Skip to contentSkip to conversation

spi-nor: intel-spi: Fix atomic sequence handling

PATCH Applied Mika Westerberg 3 participants 9 messages Feb 5, 2018 – May 18, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/2 spi-nor: intel-spi: Fix atomic sequence handling

On many older systems using SW sequencer the PREOP_OPTYPE register contains two preopcodes as following:

PREOP_OPTYPE=0xf2785006

The last two bytes are the opcodes decoded to:

0x50 - Write enable for volatile status register 0x06 - Write enable

The former is used to modify volatile bits in the status register. For non-volatile bits the latter is needed. Preopcodes are used in SW sequencer to send one command "atomically" without anything else interfering the transfer. The sequence that gets executed is:

- Send preopcode (write enable) from PREOP_OPTYPE register
- Send the actual SPI command
- Poll busy bit in the status register (0x05, RDSR)

Commit 8c473dd61bb5 ("spi-nor: intel-spi: Don't assume OPMENU0/1 to be programmed by BIOS") enabled atomic sequence handling but because both preopcodes are programmed, the following happens:

  if (preop >> 8)
  	val |= SSFSTS_CTL_SPOP;

Since on these systems preop >> 8 == 0x50 we end up picking volatile write enable instead. Because of this the actual write command is pretty much NOP unless there is a WREN latched in the chip already.

Furthermore we should not really just assume that WREN was issued in previous call to intel_spi_write_reg() because that might not be the case.

This updates driver to first check that the opcode is actually available in PREOP_OPTYPE register and if not return error back to the spi-nor core (if the controller is not locked we program it now). In addition we save the opcode to ispi->atomic_preopcode field which is checked in next call to intel_spi_sw_cycle() to actually enable atomic sequence using the requested preopcode.

Fixes 8c473dd61bb5 ("spi-nor: intel-spi: Don't assume OPMENU0/1 to be programmed by BIOS")

---
Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

 drivers/mtd/spi-nor/intel-spi.c | 76 ++++++++++++++++++++++++++++++++++++-----
 1 file changed, 67 insertions(+), 9 deletions(-)
drivers/mtd/spi-nor/intel-spi.c +67 -9
index 699951523179..8e98f4ab87c1 100644
@@ -136,6 +136,7 @@
  * @swseq_reg: Use SW sequencer in register reads/writes
  * @swseq_erase: Use SW sequencer in erase operation
  * @erase_64k: 64k erase supported
+ * @atomic_preopcode: Holds preopcode when atomic sequence is requested
  * @opcodes: Opcodes which are supported. This are programmed by BIOS
  *           before it locks down the controller.
  */
@@ -153,6 +154,7 @@ struct intel_spi {
 	bool swseq_reg;
 	bool swseq_erase;
 	bool erase_64k;
+	u8 atomic_preopcode;
 	u8 opcodes[8];
 };
 
@@ -474,7 +476,7 @@ static int intel_spi_sw_cycle(struct intel_spi *ispi, u8 opcode, int len,
 			      int optype)
 {
 	u32 val = 0, status;
-	u16 preop;
+	u8 atomic_preopcode;
 	int ret;
 
 	ret = intel_spi_opcode_index(ispi, opcode, optype);
@@ -484,17 +486,42 @@ static int intel_spi_sw_cycle(struct intel_spi *ispi, u8 opcode, int len,
 	if (len > INTEL_SPI_FIFO_SZ)
 		return -EINVAL;
 
