diff mbox series

[v2,4/4] scsi: ufs: Add history of fatal events

Message ID 1562736017-29461-5-git-send-email-stanley.chu@mediatek.com (mailing list archive)
State New, archived
Headers show
Series scsi: ufs: Provide fatal and auto-hibern8 error history | expand

Commit Message

Stanley Chu July 10, 2019, 5:20 a.m. UTC 
Currently only "interrupt-based" errors have their own history,
however there are "non-interrupt-based" errors which may be
fatal also needing history to improve debugging or help know
the health status of UFS devices.

For example,
- Link startup fail
- Suspend fail
- Resume fail
- Task or request abort event

This patch tries to add those failed events by existed UFS error
history mechanism.

Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
---
 drivers/scsi/ufs/ufshcd.c | 36 +++++++++++++++++++++++++++---------
 drivers/scsi/ufs/ufshcd.h | 10 ++++++++++
 2 files changed, 37 insertions(+), 9 deletions(-)

Comments

Avri Altman July 10, 2019, 8:04 a.m. UTC | #1 
Hi Stanley,

> 
> Currently only "interrupt-based" errors have their own history,
> however there are "non-interrupt-based" errors which may be
> fatal also needing history to improve debugging or help know
> the health status of UFS devices.
> 
> For example,
> - Link startup fail
> - Suspend fail
> - Resume fail
> - Task or request abort event
> 
> This patch tries to add those failed events by existed UFS error
> history mechanism.
> 
> Signed-off-by: Stanley Chu <stanley.chu@mediatek.com>
> ---
>  drivers/scsi/ufs/ufshcd.c | 36 +++++++++++++++++++++++++++---------
>  drivers/scsi/ufs/ufshcd.h | 10 ++++++++++
>  2 files changed, 37 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
> index a46c3d2b2ea3..969128a731e1 100644
> --- a/drivers/scsi/ufs/ufshcd.c
> +++ b/drivers/scsi/ufs/ufshcd.c
> @@ -432,6 +432,14 @@ static void ufshcd_print_host_regs(struct ufs_hba
> *hba)
>  	ufshcd_print_err_hist(hba, &hba->ufs_stats.fatal_err, "fatal_err");
>  	ufshcd_print_err_hist(hba, &hba->ufs_stats.auto_hibern8_err,
>  			      "auto_hibern8_err");
> +	ufshcd_print_err_hist(hba, &hba->ufs_stats.task_abort_err,
> +			      "task_abort");
> +	ufshcd_print_err_hist(hba, &hba->ufs_stats.link_startup_err,
> +			      "link_startup_fail");
> +	ufshcd_print_err_hist(hba, &hba->ufs_stats.suspend_err,
> +			      "suspend_fail");
> +	ufshcd_print_err_hist(hba, &hba->ufs_stats.resume_err,
> +			      "resume_fail");
> 
>  	ufshcd_print_clk_freqs(hba);
> 
> @@ -4329,6 +4337,14 @@ static inline int
> ufshcd_disable_device_tx_lcc(struct ufs_hba *hba)
>  	return ufshcd_disable_tx_lcc(hba, true);
>  }
> 
> +static void ufshcd_update_reg_hist(struct ufs_err_reg_hist *reg_hist,
> +				   u32 reg)
> +{
> +	reg_hist->reg[reg_hist->pos] = reg;
> +	reg_hist->tstamp[reg_hist->pos] = ktime_get();
> +	reg_hist->pos = (reg_hist->pos + 1) % UFS_ERR_REG_HIST_LENGTH;
> +}
> +
>  /**
>   * ufshcd_link_startup - Initialize unipro link startup
>   * @hba: per adapter instance
> @@ -4356,6 +4372,8 @@ static int ufshcd_link_startup(struct ufs_hba
> *hba)
> 
>  		/* check if device is detected by inter-connect layer */
>  		if (!ret && !ufshcd_is_device_present(hba)) {
> +			ufshcd_update_reg_hist(&hba-
> >ufs_stats.link_startup_err,
> +					       0);
>  			dev_err(hba->dev, "%s: Device not present\n",
> __func__);
>  			ret = -ENXIO;
>  			goto out;
> @@ -4366,8 +4384,11 @@ static int ufshcd_link_startup(struct ufs_hba
> *hba)
>  		 * but we can't be sure if the link is up until link startup
>  		 * succeeds. So reset the local Uni-Pro and try again.
>  		 */
> -		if (ret && ufshcd_hba_enable(hba))
> +		if (ret && ufshcd_hba_enable(hba)) {
> +			ufshcd_update_reg_hist(&hba-
> >ufs_stats.link_startup_err,
> +					       (u32)ret);
>  			goto out;
> +		}
>  	} while (ret && retries--);
> 
>  	if (ret)
Here also link startup fails...

