ring_test = &r600_ring_test,
+	.ring_ib_execute = &r600_ring_ib_execute,
 	.irq_set = NULL,
 	.irq_process = NULL,
 	.get_vblank_counter = NULL,

drivers/gpu/drm/radeon/rv770.c

@@ -236,7 +236,6 @@ void r700_cp_stop(struct radeon_device *rdev)
 	WREG32(CP_ME_CNTL, (CP_ME_HALT | CP_PFP_HALT));
 }
 
-
 static int rv770_cp_load_microcode(struct radeon_device *rdev)
 {
 	const __be32 *fw_data;
@@ -271,6 +270,11 @@ static int rv770_cp_load_microcode(struct radeon_device *rdev)
 	return 0;
 }
 
+void r700_cp_fini(struct radeon_device *rdev)
+{
+	r700_cp_stop(rdev);
+	radeon_ring_fini(rdev);
+}
 
 /*
  * Core functions
@@ -1125,7 +1129,7 @@ int rv770_init(struct radeon_device *rdev)
 	r = rv770_startup(rdev);
 	if (r) {
 		dev_err(rdev->dev, "disabling GPU acceleration\n");
-		r600_cp_fini(rdev);
+		r700_cp_fini(rdev);
 		r600_wb_fini(rdev);
 		r600_irq_fini(rdev);
 		radeon_irq_kms_fini(rdev);
@@ -1159,7 +1163,7 @@ void rv770_fini(struct radeon_device *rdev)
 {
 	radeon_pm_fini(rdev);
 	r600_blit_fini(rdev);
-	r600_cp_fini(rdev);
+	r700_cp_fini(rdev);
 	r600_wb_fini(rdev);
 	r600_irq_fini(rdev);
 	radeon_irq_kms_fini(rdev);