drm/amdgpu:determine if vPost is needed indeed

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

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 4acc92b..75f490f 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -50,6 +50,7 @@
 #include "vi.h"
 #include "bif/bif_4_1_d.h"
 #include <linux/pci.h>
+#include <linux/firmware.h>
 
 static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
 static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
@@ -651,6 +652,46 @@ bool amdgpu_card_posted(struct amdgpu_device *adev)
 
 }
 
+static bool amdgpu_vpost_needed(struct amdgpu_device *adev)
+{
+       if (amdgpu_sriov_vf(adev))
+               return false;
+
+       if (amdgpu_passthrough(adev)) {
+               /* for FIJI: In whole GPU pass-through virtualization case
+                * old smc fw won't clear some registers (e.g. MEM_SIZE, BIOS_SCRATCH)
+                * so amdgpu_card_posted return false and driver will incorrectly skip vPost.
+                * but if we force vPost do in pass-through case, the driver reload will hang.
+                * whether doing vPost depends on amdgpu_card_posted if smc version is above
+                * 00160e00 for FIJI.
+                */
+               if (adev->asic_type == CHIP_FIJI) {
+                       int err;
+                       uint32_t fw_ver;
+                       err = request_firmware(&adev->pm.fw, "amdgpu/fiji_smc.bin", adev->dev);
+                       /* force vPost if error occured */
+                       if (err)
+                               return true;
+
+                       fw_ver = *((uint32_t *)adev->pm.fw->data + 69);
+                       if (fw_ver >= 0x00160e00)
+                               return !amdgpu_card_posted(adev);
+               }
+       } else {
+               /* in bare-metal case, amdgpu_card_posted return false
+                * after system reboot/boot, and return true if driver
+                * reloaded.
+                * we shouldn't do vPost after driver reload otherwise GPU
+                * could hang.
+                */
+               if (amdgpu_card_posted(adev))
+                       return false;
+       }
+
+       /* we assume vPost is neede for all other cases */
+       return true;
+}
+
 /**
  * amdgpu_dummy_page_init - init dummy page used by the driver
  *
@@ -1649,14 +1690,13 @@ int amdgpu_device_init(struct amdgpu_device *adev,
        amdgpu_device_detect_sriov_bios(adev);
 
        /* Post card if necessary */
-       if (!amdgpu_sriov_vf(adev) &&
-               (!amdgpu_card_posted(adev) || amdgpu_passthrough(adev))) {
+       if (amdgpu_vpost_needed(adev)) {
                if (!adev->bios) {
-                       dev_err(adev->dev, "Card not posted and no BIOS - ignoring\n");
+                       dev_err(adev->dev, "no vBIOS found\n");
                        r = -EINVAL;
                        goto failed;
                }
-               DRM_INFO("GPU not posted. posting now...\n");
+               DRM_INFO("GPU posting now...\n");
                r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
                if (r) {
                        dev_err(adev->dev, "gpu post error!\n");