Skip to contentSkip to conversation

spi-nor: aspeed: add support for the 4B opcodes

PATCH Applied Cédric Le Goater 4 participants 24 messages Apr 17, 2019 – May 3, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover spi-nor: aspeed: add support for the 4B opcodes

Here is a little series providing cleanups on the Aspeed SMC Controller driver and adding support for the 4B opcodes which were recently added to the Linux kernels 5.0.x.

The use of the 4B opcodes was breaking the read of the golden buffer done at slow speed in the optimization read sequence. The code assumed that the chip was in 4B address mode, as if a EN4B opcode had been sent, but this is not the case anymore with 4B opcodes. The golden buffer is now read with a SPINOR_OP_READ_4B (0x13) when the chip supports 4B opcodes.

It should fix accesses to the palmetto PNOR and the Witherspoon BMC flash modules.

*Please test !*

Thanks,

C.

Cédric Le Goater (4):
mtd: spi-nor: aspeed: introduce a aspeed_smc_default_read() helper
mtd: spi-nor: aspeed: clarify 4BYTE address mode mask
mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization
buffer
mtd: spi-nor: aspeed: add support for the 4B opcodes

 drivers/mtd/spi-nor/aspeed-smc.c | 40 +++++++++++++++++++++++---------
 1 file changed, 29 insertions(+), 11 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH 1/4 mtd: spi-nor: aspeed: introduce a aspeed_smc_default_read() helper

---
 drivers/mtd/spi-nor/aspeed-smc.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +12 -7
index ddf7ae78aa0a..ee3059b27c07 100644
@@ -882,6 +882,17 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 };
 #define ASPEED_SMC_HCLK_DIV(i) (aspeed_smc_hclk_divs[(i) - 1] << 8)
 
+static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
+{
+	return (chip->ctl_val[smc_read] & 0x2000) |
+		(0x00 << 28) | /* Single bit */
+		(0x00 << 24) | /* CE# max */
+		(0x03 << 16) | /* use normal reads */
+		(0x00 <<  8) | /* HCLK/16 */
+		(0x00 <<  6) | /* no dummy cycle */
+		(0x00);        /* normal mode */
+}
+
 static int aspeed_smc_optimize_read(struct aspeed_smc_chip *chip,
 				     u32 max_freq)
 {
@@ -898,13 +909,7 @@ static int aspeed_smc_optimize_read(struct aspeed_smc_chip *chip,
 	/* We start with the dumbest setting (keep 4Byte bit) and read
 	 * some data
 	 */
-	chip->ctl_val[smc_read] = (chip->ctl_val[smc_read] & 0x2000) |
-		(0x00 << 28) | /* Single bit */
-		(0x00 << 24) | /* CE# max */
-		(0x03 << 16) | /* use normal reads */
-		(0x00 <<  8) | /* HCLK/16 */
-		(0x00 <<  6) | /* no dummy cycle */
-		(0x00);        /* normal read */
+	chip->ctl_val[smc_read] = aspeed_smc_default_read(chip);
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
Signature · 1 line
2.20.1
#2 lore raw

Re: [PATCH dev-5.0 1/4] mtd: spi-nor: aspeed: introduce a aspeed_smc_default_read() helper

Quoted 10 lines · Cédric Le Goater
---
 drivers/mtd/spi-nor/aspeed-smc.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +0 -0
index ddf7ae78aa0a..ee3059b27c07 100644

drivers/mtd/spi-nor/aspeed-smc.c static const uint32_t aspeed_smc_hclk_divs[] = { · line 887

Quoted 6 lines
drivers/mtd/spi-nor/aspeed-smc.c +3 -0
@@ -882,6 +882,17 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 };
 #define ASPEED_SMC_HCLK_DIV(i) (aspeed_smc_hclk_divs[(i) - 1] << 8)
 
+static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
+{
+	return (chip->ctl_val[smc_read] & 0x2000) |

Cédric, isn't this a good time to get rid of these cached ctl_val values?

Why not read/modify/store the actual register?

What's the profit of this caching?

With best regards,
Alexander Amelkin,
Leading BMC Software Engineer, YADRO
https://yadro.com

#3 lore raw

Re: [PATCH dev-5.0 1/4] mtd: spi-nor: aspeed: introduce a aspeed_smc_default_read() helper

Quoted 20 lines · Alexander Amelkin
Quoted 10 lines · Cédric Le Goater
---
 drivers/mtd/spi-nor/aspeed-smc.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +0 -0
index ddf7ae78aa0a..ee3059b27c07 100644

drivers/mtd/spi-nor/aspeed-smc.c static const uint32_t aspeed_smc_hclk_divs[] = { · line 887

Quoted 6 lines
drivers/mtd/spi-nor/aspeed-smc.c +3 -0
@@ -882,6 +882,17 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 };
 #define ASPEED_SMC_HCLK_DIV(i) (aspeed_smc_hclk_divs[(i) - 1] << 8)
 
+static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
+{
+	return (chip->ctl_val[smc_read] & 0x2000) |

Cédric, isn't this a good time to get rid of these cached ctl_val values?

Quoted 3 lines

Why not read/modify/store the actual register?

What's the profit of this caching?

Reasonable question, but I've merged the change as-is to dev-5.0 on the basis that it helps fix read corruption on AC-cycles.

We can address this in a follow-up patch if necessary.

Cheers,

Andrew

Quoted 10 lines

With best regards,
Alexander Amelkin,
Leading BMC Software Engineer, YADRO
https://yadro.com

Attachments:
* signature.asc

#4 lore raw

Re: [PATCH dev-5.0 1/4] mtd: spi-nor: aspeed: introduce a aspeed_smc_default_read() helper

Quoted 23 lines · Andrew Jeffery
Quoted 20 lines · Alexander Amelkin
Quoted 10 lines · Cédric Le Goater
---
 drivers/mtd/spi-nor/aspeed-smc.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +0 -0
index ddf7ae78aa0a..ee3059b27c07 100644

drivers/mtd/spi-nor/aspeed-smc.c static const uint32_t aspeed_smc_hclk_divs[] = { · line 887

Quoted 6 lines
drivers/mtd/spi-nor/aspeed-smc.c +3 -0
@@ -882,6 +882,17 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 };
 #define ASPEED_SMC_HCLK_DIV(i) (aspeed_smc_hclk_divs[(i) - 1] << 8)
 
+static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
+{
+	return (chip->ctl_val[smc_read] & 0x2000) |

Cédric, isn't this a good time to get rid of these cached ctl_val values?

