Skip to contentSkip to conversation

mtd: rawnand: Use the ->init_data_interface() hook

PATCH In review Rickard Andersson 2 participants 4 messages May 14, 2020 – May 19, 2020 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected];

#1 lore raw

PATCH 1/2 mtd: rawnand: Use the ->init_data_interface() hook

From: Rickard x Andersson <[email protected]>

Check if the NAND vendor provided hook works and check that the NAND controller can handle the timings.

---
 drivers/mtd/nand/raw/nand_base.c | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)
drivers/mtd/nand/raw/nand_base.c +29 -2
index 8744f0033f78..974050a04f23 100644
@@ -955,6 +955,30 @@ static int nand_setup_data_interface(struct nand_chip *chip, int chipnr)
 	return ret;
 }
 
+static int nand_try_init_data_interface(struct nand_chip *chip)
+{
+	int ret = chip->ops.init_data_interface(chip);
+
+	if (!ret) {
+		/*
+		 * Pass NAND_DATA_IFACE_CHECK_ONLY to only check if the
+		 * controller supports the requested timings.
+		 */
+		ret = chip->controller->ops->setup_data_interface(chip,
+						 NAND_DATA_IFACE_CHECK_ONLY,
+						 &chip->data_interface);
+	}
+
+	if (ret) {
+		/* The provided data interface timings did not work */
+		chip->ops.init_data_interface = NULL;
+		memset(&chip->data_interface, 0,
+		       sizeof(struct nand_data_interface));
+	}
+
+	return ret;
+}
+
 /**
  * nand_choose_data_interface - find the best data interface and timings
  * @chip: The NAND chip
@@ -980,8 +1004,11 @@ static int nand_choose_data_interface(struct nand_chip *chip)
 	 * ->init_data_interface() is expected to update the entire chip's
 	 * nand_data_interface structure.
 	 */
-	if (nand_has_init_data_interface(chip))
-		return chip->ops.init_data_interface(chip);
+	if (nand_has_init_data_interface(chip)) {
+		ret = nand_try_init_data_interface(chip);
+		if (!ret)
+			return 0;
+	}
 
 	/*
 	 * First try to identify the best timings from ONFI parameters and
Signature · 6 lines
2.11.0


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
#2 lore raw

PATCH 2/2 mtd: rawnand: Add timings for Kioxia TH58NVG2S3HBAI4

From: Rickard x Andersson <[email protected]>

The Kioxia/Toshiba TH58NVG2S3HBAI4 NAND memory is not a ONFI compliant memory. The timings of the memory is quite close to ONFI mode 4 but is breaking that spec.

Erase block read speed is increased from 6910 KiB/s to 13490 KiB/s. Erase block write speed is increased from 3350 KiB/s to 4410 KiB/s.

Tested on IMX6SX which has a NAND controller supporting EDO mode.

---
 drivers/mtd/nand/raw/nand_ids.c     |  3 ++
 drivers/mtd/nand/raw/nand_toshiba.c | 61 +++++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+)
drivers/mtd/nand/raw/nand_ids.c +3 -0
index e0dbc2e316c7..8b676e8b481b 100644
@@ -52,6 +52,9 @@ struct nand_flash_dev nand_flash_ids[] = {
 		{ .id = {0xad, 0xde, 0x94, 0xda, 0x74, 0xc4} },
 		  SZ_8K, SZ_8K, SZ_2M, NAND_NEED_SCRAMBLING, 6, 640,
 		  NAND_ECC_INFO(40, SZ_1K), 4 },
+	{"TH58NVG2S3HBAI4 4G 3.3V 8-bit",
+		{ .id = {0x98, 0xdc, 0x91, 0x15, 0x76} },
+		  SZ_2K, SZ_512, SZ_128K, 0, 5, 128, NAND_ECC_INFO(8, SZ_512) },
 
 	LEGACY_ID_NAND("NAND 4MiB 5V 8-bit",   0x6B, 4, SZ_8K, SP_OPTIONS),
 	LEGACY_ID_NAND("NAND 4MiB 3,3V 8-bit", 0xE3, 4, SZ_8K, SP_OPTIONS),
drivers/mtd/nand/raw/nand_toshiba.c +61 -0

85 lines not on this page · raw message

Signature · 6 lines
2.11.0


______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
maintainer
#3 lore raw

Re: [PATCH 2/2] mtd: rawnand: Add timings for Kioxia TH58NVG2S3HBAI4

Hi Rickard,

Rickard Andersson <[email protected]> wrote on Thu, 14 May 2020 11:13:42 +0200:

Quoted 38 lines

From: Rickard x Andersson <[email protected]>

The Kioxia/Toshiba TH58NVG2S3HBAI4 NAND memory is not a ONFI compliant memory. The timings of the memory is quite close to ONFI mode 4 but is breaking that spec.

