[PATCH v2 0/2] hwtracing: hisi_ptt: Fix reset timeout handling and clean up trace start

[PATCH v2 0/2] hwtracing: hisi_ptt: Fix reset timeout handling and clean up trace start

[Pradhan, Sanman]

From: Sanman Pradhan <psanman@juniper.net>

Patch 1: Propagate the DMA reset timeout error from
  hisi_ptt_wait_dma_reset_done() instead of discarding it. De-assert
  the reset bit and log an error on timeout. Move ctrl->started to the
  successful path so a failed start does not leave the trace marked as
  active.

Patch 2: Remove the unnecessary 16 MiB memset of trace buffers in
  hisi_ptt_trace_start(). The driver only copies data that hardware has
  written, so the zeroing is not needed.

Changes since v1:
  - Patch 1: Return bool from hisi_ptt_wait_dma_reset_done() for
    consistency with the other wait helpers
  - Patch 1: Add pci_err() on timeout
  - Patch 1: De-assert RST before returning on timeout
  - Patch 1: Move ctrl->started to the successful path
  - Dropped "Use the passed buffer index in hisi_ptt_update_aux()" patch
  - Patch 2 is unchanged

Sanman Pradhan (2):
  hwtracing: hisi_ptt: Propagate DMA reset timeout in trace_start()
  hwtracing: hisi_ptt: Remove unnecessary trace buffer zeroing in
    trace_start()

 drivers/hwtracing/ptt/hisi_ptt.c | 25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)

-- 
2.34.1

[PATCH v2 1/2] hwtracing: hisi_ptt: Propagate DMA reset timeout in trace_start()

[Pradhan, Sanman]

From: Sanman Pradhan <psanman@juniper.net>

hisi_ptt_wait_dma_reset_done() discards the return value of
readl_poll_timeout_atomic(). If the DMA engine does not complete its
reset within the timeout, hisi_ptt_trace_start() proceeds to start
tracing regardless.

Return a bool from hisi_ptt_wait_dma_reset_done(), consistent with the
other wait helpers in this driver. On timeout, log an error, de-assert
the reset bit, and return -ETIMEDOUT. Move ctrl->started to the
successful path so a failed start does not leave the trace marked as
active.

Fixes: ff0de066b463 ("hwtracing: hisi_ptt: Add trace function support for HiSilicon PCIe Tune and Trace device")
Cc: stable@vger.kernel.org
Signed-off-by: Sanman Pradhan <psanman@juniper.net>
---
v2:
  - Return bool for consistency with other wait helpers
  - Add pci_err() on timeout
  - De-assert RST before returning on timeout
  - Move ctrl->started to the successful path

 drivers/hwtracing/ptt/hisi_ptt.c | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/drivers/hwtracing/ptt/hisi_ptt.c b/drivers/hwtracing/ptt/hisi_ptt.c
index 94c371c491357..b5d851281fbf0 100644
--- a/drivers/hwtracing/ptt/hisi_ptt.c
+++ b/drivers/hwtracing/ptt/hisi_ptt.c
@@ -171,13 +171,13 @@ static bool hisi_ptt_wait_trace_hw_idle(struct hisi_ptt *hisi_ptt)
 					  HISI_PTT_WAIT_TRACE_TIMEOUT_US);
 }
 
-static void hisi_ptt_wait_dma_reset_done(struct hisi_ptt *hisi_ptt)
+static bool hisi_ptt_wait_dma_reset_done(struct hisi_ptt *hisi_ptt)
 {
 	u32 val;
 
-	readl_poll_timeout_atomic(hisi_ptt->iobase + HISI_PTT_TRACE_WR_STS,
-				  val, !val, HISI_PTT_RESET_POLL_INTERVAL_US,
-				  HISI_PTT_RESET_TIMEOUT_US);
+	return !readl_poll_timeout_atomic(hisi_ptt->iobase + HISI_PTT_TRACE_WR_STS,
+					  val, !val, HISI_PTT_RESET_POLL_INTERVAL_US,
+					  HISI_PTT_RESET_TIMEOUT_US);
 }
 
 static void hisi_ptt_trace_end(struct hisi_ptt *hisi_ptt)
@@ -202,14 +202,18 @@ static int hisi_ptt_trace_start(struct hisi_ptt *hisi_ptt)
 		return -EBUSY;
 	}
 
-	ctrl->started = true;
-
 	/* Reset the DMA before start tracing */
 	val = readl(hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
 	val |= HISI_PTT_TRACE_CTRL_RST;
 	writel(val, hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
 