+	/*
+	 * Always clear it after each SW sequencer operation regardless
+	 * of whether it is successful or not.
+	 */
+	atomic_preopcode = ispi->atomic_preopcode;
+	ispi->atomic_preopcode = 0;
+
 	/* Only mark 'Data Cycle' bit when there is data to be transferred */
 	if (len > 0)
 		val = ((len - 1) << SSFSTS_CTL_DBC_SHIFT) | SSFSTS_CTL_DS;
 	val |= ret << SSFSTS_CTL_COP_SHIFT;
 	val |= SSFSTS_CTL_FCERR | SSFSTS_CTL_FDONE;
 	val |= SSFSTS_CTL_SCGO;
-	preop = readw(ispi->sregs + PREOP_OPTYPE);
-	if (preop) {
-		val |= SSFSTS_CTL_ACS;
-		if (preop >> 8)
-			val |= SSFSTS_CTL_SPOP;
+	if (atomic_preopcode) {
+		u16 preop;
+
+		switch (optype) {
+		case OPTYPE_WRITE_NO_ADDR:
+		case OPTYPE_WRITE_WITH_ADDR:
+			/* Pick matching preopcode for the atomic sequence */
+			preop = readw(ispi->sregs + PREOP_OPTYPE);
+			if ((preop & 0xff) == atomic_preopcode)
+				; /* Do nothing */
+			else if ((preop >> 8) == atomic_preopcode)
+				val |= SSFSTS_CTL_SPOP;
+			else
+				return -EINVAL;
+
+			/* Enable atomic sequence */
+			val |= SSFSTS_CTL_ACS;
+			break;
+
+		default:
+			return -EINVAL;
+		}
+
 	}
 	writel(val, ispi->sregs + SSFSTS_CTL);
 
@@ -538,13 +565,31 @@ static int intel_spi_write_reg(struct spi_nor *nor, u8 opcode, u8 *buf, int len)
 
 	/*
 	 * This is handled with atomic operation and preop code in Intel
-	 * controller so skip it here now. If the controller is not locked,
-	 * program the opcode to the PREOP register for later use.
+	 * controller so we only verify that it is available. If the
+	 * controller is not locked, program the opcode to the PREOP
+	 * register for later use.
+	 *
+	 * When hardware sequencer is used there is no need to program
+	 * any opcodes (it handles them automatically as part of a command).
 	 */
 	if (opcode == SPINOR_OP_WREN) {
-		if (!ispi->locked)
+		u16 preop;
+
+		if (!ispi->swseq_reg)
+			return 0;
+
+		preop = readw(ispi->sregs + PREOP_OPTYPE);
+		if ((preop & 0xff) != opcode && (preop >> 8) != opcode) {
+			if (ispi->locked)
+				return -EINVAL;
 			writel(opcode, ispi->sregs + PREOP_OPTYPE);
+		}
 
+		/*
+		 * This enables atomic sequence on next SW sycle. Will
+		 * be cleared after next operation.
+		 */
+		ispi->atomic_preopcode = opcode;
 		return 0;
 	}
 
@@ -569,6 +614,13 @@ static ssize_t intel_spi_read(struct spi_nor *nor, loff_t from, size_t len,
 	u32 val, status;
 	ssize_t ret;
 
+	/*
+	 * Atomic sequence is not expected with HW sequencer reads. Make
+	 * sure it is cleared regardless.
+	 */
+	if (WARN_ON_ONCE(ispi->atomic_preopcode))
+		ispi->atomic_preopcode = 0;
+
 	switch (nor->read_opcode) {
 	case SPINOR_OP_READ:
 	case SPINOR_OP_READ_FAST:
@@ -627,6 +679,9 @@ static ssize_t intel_spi_write(struct spi_nor *nor, loff_t to, size_t len,
 	u32 val, status;
 	ssize_t ret;
 
+	/* Not needed with HW sequencer write, make sure it is cleared */
+	ispi->atomic_preopcode = 0;
+
 	while (len > 0) {
 		block_size = min_t(size_t, len, INTEL_SPI_FIFO_SZ);
 
@@ -707,6 +762,9 @@ static int intel_spi_erase(struct spi_nor *nor, loff_t offs)
 		return 0;
 	}
 
+	/* Not needed with HW sequencer erase, make sure it is cleared */
+	ispi->atomic_preopcode = 0;
+
 	while (len > 0) {
 		writel(offs, ispi->base + FADDR);
 
Signature · 1 line
2.15.1
#1 lore raw

PATCH v2 2/2 spi-nor: intel-spi: Explicitly mark the driver as dangerous in Kconfig

The driver is not meant for normal users at all but instead such users who really know what they are doing and are able to build their own kernel to enable it. Mark both driver Kconfig entries as dangerous to make sure the driver is not accidentally enabled without understanding possible consequences in doing so.