> @@ -5350,14 +5371,6 @@ static void ufshcd_err_handler(struct
> work_struct *work)
>  	pm_runtime_put_sync(hba->dev);
>  }
> 
> -static void ufshcd_update_reg_hist(struct ufs_err_reg_hist *reg_hist,
> -				   u32 reg)
> -{
> -	reg_hist->reg[reg_hist->pos] = reg;
> -	reg_hist->tstamp[reg_hist->pos] = ktime_get();
> -	reg_hist->pos = (reg_hist->pos + 1) % UFS_ERR_REG_HIST_LENGTH;
> -}
> -
>  /**
>   * ufshcd_update_uic_error - check and set fatal UIC error flags.
>   * @hba: per-adapter instance
> @@ -6043,6 +6056,7 @@ static int ufshcd_abort(struct scsi_cmnd *cmd)
>  	 */
>  	scsi_print_command(hba->lrb[tag].cmd);
>  	if (!hba->req_abort_count) {
> +		ufshcd_update_reg_hist(&hba->ufs_stats.task_abort_err,
> 0);
Here you are collecting abort events statistics, not abort errors.
If this is what you meant, then it's not task_abort_err, but task_abort.
And if indeed you are tracking task aborts, maybe add lun resets as well?


>  		ufshcd_print_host_regs(hba);
>  		ufshcd_print_host_state(hba);
>  		ufshcd_print_pwr_info(hba);
> @@ -7819,6 +7833,8 @@ static int ufshcd_suspend(struct ufs_hba *hba,
> enum ufs_pm_op pm_op)
>  	ufshcd_release(hba);
>  out:
>  	hba->pm_op_in_progress = 0;
> +	if (ret)
> +		ufshcd_update_reg_hist(&hba->ufs_stats.suspend_err,
> (u32)ret);
>  	return ret;
>  }
> 
> @@ -7921,6 +7937,8 @@ static int ufshcd_resume(struct ufs_hba *hba,
> enum ufs_pm_op pm_op)
>  	ufshcd_setup_clocks(hba, false);
>  out:
>  	hba->pm_op_in_progress = 0;
> +	if (ret)
> +		ufshcd_update_reg_hist(&hba->ufs_stats.resume_err,
> (u32)ret);
>  	return ret;
>  }
> 
> diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> index c6ec5c749ceb..f9f109da7f18 100644
> --- a/drivers/scsi/ufs/ufshcd.h
> +++ b/drivers/scsi/ufs/ufshcd.h
> @@ -438,6 +438,10 @@ struct ufs_err_reg_hist {
>   * @dme_err: tracks dme errors
>   * @fatal_err: tracks fatal errors
>   * @auto_hibern8_err: tracks auto-hibernate errors
> + * @tsk_abort_err: tracks task abort events
> + * @linkup_err: tracks link-startup fail events
> + * @suspend_err: tracks suspend fail events
> + * @resume_err: tracks resume fail events
>   */
>  struct ufs_stats {
>  	u32 hibern8_exit_cnt;
> @@ -453,6 +457,12 @@ struct ufs_stats {
>  	/* fatal errors */
>  	struct ufs_err_reg_hist fatal_err;
>  	struct ufs_err_reg_hist auto_hibern8_err;
> +
> +	/* fatal events */
Maybe move here fatal_err as well?

> +	struct ufs_err_reg_hist task_abort_err;
> +	struct ufs_err_reg_hist link_startup_err;
> +	struct ufs_err_reg_hist suspend_err;
> +	struct ufs_err_reg_hist resume_err;
>  };
> 
>  /**
> --
> 2.18.0


Thanks,
Avri
Stanley Chu July 10, 2019, 9:28 a.m. UTC | #2 
Hi Avri,

On Wed, 2019-07-10 at 08:04 +0000, Avri Altman wrote:
> Hi Stanley,
> 
> > +					       (u32)ret);
> >  			goto out;
> > +		}
> >  	} while (ret && retries--);
> > 
> >  	if (ret)
> Here also link startup fails...