Quoted 3 lines
Quoted 3 lines

Why not read/modify/store the actual register?

What's the profit of this caching?

These values are computed when the controller is probed and when the chip devices are scanned (which takes a couple of seconds to do the training). They are the default values used when the flash contents is accessed through the flash MMIO window.

When in User mode, for register operation, erase, write, other values are used, see read/write_reg operations, and we don't want to recompute the same values, it takes time and the HW has not changed anyhow.

It is also safer as some FW still hack the value in the back of the Linux driver so we should not trust how the HW has been configured. I believe HIOMAP is available for P8 also now ?

The state of chips is another problem. If the 4B enablement in the chip was dropped and if the controller is still configured to use 4B, then that's a problem. I agree. Using 4B opcodes only should help in the future.

Quoted 4 lines

Reasonable question, but I've merged the change as-is to dev-5.0 on the basis that it helps fix read corruption on AC-cycles.

We can address this in a follow-up patch if necessary.

spi-mem should be taken into account now, which requires a large rewrite of the driver.

Cheers,

C.

Quoted 15 lines

Cheers,

Andrew

Quoted 10 lines

With best regards,
Alexander Amelkin,
Leading BMC Software Engineer, YADRO
https://yadro.com

Attachments:
* signature.asc

#5 lore raw

PATCH 2/4 mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

In reply to Cédric Le Goater
---
 drivers/mtd/spi-nor/aspeed-smc.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/mtd/spi-nor/aspeed-smc.c +9 -1
index ee3059b27c07..1437732fdea1 100644
@@ -884,7 +884,15 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 
 static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
 {
-	return (chip->ctl_val[smc_read] & 0x2000) |
+	/*
+	 * Keep the 4Byte address mode on the AST2400 SPI controller.
+	 * Other controllers set the 4Byte mode in the CE Control
+	 * Register
+	 */
+	u32 ctl_mask = chip->controller->info == &spi_2400_info ?
+		 CONTROL_IO_ADDRESS_4B : 0;
+
+	return (chip->ctl_val[smc_read] & ctl_mask) |
 		(0x00 << 28) | /* Single bit */
 		(0x00 << 24) | /* CE# max */
 		(0x03 << 16) | /* use normal reads */
Signature · 1 line
2.20.1
#6 lore raw

Re: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

Quoted 30 lines · Cédric Le Goater
---
 drivers/mtd/spi-nor/aspeed-smc.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/mtd/spi-nor/aspeed-smc.c +9 -1
index ee3059b27c07..1437732fdea1 100644
@@ -884,7 +884,15 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 
 static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
 {
-	return (chip->ctl_val[smc_read] & 0x2000) |
+	/*
+	 * Keep the 4Byte address mode on the AST2400 SPI controller.
+	 * Other controllers set the 4Byte mode in the CE Control
+	 * Register
+	 */
+	u32 ctl_mask = chip->controller->info == &spi_2400_info ?
+		 CONTROL_IO_ADDRESS_4B : 0;
+
+	return (chip->ctl_val[smc_read] & ctl_mask) |
 		(0x00 << 28) | /* Single bit */
 		(0x00 << 24) | /* CE# max */
 		(0x03 << 16) | /* use normal reads */
Signature · 1 line
2.20.1

Applied to dev-5.0.

Cheers,

Andrew

#7 lore raw

Re: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

In reply to Cédric Le Goater
Quoted 2 lines · Cédric Le Goater

Subject: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE>address mode mask

Missing  change log here.

Quoted 12 lines

drivers/mtd/spi-nor/aspeed-smc.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
drivers/mtd/spi-nor/aspeed-smc.c +0 -0
index ee3059b27c07..1437732fdea1 100644
@@ -884,7 +884,15 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 
 static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)

drivers/mtd/spi-nor/aspeed-smc.c static const uint32_t aspeed_smc_hclk_divs[] = { · line 894

Quoted 10 lines
drivers/mtd/spi-nor/aspeed-smc.c +8 -1
 {
- return (chip->ctl_val[smc_read] & 0x2000) |
+ /*
+ * Keep the 4Byte address mode on the AST2400 SPI controller.
+ * Other controllers set the 4Byte mode in the CE Control
+ * Register
+ */
+ u32 ctl_mask = chip->controller->info == &spi_2400_info ?
+ CONTROL_IO_ADDRESS_4B : 0;
+

I dislike this patch because it violates the data driven model of the types.
Either a dupicate method should be created or at least a member of thetype structure should be used instead of a compare to a specific instance.

Quoted 8 lines
drivers/mtd/spi-nor/aspeed-smc.c +1 -0
+ return (chip->ctl_val[smc_read] & ctl_mask) |
(0x00 << 28) | /* Single bit */
(0x00 << 24) | /* CE# max */
(0x03 << 16) | /* use normal reads */

--
2.20.1

#8 lore raw

Re: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

Quoted 7 lines · Milton Miller II
Quoted 3 lines · Cédric Le Goater

Subject: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

Missing  change log here.

This is a oneliner patch.

Quoted 27 lines
Quoted 14 lines
Signed-off-by Cédric Le Goater <[email protected] <mailto:[email protected]>>

---
drivers/mtd/spi-nor/aspeed-smc.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
drivers/mtd/spi-nor/aspeed-smc.c +0 -0
index ee3059b27c07..1437732fdea1 100644
@@ -884,7 +884,15 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 
 static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)

drivers/mtd/spi-nor/aspeed-smc.c static const uint32_t aspeed_smc_hclk_divs[] = { · line 894

Quoted 10 lines
drivers/mtd/spi-nor/aspeed-smc.c +8 -1
 {
- return (chip->ctl_val[smc_read] & 0x2000) |
+ /*
+ * Keep the 4Byte address mode on the AST2400 SPI controller.
+ * Other controllers set the 4Byte mode in the CE Control
+ * Register
+ */
+ u32 ctl_mask = chip->controller->info == &spi_2400_info ?
+ CONTROL_IO_ADDRESS_4B : 0;
+

I dislike this patch because it violates the data driven model of the types.

