From c9a6b82f45e261d247b980a7949aaa6a9bfffe01 Mon Sep 17 00:00:00 2001
From: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Date: Wed, 29 Jul 2020 12:59:45 -0400
Subject: [PATCH] drm/amdgpu: Implement DPC recovery
Add PCI Downstream Port Containment (DPC) with
basic recovery functionality
v2: remove pci_save_state to avoid breaking suspend/resume
v3: Fix style comments
v4: Improve description.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h | 8 +
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 161 +++++++++++++++++++++
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 9 +-
3 files changed, 177 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 4009d2e307275..29c0b4c872315 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -49,6 +49,8 @@
#include <linux/rbtree.h>
#include <linux/hashtable.h>
#include <linux/dma-fence.h>
+#include <linux/pci.h>
+#include <linux/aer.h>
#include <drm/ttm/ttm_bo_api.h>
#include <drm/ttm/ttm_bo_driver.h>
@@ -1260,6 +1262,12 @@ static inline int amdgpu_dm_display_resume(struct amdgpu_device *adev) { return
void amdgpu_register_gpu_instance(struct amdgpu_device *adev);
void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev);
+pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev,
+ pci_channel_state_t state);
+pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev);
+pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev);
+void amdgpu_pci_resume(struct pci_dev *pdev);
+
#include "amdgpu_object.h"
/* used by df_v3_6.c and amdgpu_pmu.c */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index f7307af764521..99c0e6e53e848 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2999,6 +2999,7 @@ static const struct attribute *amdgpu_dev_attributes[] = {
NULL
};
+
/**
* amdgpu_device_init - initialize the driver
*
@@ -3217,6 +3218,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
}
}
+ pci_enable_pcie_error_reporting(adev->ddev.pdev);
+
/* Post card if necessary */
if (amdgpu_device_need_post(adev)) {
if (!adev->bios) {
@@ -4705,3 +4708,161 @@ int amdgpu_device_baco_exit(struct drm_device *dev)
return 0;
}
+
+/**
+ * amdgpu_pci_error_detected - Called when a PCI error is detected.
+ * @pdev: PCI device struct
+ * @state: PCI channel state
+ *
+ * Description: Called when a PCI error is detected.
+ *
+ * Return: PCI_ERS_RESULT_NEED_RESET or PCI_ERS_RESULT_DISCONNECT.
+ */
+pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_state_t state)
+{
+ struct drm_device *dev = pci_get_drvdata(pdev);
+ struct amdgpu_device *adev = drm_to_adev(dev);
+
+ DRM_INFO("PCI error: detected callback, state(%d)!!\n", state);
+
+ switch (state) {
+ case pci_channel_io_normal:
+ return PCI_ERS_RESULT_CAN_RECOVER;
+ case pci_channel_io_frozen:
+ /* Fatal error, prepare for slot reset */
+ amdgpu_device_lock_adev(adev);
+ return PCI_ERS_RESULT_NEED_RESET;
+ case pci_channel_io_perm_failure:
+ /* Permanent error, prepare for device removal */
+ return PCI_ERS_RESULT_DISCONNECT;
+ }
+
+ return PCI_ERS_RESULT_NEED_RESET;
+}
+
+/**
+ * amdgpu_pci_mmio_enabled - Enable MMIO and dump debug registers
+ * @pdev: pointer to PCI device
+ */
+pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev)
+{
+
+ DRM_INFO("PCI error: mmio enabled callback!!\n");
+
+ /* TODO - dump whatever for debugging purposes */
+
+ /* This called only if amdgpu_pci_error_detected returns
+ * PCI_ERS_RESULT_CAN_RECOVER. Read/write to the device still
+ * works, no need to reset slot.
+ */
+
+ return PCI_ERS_RESULT_RECOVERED;
+}
+
+/**
+ * amdgpu_pci_slot_reset - Called when PCI slot has been reset.
+ * @pdev: PCI device struct
+ *
+ * Description: This routine is called by the pci error recovery
+ * code after the PCI slot has been reset, just before we
+ * should resume normal operations.
