[PATCH v2 0/4] Improve SPI target mode support and error handling

[PATCH v2 0/4] Improve SPI target mode support and error handling

[Praveen Talari]

Enhance the Qualcomm GENI SPI driver's target mode support and improve 
error handling for serial engine operations.

The current implementation has issues with incorrect controller allocation
for target mode, missing abort sequence execution during error recovery,
and lack of graceful abort mechanism for target operations. These problems
can lead to serial engine undefined behavior and improper resource
cleanup.

Fix controller allocation to use proper target APIs, ensure abort sequence
always executes for error recovery, and add target abort support for 
graceful transfer cancellation.

Praveen Talari (4):
  spi: geni-qcom: Improve target mode allocation by using proper
    allocation functions
  spi: geni-qcom: Fix abort sequence execution for serial engine errors
  spi: qcom-geni: Drop unused msg parameter from timeout handlers
  spi: geni-qcom: Add target abort support
---
v1->v2
- Added a new patch to remove unused param.

 drivers/spi/spi-geni-qcom.c | 56 ++++++++++++++++++++++---------------
 1 file changed, 33 insertions(+), 23 deletions(-)


base-commit: 193579fe01389bc21aff0051d13f24e8ea95b47d
-- 
2.34.1

[PATCH v2 1/4] spi: geni-qcom: Improve target mode allocation by using proper allocation functions

[Praveen Talari]

The current implementation always allocates a host controller and sets the
target flag later when the "spi-slave" device tree property is present.
This approach is suboptimal as it doesn't utilize the dedicated allocation
functions designed for target mode.

Use devm_spi_alloc_target() when "spi-slave" device tree property is
present, otherwise use devm_spi_alloc_host(). This replaces the previous
approach of always allocating a host controller and setting target flag
later.

Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
---
v1->v2
konrad
- Placed changes where host is allocated.
---
 drivers/spi/spi-geni-qcom.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/spi/spi-geni-qcom.c b/drivers/spi/spi-geni-qcom.c
index 0e5fd9df1a8f..f886a9ba56e6 100644
--- a/drivers/spi/spi-geni-qcom.c
+++ b/drivers/spi/spi-geni-qcom.c
@@ -1033,7 +1033,11 @@ static int spi_geni_probe(struct platform_device *pdev)
 	if (IS_ERR(clk))
 		return PTR_ERR(clk);
 
-	spi = devm_spi_alloc_host(dev, sizeof(*mas));
+	if (device_property_read_bool(dev, "spi-slave"))
+		spi = devm_spi_alloc_target(dev, sizeof(*mas));
+	else
+		spi = devm_spi_alloc_host(dev, sizeof(*mas));
+
 	if (!spi)
 		return -ENOMEM;
 
@@ -1087,9 +1091,6 @@ static int spi_geni_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
-	if (device_property_read_bool(&pdev->dev, "spi-slave"))
-		spi->target = true;
-
 	/* Set the bus quota to a reasonable value for register access */
 	mas->se.icc_paths[GENI_TO_CORE].avg_bw = Bps_to_icc(CORE_2X_50_MHZ);
 	mas->se.icc_paths[CPU_TO_GENI].avg_bw = GENI_DEFAULT_BW;
-- 
2.34.1

[PATCH v2 2/4] spi: geni-qcom: Fix abort sequence execution for serial engine errors

[Praveen Talari]

The driver currently skips the abort sequence for target mode when serial
engine errors occur. This leads to improper error recovery as the serial
engine may remain in an undefined state without proper cleanup, potentially
causing subsequent operations to fail or behave unpredictably.

Fix this by ensuring the abort sequence and DMA reset always execute during
error recovery, as both are required for proper serial engine error
handling.