-	hisi_ptt_wait_dma_reset_done(hisi_ptt);
+	if (!hisi_ptt_wait_dma_reset_done(hisi_ptt)) {
+		pci_err(hisi_ptt->pdev, "timed out waiting for DMA reset\n");
+		val = readl(hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
+		val &= ~HISI_PTT_TRACE_CTRL_RST;
+		writel(val, hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
+		return -ETIMEDOUT;
+	}
 
 	val = readl(hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
 	val &= ~HISI_PTT_TRACE_CTRL_RST;
@@ -234,6 +238,8 @@ static int hisi_ptt_trace_start(struct hisi_ptt *hisi_ptt)
 	if (!hisi_ptt->trace_ctrl.is_port)
 		val |= HISI_PTT_TRACE_CTRL_FILTER_MODE;
 
+	ctrl->started = true;
+
 	/* Start the Trace */
 	val |= HISI_PTT_TRACE_CTRL_EN;
 	writel(val, hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
-- 
2.34.1

Re: [PATCH v2 1/2] hwtracing: hisi_ptt: Propagate DMA reset timeout in trace_start()

[Yicong Yang]

+cc Suzuki and Sizhe..

On 2026/4/15 01:25, Pradhan, Sanman wrote:
> From: Sanman Pradhan <psanman@juniper.net>
>
> hisi_ptt_wait_dma_reset_done() discards the return value of
> readl_poll_timeout_atomic(). If the DMA engine does not complete its
> reset within the timeout, hisi_ptt_trace_start() proceeds to start
> tracing regardless.
>
> Return a bool from hisi_ptt_wait_dma_reset_done(), consistent with the
> other wait helpers in this driver. On timeout, log an error, de-assert
> the reset bit, and return -ETIMEDOUT. Move ctrl->started to the
> successful path so a failed start does not leave the trace marked as
> active.
>
> Fixes: ff0de066b463 ("hwtracing: hisi_ptt: Add trace function support for HiSilicon PCIe Tune and Trace device")
> Cc: stable@vger.kernel.org
> Signed-off-by: Sanman Pradhan <psanman@juniper.net>

looks good to me.

Reviewed-by: Yicong Yang <yangyccccc@gmail.com>

I see the Suzuki has sent out the PR for 7.1, so this may wait after the merge window...

thanks.

> ---
> v2:
>   - Return bool for consistency with other wait helpers
>   - Add pci_err() on timeout
>   - De-assert RST before returning on timeout
>   - Move ctrl->started to the successful path
>
>  drivers/hwtracing/ptt/hisi_ptt.c | 20 +++++++++++++-------
>  1 file changed, 13 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/hwtracing/ptt/hisi_ptt.c b/drivers/hwtracing/ptt/hisi_ptt.c
> index 94c371c491357..b5d851281fbf0 100644
> --- a/drivers/hwtracing/ptt/hisi_ptt.c
> +++ b/drivers/hwtracing/ptt/hisi_ptt.c
> @@ -171,13 +171,13 @@ static bool hisi_ptt_wait_trace_hw_idle(struct hisi_ptt *hisi_ptt)
>  					  HISI_PTT_WAIT_TRACE_TIMEOUT_US);
>  }
>  
> -static void hisi_ptt_wait_dma_reset_done(struct hisi_ptt *hisi_ptt)
> +static bool hisi_ptt_wait_dma_reset_done(struct hisi_ptt *hisi_ptt)
>  {
>  	u32 val;
>  
> -	readl_poll_timeout_atomic(hisi_ptt->iobase + HISI_PTT_TRACE_WR_STS,
> -				  val, !val, HISI_PTT_RESET_POLL_INTERVAL_US,
> -				  HISI_PTT_RESET_TIMEOUT_US);
> +	return !readl_poll_timeout_atomic(hisi_ptt->iobase + HISI_PTT_TRACE_WR_STS,
> +					  val, !val, HISI_PTT_RESET_POLL_INTERVAL_US,
> +					  HISI_PTT_RESET_TIMEOUT_US);
>  }
>  
>  static void hisi_ptt_trace_end(struct hisi_ptt *hisi_ptt)
> @@ -202,14 +202,18 @@ static int hisi_ptt_trace_start(struct hisi_ptt *hisi_ptt)
>  		return -EBUSY;
>  	}
>  
> -	ctrl->started = true;
> -
>  	/* Reset the DMA before start tracing */
>  	val = readl(hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
>  	val |= HISI_PTT_TRACE_CTRL_RST;
>  	writel(val, hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
>  
> -	hisi_ptt_wait_dma_reset_done(hisi_ptt);
> +	if (!hisi_ptt_wait_dma_reset_done(hisi_ptt)) {
> +		pci_err(hisi_ptt->pdev, "timed out waiting for DMA reset\n");
> +		val = readl(hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
> +		val &= ~HISI_PTT_TRACE_CTRL_RST;
> +		writel(val, hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
> +		return -ETIMEDOUT;
> +	}
>  
>  	val = readl(hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);
>  	val &= ~HISI_PTT_TRACE_CTRL_RST;
> @@ -234,6 +238,8 @@ static int hisi_ptt_trace_start(struct hisi_ptt *hisi_ptt)
>  	if (!hisi_ptt->trace_ctrl.is_port)
>  		val |= HISI_PTT_TRACE_CTRL_FILTER_MODE;
>  
> +	ctrl->started = true;
> +
>  	/* Start the Trace */
>  	val |= HISI_PTT_TRACE_CTRL_EN;
>  	writel(val, hisi_ptt->iobase + HISI_PTT_TRACE_CTRL);

