Skip to contentSkip to conversation

i3c: mipi-i3c-hci: Allow only relevant INTR_STATUS bit updates

PATCH Applied Jarkko Nikula 3 participants 9 messages Apr 9, 2025 – May 15, 2025 I3C

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/5 i3c: mipi-i3c-hci: Allow only relevant INTR_STATUS bit updates

Since MIPI I3C HCI specification version v0.8 INTR_STATUS bits 9:0 are reserved. Version v0.5 has bits 9 and 5:0 in use but not handled by the current driver code and not needed in DMA transfers.

PIO transfers with v0.5 would require changes to both core.c: i3c_hci_irq_handler() and pio.c: hci_pio_irq_handler() though.

For these reasons don't enable signal updates from INTR_STATUS bits 9:0.

It allow to get rid of "unexpected INTR_STATUS" error messages on old v0.5 IP version and is a no-op for later versions starting from v0.8.

---
v3: Changed the last commit log sentence and clarified more the comment about INTR_STATUS_ENABLE signal updates according to Frank Li's suggestion.

---
 drivers/i3c/master/mipi-i3c-hci/core.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +7 -2
index a71226d7ca59..ba7aa6bbcec5 100644
@@ -699,9 +699,14 @@ static int i3c_hci_init(struct i3c_hci *hci)
 	if (ret)
 		return -ENXIO;
 
-	/* Disable all interrupts and allow all signal updates */
+	/* Disable all interrupts */
 	reg_write(INTR_SIGNAL_ENABLE, 0x0);
-	reg_write(INTR_STATUS_ENABLE, 0xffffffff);
+	/*
+	 * Only allow bit 31:10 signal updates because
+	 * Bit 0:9 are reserved in IP version >= 0.8
+	 * Bit 0:5 are defined in IP version < 0.8 but not handled by PIO code
+	 */
+	reg_write(INTR_STATUS_ENABLE, GENMASK(31, 10));
 
 	/* Make sure our data ordering fits the host's */
 	regval = reg_read(HC_CONTROL);
Signature · 7 lines
2.47.2


-- 
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
#1 lore raw

PATCH v3 2/5 i3c: mipi-i3c-hci: Fix handling status of i3c_hci_irq_handler()

Return IRQ_HANDLED from the i3c_hci_irq_handler() only if some INTR_STATUS bit was set or if DMA/PIO handler handled it.

Currently it returns IRQ_HANDLED in case INTR_STATUS is zero and IO handler returns false. Which could be the case if interrupt comes from other device or is spurious.

---
v3: Added Reviewed-by

---
 drivers/i3c/master/mipi-i3c-hci/core.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +3 -3
index ba7aa6bbcec5..780e9db7e21e 100644
@@ -594,6 +594,7 @@ static irqreturn_t i3c_hci_irq_handler(int irq, void *dev_id)
 
 	if (val) {
 		reg_write(INTR_STATUS, val);
+		result = IRQ_HANDLED;
 	}
 
 	if (val & INTR_HC_RESET_CANCEL) {
@@ -605,12 +606,11 @@ static irqreturn_t i3c_hci_irq_handler(int irq, void *dev_id)
 		val &= ~INTR_HC_INTERNAL_ERR;
 	}
 
-	hci->io->irq_handler(hci);
+	if (hci->io->irq_handler(hci))
+		result = IRQ_HANDLED;
 
 	if (val)
 		dev_err(&hci->master.dev, "unexpected INTR_STATUS %#x\n", val);
-	else
-		result = IRQ_HANDLED;
 
 	return result;
 }
Signature · 7 lines
2.47.2


-- 
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
#2 lore raw

PATCH v3 3/5 i3c: mipi-i3c-hci: Clear INTR_STATUS unconditionally

In reply to Jarkko Nikula

Status fields in INTR_STATUS register are write 1 to clear so do it unconditionally and move clearing of them out of an if block.

---
 drivers/i3c/master/mipi-i3c-hci/core.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +2 -3
index 780e9db7e21e..4c4100d2d9af 100644
@@ -590,12 +590,11 @@ static irqreturn_t i3c_hci_irq_handler(int irq, void *dev_id)
 	u32 val;
 
 	val = reg_read(INTR_STATUS);
+	reg_write(INTR_STATUS, val);
 	DBG("INTR_STATUS = %#x", val);
 
-	if (val) {
-		reg_write(INTR_STATUS, val);
+	if (val)
 		result = IRQ_HANDLED;
-	}
 
 	if (val & INTR_HC_RESET_CANCEL) {
 		DBG("cancelled reset");
Signature · 7 lines
2.47.2


-- 
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
reviewer Reviewed-by on v3 3/5
#3 lore raw

Re: [PATCH v3 3/5] i3c: mipi-i3c-hci: Clear INTR_STATUS unconditionally

Quoted 5 lines · Jarkko Nikula

Status fields in INTR_STATUS register are write 1 to clear so do it unconditionally and move clearing of them out of an if block.

