drm/amdgpu/vcn: Add vcn firmware log

vcn fwlog is for debugging purpose only,
by default, it is disabled.

Signed-off-by: Ruijing Dong <ruijing.dong@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -233,6 +233,9 @@ extern int amdgpu_cik_support;
 #endif
 extern int amdgpu_num_kcq;
 
+#define AMDGPU_VCNFW_LOG_SIZE (32 * 1024)
+extern int amdgpu_vcnfw_log;
+
 #define AMDGPU_VM_MAX_NUM_CTX			4096
 #define AMDGPU_SG_THRESHOLD			(256*1024*1024)
 #define AMDGPU_DEFAULT_GTT_SIZE_MB		3072ULL /* 3GB by default */

drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c

@@ -1778,6 +1778,16 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev)
 		amdgpu_debugfs_ring_init(adev, ring);
 	}
 
+	for ( i = 0; i < adev->vcn.num_vcn_inst; i++) {
+		if (!amdgpu_vcnfw_log)
+			break;
+
+		if (adev->vcn.harvest_config & (1 << i))
+			continue;
+
+		amdgpu_debugfs_vcn_fwlog_init(adev, i, &adev->vcn.inst[i]);
+	}
+
 	amdgpu_ras_debugfs_create_all(adev);
 	amdgpu_rap_debugfs_init(adev);
 	amdgpu_securedisplay_debugfs_init(adev);

drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c

@@ -178,6 +178,7 @@ int amdgpu_reset_method = -1; /* auto */
 int amdgpu_num_kcq = -1;
 int amdgpu_smartshift_bias;
 int amdgpu_use_xgmi_p2p = 1;
+int amdgpu_vcnfw_log;
 
 static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work);
 
@@ -855,6 +856,13 @@ module_param_named(bad_page_threshold, amdgpu_bad_page_threshold, int, 0444);
 MODULE_PARM_DESC(num_kcq, "number of kernel compute queue user want to setup (8 if set to greater than 8 or less than 0, only affect gfx 8+)");
 module_param_named(num_kcq, amdgpu_num_kcq, int, 0444);
 
+/**
+ * DOC: vcnfw_log (int)
+ * Enable vcnfw log output for debugging, the default is disabled.
+ */
+MODULE_PARM_DESC(vcnfw_log, "Enable vcnfw log(0 = disable (default value), 1 = enable)");
+module_param_named(vcnfw_log, amdgpu_vcnfw_log, int, 0444);
+
 /**
  * DOC: smu_pptable_id (int)
  * Used to override pptable id. id = 0 use VBIOS pptable.

drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c

@@ -27,6 +27,7 @@
 #include <linux/firmware.h>
 #include <linux/module.h>
 #include <linux/pci.h>
+#include <linux/debugfs.h>
 #include <drm/drm_drv.h>
 
 #include "amdgpu.h"
@@ -79,7 +80,7 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
 	const char *fw_name;
 	const struct common_firmware_header *hdr;
 	unsigned char fw_check;
-	unsigned int fw_shared_size;
+	unsigned int fw_shared_size, log_offset;
 	int i, r;
 
 	INIT_DELAYED_WORK(&adev->vcn.idle_work, amdgpu_vcn_idle_work_handler);
@@ -228,8 +229,12 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
 	if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
 		bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
 	fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared));
+	log_offset = offsetof(struct amdgpu_fw_shared, fw_log);
 	bo_size += fw_shared_size;
 
+	if (amdgpu_vcnfw_log)
+		bo_size += AMDGPU_VCNFW_LOG_SIZE;
+
 	for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
 		if (adev->vcn.harvest_config & (1 << i))
 			continue;
@@ -249,6 +254,12 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
 
 		adev->vcn.inst[i].fw_shared.mem_size = fw_shared_size;
 