[PATCH v2 2/2] hwtracing: hisi_ptt: Remove unnecessary trace buffer zeroing in trace_start()

[Pradhan, Sanman]

From: Sanman Pradhan <psanman@juniper.net>

hisi_ptt_trace_start() clears all four trace buffers before enabling
tracing.

This is unnecessary. On trace stop, hisi_ptt_update_aux() copies only
the number of bytes reported in HISI_PTT_TRACE_WR_STS. On buffer-full
interrupts, it copies a full completed buffer. In both cases the driver
only consumes data written by hardware.

Remove the buffer clearing from the trace start path.

Signed-off-by: Sanman Pradhan <psanman@juniper.net>
Reviewed-by: Yicong Yang <yangyccccc@gmail.com>
---
v2:
  - No changes

 drivers/hwtracing/ptt/hisi_ptt.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/drivers/hwtracing/ptt/hisi_ptt.c b/drivers/hwtracing/ptt/hisi_ptt.c
index b5d851281fbf0..a8f6986c8e1f7 100644
--- a/drivers/hwtracing/ptt/hisi_ptt.c
+++ b/drivers/hwtracing/ptt/hisi_ptt.c
@@ -194,7 +194,6 @@ static int hisi_ptt_trace_start(struct hisi_ptt *hisi_ptt)
 {
 	struct hisi_ptt_trace_ctrl *ctrl = &hisi_ptt->trace_ctrl;
 	u32 val;
-	int i;
 
 	/* Check device idle before start trace */
 	if (!hisi_ptt_wait_trace_hw_idle(hisi_ptt)) {
@@ -222,10 +221,6 @@ static int hisi_ptt_trace_start(struct hisi_ptt *hisi_ptt)
 	/* Reset the index of current buffer */
 	hisi_ptt->trace_ctrl.buf_index = 0;
 
-	/* Zero the trace buffers */
-	for (i = 0; i < HISI_PTT_TRACE_BUF_CNT; i++)
-		memset(ctrl->trace_buf[i].addr, 0, HISI_PTT_TRACE_BUF_SIZE);
-
 	/* Clear the interrupt status */
 	writel(HISI_PTT_TRACE_INT_STAT_MASK, hisi_ptt->iobase + HISI_PTT_TRACE_INT_STAT);
 	writel(0, hisi_ptt->iobase + HISI_PTT_TRACE_INT_MASK);
-- 
2.34.1

Re: [PATCH v2 0/2] hwtracing: hisi_ptt: Fix reset timeout handling and clean up trace start

[Jonathan Cameron]

On Tue, 14 Apr 2026 17:25:03 +0000
"Pradhan, Sanman" <sanman.pradhan@hpe.com> wrote:

> From: Sanman Pradhan <psanman@juniper.net>
+CC Sizhe Liu

> 
> Patch 1: Propagate the DMA reset timeout error from
>   hisi_ptt_wait_dma_reset_done() instead of discarding it. De-assert
>   the reset bit and log an error on timeout. Move ctrl->started to the
>   successful path so a failed start does not leave the trace marked as
>   active.
> 
> Patch 2: Remove the unnecessary 16 MiB memset of trace buffers in
>   hisi_ptt_trace_start(). The driver only copies data that hardware has
>   written, so the zeroing is not needed.
> 
> Changes since v1:
>   - Patch 1: Return bool from hisi_ptt_wait_dma_reset_done() for
>     consistency with the other wait helpers
>   - Patch 1: Add pci_err() on timeout
>   - Patch 1: De-assert RST before returning on timeout
>   - Patch 1: Move ctrl->started to the successful path
>   - Dropped "Use the passed buffer index in hisi_ptt_update_aux()" patch
>   - Patch 2 is unchanged
> 
> Sanman Pradhan (2):
>   hwtracing: hisi_ptt: Propagate DMA reset timeout in trace_start()
>   hwtracing: hisi_ptt: Remove unnecessary trace buffer zeroing in
>     trace_start()
> 
>  drivers/hwtracing/ptt/hisi_ptt.c | 25 +++++++++++++------------
>  1 file changed, 13 insertions(+), 12 deletions(-)
>