Quoted 31 lines
---
 drivers/i3c/master/mipi-i3c-hci/core.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +2 -4
index 780e9db7e21e..4c4100d2d9af 100644
@@ -590,12 +590,11 @@ static irqreturn_t i3c_hci_irq_handler(int irq, void *dev_id)
 	u32 val;
 
 	val = reg_read(INTR_STATUS);
+	reg_write(INTR_STATUS, val);
 	DBG("INTR_STATUS = %#x", val);
 
-	if (val) {
-		reg_write(INTR_STATUS, val);
+	if (val)
 		result = IRQ_HANDLED;
-	}
 
 	if (val & INTR_HC_RESET_CANCEL) {
 		DBG("cancelled reset");
--

2.47.2

--
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c

Signature · 3 lines
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
#4 lore raw

PATCH v3 4/5 i3c: mipi-i3c-hci: Change name of INTR_STATUS bit 11

In reply to Jarkko Nikula

INTR_STATUS bit 11 INTR_HC_RESET_CANCEL was probably projected for the MIPI I3C HCI specification version 2 but was not ever implemented.

This bit is first time specified in the v1.2 as HC_SEQ_CANCEL_STAT "Host Controller Cancelled Transaction Sequence". Update the definition and debug print of it accordingly.

While at it, change DBG() print to dev_dbg().

---
v3: Added Reviewed-by

---
 drivers/i3c/master/mipi-i3c-hci/core.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +5 -4
index 4c4100d2d9af..0bb74173ca94 100644
@@ -78,7 +78,7 @@
 #define INTR_SIGNAL_ENABLE		0x28
 #define INTR_FORCE			0x2c
 #define INTR_HC_CMD_SEQ_UFLOW_STAT	BIT(12)	/* Cmd Sequence Underflow */
-#define INTR_HC_RESET_CANCEL		BIT(11)	/* HC Cancelled Reset */
+#define INTR_HC_SEQ_CANCEL		BIT(11)	/* HC Cancelled Transaction Sequence */
 #define INTR_HC_INTERNAL_ERR		BIT(10)	/* HC Internal Error */
 
 #define DAT_SECTION			0x30	/* Device Address Table */
@@ -596,9 +596,10 @@ static irqreturn_t i3c_hci_irq_handler(int irq, void *dev_id)
 	if (val)
 		result = IRQ_HANDLED;
 
-	if (val & INTR_HC_RESET_CANCEL) {
-		DBG("cancelled reset");
-		val &= ~INTR_HC_RESET_CANCEL;
+	if (val & INTR_HC_SEQ_CANCEL) {
+		dev_dbg(&hci->master.dev,
+			"Host Controller Cancelled Transaction Sequence\n");
+		val &= ~INTR_HC_SEQ_CANCEL;
 	}
 	if (val & INTR_HC_INTERNAL_ERR) {
 		dev_err(&hci->master.dev, "Host Controller Internal Error\n");
Signature · 7 lines
2.47.2


-- 
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
#5 lore raw

PATCH v3 5/5 i3c: mipi-i3c-hci: Move unexpected INTR_STATUS print before IO handler

In reply to Jarkko Nikula

Move "unexpected INTR_STATUS" error print before calling the IO handler as it is more consistent that way. Otherwise it may be confusing if generic interrupt related prints are mixed with IO handler prints.

Since this error print is more indication of missing code rather than runtime error downgrade it to dev_warn_once().

---
v2: Added Reviewed-by

---
 drivers/i3c/master/mipi-i3c-hci/core.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +4 -3
index 0bb74173ca94..bc4538694540 100644
@@ -606,12 +606,13 @@ static irqreturn_t i3c_hci_irq_handler(int irq, void *dev_id)
 		val &= ~INTR_HC_INTERNAL_ERR;
 	}
 
+	if (val)
+		dev_warn_once(&hci->master.dev,
+			      "unexpected INTR_STATUS %#x\n", val);
+
 	if (hci->io->irq_handler(hci))
 		result = IRQ_HANDLED;
 
-	if (val)
-		dev_err(&hci->master.dev, "unexpected INTR_STATUS %#x\n", val);
-
 	return result;
 }
 
Signature · 7 lines
2.47.2


-- 
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
#6 lore raw
In reply to Jarkko Nikula

Hi Frank

Quoted 12 lines · Jarkko Nikula

Since MIPI I3C HCI specification version v0.8 INTR_STATUS bits 9:0 are reserved. Version v0.5 has bits 9 and 5:0 in use but not handled by the current driver code and not needed in DMA transfers.