Quoted 2 lines

Either a dupicate method should be created or at least a member of the type structure should be used instead of a compare to a specific instance.

we could add a ops returning the bitmask. yes. I think this is a efficient considering the patch is not in mainline and was not reviewed by the Linux maintainers.

C.

Quoted 9 lines
Quoted 9 lines
drivers/mtd/spi-nor/aspeed-smc.c +1 -0
+ return (chip->ctl_val[smc_read] & ctl_mask) |
(0x00 << 28) | /* Single bit */
(0x00 << 24) | /* CE# max */
(0x03 << 16) | /* use normal reads */

--
2.20.1

#9 lore raw

Re: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

Quoted 10 lines · Cédric Le Goater
Quoted 7 lines · Milton Miller II
Quoted 3 lines · Cédric Le Goater

Subject: [PATCH dev-5.0 2/4] mtd: spi-nor: aspeed: clarify 4BYTE address mode mask

Missing  change log here.

This is a oneliner patch.

Yes, I think the comment more than made up for the lack of a changelog.

Easy enough to view both together, and I find that acceptable for now.

Andrew

Quoted 49 lines
Quoted 28 lines
Quoted 15 lines
Signed-off-by Cédric Le Goater <[email protected] <mailto:[email protected]>>

---
drivers/mtd/spi-nor/aspeed-smc.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
drivers/mtd/spi-nor/aspeed-smc.c +0 -0
index ee3059b27c07..1437732fdea1 100644
@@ -884,7 +884,15 @@ static const uint32_t aspeed_smc_hclk_divs[] = {
 
 static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)

drivers/mtd/spi-nor/aspeed-smc.c static const uint32_t aspeed_smc_hclk_divs[] = { · line 894

Quoted 10 lines
drivers/mtd/spi-nor/aspeed-smc.c +8 -1
 {
- return (chip->ctl_val[smc_read] & 0x2000) |
+ /*
+ * Keep the 4Byte address mode on the AST2400 SPI controller.
+ * Other controllers set the 4Byte mode in the CE Control
+ * Register
+ */
+ u32 ctl_mask = chip->controller->info == &spi_2400_info ?
+ CONTROL_IO_ADDRESS_4B : 0;
+

I dislike this patch because it violates the data driven model of the types.

Quoted 2 lines

Either a dupicate method should be created or at least a member of the type structure should be used instead of a compare to a specific instance.

we could add a ops returning the bitmask. yes. I think this is a efficient considering the patch is not in mainline and was not reviewed by the Linux maintainers.

C.

Quoted 9 lines
Quoted 9 lines
drivers/mtd/spi-nor/aspeed-smc.c +1 -0
+ return (chip->ctl_val[smc_read] & ctl_mask) |
(0x00 << 28) | /* Single bit */
(0x00 << 24) | /* CE# max */
(0x03 << 16) | /* use normal reads */

--
2.20.1

#10 lore raw

PATCH 3/4 mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

In reply to Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

---
 drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +2 -4
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
 	}
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
 	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#11 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Quoted 35 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

---
 drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct 
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
 	}
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct 
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
Quoted 5 lines

chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;

Signature · 1 line
2.20.1

Applied to dev-5.0.

Cheers,

Andrew

#12 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

In reply to Cédric Le Goater
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

Quoted 36 lines

---
drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----

1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
	}
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#13 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Quoted 14 lines · Milton Miller II
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

Quoted 8 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

Yeah, I had a briefly nagging thought about this when I applied it. I should have gone with that gut instinct and inspected it in more depth.

Thanks for calling it out.

Reverting for dev-5.0, this is an isolated issue (arguably should have been separated from the current series).

Andrew

Quoted 39 lines
Quoted 37 lines

---
drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----

1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
	}
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#14 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Quoted 12 lines · Andrew Jeffery
Quoted 9 lines · Milton Miller II
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

Quoted 4 lines
Quoted 4 lines

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

It doesn't AFAICT looking at other drivers and also from my test.

Quoted 2 lines
Quoted 2 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit

That might have been the case on Linux 4.7. It doesn't seem to be the case anymore. See below.

Quoted 6 lines
Quoted 6 lines

arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

I don't think you have enough convincing arguments for that.

Quoted 5 lines

Yeah, I had a briefly nagging thought about this when I applied it. I should have gone with that gut instinct and inspected it in more depth.

Thanks for calling it out.

I think this needs more thinking and tests. Witherspoon and palmetto are fine. What's the problem.

Quoted 2 lines

Reverting for dev-5.0, this is an isolated issue (arguably should have been separated from the current series).

We have been using memcpy_fromio for reads for a very long time since 4.13 in the Aspeed SMC driver on *all* platforms*.

See aspeed_smc_read() for the "Command mode".

C.

Quoted 42 lines

Andrew

Quoted 39 lines
Quoted 37 lines

---
drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----

1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
	}
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#15 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Hi Cédric

Quoted 26 lines · Cédric Le Goater
Quoted 12 lines · Andrew Jeffery
Quoted 9 lines · Milton Miller II
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

Quoted 4 lines
Quoted 4 lines

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

It doesn't AFAICT looking at other drivers and also from my test.

Quoted 2 lines
Quoted 2 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit

That might have been the case on Linux 4.7. It doesn't seem to be the case anymore. See below.

Quoted 8 lines
Quoted 6 lines
Quoted 6 lines

arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

I don't think you have enough convincing arguments for that.

That may be the case, but having seen the pain of the original corruption problems that drove the ioreadX_rep() implementation above, Milton's protest combined with my initial, briefly nagging concern was enough for me to revert. Two things here:

1. We've run without this patch for quite some time. Despite oddities, the existing implementation has been stable
2. With patch 4/4, you've resolved the 4B corruption problem. This was the immediate concern, as it was impacting teams developing and testing OpenBMC master. I appreciate the effort you put into hunting that down, the root cause was certainly not obvious.

From *my* testing we appear to be stable both with and without this change, however my concern is that *my* testing is not complete enough to be confident that we're not going to hit the subtle corruption problems that drove the introduction of the existing code.

For some additional context, I'm now on leave until the 30th, and Joel to the 29th. These patches have been through a process that has proceeded much more hastily than I would have liked, and that's lead to where we are now.

