drm/amdgpu: Switch to VRAM buffer for USBC PD FW.

System memory-based implementation for updating the
USBCPD is deprecated for so switching
to LFB based implementation for all the ASICs.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Alexander Deucher <Alexander.Deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -24,7 +24,6 @@
  */
 
 #include <linux/firmware.h>
-#include <linux/dma-mapping.h>
 #include <drm/drm_drv.h>
 
 #include "amdgpu.h"
@@ -3273,11 +3272,12 @@ static ssize_t psp_usbc_pd_fw_sysfs_write(struct device *dev,
 {
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct amdgpu_device *adev = drm_to_adev(ddev);
-	void *cpu_addr;
-	dma_addr_t dma_addr;
 	int ret, idx;
 	char fw_name[100];
 	const struct firmware *usbc_pd_fw;
+	struct amdgpu_bo *fw_buf_bo = NULL;
+	uint64_t fw_pri_mc_addr;
+	void *fw_pri_cpu_addr;
 
 	if (!adev->ip_blocks[AMD_IP_BLOCK_TYPE_PSP].status.late_initialized) {
 		DRM_INFO("PSP block is not ready yet.");
@@ -3292,31 +3292,24 @@ static ssize_t psp_usbc_pd_fw_sysfs_write(struct device *dev,
 	if (ret)
 		goto fail;
 
-	/* We need contiguous physical mem to place the FW  for psp to access */
-	cpu_addr = dma_alloc_coherent(adev->dev, usbc_pd_fw->size, &dma_addr, GFP_KERNEL);
-
-	ret = dma_mapping_error(adev->dev, dma_addr);
+	/* LFB address which is aligned to 1MB boundary per PSP request */
+	ret = amdgpu_bo_create_kernel(adev, usbc_pd_fw->size, 0x100000,
+						AMDGPU_GEM_DOMAIN_VRAM,
+						&fw_buf_bo,
+						&fw_pri_mc_addr,
+						&fw_pri_cpu_addr);
 	if (ret)
 		goto rel_buf;
 
-	memcpy_toio(cpu_addr, usbc_pd_fw->data, usbc_pd_fw->size);
-
-	/*
-	 * x86 specific workaround.
-	 * Without it the buffer is invisible in PSP.
-	 *
-	 * TODO Remove once PSP starts snooping CPU cache
-	 */
-#ifdef CONFIG_X86
-	clflush_cache_range(cpu_addr, (usbc_pd_fw->size & ~(L1_CACHE_BYTES - 1)));
-#endif
+	memcpy_toio(fw_pri_cpu_addr, usbc_pd_fw->data, usbc_pd_fw->size);
 
 	mutex_lock(&adev->psp.mutex);
-	ret = psp_load_usbc_pd_fw(&adev->psp, dma_addr);
+	ret = psp_load_usbc_pd_fw(&adev->psp, fw_pri_mc_addr);
 	mutex_unlock(&adev->psp.mutex);
 
+	amdgpu_bo_free_kernel(&fw_buf_bo, &fw_pri_mc_addr, &fw_pri_cpu_addr);
+
 rel_buf:
-	dma_free_coherent(adev->dev, usbc_pd_fw->size, cpu_addr, dma_addr);
 	release_firmware(usbc_pd_fw);
 fail:
 	if (ret) {

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

@@ -106,7 +106,7 @@ struct psp_funcs
 	int (*mem_training)(struct psp_context *psp, uint32_t ops);
 	uint32_t (*ring_get_wptr)(struct psp_context *psp);
 	void (*ring_set_wptr)(struct psp_context *psp, uint32_t value);
-	int (*load_usbc_pd_fw)(struct psp_context *psp, dma_addr_t dma_addr);
+	int (*load_usbc_pd_fw)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
 	int (*read_usbc_pd_fw)(struct psp_context *psp, uint32_t *fw_ver);
 };
 
@@ -414,9 +414,9 @@ struct amdgpu_psp_funcs {
 #define psp_ring_get_wptr(psp) (psp)->funcs->ring_get_wptr((psp))
 #define psp_ring_set_wptr(psp, value) (psp)->funcs->ring_set_wptr((psp), (value))
 
-#define psp_load_usbc_pd_fw(psp, dma_addr) \
+#define psp_load_usbc_pd_fw(psp, fw_pri_mc_addr) \
 	((psp)->funcs->load_usbc_pd_fw ? \
-	(psp)->funcs->load_usbc_pd_fw((psp), (dma_addr)) : -EINVAL)
+	(psp)->funcs->load_usbc_pd_fw((psp), (fw_pri_mc_addr)) : -EINVAL)
 
 #define psp_read_usbc_pd_fw(psp, fw_ver) \
 	((psp)->funcs->read_usbc_pd_fw ? \

drivers/gpu/drm/amd/amdgpu/psp_v11_0.c

@@ -80,6 +80,9 @@ MODULE_FIRMWARE("amdgpu/beige_goby_ta.bin");
 /* For large FW files the time to complete can be very long */
 #define USBC_PD_POLLING_LIMIT_S 240
 
+/* Read USB-PD from LFB */
+#define GFX_CMD_USB_PD_USE_LFB 0x480
+
 static int psp_v11_0_init_microcode(struct psp_context *psp)
 {
 	struct amdgpu_device *adev = psp->adev;
@@ -753,44 +756,26 @@ static void psp_v11_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
 		WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, value);
 }
 
-static int psp_v11_0_load_usbc_pd_fw(struct psp_context *psp, dma_addr_t dma_addr)
+static int psp_v11_0_load_usbc_pd_fw(struct psp_context *psp, uint64_t fw_pri_mc_addr)
 {
 	struct amdgpu_device *adev = psp->adev;
 	uint32_t reg_status;
 	int ret, i = 0;
 
-	/* Write lower 32-bit address of the PD Controller FW */
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, lower_32_bits(dma_addr));
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			     0x80000000, 0x80000000, false);
-	if (ret)
-		return ret;
-
-	/* Fireup interrupt so PSP can pick up the lower address */
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35, 0x800000);
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
-			     0x80000000, 0x80000000, false);
-	if (ret)
-		return ret;
-
-	reg_status = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35);
-
-	if ((reg_status & 0xFFFF) != 0) {
-		DRM_ERROR("Lower address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = %02x...\n",
-				reg_status & 0xFFFF);
-		return -EIO;
-	}
-
-	/* Write upper 32-bit address of the PD Controller FW */
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, upper_32_bits(dma_addr));
+	/*
+	 * LFB address which is aligned to 1MB address and has to be
+	 * right-shifted by 20 so that LFB address can be passed on a 32-bit C2P
+	 * register
+	 */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_36, (fw_pri_mc_addr >> 20));
 
 	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
 			     0x80000000, 0x80000000, false);
 	if (ret)
 		return ret;
 
-	/* Fireup interrupt so PSP can pick up the upper address */
-	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35, 0x4000000);
+	/* Fireup interrupt so PSP can pick up the address */
+	WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_35, (GFX_CMD_USB_PD_USE_LFB << 16));
 
 	/* FW load takes very long time */
 	do {
@@ -806,7 +791,7 @@ static int psp_v11_0_load_usbc_pd_fw(struct psp_context *psp, dma_addr_t dma_add
 done:
 
 	if ((reg_status & 0xFFFF) != 0) {
-		DRM_ERROR("Upper address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = x%04x\n",
+		DRM_ERROR("Address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = 0x%04x\n",
 				reg_status & 0xFFFF);
 		return -EIO;
 	}