Thanks! Will track this place as well in next version.

> >   * ufshcd_update_uic_error - check and set fatal UIC error flags.
> >   * @hba: per-adapter instance
> > @@ -6043,6 +6056,7 @@ static int ufshcd_abort(struct scsi_cmnd *cmd)
> >  	 */
> >  	scsi_print_command(hba->lrb[tag].cmd);
> >  	if (!hba->req_abort_count) {
> > +		ufshcd_update_reg_hist(&hba->ufs_stats.task_abort_err,
> > 0);
> Here you are collecting abort events statistics, not abort errors.
> If this is what you meant, then it's not task_abort_err, but task_abort.
> And if indeed you are tracking task aborts, maybe add lun resets as well?

Good suggestion! I would add history of lun reset and host reset as well
in next version.

> > diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
> > index c6ec5c749ceb..f9f109da7f18 100644
> > --- a/drivers/scsi/ufs/ufshcd.h
> > +++ b/drivers/scsi/ufs/ufshcd.h
> > @@ -438,6 +438,10 @@ struct ufs_err_reg_hist {
> >   * @dme_err: tracks dme errors
> >   * @fatal_err: tracks fatal errors
> >   * @auto_hibern8_err: tracks auto-hibernate errors
> > + * @tsk_abort_err: tracks task abort events
> > + * @linkup_err: tracks link-startup fail events
> > + * @suspend_err: tracks suspend fail events
> > + * @resume_err: tracks resume fail events
> >   */
> >  struct ufs_stats {
> >  	u32 hibern8_exit_cnt;
> > @@ -453,6 +457,12 @@ struct ufs_stats {
> >  	/* fatal errors */
> >  	struct ufs_err_reg_hist fatal_err;
> >  	struct ufs_err_reg_hist auto_hibern8_err;
> > +
> > +	/* fatal events */
> Maybe move here fatal_err as well?

OK! these could be classified as fatal errors as well.
Will fix them in next version.

> 
> > +	struct ufs_err_reg_hist task_abort_err;
> > +	struct ufs_err_reg_hist link_startup_err;
> > +	struct ufs_err_reg_hist suspend_err;
> > +	struct ufs_err_reg_hist resume_err;
> >  };
> > 
> >  /**
> > --
> > 2.18.0
> 
> 
> Thanks,
> Avri
> 

Thanks,
Stanley
diff mbox series

Patch

diff --git a/drivers/scsi/ufs/ufshcd.c b/drivers/scsi/ufs/ufshcd.c
index a46c3d2b2ea3..969128a731e1 100644
--- a/drivers/scsi/ufs/ufshcd.c
+++ b/drivers/scsi/ufs/ufshcd.c
@@ -432,6 +432,14 @@  static void ufshcd_print_host_regs(struct ufs_hba *hba)
 	ufshcd_print_err_hist(hba, &hba->ufs_stats.fatal_err, "fatal_err");
 	ufshcd_print_err_hist(hba, &hba->ufs_stats.auto_hibern8_err,
 			      "auto_hibern8_err");
+	ufshcd_print_err_hist(hba, &hba->ufs_stats.task_abort_err,
+			      "task_abort");
+	ufshcd_print_err_hist(hba, &hba->ufs_stats.link_startup_err,
+			      "link_startup_fail");
+	ufshcd_print_err_hist(hba, &hba->ufs_stats.suspend_err,
+			      "suspend_fail");
+	ufshcd_print_err_hist(hba, &hba->ufs_stats.resume_err,
+			      "resume_fail");
 
 	ufshcd_print_clk_freqs(hba);
 
@@ -4329,6 +4337,14 @@  static inline int ufshcd_disable_device_tx_lcc(struct ufs_hba *hba)
 	return ufshcd_disable_tx_lcc(hba, true);
 }
 
+static void ufshcd_update_reg_hist(struct ufs_err_reg_hist *reg_hist,
+				   u32 reg)
+{
+	reg_hist->reg[reg_hist->pos] = reg;
+	reg_hist->tstamp[reg_hist->pos] = ktime_get();
+	reg_hist->pos = (reg_hist->pos + 1) % UFS_ERR_REG_HIST_LENGTH;
+}
+
 /**
  * ufshcd_link_startup - Initialize unipro link startup
  * @hba: per adapter instance
@@ -4356,6 +4372,8 @@  static int ufshcd_link_startup(struct ufs_hba *hba)
 