With that in mind, less change is better, and so I have decided to back this patch out. It's a risk-based decision, not a reflection of the effort, desires or technicalities involved.

So there are lessons in this for me too as I've felt squeezed by the impact the 4B corruption bug has had on teams testing master, and my impending leave. With that reasoning and 20/20 hindsight I should probably have held off to begin with. Next time. Some reflection is in order.

Anyway, I hope that helps clarify. Time to resolve the disagreement here.

Milton: Given your protest, can you please verify whether there continue to be issues with memcpy_fromio() that prevent us from making use of it. As Cédric points out we do make heavy use of it anyway, and using a well-worn implementation rather than something bespoke for this particular case would improve maintainability.

Cheers,

Andrew

Quoted 67 lines
Quoted 6 lines

Yeah, I had a briefly nagging thought about this when I applied it. I should have gone with that gut instinct and inspected it in more depth.

Thanks for calling it out.

I think this needs more thinking and tests. Witherspoon and palmetto are fine. What's the problem.

Quoted 2 lines

Reverting for dev-5.0, this is an isolated issue (arguably should have been separated from the current series).

We have been using memcpy_fromio for reads for a very long time since 4.13 in the Aspeed SMC driver on *all* platforms*.

See aspeed_smc_read() for the "Command mode".

C.

Quoted 42 lines

Andrew

Quoted 39 lines
Quoted 37 lines

---
drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----

1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
	}
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#16 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Quoted 50 lines · Andrew Jeffery

Hi Cédric

Quoted 26 lines · Cédric Le Goater
Quoted 12 lines · Andrew Jeffery
Quoted 9 lines · Milton Miller II
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

Quoted 4 lines
Quoted 4 lines

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

It doesn't AFAICT looking at other drivers and also from my test.

Quoted 2 lines
Quoted 2 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit

That might have been the case on Linux 4.7. It doesn't seem to be the case anymore. See below.

Quoted 8 lines
Quoted 6 lines
Quoted 6 lines

arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

I don't think you have enough convincing arguments for that.

That may be the case, but having seen the pain of the original corruption problems that drove the ioreadX_rep() implementation above, Milton's protest combined with my initial, briefly nagging concern was enough for me to revert. Two things here:

1. We've run without this patch for quite some time. Despite oddities, the existing implementation has been stable
2. With patch 4/4, you've resolved the 4B corruption problem. This was the immediate concern, as it was impacting teams developing and testing OpenBMC master. I appreciate the effort you put into hunting that down, the root cause was certainly not obvious.

Quoted 4 lines

From *my* testing we appear to be stable both with and without this change, however my concern is that *my* testing is not complete enough to be confident that we're not going to hit the subtle corruption problems that drove the introduction of the existing code.

QEMU would have caught this regression if SFDP was modeled. It does today if 4B opcodes are forced on the mx25l25635e. Given the time the team spent on this, I would say 1 or 2PM overall, QEMU is a good investment.

                                                  ^
                                                  |

Managers are you reading this ? ------------------+

Quoted 8 lines

For some additional context, I'm now on leave until the 30th, and Joel to the 29th. These patches have been through a process that has proceeded much more hastily than I would have liked, and that's lead to where we are now.

With that in mind, less change is better, and so I have decided to back this patch out. It's a risk-based decision, not a reflection of the effort, desires or technicalities involved.

Back to where we were before, it's fine as it works.

The optimize reads are just reading the first 4 bytes :

[   14.130480] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   14.136664] aspeed-smc 1e630000.spi: write control register: 00122302
[   14.143326] aspeed-smc 1e630000.spi: read control register: 203c2341
[   14.149750] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   14.181561] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.188894] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.196230] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.203558] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.210751] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.218067] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.225397] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.232722] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.239912] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.247232] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART

with memcpy_fromio :

[   13.779087] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   13.785255] aspeed-smc 1e630000.spi: write control register: 00122302
[   13.791762] aspeed-smc 1e630000.spi: read control register: 203c2341
[   13.798326] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   13.815420] 50 41 52 54 00 00 00 01 00 00 00 01 00 00 00 80  PART............
[   13.822759] 00 00 00 1b 00 00 10 00 00 00 20 00 00 00 00 00  .......... .....
[   13.829946] 00 00 00 00 00 00 00 00 00 00 00 00 50 41 62 cf  ............PAb.
[   13.837266] 70 61 72 74 00 00 00 00 00 00 00 00 00 00 00 00  part............
[   13.844597] 00 00 00 00 00 00 00 01 ff ff ff ff 00 00 00 01  ................
[   13.851788] 00 00 00 03 00 00 00 01 00 00 10 00 00 00 00 00  ................
[   13.859105] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.866433] 00 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
[   13.873759] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.880951] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.888267] 00 00 00 00 00 00 00 00 00 00 00 00 8f de 9d 89  ................
[   13.895591] 48 42 49 00 00 00 00 00 00 00 00 00 00 00 00 00  HBI.............
[   13.902917] 00 00 00 10 00 00 05 a0 ff ff ff ff 00 00 00 02  ................

I should have added these tests in the commit log. Sorry about that. We will see later on. There are no hurries for this fix. Optimization is still being done.

Thanks,

C.

Quoted 85 lines

So there are lessons in this for me too as I've felt squeezed by the impact the 4B corruption bug has had on teams testing master, and my impending leave. With that reasoning and 20/20 hindsight I should probably have held off to begin with. Next time. Some reflection is in order.

Anyway, I hope that helps clarify. Time to resolve the disagreement here.

Milton: Given your protest, can you please verify whether there continue to be issues with memcpy_fromio() that prevent us from making use of it. As Cédric points out we do make heavy use of it anyway, and using a well-worn implementation rather than something bespoke for this particular case would improve maintainability.

Cheers,

Andrew

Quoted 67 lines
Quoted 6 lines

Yeah, I had a briefly nagging thought about this when I applied it. I should have gone with that gut instinct and inspected it in more depth.

Thanks for calling it out.

I think this needs more thinking and tests. Witherspoon and palmetto are fine. What's the problem.

Quoted 2 lines

Reverting for dev-5.0, this is an isolated issue (arguably should have been separated from the current series).