---
No changes from v1.

 drivers/mtd/spi-nor/Kconfig | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/mtd/spi-nor/Kconfig +2 -2
index 89da88e59121..f480b227a6b8 100644
@@ -90,7 +90,7 @@ config SPI_INTEL_SPI
 	tristate
 
 config SPI_INTEL_SPI_PCI
-	tristate "Intel PCH/PCU SPI flash PCI driver"
+	tristate "Intel PCH/PCU SPI flash PCI driver (DANGEROUS)"
 	depends on X86 && PCI
 	select SPI_INTEL_SPI
 	help
@@ -106,7 +106,7 @@ config SPI_INTEL_SPI_PCI
 	  will be called intel-spi-pci.
 
 config SPI_INTEL_SPI_PLATFORM
-	tristate "Intel PCH/PCU SPI flash platform driver"
+	tristate "Intel PCH/PCU SPI flash platform driver (DANGEROUS)"
 	depends on X86
 	select SPI_INTEL_SPI
 	help
Signature · 1 line
2.15.1
#2 lore raw
In reply to Mika Westerberg
Quoted 7 lines · Mika Westerberg

Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

Any comments on these two? I did the changes Cyrille asked me to do.

Thanks!

#3 lore raw
Quoted 10 lines · Mika Westerberg
Quoted 7 lines · Mika Westerberg

Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

Any comments on these two? I did the changes Cyrille asked me to do.

Ping. If no comments, could someone pick these up please.

Reviewed-by
#4 lore raw
Quoted 10 lines · Mika Westerberg
Quoted 9 lines · Mika Westerberg
Quoted 7 lines · Mika Westerberg

Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

Quoted 4 lines
Quoted 1 line

Any comments on these two? I did the changes Cyrille asked me to do.

Ping. If no comments, could someone pick these up please.

Signature · 2 lines
Best regards,
Marek Vasut
#5 lore raw
Quoted 11 lines · On Wed, 4 Apr 2018 10:21:45 +0200 Marek Vasut <[email protected]> wrote:
Quoted 10 lines · Mika Westerberg
Quoted 9 lines · Mika Westerberg
Quoted 7 lines · Mika Westerberg

Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

Any comments on these two? I did the changes Cyrille asked me to do.

Ping. If no comments, could someone pick these up please.

Will queue this patches to mtd/fixes just after 4.17-rc1 is out. Sorry for the huge delay.

Regards,

Boris

#6 lore raw
Quoted 16 lines · Boris Brezillon
Quoted 13 lines · On Wed, 4 Apr 2018 10:21:45 +0200 Marek Vasut <[email protected]> wrote:
Quoted 12 lines · Mika Westerberg
Quoted 10 lines · Mika Westerberg
Quoted 7 lines · Mika Westerberg

Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

Any comments on these two? I did the changes Cyrille asked me to do.

Quoted 7 lines
Quoted 4 lines
Quoted 2 lines

Ping. If no comments, could someone pick these up please.

Will queue this patches to mtd/fixes just after 4.17-rc1 is out. Sorry for the huge delay.

Hi,

Did these patches end up somewhere? I can't see them in current v4.17-rc5.

Thanks!

#7 lore raw
Quoted 22 lines · On Fri, 18 May 2018 08:20:17 +0300 Mika Westerberg…
Quoted 21 lines · Boris Brezillon
Quoted 17 lines · On Wed, 4 Apr 2018 10:21:45 +0200 Marek Vasut <[email protected]> wrote:
Quoted 10 lines · Mika Westerberg
Quoted 9 lines · Mika Westerberg
Quoted 7 lines · Mika Westerberg

Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

Quoted 4 lines
Quoted 1 line

Any comments on these two? I did the changes Cyrille asked me to do.

Ping. If no comments, could someone pick these up please.

Quoted 7 lines
Quoted 2 lines

Will queue this patches to mtd/fixes just after 4.17-rc1 is out. Sorry for the huge delay.

Hi,

Did these patches end up somewhere? I can't see them in current v4.17-rc5.