 		/* check if device is detected by inter-connect layer */
 		if (!ret && !ufshcd_is_device_present(hba)) {
+			ufshcd_update_reg_hist(&hba->ufs_stats.link_startup_err,
+					       0);
 			dev_err(hba->dev, "%s: Device not present\n", __func__);
 			ret = -ENXIO;
 			goto out;
@@ -4366,8 +4384,11 @@  static int ufshcd_link_startup(struct ufs_hba *hba)
 		 * but we can't be sure if the link is up until link startup
 		 * succeeds. So reset the local Uni-Pro and try again.
 		 */
-		if (ret && ufshcd_hba_enable(hba))
+		if (ret && ufshcd_hba_enable(hba)) {
+			ufshcd_update_reg_hist(&hba->ufs_stats.link_startup_err,
+					       (u32)ret);
 			goto out;
+		}
 	} while (ret && retries--);
 
 	if (ret)
@@ -5350,14 +5371,6 @@  static void ufshcd_err_handler(struct work_struct *work)
 	pm_runtime_put_sync(hba->dev);
 }
 
-static void ufshcd_update_reg_hist(struct ufs_err_reg_hist *reg_hist,
-				   u32 reg)
-{
-	reg_hist->reg[reg_hist->pos] = reg;
-	reg_hist->tstamp[reg_hist->pos] = ktime_get();
-	reg_hist->pos = (reg_hist->pos + 1) % UFS_ERR_REG_HIST_LENGTH;
-}
-
 /**
  * ufshcd_update_uic_error - check and set fatal UIC error flags.
  * @hba: per-adapter instance
@@ -6043,6 +6056,7 @@  static int ufshcd_abort(struct scsi_cmnd *cmd)
 	 */
 	scsi_print_command(hba->lrb[tag].cmd);
 	if (!hba->req_abort_count) {
+		ufshcd_update_reg_hist(&hba->ufs_stats.task_abort_err, 0);
 		ufshcd_print_host_regs(hba);
 		ufshcd_print_host_state(hba);
 		ufshcd_print_pwr_info(hba);
@@ -7819,6 +7833,8 @@  static int ufshcd_suspend(struct ufs_hba *hba, enum ufs_pm_op pm_op)
 	ufshcd_release(hba);
 out:
 	hba->pm_op_in_progress = 0;
+	if (ret)
+		ufshcd_update_reg_hist(&hba->ufs_stats.suspend_err, (u32)ret);
 	return ret;
 }
 
@@ -7921,6 +7937,8 @@  static int ufshcd_resume(struct ufs_hba *hba, enum ufs_pm_op pm_op)
 	ufshcd_setup_clocks(hba, false);
 out:
 	hba->pm_op_in_progress = 0;
+	if (ret)
+		ufshcd_update_reg_hist(&hba->ufs_stats.resume_err, (u32)ret);
 	return ret;
 }
 
diff --git a/drivers/scsi/ufs/ufshcd.h b/drivers/scsi/ufs/ufshcd.h
index c6ec5c749ceb..f9f109da7f18 100644
--- a/drivers/scsi/ufs/ufshcd.h
+++ b/drivers/scsi/ufs/ufshcd.h
@@ -438,6 +438,10 @@  struct ufs_err_reg_hist {
  * @dme_err: tracks dme errors
  * @fatal_err: tracks fatal errors
  * @auto_hibern8_err: tracks auto-hibernate errors
+ * @tsk_abort_err: tracks task abort events
+ * @linkup_err: tracks link-startup fail events
+ * @suspend_err: tracks suspend fail events
+ * @resume_err: tracks resume fail events
  */
 struct ufs_stats {
 	u32 hibern8_exit_cnt;
@@ -453,6 +457,12 @@  struct ufs_stats {
 	/* fatal errors */
 	struct ufs_err_reg_hist fatal_err;
 	struct ufs_err_reg_hist auto_hibern8_err;
+
+	/* fatal events */
+	struct ufs_err_reg_hist task_abort_err;
+	struct ufs_err_reg_hist link_startup_err;
+	struct ufs_err_reg_hist suspend_err;
+	struct ufs_err_reg_hist resume_err;
 };
 
 /**