We have been using memcpy_fromio for reads for a very long time since 4.13 in the Aspeed SMC driver on *all* platforms*.

See aspeed_smc_read() for the "Command mode".

C.

Quoted 42 lines

Andrew

Quoted 39 lines
Quoted 37 lines

---
drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----

1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
	}
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#17 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Quoted 112 lines · Cédric Le Goater
Quoted 50 lines · Andrew Jeffery

Hi Cédric

Quoted 26 lines · Cédric Le Goater
Quoted 12 lines · Andrew Jeffery
Quoted 9 lines · Milton Miller II
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

Quoted 4 lines
Quoted 4 lines

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

It doesn't AFAICT looking at other drivers and also from my test.

Quoted 2 lines
Quoted 2 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit

That might have been the case on Linux 4.7. It doesn't seem to be the case anymore. See below.

Quoted 8 lines
Quoted 6 lines
Quoted 6 lines

arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

I don't think you have enough convincing arguments for that.

That may be the case, but having seen the pain of the original corruption problems that drove the ioreadX_rep() implementation above, Milton's protest combined with my initial, briefly nagging concern was enough for me to revert. Two things here:

1. We've run without this patch for quite some time. Despite oddities, the existing implementation has been stable
2. With patch 4/4, you've resolved the 4B corruption problem. This was the immediate concern, as it was impacting teams developing and testing OpenBMC master. I appreciate the effort you put into hunting that down, the root cause was certainly not obvious.

Quoted 4 lines

From *my* testing we appear to be stable both with and without this change, however my concern is that *my* testing is not complete enough to be confident that we're not going to hit the subtle corruption problems that drove the introduction of the existing code.

QEMU would have caught this regression if SFDP was modeled. It does today if 4B opcodes are forced on the mx25l25635e. Given the time the team spent on this, I would say 1 or 2PM overall, QEMU is a good investment.

                                                  ^
                                                  |

Managers are you reading this ? ------------------+

Quoted 8 lines

For some additional context, I'm now on leave until the 30th, and Joel to the 29th. These patches have been through a process that has proceeded much more hastily than I would have liked, and that's lead to where we are now.

With that in mind, less change is better, and so I have decided to back this patch out. It's a risk-based decision, not a reflection of the effort, desires or technicalities involved.

Back to where we were before, it's fine as it works.

The optimize reads are just reading the first 4 bytes :

[   14.130480] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   14.136664] aspeed-smc 1e630000.spi: write control register: 00122302
[   14.143326] aspeed-smc 1e630000.spi: read control register: 203c2341
[   14.149750] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   14.181561] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.188894] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.196230] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.203558] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.210751] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.218067] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.225397] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.232722] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.239912] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.247232] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART

with memcpy_fromio :

[   13.779087] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   13.785255] aspeed-smc 1e630000.spi: write control register: 00122302
[   13.791762] aspeed-smc 1e630000.spi: read control register: 203c2341
[   13.798326] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   13.815420] 50 41 52 54 00 00 00 01 00 00 00 01 00 00 00 80  PART............
[   13.822759] 00 00 00 1b 00 00 10 00 00 00 20 00 00 00 00 00  .......... .....
[   13.829946] 00 00 00 00 00 00 00 00 00 00 00 00 50 41 62 cf  ............PAb.
[   13.837266] 70 61 72 74 00 00 00 00 00 00 00 00 00 00 00 00  part............
[   13.844597] 00 00 00 00 00 00 00 01 ff ff ff ff 00 00 00 01  ................
[   13.851788] 00 00 00 03 00 00 00 01 00 00 10 00 00 00 00 00  ................
[   13.859105] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.866433] 00 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
[   13.873759] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.880951] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.888267] 00 00 00 00 00 00 00 00 00 00 00 00 8f de 9d 89  ................
[   13.895591] 48 42 49 00 00 00 00 00 00 00 00 00 00 00 00 00  HBI.............
[   13.902917] 00 00 00 10 00 00 05 a0 ff ff ff ff 00 00 00 02  ................
Quoted 1 line

I should have added these tests in the commit log. Sorry about that.

Not to worry! Thanks for the data. It paints a good picture of why the change is desirable :)

Quoted 2 lines

We will see later on. There are no hurries for this fix. Optimization is still being done.

Yes. A silly situation all told. Thanks for your effort again.

Andrew

Quoted 93 lines

Thanks,

C.

Quoted 85 lines

So there are lessons in this for me too as I've felt squeezed by the impact the 4B corruption bug has had on teams testing master, and my impending leave. With that reasoning and 20/20 hindsight I should probably have held off to begin with. Next time. Some reflection is in order.

Anyway, I hope that helps clarify. Time to resolve the disagreement here.

Milton: Given your protest, can you please verify whether there continue to be issues with memcpy_fromio() that prevent us from making use of it. As Cédric points out we do make heavy use of it anyway, and using a well-worn implementation rather than something bespoke for this particular case would improve maintainability.

Cheers,

Andrew

Quoted 67 lines
Quoted 6 lines

Yeah, I had a briefly nagging thought about this when I applied it. I should have gone with that gut instinct and inspected it in more depth.

Thanks for calling it out.

I think this needs more thinking and tests. Witherspoon and palmetto are fine. What's the problem.

Quoted 2 lines

Reverting for dev-5.0, this is an isolated issue (arguably should have been separated from the current series).

We have been using memcpy_fromio for reads for a very long time since 4.13 in the Aspeed SMC driver on *all* platforms*.

See aspeed_smc_read() for the "Command mode".

C.