Co-developed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
---
v1->v2
Konrad
- Updated code changes as suggested by Konrad
---
 drivers/spi/spi-geni-qcom.c | 24 ++++++++++--------------
 1 file changed, 10 insertions(+), 14 deletions(-)

diff --git a/drivers/spi/spi-geni-qcom.c b/drivers/spi/spi-geni-qcom.c
index f886a9ba56e6..e48aea7ce681 100644
--- a/drivers/spi/spi-geni-qcom.c
+++ b/drivers/spi/spi-geni-qcom.c
@@ -161,24 +161,20 @@ static void handle_se_timeout(struct spi_controller *spi,
 	xfer = mas->cur_xfer;
 	mas->cur_xfer = NULL;
 
-	if (spi->target) {
-		/*
-		 * skip CMD Cancel sequnece since spi target
-		 * doesn`t support CMD Cancel sequnece
-		 */
+	/* The controller doesn't support the Cancel commnand in target mode */
+	if (!spi->target) {
+		reinit_completion(&mas->cancel_done);
+		geni_se_cancel_m_cmd(se);
+
 		spin_unlock_irq(&mas->lock);
-		goto reset_if_dma;
-	}
 
-	reinit_completion(&mas->cancel_done);
-	geni_se_cancel_m_cmd(se);
-	spin_unlock_irq(&mas->lock);
+		time_left = wait_for_completion_timeout(&mas->cancel_done, HZ);
+		if (time_left)
+			goto reset_if_dma;
 
-	time_left = wait_for_completion_timeout(&mas->cancel_done, HZ);
-	if (time_left)
-		goto reset_if_dma;
+		spin_lock_irq(&mas->lock);
+	}
 
-	spin_lock_irq(&mas->lock);
 	reinit_completion(&mas->abort_done);
 	geni_se_abort_m_cmd(se);
 	spin_unlock_irq(&mas->lock);
-- 
2.34.1

[PATCH v2 3/4] spi: qcom-geni: Drop unused msg parameter from timeout handlers

[Praveen Talari]

The timeout helpers (handle_se_timeout() and handle_gpi_timeout())
took a 'struct spi_message *msg' that was never used. Remove this
parameter and adjust the callers to simplify the interfaces.

Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
---
 drivers/spi/spi-geni-qcom.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/drivers/spi/spi-geni-qcom.c b/drivers/spi/spi-geni-qcom.c
index e48aea7ce681..5077dc041e3a 100644
--- a/drivers/spi/spi-geni-qcom.c
+++ b/drivers/spi/spi-geni-qcom.c
@@ -146,8 +146,7 @@ static int get_spi_clk_cfg(unsigned int speed_hz,
 	return ret;
 }
 
-static void handle_se_timeout(struct spi_controller *spi,
-			      struct spi_message *msg)
+static void handle_se_timeout(struct spi_controller *spi)
 {
 	struct spi_geni_master *mas = spi_controller_get_devdata(spi);
 	unsigned long time_left;
@@ -222,7 +221,7 @@ static void handle_se_timeout(struct spi_controller *spi,
 	}
 }
 
-static void handle_gpi_timeout(struct spi_controller *spi, struct spi_message *msg)
+static void handle_gpi_timeout(struct spi_controller *spi)
 {
 	struct spi_geni_master *mas = spi_controller_get_devdata(spi);
 
@@ -237,10 +236,10 @@ static void spi_geni_handle_err(struct spi_controller *spi, struct spi_message *
 	switch (mas->cur_xfer_mode) {
 	case GENI_SE_FIFO:
 	case GENI_SE_DMA:
-		handle_se_timeout(spi, msg);
+		handle_se_timeout(spi);
 		break;
 	case GENI_GPI_DMA:
-		handle_gpi_timeout(spi, msg);
+		handle_gpi_timeout(spi);
 		break;
 	default:
 		dev_err(mas->dev, "Abort on Mode:%d not supported", mas->cur_xfer_mode);
-- 
2.34.1

[PATCH v2 4/4] spi: geni-qcom: Add target abort support

[Praveen Talari]

SPI target mode currently lacks a mechanism to gracefully abort ongoing
transfers when the client or core needs to cancel active transactions.

Implement spi_geni_target_abort() to handle aborting SPI target
operations when the client and core want to cancel ongoing transfers.
This provides a mechanism for graceful termination of active SPI
transactions in target mode.

Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
---
v1->v2
- Removed unused param from time out handlers.
---
 drivers/spi/spi-geni-qcom.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/drivers/spi/spi-geni-qcom.c b/drivers/spi/spi-geni-qcom.c
index 5077dc041e3a..43ce47f2454c 100644
--- a/drivers/spi/spi-geni-qcom.c
+++ b/drivers/spi/spi-geni-qcom.c
@@ -1003,6 +1003,17 @@ static irqreturn_t geni_spi_isr(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
+static int spi_geni_target_abort(struct spi_controller *spi)
+{
+	if (!spi->cur_msg)
+		return 0;
+
+	handle_se_timeout(spi);
+	spi_finalize_current_transfer(spi);
+
+	return 0;
+}
+
 static int spi_geni_probe(struct platform_device *pdev)
 {
 	int ret, irq;
@@ -1076,6 +1087,9 @@ static int spi_geni_probe(struct platform_device *pdev)
 	init_completion(&mas->rx_reset_done);
 	spin_lock_init(&mas->lock);
 
+	if (spi->target)
+		spi->target_abort = spi_geni_target_abort;
+
 	ret = geni_icc_get(&mas->se, NULL);
 	if (ret)
 		return ret;
-- 
2.34.1

Re: [PATCH v2 1/4] spi: geni-qcom: Improve target mode allocation by using proper allocation functions

[Konrad Dybcio]

On 2/4/26 5:28 PM, Praveen Talari wrote:
> The current implementation always allocates a host controller and sets the
> target flag later when the "spi-slave" device tree property is present.
> This approach is suboptimal as it doesn't utilize the dedicated allocation
> functions designed for target mode.
> 
> Use devm_spi_alloc_target() when "spi-slave" device tree property is
> present, otherwise use devm_spi_alloc_host(). This replaces the previous
> approach of always allocating a host controller and setting target flag
> later.
> 
> Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
> ---

Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Konrad

Re: [PATCH v2 3/4] spi: qcom-geni: Drop unused msg parameter from timeout handlers

[Konrad Dybcio]

On 2/4/26 5:28 PM, Praveen Talari wrote:
> The timeout helpers (handle_se_timeout() and handle_gpi_timeout())
> took a 'struct spi_message *msg' that was never used. Remove this
> parameter and adjust the callers to simplify the interfaces.
> 
> Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
> ---

Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Konrad

Re: [PATCH v2 4/4] spi: geni-qcom: Add target abort support

[Konrad Dybcio]

On 2/4/26 5:28 PM, Praveen Talari wrote:
> SPI target mode currently lacks a mechanism to gracefully abort ongoing
> transfers when the client or core needs to cancel active transactions.
> 
> Implement spi_geni_target_abort() to handle aborting SPI target
> operations when the client and core want to cancel ongoing transfers.
> This provides a mechanism for graceful termination of active SPI
> transactions in target mode.
> 
> Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
> ---


> v1->v2
> - Removed unused param from time out handlers.
> ---
>  drivers/spi/spi-geni-qcom.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/drivers/spi/spi-geni-qcom.c b/drivers/spi/spi-geni-qcom.c
> index 5077dc041e3a..43ce47f2454c 100644
> --- a/drivers/spi/spi-geni-qcom.c
> +++ b/drivers/spi/spi-geni-qcom.c
> @@ -1003,6 +1003,17 @@ static irqreturn_t geni_spi_isr(int irq, void *data)
>  	return IRQ_HANDLED;
>  }
>  
> +static int spi_geni_target_abort(struct spi_controller *spi)
> +{
> +	if (!spi->cur_msg)
> +		return 0;
> +
> +	handle_se_timeout(spi);
> +	spi_finalize_current_transfer(spi);
> +
> +	return 0;
> +}
> +
>  static int spi_geni_probe(struct platform_device *pdev)
>  {
>  	int ret, irq;
> @@ -1076,6 +1087,9 @@ static int spi_geni_probe(struct platform_device *pdev)
>  	init_completion(&mas->rx_reset_done);
>  	spin_lock_init(&mas->lock);
>  
> +	if (spi->target)
> +		spi->target_abort = spi_geni_target_abort;

The same check is made in core:

spi_target_abort()
-> spi_controller_is_target()

So I'm assuming the intention was to allow assigning the func pointer
indiscriminately. Other drivers seem to do it both ways.