+ */
+pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
+{
+ struct drm_device *dev = pci_get_drvdata(pdev);
+ struct amdgpu_device *adev = drm_to_adev(dev);
+ int r;
+ bool vram_lost;
+
+ DRM_INFO("PCI error: slot reset callback!!\n");
+
+ pci_restore_state(pdev);
+
+ r = amdgpu_device_ip_suspend(adev);
+ if (r)
+ goto out;
+
+
+ /* post card */
+ r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
+ if (r)
+ goto out;
+
+ r = amdgpu_device_ip_resume_phase1(adev);
+ if (r)
+ goto out;
+
+ vram_lost = amdgpu_device_check_vram_lost(adev);
+ if (vram_lost) {
+ DRM_INFO("VRAM is lost due to GPU reset!\n");
+ amdgpu_inc_vram_lost(adev);
+ }
+
+ r = amdgpu_gtt_mgr_recover(
+ &adev->mman.bdev.man[TTM_PL_TT]);
+ if (r)
+ goto out;
+
+ r = amdgpu_device_fw_loading(adev);
+ if (r)
+ return r;
+
+ r = amdgpu_device_ip_resume_phase2(adev);
+ if (r)
+ goto out;
+
+ if (vram_lost)
+ amdgpu_device_fill_reset_magic(adev);
+
+ /*
+ * Add this ASIC as tracked as reset was already
+ * complete successfully.
+ */
+ amdgpu_register_gpu_instance(adev);
+
+ r = amdgpu_device_ip_late_init(adev);
+ if (r)
+ goto out;
+
+ amdgpu_fbdev_set_suspend(adev, 0);
+
+ /* must succeed. */
+ amdgpu_ras_resume(adev);
+
+
+ amdgpu_irq_gpu_reset_resume_helper(adev);
+ r = amdgpu_ib_ring_tests(adev);
+ if (r)
+ goto out;
+
+ r = amdgpu_device_recover_vram(adev);
+
+out:
+
+ if (!r) {
+ DRM_INFO("PCIe error recovery succeeded\n");
+ } else {
+ DRM_ERROR("PCIe error recovery failed, err:%d", r);
+ amdgpu_device_unlock_adev(adev);
+ }
+
+ return r ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED;
+}
+
+/**
+ * amdgpu_pci_resume() - resume normal ops after PCI reset
+ * @pdev: pointer to PCI device
+ *
+ * Called when the error recovery driver tells us that its
+ * OK to resume normal operation. Use completion to allow
+ * halted scsi ops to resume.
+ */
+void amdgpu_pci_resume(struct pci_dev *pdev)
+{
+ struct drm_device *dev = pci_get_drvdata(pdev);
+ struct amdgpu_device *adev = drm_to_adev(dev);
+
+ amdgpu_device_unlock_adev(adev);
+
+ DRM_INFO("PCI error: resume callback!!\n");
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 6edde2b9e4020..e9daab175d170 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -32,7 +32,6 @@
#include <drm/drm_pciids.h>
#include <linux/console.h>
#include <linux/module.h>
-#include <linux/pci.h>
#include <linux/pm_runtime.h>
#include <linux/vga_switcheroo.h>
#include <drm/drm_probe_helper.h>
@@ -1528,6 +1527,13 @@ static struct drm_driver kms_driver = {
.patchlevel = KMS_DRIVER_PATCHLEVEL,
};
+static struct pci_error_handlers amdgpu_pci_err_handler = {
+ .error_detected = amdgpu_pci_error_detected,
+ .mmio_enabled = amdgpu_pci_mmio_enabled,
+ .slot_reset = amdgpu_pci_slot_reset,
+ .resume = amdgpu_pci_resume,
+};
+
static struct pci_driver amdgpu_kms_pci_driver = {
.name = DRIVER_NAME,
.id_table = pciidlist,
@@ -1535,6 +1541,7 @@ static struct pci_driver amdgpu_kms_pci_driver = {
.remove = amdgpu_pci_remove,
.shutdown = amdgpu_pci_shutdown,
.driver.pm = &amdgpu_pm_ops,
+ .err_handler = &amdgpu_pci_err_handler,
};
static int __init amdgpu_init(void)
--
GitLab