Quoted 42 lines

Andrew

Quoted 39 lines
Quoted 37 lines

---
drivers/mtd/spi-nor/aspeed-smc.c | 6 ++----

1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
index 1437732fdea1..7e289ecb1c99 100644
@@ -796,8 +796,7 @@ static bool aspeed_smc_check_reads(struct
 aspeed_smc_chip *chip,
 	int i;
 
 	for (i = 0; i < 10; i++) {
-		aspeed_smc_read_from_ahb(test_buf, chip->ahb_base,
-					 CALIBRATE_BUF_SIZE);
+		memcpy_fromio(test_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 		if (memcmp(test_buf, golden_buf, CALIBRATE_BUF_SIZE) != 0)
 			return false;
	}
drivers/mtd/spi-nor/aspeed-smc.c +1 -2
@@ -921,8 +920,7 @@ static int aspeed_smc_optimize_read(struct
 aspeed_smc_chip *chip,
 
 	writel(chip->ctl_val[smc_read], chip->ctl);
 
-	aspeed_smc_read_from_ahb(golden_buf, chip->ahb_base,
-				 CALIBRATE_BUF_SIZE);
+	memcpy_fromio(golden_buf, chip->ahb_base, CALIBRATE_BUF_SIZE);
 
 	/* Establish our read mode with freq field set to 0 (HCLK/16) */
	chip->ctl_val[smc_read] = save_read_val & 0xfffff0ff;
Signature · 1 line
2.20.1
#18 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

In reply to Cédric Le Goater
Quoted 112 lines · Cédric Le Goater
Quoted 50 lines · Andrew Jeffery

Hi Cédric

Quoted 26 lines · Cédric Le Goater
Quoted 12 lines · Andrew Jeffery
Quoted 9 lines · Milton Miller II
Quoted 5 lines · Cédric Le Goater

aspeed_smc_read_from_ahb() only reads the first word which is not what
we want. We want to capture a CALIBRATE_BUF_SIZE size window of the flash contents to optimize the read.

Quoted 4 lines
Quoted 4 lines

NACK

This justifcation is false. The routine reads the whole buffer because it calls the _rep routine and takes the size.

It doesn't AFAICT looking at other drivers and also from my test.

Quoted 2 lines
Quoted 2 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit

That might have been the case on Linux 4.7. It doesn't seem to be the case anymore. See below.

Quoted 8 lines
Quoted 6 lines
Quoted 6 lines

arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1] https://github.com/openbmc/openbmc/issues/3521

Andrew, Please revert this patch.

I don't think you have enough convincing arguments for that.

That may be the case, but having seen the pain of the original corruption problems that drove the ioreadX_rep() implementation above, Milton's protest combined with my initial, briefly nagging concern was enough for me to revert. Two things here:

1. We've run without this patch for quite some time. Despite oddities, the existing implementation has been stable
2. With patch 4/4, you've resolved the 4B corruption problem. This was the immediate concern, as it was impacting teams developing and testing OpenBMC master. I appreciate the effort you put into hunting that down, the root cause was certainly not obvious.

Quoted 4 lines

From *my* testing we appear to be stable both with and without this change, however my concern is that *my* testing is not complete enough to be confident that we're not going to hit the subtle corruption problems that drove the introduction of the existing code.

QEMU would have caught this regression if SFDP was modeled. It does today if 4B opcodes are forced on the mx25l25635e. Given the time the team spent on this, I would say 1 or 2PM overall, QEMU is a good investment.

                                                  ^
                                                  |

Managers are you reading this ? ------------------+

Quoted 8 lines

For some additional context, I'm now on leave until the 30th, and Joel to the 29th. These patches have been through a process that has proceeded much more hastily than I would have liked, and that's lead to where we are now.

With that in mind, less change is better, and so I have decided to back this patch out. It's a risk-based decision, not a reflection of the effort, desires or technicalities involved.

Back to where we were before, it's fine as it works.

The optimize reads are just reading the first 4 bytes :

[   14.130480] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   14.136664] aspeed-smc 1e630000.spi: write control register: 00122302
[   14.143326] aspeed-smc 1e630000.spi: read control register: 203c2341
[   14.149750] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   14.181561] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.188894] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.196230] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.203558] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.210751] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.218067] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.225397] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.232722] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.239912] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART
[   14.247232] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54  PARTPARTPARTPART

with memcpy_fromio :

[   13.779087] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   13.785255] aspeed-smc 1e630000.spi: write control register: 00122302
[   13.791762] aspeed-smc 1e630000.spi: read control register: 203c2341
[   13.798326] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   13.815420] 50 41 52 54 00 00 00 01 00 00 00 01 00 00 00 80  PART............
[   13.822759] 00 00 00 1b 00 00 10 00 00 00 20 00 00 00 00 00  .......... .....
[   13.829946] 00 00 00 00 00 00 00 00 00 00 00 00 50 41 62 cf  ............PAb.
[   13.837266] 70 61 72 74 00 00 00 00 00 00 00 00 00 00 00 00  part............
[   13.844597] 00 00 00 00 00 00 00 01 ff ff ff ff 00 00 00 01  ................
[   13.851788] 00 00 00 03 00 00 00 01 00 00 10 00 00 00 00 00  ................
[   13.859105] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.866433] 00 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00  .@..............
[   13.873759] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.880951] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
[   13.888267] 00 00 00 00 00 00 00 00 00 00 00 00 8f de 9d 89  ................
[   13.895591] 48 42 49 00 00 00 00 00 00 00 00 00 00 00 00 00  HBI.............
[   13.902917] 00 00 00 10 00 00 05 a0 ff ff ff ff 00 00 00 02  ................
Quoted 3 lines

I should have added these tests in the commit log. Sorry about that. We will see later on. There are no hurries for this fix. Optimization is still being done.

Milton: Given you NACK'ed the patch I'd appreciate a follow-up in light of this data.

Cheers,

Andrew

#19 lore raw

Re: [PATCH dev-5.0 3/4] mtd: spi-nor: aspeed: use memcpy_fromio() to capture the optimization buffer

Quoted 166 lines · Andrew Jeffery
Quoted 8 lines · Andrew Jeffery
Quoted 8 lines

Hi Cédric

Quoted 5 lines · Cédric Le Goater
Quoted 4 lines · Andrew Jeffery
Quoted 1 line · Milton Miller II

About 04/17/2019 09:20AM in some timezone, Cédric Le Goater

Quoted 3 lines · wrote:
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

aspeed_smc_read_from_ahb() only reads the first word which is

not

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

what
we want. We want to capture a CALIBRATE_BUF_SIZE size window

of the

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

flash contents to optimize the read.

NACK

This justifcation is false. The routine reads the whole

buffer

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line
Quoted 1 line

because it calls the _rep routine and takes the size.

It doesn't AFAICT looking at other drivers and also from my

test.

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 3 lines
Quoted 3 lines

In addition, the comment just before aspeed_smc_read_from_ahb tells why memcpy_fromio and memcpy_toio are broken on 32 bit

That might have been the case on Linux 4.7. It doesn't seem to

be

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

the case anymore. See below.

Quoted 4 lines
Quoted 4 lines

arm, and this is still the case judging from the recent bug reportfrom a Nuvaton user [1].

[1]

https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_openb
mc_openbmc_issues_3521&d=DwIFaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=bvv7AJEECo RKBU02rcu4F5DWd-EwX8As2xrXeO9ZSo4&m=sK1b4XTLYG8JeD8M-9ido3CQX_AOERqbR DK4EyTZHWc&s=fYtUMc0yvOgIU3iNg2S3anMU3YSmstjFPxQR3JpCtco&e=

Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

Andrew, Please revert this patch.

I don't think you have enough convincing arguments for that.

That may be the case, but having seen the pain of the original

corruption

Quoted 1 line
Quoted 1 line

problems that drove the ioreadX_rep() implementation above,

Milton's

Quoted 1 line
Quoted 1 line

protest combined with my initial, briefly nagging concern was

enough for

Quoted 3 lines
Quoted 3 lines

me to revert. Two things here:

1. We've run without this patch for quite some time. Despite

oddities,

Quoted 2 lines
Quoted 2 lines

the existing implementation has been stable
2. With patch 4/4, you've resolved the 4B corruption problem.

This was

Quoted 2 lines
Quoted 2 lines

the immediate concern, as it was impacting teams developing and testing OpenBMC master. I appreciate the effort you put into

hunting

Quoted 3 lines
Quoted 3 lines

that down, the root cause was certainly not obvious.

From *my* testing we appear to be stable both with and without

this

Quoted 1 line
Quoted 1 line

change, however my concern is that *my* testing is not complete

enough

Quoted 1 line
Quoted 1 line

to be confident that we're not going to hit the subtle corruption

problems

Quoted 3 lines
Quoted 1 line

that drove the introduction of the existing code.

QEMU would have caught this regression if SFDP was modeled. It does

today

Quoted 1 line

if 4B opcodes are forced on the mx25l25635e. Given the time the

team spent

Quoted 6 lines

on this, I would say 1 or 2PM overall, QEMU is a good investment.

                                                  ^
                                                  |

Managers are you reading this ? ------------------+

Quoted 1 line

For some additional context, I'm now on leave until the 30th, and

Joel to

Quoted 1 line
Quoted 1 line

the 29th. These patches have been through a process that has

proceeded

Quoted 1 line
Quoted 1 line

much more hastily than I would have liked, and that's lead to

where we

Quoted 3 lines
Quoted 3 lines

are now.

With that in mind, less change is better, and so I have decided

to back

Quoted 1 line
Quoted 1 line

this patch out. It's a risk-based decision, not a reflection of

the effort,

Quoted 8 lines
Quoted 1 line

desires or technicalities involved.

Back to where we were before, it's fine as it works.

The optimize reads are just reading the first 4 bytes :

[   14.130480] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   14.136664] aspeed-smc 1e630000.spi: write control register:

00122302

Quoted 1 line
[   14.143326] aspeed-smc 1e630000.spi: read control register:

203c2341

Quoted 2 lines
[   14.149750] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   14.181561] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.188894] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.196230] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.203558] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.210751] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.218067] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.225397] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.232722] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.239912] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 1 line
[   14.247232] 50 41 52 54 50 41 52 54 50 41 52 54 50 41 52 54

PARTPARTPARTPART

Quoted 5 lines

with memcpy_fromio :

[   13.779087] aspeed-smc 1e630000.spi: mx25l25635e (32768 Kbytes)
[   13.785255] aspeed-smc 1e630000.spi: write control register:

00122302

Quoted 1 line
[   13.791762] aspeed-smc 1e630000.spi: read control register:

203c2341

Quoted 2 lines
[   13.798326] aspeed-smc 1e630000.spi: AHB frequency: 192 MHz
[   13.815420] 50 41 52 54 00 00 00 01 00 00 00 01 00 00 00 80

PART............

Quoted 1 line
[   13.822759] 00 00 00 1b 00 00 10 00 00 00 20 00 00 00 00 00

.......... .....

Quoted 1 line
[   13.829946] 00 00 00 00 00 00 00 00 00 00 00 00 50 41 62 cf

............PAb.

Quoted 1 line
[   13.837266] 70 61 72 74 00 00 00 00 00 00 00 00 00 00 00 00

part............

Quoted 1 line
[   13.844597] 00 00 00 00 00 00 00 01 ff ff ff ff 00 00 00 01

................

Quoted 1 line
[   13.851788] 00 00 00 03 00 00 00 01 00 00 10 00 00 00 00 00

................

Quoted 1 line
[   13.859105] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

................

Quoted 1 line
[   13.866433] 00 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00

.@..............

Quoted 1 line
[   13.873759] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

................

Quoted 1 line
[   13.880951] 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

................

Quoted 1 line
[   13.888267] 00 00 00 00 00 00 00 00 00 00 00 00 8f de 9d 89

................

Quoted 1 line
[   13.895591] 48 42 49 00 00 00 00 00 00 00 00 00 00 00 00 00

HBI.............

Quoted 1 line
[   13.902917] 00 00 00 10 00 00 05 a0 ff ff ff ff 00 00 00 02

................

Quoted 3 lines

I should have added these tests in the commit log. Sorry about

Quoted 9 lines

that.

Quoted 1 line

We will see later on. There are no hurries for this fix.

Optimization

Quoted 1 line

is still being done.

Milton: Given you NACK'ed the patch I'd appreciate a follow-up in light of
this data.

Yes, I should have replied last week.

I accept the difference between the case where it fails and the case where its used is the difference between read mode, where the chip verifies the address of each access, and user mode, where the region is decoded at the block level and the data is routed through a fifo, where extra reads or writes are problematic.

That said, it will work for read, or where the source and destination have the same alignemnt. The memcpy routine will cause two transactions in the spi controller, but the result will function. I'm not sure a write would work.

I withdraw my NACK.

milton

#20 lore raw

PATCH 4/4 mtd: spi-nor: aspeed: add support for the 4B opcodes

In reply to Cédric Le Goater

Switch the default controller value to use the read mode in order to customize the command and use SPINOR_OP_READ_4B (0x13) when the chip supports 4B opcodes.