Oops, my bad. I'll queue it in spi-nor/next (it's too late for -rc6). Sorry for the inconvenience.

#8 lore raw
In reply to Mika Westerberg
Quoted 41 lines · On Mon, 5 Feb 2018 14:32:59 +0300 Mika Westerberg…

On many older systems using SW sequencer the PREOP_OPTYPE register contains two preopcodes as following:

PREOP_OPTYPE=0xf2785006

The last two bytes are the opcodes decoded to:

0x50 - Write enable for volatile status register 0x06 - Write enable

The former is used to modify volatile bits in the status register. For non-volatile bits the latter is needed. Preopcodes are used in SW sequencer to send one command "atomically" without anything else interfering the transfer. The sequence that gets executed is:

- Send preopcode (write enable) from PREOP_OPTYPE register
- Send the actual SPI command
- Poll busy bit in the status register (0x05, RDSR)

Commit 8c473dd61bb5 ("spi-nor: intel-spi: Don't assume OPMENU0/1 to be programmed by BIOS") enabled atomic sequence handling but because both preopcodes are programmed, the following happens:

  if (preop >> 8)
  	val |= SSFSTS_CTL_SPOP;

Since on these systems preop >> 8 == 0x50 we end up picking volatile write enable instead. Because of this the actual write command is pretty much NOP unless there is a WREN latched in the chip already.

Furthermore we should not really just assume that WREN was issued in previous call to intel_spi_write_reg() because that might not be the case.

This updates driver to first check that the opcode is actually available in PREOP_OPTYPE register and if not return error back to the spi-nor core (if the controller is not locked we program it now). In addition we save the opcode to ispi->atomic_preopcode field which is checked in next call to intel_spi_sw_cycle() to actually enable atomic sequence using the requested preopcode.

Quoted 3 lines
Fixes 8c473dd61bb5 ("spi-nor: intel-spi: Don't assume OPMENU0/1 to be programmed by BIOS")

Queued to spi-nor/next.

Thanks,

Boris

Quoted 158 lines

---
Changes from v1:

- Added atomic_preopcode field and only when it is set, enable atomic sequence.
- Return -EINVAL if the requested WREN operation cannot be found from PREOP register instead of silently pretending it is supported.
- Updated patch subject to match more closely what is being fixed.

 drivers/mtd/spi-nor/intel-spi.c | 76 ++++++++++++++++++++++++++++++++++++-----
 1 file changed, 67 insertions(+), 9 deletions(-)
drivers/mtd/spi-nor/intel-spi.c +67 -9
index 699951523179..8e98f4ab87c1 100644
@@ -136,6 +136,7 @@
  * @swseq_reg: Use SW sequencer in register reads/writes
  * @swseq_erase: Use SW sequencer in erase operation
  * @erase_64k: 64k erase supported
+ * @atomic_preopcode: Holds preopcode when atomic sequence is requested
  * @opcodes: Opcodes which are supported. This are programmed by BIOS
  *           before it locks down the controller.
  */
@@ -153,6 +154,7 @@ struct intel_spi {
 	bool swseq_reg;
 	bool swseq_erase;
 	bool erase_64k;
+	u8 atomic_preopcode;
 	u8 opcodes[8];
 };
 
@@ -474,7 +476,7 @@ static int intel_spi_sw_cycle(struct intel_spi *ispi, u8 opcode, int len,
 			      int optype)
 {
 	u32 val = 0, status;
-	u16 preop;
+	u8 atomic_preopcode;
 	int ret;
 
 	ret = intel_spi_opcode_index(ispi, opcode, optype);
@@ -484,17 +486,42 @@ static int intel_spi_sw_cycle(struct intel_spi *ispi, u8 opcode, int len,
 	if (len > INTEL_SPI_FIFO_SZ)
 		return -EINVAL;
 