Mark, any specific preference?

Konrad

Re: [PATCH v2 4/4] spi: geni-qcom: Add target abort support

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 660 bytes --]

On Wed, Feb 04, 2026 at 05:41:46PM +0100, Konrad Dybcio wrote:
> On 2/4/26 5:28 PM, Praveen Talari wrote:

> > +	if (spi->target)
> > +		spi->target_abort = spi_geni_target_abort;

> The same check is made in core:

> spi_target_abort()
> -> spi_controller_is_target()

> So I'm assuming the intention was to allow assigning the func pointer
> indiscriminately. Other drivers seem to do it both ways.

> Mark, any specific preference?

Not really TBH, it's redundant in the driver but if the people working
on the driver find having the check there saves them having to check
that the core does the right thing that's fine.  It's not like this is a
fast path.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH v2 4/4] spi: geni-qcom: Add target abort support

[Konrad Dybcio]

On 2/4/26 5:57 PM, Mark Brown wrote:
> On Wed, Feb 04, 2026 at 05:41:46PM +0100, Konrad Dybcio wrote:
>> On 2/4/26 5:28 PM, Praveen Talari wrote:
> 
>>> +	if (spi->target)
>>> +		spi->target_abort = spi_geni_target_abort;
> 
>> The same check is made in core:
> 
>> spi_target_abort()
>> -> spi_controller_is_target()
> 
>> So I'm assuming the intention was to allow assigning the func pointer
>> indiscriminately. Other drivers seem to do it both ways.
> 
>> Mark, any specific preference?
> 
> Not really TBH, it's redundant in the driver but if the people working
> on the driver find having the check there saves them having to check
> that the core does the right thing that's fine.  It's not like this is a
> fast path.

Thanks

Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Konrad

Re: [PATCH v2 2/4] spi: geni-qcom: Fix abort sequence execution for serial engine errors

[Konrad Dybcio]

On 2/4/26 5:28 PM, Praveen Talari wrote:
> The driver currently skips the abort sequence for target mode when serial
> engine errors occur. This leads to improper error recovery as the serial
> engine may remain in an undefined state without proper cleanup, potentially
> causing subsequent operations to fail or behave unpredictably.
> 
> Fix this by ensuring the abort sequence and DMA reset always execute during
> error recovery, as both are required for proper serial engine error
> handling.
> 
> Co-developed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Well I never left such a tag, but let's say I won't object too much..

(it potentially carries legal meaning)

it's fine not to leave any sort of attribution for small changes or
suggestions you receive during review, perhaps suggested-by if someone
says "oh btw could you please look into fixing x in xyz.c too"

for the code:

Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Konrad

Re: [PATCH v2 0/4] Improve SPI target mode support and error handling

[Mark Brown]

On Wed, 04 Feb 2026 21:58:50 +0530, Praveen Talari wrote:
> Enhance the Qualcomm GENI SPI driver's target mode support and improve
> error handling for serial engine operations.
> 
> The current implementation has issues with incorrect controller allocation
> for target mode, missing abort sequence execution during error recovery,
> and lack of graceful abort mechanism for target operations. These problems
> can lead to serial engine undefined behavior and improper resource
> cleanup.
> 
> [...]

Applied to

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi.git for-next

Thanks!

[1/4] spi: geni-qcom: Improve target mode allocation by using proper allocation functions
      commit: 50464497b5874b7c2f3ab991cec81e916d40743c
[2/4] spi: geni-qcom: Fix abort sequence execution for serial engine errors
      commit: 96e041647bb0f9d92f95df1d69cb7442d7408b79
[3/4] spi: qcom-geni: Drop unused msg parameter from timeout handlers
      commit: 9934383f98c91d8dfdf16d0e980431041c2d17ee
[4/4] spi: geni-qcom: Add target abort support
      commit: 6516169fa7f6110ce42a5958c921255f2d0968bc

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark