|
Jiri Slaby |
ef7db2 |
From: Moti Haimovski <mhaimovski@habana.ai>
|
|
Jiri Slaby |
ef7db2 |
Date: Thu, 29 Dec 2022 12:44:09 +0200
|
|
Jiri Slaby |
ef7db2 |
Subject: [PATCH] habanalabs: extend fatal messages to contain PCI info
|
|
Jiri Slaby |
ef7db2 |
References: bsc#1012628
|
|
Jiri Slaby |
ef7db2 |
Patch-mainline: 6.2.3
|
|
Jiri Slaby |
ef7db2 |
Git-commit: 2a0a839b6a28f7c4c528bb75b740c7f38ef79a37
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
[ Upstream commit 2a0a839b6a28f7c4c528bb75b740c7f38ef79a37 ]
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
This commit attaches the PCI device address to driver fatal messages
|
|
Jiri Slaby |
ef7db2 |
in order to ease debugging in multi-device setups.
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
Signed-off-by: Moti Haimovski <mhaimovski@habana.ai>
|
|
Jiri Slaby |
ef7db2 |
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
|
|
Jiri Slaby |
ef7db2 |
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
|
|
Jiri Slaby |
ef7db2 |
Signed-off-by: Sasha Levin <sashal@kernel.org>
|
|
Jiri Slaby |
ef7db2 |
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
|
|
Jiri Slaby |
ef7db2 |
---
|
|
Jiri Slaby |
ef7db2 |
drivers/misc/habanalabs/common/device.c | 38 ++++++++++++++++---------
|
|
Jiri Slaby |
ef7db2 |
1 file changed, 25 insertions(+), 13 deletions(-)
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
diff --git a/drivers/misc/habanalabs/common/device.c b/drivers/misc/habanalabs/common/device.c
|
|
Jiri Slaby |
ef7db2 |
index 87ab329e..f7b9c387 100644
|
|
Jiri Slaby |
ef7db2 |
--- a/drivers/misc/habanalabs/common/device.c
|
|
Jiri Slaby |
ef7db2 |
+++ b/drivers/misc/habanalabs/common/device.c
|
|
Jiri Slaby |
ef7db2 |
@@ -1566,7 +1566,8 @@ int hl_device_reset(struct hl_device *hdev, u32 flags)
|
|
Jiri Slaby |
ef7db2 |
if (rc == -EBUSY) {
|
|
Jiri Slaby |
ef7db2 |
if (hdev->device_fini_pending) {
|
|
Jiri Slaby |
ef7db2 |
dev_crit(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "Failed to kill all open processes, stopping hard reset\n");
|
|
Jiri Slaby |
ef7db2 |
+ "%s Failed to kill all open processes, stopping hard reset\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
goto out_err;
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
@@ -1576,7 +1577,8 @@ int hl_device_reset(struct hl_device *hdev, u32 flags)
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
if (rc) {
|
|
Jiri Slaby |
ef7db2 |
dev_crit(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "Failed to kill all open processes, stopping hard reset\n");
|
|
Jiri Slaby |
ef7db2 |
+ "%s Failed to kill all open processes, stopping hard reset\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
goto out_err;
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
@@ -1627,14 +1629,16 @@ int hl_device_reset(struct hl_device *hdev, u32 flags)
|
|
Jiri Slaby |
ef7db2 |
* ensure driver puts the driver in a unusable state
|
|
Jiri Slaby |
ef7db2 |
*/
|
|
Jiri Slaby |
ef7db2 |
dev_crit(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "Consecutive FW fatal errors received, stopping hard reset\n");
|
|
Jiri Slaby |
ef7db2 |
+ "%s Consecutive FW fatal errors received, stopping hard reset\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
rc = -EIO;
|
|
Jiri Slaby |
ef7db2 |
goto out_err;
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
if (hdev->kernel_ctx) {
|
|
Jiri Slaby |
ef7db2 |
dev_crit(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "kernel ctx was alive during hard reset, something is terribly wrong\n");
|
|
Jiri Slaby |
ef7db2 |
+ "%s kernel ctx was alive during hard reset, something is terribly wrong\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
rc = -EBUSY;
|
|
Jiri Slaby |
ef7db2 |