+	/*
+	 * Always clear it after each SW sequencer operation regardless
+	 * of whether it is successful or not.
+	 */
+	atomic_preopcode = ispi->atomic_preopcode;
+	ispi->atomic_preopcode = 0;
+
 	/* Only mark 'Data Cycle' bit when there is data to be transferred */
 	if (len > 0)
 		val = ((len - 1) << SSFSTS_CTL_DBC_SHIFT) | SSFSTS_CTL_DS;
 	val |= ret << SSFSTS_CTL_COP_SHIFT;
 	val |= SSFSTS_CTL_FCERR | SSFSTS_CTL_FDONE;
 	val |= SSFSTS_CTL_SCGO;
-	preop = readw(ispi->sregs + PREOP_OPTYPE);
-	if (preop) {
-		val |= SSFSTS_CTL_ACS;
-		if (preop >> 8)
-			val |= SSFSTS_CTL_SPOP;
+	if (atomic_preopcode) {
+		u16 preop;
+
+		switch (optype) {
+		case OPTYPE_WRITE_NO_ADDR:
+		case OPTYPE_WRITE_WITH_ADDR:
+			/* Pick matching preopcode for the atomic sequence */
+			preop = readw(ispi->sregs + PREOP_OPTYPE);
+			if ((preop & 0xff) == atomic_preopcode)
+				; /* Do nothing */
+			else if ((preop >> 8) == atomic_preopcode)
+				val |= SSFSTS_CTL_SPOP;
+			else
+				return -EINVAL;
+
+			/* Enable atomic sequence */
+			val |= SSFSTS_CTL_ACS;
+			break;
+
+		default:
+			return -EINVAL;
+		}
+
 	}
 	writel(val, ispi->sregs + SSFSTS_CTL);
 
@@ -538,13 +565,31 @@ static int intel_spi_write_reg(struct spi_nor *nor, u8 opcode, u8 *buf, int len)
 
 	/*
 	 * This is handled with atomic operation and preop code in Intel
-	 * controller so skip it here now. If the controller is not locked,
-	 * program the opcode to the PREOP register for later use.
+	 * controller so we only verify that it is available. If the
+	 * controller is not locked, program the opcode to the PREOP
+	 * register for later use.
+	 *
+	 * When hardware sequencer is used there is no need to program
+	 * any opcodes (it handles them automatically as part of a command).
 	 */
 	if (opcode == SPINOR_OP_WREN) {
-		if (!ispi->locked)
+		u16 preop;
+
+		if (!ispi->swseq_reg)
+			return 0;
+
+		preop = readw(ispi->sregs + PREOP_OPTYPE);
+		if ((preop & 0xff) != opcode && (preop >> 8) != opcode) {
+			if (ispi->locked)
+				return -EINVAL;
 			writel(opcode, ispi->sregs + PREOP_OPTYPE);
+		}
 
+		/*
+		 * This enables atomic sequence on next SW sycle. Will
+		 * be cleared after next operation.
+		 */
+		ispi->atomic_preopcode = opcode;
 		return 0;
 	}
 
@@ -569,6 +614,13 @@ static ssize_t intel_spi_read(struct spi_nor *nor, loff_t from, size_t len,
 	u32 val, status;
 	ssize_t ret;
 
+	/*
+	 * Atomic sequence is not expected with HW sequencer reads. Make
+	 * sure it is cleared regardless.
+	 */
+	if (WARN_ON_ONCE(ispi->atomic_preopcode))
+		ispi->atomic_preopcode = 0;
+
 	switch (nor->read_opcode) {
 	case SPINOR_OP_READ:
 	case SPINOR_OP_READ_FAST:
@@ -627,6 +679,9 @@ static ssize_t intel_spi_write(struct spi_nor *nor, loff_t to, size_t len,
 	u32 val, status;
 	ssize_t ret;
 
+	/* Not needed with HW sequencer write, make sure it is cleared */
+	ispi->atomic_preopcode = 0;
+
 	while (len > 0) {
 		block_size = min_t(size_t, len, INTEL_SPI_FIFO_SZ);
 
@@ -707,6 +762,9 @@ static int intel_spi_erase(struct spi_nor *nor, loff_t offs)
 		return 0;
 	}
 
+	/* Not needed with HW sequencer erase, make sure it is cleared */
+	ispi->atomic_preopcode = 0;
+
 	while (len > 0) {
 		writel(offs, ispi->base + FADDR);