---
 drivers/mtd/spi-nor/aspeed-smc.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +9 -2
index 7e289ecb1c99..1200b23416e3 100644
@@ -890,14 +890,21 @@ static u32 aspeed_smc_default_read(struct aspeed_smc_chip *chip)
 	 */
 	u32 ctl_mask = chip->controller->info == &spi_2400_info ?
 		 CONTROL_IO_ADDRESS_4B : 0;
+	u8 cmd = chip->nor.flags & SNOR_F_4B_OPCODES ? SPINOR_OP_READ_4B :
+		SPINOR_OP_READ;
 
+	/*
+	 * Use the "read command" mode to customize the opcode. In
+	 * normal command mode, the value is necessarily READ (0x3) on
+	 * the AST2400/2500 SoCs.
+	 */
 	return (chip->ctl_val[smc_read] & ctl_mask) |
 		(0x00 << 28) | /* Single bit */
 		(0x00 << 24) | /* CE# max */
-		(0x03 << 16) | /* use normal reads */
+		(cmd  << 16) | /* use read mode to support 4B opcode */
 		(0x00 <<  8) | /* HCLK/16 */
 		(0x00 <<  6) | /* no dummy cycle */
-		(0x00);        /* normal mode */
+		(0x01);        /* read mode */
 }
 
 static int aspeed_smc_optimize_read(struct aspeed_smc_chip *chip,
Signature · 1 line
2.20.1
#21 lore raw

Re: [PATCH dev-5.0 4/4] mtd: spi-nor: aspeed: add support for the 4B opcodes

Quoted 38 lines · Cédric Le Goater

Switch the default controller value to use the read mode in order to customize the command and use SPINOR_OP_READ_4B (0x13) when the chip supports 4B opcodes.

---
 drivers/mtd/spi-nor/aspeed-smc.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/mtd/spi-nor/aspeed-smc.c +9 -2
index 7e289ecb1c99..1200b23416e3 100644
@@ -890,14 +890,21 @@ static u32 aspeed_smc_default_read(struct 
 aspeed_smc_chip *chip)
 	 */
 	u32 ctl_mask = chip->controller->info == &spi_2400_info ?
 		 CONTROL_IO_ADDRESS_4B : 0;
+	u8 cmd = chip->nor.flags & SNOR_F_4B_OPCODES ? SPINOR_OP_READ_4B :
+		SPINOR_OP_READ;
 
+	/*
+	 * Use the "read command" mode to customize the opcode. In
+	 * normal command mode, the value is necessarily READ (0x3) on
+	 * the AST2400/2500 SoCs.
+	 */
 	return (chip->ctl_val[smc_read] & ctl_mask) |
 		(0x00 << 28) | /* Single bit */
 		(0x00 << 24) | /* CE# max */
-		(0x03 << 16) | /* use normal reads */
+		(cmd  << 16) | /* use read mode to support 4B opcode */
 		(0x00 <<  8) | /* HCLK/16 */
 		(0x00 <<  6) | /* no dummy cycle */
-		(0x00);        /* normal mode */
+		(0x01);        /* read mode */
 }
 
Quoted 5 lines

static int aspeed_smc_optimize_read(struct aspeed_smc_chip *chip,

Signature · 1 line
2.20.1

Applied to dev-5.0.

Cheers,

Andrew

#22 lore raw
In reply to Cédric Le Goater
Quoted 13 lines · Cédric Le Goater

Here is a little series providing cleanups on the Aspeed SMC>Controller driver and adding support for the 4B opcodes which were recently added
to the Linux kernels 5.0.x.

The use of the 4B opcodes was breaking the read of the golden buffer done at slow speed in the optimization read sequence. The code assumed
that the chip was in 4B address mode, as if a EN4B opcode had been sent, but this is not the case anymore with 4B opcodes. The golden buffer is now read with a SPINOR_OP_READ_4B (0x13) when the chip supports 4B opcodes.

Quoted 5 lines

It should fix accesses to the palmetto PNOR and the Witherspoon BMC flash modules.

*Please test !*

Cedric,
did you notice that spi_nor_read_sfdp is now calling the nor->read method?
Since the read method is hard-coding the opcode, this means that we arereturning random flash contents instead of the flash parameter block, whichprobably confuses the kernel. milton

#23 lore raw
Quoted 21 lines · Milton Miller II
Quoted 18 lines · Cédric Le Goater

Here is a little series providing cleanups on the Aspeed SMC Controller
driver and adding support for the 4B opcodes which were recently added
to the Linux kernels 5.0.x.

The use of the 4B opcodes was breaking the read of the golden buffer done at slow speed in the optimization read sequence. The code assumed
that the chip was in 4B address mode, as if a EN4B opcode had been sent, but this is not the case anymore with 4B opcodes. The golden buffer is now read with a SPINOR_OP_READ_4B (0x13) when the chip supports 4B opcodes.

It should fix accesses to the palmetto PNOR and the Witherspoon BMC flash modules.

Quoted 6 lines
Quoted 1 line

*Please test !*

Cedric,

did you notice that spi_nor_read_sfdp is now calling the nor->read method?

yes.

Quoted 3 lines

Since the read method is hard-coding the opcode, this means that we are returning random flash contents instead of the flash parameter block, which probably confuses the kernel.

I see your point. but, at the time of the sfdp read, the controller settings are very basic :

  [    7.537426] aspeed-smc 1e630000.spi: default control register: 00000300

and the sfdp values are fine.

Here is what we get for sfdp on a palmetto PNOR "mx25l25635e" :

w5:fffffffe w6:ff00ffff w7:eb44ffff w8:520f200c w9:ff00d810 w10:00000000

This looks consistent with the mx25l25635f specs and with the test being done in Linux in mx25l25635_post_bfpt_fixups().

This driver needs a rewrite because the SPI-NOR layer has been changing a lot. I have started a new one on top of spi-mem but the core layer misses support for link training. So this is still work in progress. Help welcomed. The initial target should be *mainline*.

Also,

If someone had some time to model SFDP in QEMU, it would really help. We would have caught this regression earlier.

One way to reproduce the problem in QEMU is to force 4B opcode in Linux but that's only useful when one knows about the problem already. useful to work on the fix but not to track regressions.

Thanks,

C.

Quoted 4 lines

milton