goto out_err;
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
@@ -1752,9 +1756,13 @@ int hl_device_reset(struct hl_device *hdev, u32 flags)
|
|
Jiri Slaby |
ef7db2 |
hdev->reset_info.needs_reset = false;
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
if (hard_reset)
|
|
Jiri Slaby |
ef7db2 |
- dev_info(hdev->dev, "Successfully finished resetting the device\n");
|
|
Jiri Slaby |
ef7db2 |
+ dev_info(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
+ "Successfully finished resetting the %s device\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
else
|
|
Jiri Slaby |
ef7db2 |
- dev_dbg(hdev->dev, "Successfully finished resetting the device\n");
|
|
Jiri Slaby |
ef7db2 |
+ dev_dbg(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
+ "Successfully finished resetting the %s device\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
if (hard_reset) {
|
|
Jiri Slaby |
ef7db2 |
hdev->reset_info.hard_reset_cnt++;
|
|
Jiri Slaby |
ef7db2 |
@@ -1789,7 +1797,9 @@ int hl_device_reset(struct hl_device *hdev, u32 flags)
|
|
Jiri Slaby |
ef7db2 |
hdev->reset_info.in_compute_reset = 0;
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
if (hard_reset) {
|
|
Jiri Slaby |
ef7db2 |
- dev_err(hdev->dev, "Failed to reset! Device is NOT usable\n");
|
|
Jiri Slaby |
ef7db2 |
+ dev_err(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
+ "%s Failed to reset! Device is NOT usable\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
hdev->reset_info.hard_reset_cnt++;
|
|
Jiri Slaby |
ef7db2 |
} else if (reset_upon_device_release) {
|
|
Jiri Slaby |
ef7db2 |
spin_unlock(&hdev->reset_info.lock);
|
|
Jiri Slaby |
ef7db2 |
@@ -2186,7 +2196,8 @@ int hl_device_init(struct hl_device *hdev, struct class *hclass)
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
dev_notice(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "Successfully added device to habanalabs driver\n");
|
|
Jiri Slaby |
ef7db2 |
+ "Successfully added device %s to habanalabs driver\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
hdev->init_done = true;
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
@@ -2235,11 +2246,11 @@ int hl_device_init(struct hl_device *hdev, struct class *hclass)
|
|
Jiri Slaby |
ef7db2 |
device_cdev_sysfs_add(hdev);
|
|
Jiri Slaby |
ef7db2 |
if (hdev->pdev)
|
|
Jiri Slaby |
ef7db2 |
dev_err(&hdev->pdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "Failed to initialize hl%d. Device is NOT usable !\n",
|
|
Jiri Slaby |
ef7db2 |
- hdev->cdev_idx);
|
|
Jiri Slaby |
ef7db2 |
+ "Failed to initialize hl%d. Device %s is NOT usable !\n",
|
|
Jiri Slaby |
ef7db2 |
+ hdev->cdev_idx, dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
else
|
|
Jiri Slaby |
ef7db2 |
- pr_err("Failed to initialize hl%d. Device is NOT usable !\n",
|
|
Jiri Slaby |
ef7db2 |
- hdev->cdev_idx);
|
|
Jiri Slaby |
ef7db2 |
+ pr_err("Failed to initialize hl%d. Device %s is NOT usable !\n",
|
|
Jiri Slaby |
ef7db2 |
+ hdev->cdev_idx, dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
return rc;
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
@@ -2295,7 +2306,8 @@ void hl_device_fini(struct hl_device *hdev)
|
|
Jiri Slaby |
ef7db2 |
|
|
Jiri Slaby |
ef7db2 |
if (ktime_compare(ktime_get(), timeout) > 0) {
|
|
Jiri Slaby |
ef7db2 |
dev_crit(hdev->dev,
|
|
Jiri Slaby |
ef7db2 |
- "Failed to remove device because reset function did not finish\n");
|
|
Jiri Slaby |
ef7db2 |
+ "%s Failed to remove device because reset function did not finish\n",
|
|
Jiri Slaby |
ef7db2 |
+ dev_name(&(hdev)->pdev->dev));
|
|
Jiri Slaby |
ef7db2 |
return;
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
}
|
|
Jiri Slaby |
ef7db2 |
--
|
|
Jiri Slaby |
ef7db2 |
2.35.3
|
|
Jiri Slaby |
ef7db2 |
|