+		if (amdgpu_vcnfw_log) {
+			adev->vcn.inst[i].fw_shared.cpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
+			adev->vcn.inst[i].fw_shared.gpu_addr -= AMDGPU_VCNFW_LOG_SIZE;
+			adev->vcn.inst[i].fw_shared.log_offset = log_offset;
+		}
+
 		if (adev->vcn.indirect_sram) {
 			r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
 					AMDGPU_GEM_DOMAIN_VRAM, &adev->vcn.inst[i].dpg_sram_bo,
@@ -983,3 +994,112 @@ void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev)
 		dev_info(adev->dev, "Will use PSP to load VCN firmware\n");
 	}
 }
+
+/*
+ * debugfs for mapping vcn firmware log buffer.
+ */
+#if defined(CONFIG_DEBUG_FS)
+static ssize_t amdgpu_debugfs_vcn_fwlog_read(struct file *f, char __user *buf,
+                                             size_t size, loff_t *pos)
+{
+	struct amdgpu_vcn_inst *vcn;
+	void *log_buf;
+	volatile struct amdgpu_vcn_fwlog *plog;
+	unsigned int read_pos, write_pos, available, i, read_bytes = 0;
+	unsigned int read_num[2] = {0};
+
+	vcn = file_inode(f)->i_private;
+	if (!vcn)
+		return -ENODEV;
+
+	if (!vcn->fw_shared.cpu_addr || !amdgpu_vcnfw_log)
+		return -EFAULT;
+
+	log_buf = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size;
+
+	plog = (volatile struct amdgpu_vcn_fwlog *)log_buf;
+	read_pos = plog->rptr;
+	write_pos = plog->wptr;
+
+	if (read_pos > AMDGPU_VCNFW_LOG_SIZE || write_pos > AMDGPU_VCNFW_LOG_SIZE)
+		return -EFAULT;
+
+	if (!size || (read_pos == write_pos))
+		return 0;
+
+	if (write_pos > read_pos) {
+		available = write_pos - read_pos;
+		read_num[0] = min(size, (size_t)available);
+	} else {
+		read_num[0] = AMDGPU_VCNFW_LOG_SIZE - read_pos;
+		available = read_num[0] + write_pos - plog->header_size;
+		if (size > available)
+			read_num[1] = write_pos - plog->header_size;
+		else if (size > read_num[0])
+			read_num[1] = size - read_num[0];
+		else
+			read_num[0] = size;
+	}
+
+	for (i = 0; i < 2; i++) {
+		if (read_num[i]) {
+			if (read_pos == AMDGPU_VCNFW_LOG_SIZE)
+				read_pos = plog->header_size;
+			if (read_num[i] == copy_to_user((buf + read_bytes),
+			                                (log_buf + read_pos), read_num[i]))
+				return -EFAULT;
+
+			read_bytes += read_num[i];
+			read_pos += read_num[i];
+		}
+	}
+
+	plog->rptr = read_pos;
+	*pos += read_bytes;
+	return read_bytes;
+}
+
+static const struct file_operations amdgpu_debugfs_vcnfwlog_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_vcn_fwlog_read,
+	.llseek = default_llseek
+};
+#endif
+
+void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev, uint8_t i,
+                                   struct amdgpu_vcn_inst *vcn)
+{
+#if defined(CONFIG_DEBUG_FS)
+	struct drm_minor *minor = adev_to_drm(adev)->primary;
+	struct dentry *root = minor->debugfs_root;
+	char name[32];
+
+	sprintf(name, "amdgpu_vcn_%d_fwlog", i);
+	debugfs_create_file_size(name, S_IFREG | S_IRUGO, root, vcn,
+				 &amdgpu_debugfs_vcnfwlog_fops,
+				 AMDGPU_VCNFW_LOG_SIZE);
+#endif
+}
+
+void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn)
+{
+#if defined(CONFIG_DEBUG_FS)
+	volatile uint32_t *flag = vcn->fw_shared.cpu_addr;
+	void *fw_log_cpu_addr = vcn->fw_shared.cpu_addr + vcn->fw_shared.mem_size;
+	uint64_t fw_log_gpu_addr = vcn->fw_shared.gpu_addr + vcn->fw_shared.mem_size;
+	volatile struct amdgpu_vcn_fwlog *log_buf = fw_log_cpu_addr;
+	volatile struct amdgpu_fw_shared_fw_logging *fw_log = vcn->fw_shared.cpu_addr
+                                                         + vcn->fw_shared.log_offset;
+	*flag |= cpu_to_le32(AMDGPU_VCN_FW_LOGGING_FLAG);
+	fw_log->is_enabled = 1;
+	fw_log->addr_lo = cpu_to_le32(fw_log_gpu_addr & 0xFFFFFFFF);
+	fw_log->addr_hi = cpu_to_le32(fw_log_gpu_addr >> 32);
+	fw_log->size = cpu_to_le32(AMDGPU_VCNFW_LOG_SIZE);
+
+	log_buf->header_size = sizeof(struct amdgpu_vcn_fwlog);
+	log_buf->buffer_size = AMDGPU_VCNFW_LOG_SIZE;
+	log_buf->rptr = log_buf->header_size;
+	log_buf->wptr = log_buf->header_size;
+	log_buf->wrapped = 0;
+#endif
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h

@@ -158,6 +158,7 @@
 #define AMDGPU_VCN_FW_SHARED_FLAG_0_RB	(1 << 6)
 #define AMDGPU_VCN_MULTI_QUEUE_FLAG	(1 << 8)
 #define AMDGPU_VCN_SW_RING_FLAG		(1 << 9)
+#define AMDGPU_VCN_FW_LOGGING_FLAG	(1 << 10)
 
 #define AMDGPU_VCN_IB_FLAG_DECODE_BUFFER	0x00000001
 #define AMDGPU_VCN_CMD_FLAG_MSG_BUFFER		0x00000001
@@ -335,4 +336,7 @@ enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring);
 
 void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev);
 
+void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn);
+void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev,
+                                   uint8_t i, struct amdgpu_vcn_inst *vcn);
 #endif

drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c

@@ -148,6 +148,13 @@ static int vcn_v1_0_sw_init(void *handle)
 
 	adev->vcn.pause_dpg_mode = vcn_v1_0_pause_dpg_mode;
 
+	if (amdgpu_vcnfw_log) {
+		volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
+
+		fw_shared->present_flag_0 = 0;
+		amdgpu_vcn_fwlog_init(adev->vcn.inst);
+	}
+
 	r = jpeg_v1_0_sw_init(handle);
 
 	return r;

drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c

@@ -174,6 +174,10 @@ static int vcn_v2_0_sw_init(void *handle)
 
 	fw_shared = adev->vcn.inst->fw_shared.cpu_addr;
 	fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG);
+
+	if (amdgpu_vcnfw_log)
+		amdgpu_vcn_fwlog_init(adev->vcn.inst);
+
 	return 0;
 }
 

drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c

@@ -198,6 +198,9 @@ static int vcn_v2_5_sw_init(void *handle)
 
 		fw_shared = adev->vcn.inst[j].fw_shared.cpu_addr;
 		fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG);
+
+		if (amdgpu_vcnfw_log)
+			amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
 	}
 
 	if (amdgpu_sriov_vf(adev)) {

drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c

@@ -218,6 +218,9 @@ static int vcn_v3_0_sw_init(void *handle)
 					     cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG) |
 					     cpu_to_le32(AMDGPU_VCN_FW_SHARED_FLAG_0_RB);
 		fw_shared->sw_ring.is_enabled = cpu_to_le32(DEC_SW_RING_ENABLED);
+
+		if (amdgpu_vcnfw_log)
+			amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
 	}
 
 	if (amdgpu_sriov_vf(adev)) {