drm/amdgpu: add RAS CPER ring buffer

And initialize it, this is a pure software ring to store RAS CPER data.

v2: change ring size to 0x100000
v2: update the initialization of count_dw of cper ring, it's dword
variable
v3: skip VM inv eng for cper
v3: init/fini when aca enabled

Signed-off-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Xiang Liu <xiang.liu@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -382,6 +382,39 @@ int amdgpu_cper_generate_ce_records(struct amdgpu_device *adev,
 	return 0;
 }
 
+static u64 amdgpu_cper_ring_get_rptr(struct amdgpu_ring *ring)
+{
+	return *(ring->rptr_cpu_addr);
+}
+
+static u64 amdgpu_cper_ring_get_wptr(struct amdgpu_ring *ring)
+{
+	return ring->wptr;
+}
+
+static const struct amdgpu_ring_funcs cper_ring_funcs = {
+	.type = AMDGPU_RING_TYPE_CPER,
+	.align_mask = 0xff,
+	.support_64bit_ptrs = false,
+	.get_rptr = amdgpu_cper_ring_get_rptr,
+	.get_wptr = amdgpu_cper_ring_get_wptr,
+};
+
+static int amdgpu_cper_ring_init(struct amdgpu_device *adev)
+{
+	struct amdgpu_ring *ring = &(adev->cper.ring_buf);
+
+	ring->adev = NULL;
+	ring->ring_obj = NULL;
+	ring->use_doorbell = false;
+	ring->no_scheduler = true;
+	ring->funcs = &cper_ring_funcs;
+
+	sprintf(ring->name, "cper");
+	return amdgpu_ring_init(adev, ring, CPER_MAX_RING_SIZE, NULL, 0,
+				AMDGPU_RING_PRIO_DEFAULT, NULL);
+}
+
 int amdgpu_cper_init(struct amdgpu_device *adev)
 {
 	mutex_init(&adev->cper.cper_lock);
@@ -389,16 +422,14 @@ int amdgpu_cper_init(struct amdgpu_device *adev)
 	adev->cper.enabled = true;
 	adev->cper.max_count = CPER_MAX_ALLOWED_COUNT;
 
-	/*TODO: initialize cper ring*/
-
-	return 0;
+	return amdgpu_cper_ring_init(adev);
 }
 
 int amdgpu_cper_fini(struct amdgpu_device *adev)
 {
 	adev->cper.enabled = false;
 
-	/*TODO: free cper ring */
+	amdgpu_ring_fini(&(adev->cper.ring_buf));
 	adev->cper.count = 0;
 	adev->cper.wptr = 0;
 

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

@@ -29,6 +29,7 @@
 #include "amdgpu_aca.h"
 
 #define CPER_MAX_ALLOWED_COUNT		0x1000
+#define CPER_MAX_RING_SIZE		0X100000
 #define HDR_LEN				(sizeof(struct cper_hdr))
 #define SEC_DESC_LEN			(sizeof(struct cper_sec_desc))
 
@@ -62,6 +63,7 @@ struct amdgpu_cper {
 	uint32_t wptr;
 
 	void *ring[CPER_MAX_ALLOWED_COUNT];
+	struct amdgpu_ring ring_buf;
 };
 
 void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,

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

@@ -3091,7 +3091,8 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 
 	amdgpu_fru_get_product_info(adev);
 
-	r = amdgpu_cper_init(adev);
+	if (amdgpu_aca_is_enabled(adev))
+		r = amdgpu_cper_init(adev);
 
 init_failed:
 
@@ -3453,7 +3454,8 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
 {
 	int i, r;
 
-	amdgpu_cper_fini(adev);
+	if (amdgpu_aca_is_enabled(adev))
+		amdgpu_cper_fini(adev);
 
 	if (amdgpu_sriov_vf(adev) && adev->virt.ras_init_done)
 		amdgpu_virt_release_ras_err_handler_data(adev);

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

@@ -591,7 +591,8 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev)
 
 		if (ring == &adev->mes.ring[0] ||
 		    ring == &adev->mes.ring[1] ||
-		    ring == &adev->umsch_mm.ring)
+		    ring == &adev->umsch_mm.ring ||
+		    ring == &adev->cper.ring_buf)
 			continue;
 
 		inv_eng = ffs(vm_inv_engs[vmhub]);

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

@@ -324,21 +324,28 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 	/* always set cond_exec_polling to CONTINUE */
 	*ring->cond_exe_cpu_addr = 1;
 
-	r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
-	if (r) {
-		dev_err(adev->dev, "failed initializing fences (%d).\n", r);
-		return r;
+	if (ring->funcs->type != AMDGPU_RING_TYPE_CPER) {
+		r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
+		if (r) {
+			dev_err(adev->dev, "failed initializing fences (%d).\n", r);
+			return r;
+		}
+
+		max_ibs_dw = ring->funcs->emit_frame_size +
+			     amdgpu_ring_max_ibs(ring->funcs->type) * ring->funcs->emit_ib_size;
+		max_ibs_dw = (max_ibs_dw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
+
+		if (WARN_ON(max_ibs_dw > max_dw))
+			max_dw = max_ibs_dw;
+
+		ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
+	} else {
+		ring->ring_size = roundup_pow_of_two(max_dw * 4);
+		ring->count_dw = (ring->ring_size - 4) >> 2;
+		/* ring buffer is empty now */
+		ring->wptr = *ring->rptr_cpu_addr = 0;
 	}
 
-	max_ibs_dw = ring->funcs->emit_frame_size +
-		     amdgpu_ring_max_ibs(ring->funcs->type) * ring->funcs->emit_ib_size;
-	max_ibs_dw = (max_ibs_dw + ring->funcs->align_mask) & ~ring->funcs->align_mask;
-
-	if (WARN_ON(max_ibs_dw > max_dw))
-		max_dw = max_ibs_dw;
-
-	ring->ring_size = roundup_pow_of_two(max_dw * 4 * sched_hw_submission);
-
 	ring->buf_mask = (ring->ring_size / 4) - 1;
 	ring->ptr_mask = ring->funcs->support_64bit_ptrs ?
 		0xffffffffffffffff : ring->buf_mask;

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

@@ -82,6 +82,7 @@ enum amdgpu_ring_type {
 	AMDGPU_RING_TYPE_KIQ,
 	AMDGPU_RING_TYPE_MES,
 	AMDGPU_RING_TYPE_UMSCH_MM,
+	AMDGPU_RING_TYPE_CPER,
 };
 
 enum amdgpu_ib_pool_type {

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

@@ -77,7 +77,8 @@ static void aqua_vanjaram_set_xcp_id(struct amdgpu_device *adev,
 	ring->xcp_id = AMDGPU_XCP_NO_PARTITION;
 	if (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)
 		adev->gfx.enforce_isolation[0].xcp_id = ring->xcp_id;
-	if (adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE)
+	if ((adev->xcp_mgr->mode == AMDGPU_XCP_MODE_NONE) ||
+	    (ring->funcs->type == AMDGPU_RING_TYPE_CPER))
 		return;
 
 	inst_mask = 1 << inst_idx;