PIO transfers with v0.5 would require changes to both core.c: i3c_hci_irq_handler() and pio.c: hci_pio_irq_handler() though.

For these reasons don't enable signal updates from INTR_STATUS bits 9:0.

It allow to get rid of "unexpected INTR_STATUS" error messages on old v0.5 IP version and is a no-op for later versions starting from v0.8.

Quoted 8 lines

---
v3: Changed the last commit log sentence and clarified more the comment about INTR_STATUS_ENABLE signal updates according to Frank Li's suggestion.

---
  drivers/i3c/master/mipi-i3c-hci/core.c | 9 +++++++--
  1 file changed, 7 insertions(+), 2 deletions(-)

Was this and patch 3/5 now ok in your point of view?

Signature · 3 lines
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
reviewer Reviewed-by on v3 1/5
#7 lore raw
In reply to Jarkko Nikula
Quoted 7 lines · Jarkko Nikula

Since MIPI I3C HCI specification version v0.8 INTR_STATUS bits 9:0 are reserved. Version v0.5 has bits 9 and 5:0 in use but not handled by the current driver code and not needed in DMA transfers.

PIO transfers with v0.5 would require changes to both core.c: i3c_hci_irq_handler() and pio.c: hci_pio_irq_handler() though.

Quoted 6 lines

For these reasons don't enable signal updates from INTR_STATUS bits 9:0.

It allow to get rid of "unexpected INTR_STATUS" error messages on old v0.5 IP version and is a no-op for later versions starting from v0.8.

Quoted 36 lines

---
v3: Changed the last commit log sentence and clarified more the comment about INTR_STATUS_ENABLE signal updates according to Frank Li's suggestion.

---
 drivers/i3c/master/mipi-i3c-hci/core.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/i3c/master/mipi-i3c-hci/core.c +7 -3
index a71226d7ca59..ba7aa6bbcec5 100644
@@ -699,9 +699,14 @@ static int i3c_hci_init(struct i3c_hci *hci)
 	if (ret)
 		return -ENXIO;
 
-	/* Disable all interrupts and allow all signal updates */
+	/* Disable all interrupts */
 	reg_write(INTR_SIGNAL_ENABLE, 0x0);
-	reg_write(INTR_STATUS_ENABLE, 0xffffffff);
+	/*
+	 * Only allow bit 31:10 signal updates because
+	 * Bit 0:9 are reserved in IP version >= 0.8
+	 * Bit 0:5 are defined in IP version < 0.8 but not handled by PIO code
+	 */
+	reg_write(INTR_STATUS_ENABLE, GENMASK(31, 10));
 
 	/* Make sure our data ordering fits the host's */
 	regval = reg_read(HC_CONTROL);
--

2.47.2

--
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c

Signature · 3 lines
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c
maintainer Applied
#8 lore raw
In reply to Jarkko Nikula
Quoted 8 lines · Jarkko Nikula

Since MIPI I3C HCI specification version v0.8 INTR_STATUS bits 9:0 are reserved. Version v0.5 has bits 9 and 5:0 in use but not handled by the current driver code and not needed in DMA transfers.

PIO transfers with v0.5 would require changes to both core.c: i3c_hci_irq_handler() and pio.c: hci_pio_irq_handler() though.

[...]

Applied, thanks!

[1/5] i3c: mipi-i3c-hci: Allow only relevant INTR_STATUS bit updates

      https://git.kernel.org/abelloni/c/eeeec6c5475e
[2/5] i3c: mipi-i3c-hci: Fix handling status of i3c_hci_irq_handler()
      https://git.kernel.org/abelloni/c/279c24021b83
[3/5] i3c: mipi-i3c-hci: Clear INTR_STATUS unconditionally
      https://git.kernel.org/abelloni/c/a7035a8ee966
[4/5] i3c: mipi-i3c-hci: Change name of INTR_STATUS bit 11
      https://git.kernel.org/abelloni/c/7479d2675c50
[5/5] i3c: mipi-i3c-hci: Move unexpected INTR_STATUS print before IO handler
      https://git.kernel.org/abelloni/c/bd916806632d

Best regards,

Signature · 8 lines
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

-- 
linux-i3c mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-i3c