Erase block read speed is increased from 6910 KiB/s to 13490 KiB/s. Erase block write speed is increased from 3350 KiB/s to 4410 KiB/s.

Tested on IMX6SX which has a NAND controller supporting EDO mode.

---
 drivers/mtd/nand/raw/nand_ids.c     |  3 ++
 drivers/mtd/nand/raw/nand_toshiba.c | 61 +++++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+)
drivers/mtd/nand/raw/nand_ids.c +3 -0
index e0dbc2e316c7..8b676e8b481b 100644
@@ -52,6 +52,9 @@ struct nand_flash_dev nand_flash_ids[] = {
 		{ .id = {0xad, 0xde, 0x94, 0xda, 0x74, 0xc4} },
 		  SZ_8K, SZ_8K, SZ_2M, NAND_NEED_SCRAMBLING, 6, 640,
 		  NAND_ECC_INFO(40, SZ_1K), 4 },
+	{"TH58NVG2S3HBAI4 4G 3.3V 8-bit",
+		{ .id = {0x98, 0xdc, 0x91, 0x15, 0x76} },
+		  SZ_2K, SZ_512, SZ_128K, 0, 5, 128, NAND_ECC_INFO(8, SZ_512) },
 
 	LEGACY_ID_NAND("NAND 4MiB 5V 8-bit",   0x6B, 4, SZ_8K, SP_OPTIONS),
 	LEGACY_ID_NAND("NAND 4MiB 3,3V 8-bit", 0xE3, 4, SZ_8K, SP_OPTIONS),
drivers/mtd/nand/raw/nand_toshiba.c +0 -0
index b6efaf5195bb..380499cfa491 100644

drivers/mtd/nand/raw/nand_toshiba.c line 31

Quoted 6 lines
drivers/mtd/nand/raw/nand_toshiba.c +3 -0
@@ -26,6 +26,52 @@
 /* Max ECC Steps for BENAND */
 #define TOSHIBA_NAND_MAX_ECC_STEPS		8
 
+static const struct nand_data_interface th58nvg2s3hbai4_timings = {
+	.type = NAND_SDR_IFACE,
+	.timings.mode = 0,

I just a branch with a new helper to fill this field, it is called onfi_find_equivalent_sdr_mode().

It is only compile tested so please verify it works.

Quoted 47 lines
drivers/mtd/nand/raw/nand_toshiba.c +43 -0
+	.timings.sdr = {
+		.tPROG_max = 700000000,
+		.tBERS_max = 5000000000,
+		.tCCS_min = 500000,
+		.tR_max = 200000000,
+		.tADL_min = 400000,
+		.tALH_min = 5000,
+		.tALS_min = 12000,
+		.tAR_min = 10000,
+		.tCEA_max = 25000,
+		.tCEH_min = 20000,
+		.tCH_min = 5000,
+		.tCHZ_max = 20000,
+		.tCLH_min = 5000,
+		.tCLR_min = 10000,
+		.tCLS_min = 12000,
+		.tCOH_min = 0,
+		.tCS_min = 20000,
+		.tDH_min = 5000,
+		.tDS_min = 12000,
+		.tFEAT_max = 1000000,
+		.tIR_min = 0,
+		.tITC_max = 1000000,
+		.tRC_min = 25000,
+		.tREA_max = 20000,
+		.tREH_min = 10000,
+		.tRHOH_min = 25000,
+		.tRHW_min = 30000,
+		.tRHZ_max = 60000,
+		.tRLOH_min = 5000,
+		.tRP_min = 12000,
+		.tRR_min = 20000,
+		.tRST_max = 500000000,
+		.tWB_max = 100000,
+		.tWC_min = 25000,
+		.tWH_min = 10000,
+		.tWHR_min = 60000,
+		.tWP_min = 12000,
+		.tWW_min = 100000,
+	}
+};
+
+
 static int toshiba_nand_benand_read_eccstatus_op(struct nand_chip
 *chip, u8 *buf)
 {

drivers/mtd/nand/raw/nand_toshiba.c static void toshiba_nand_decode_id(struct · line 243

Quoted 4 lines
drivers/mtd/nand/raw/nand_toshiba.c +1 -0
@@ -194,6 +240,13 @@ static void toshiba_nand_decode_id(struct
 nand_chip *chip) }
 }
 
+static int th58nvg2s3hbai4_init_data_interface(struct nand_chip

I renamed the hook so the helper should be
th58..._choose_data_interface()

drivers/mtd/nand/raw/nand_toshiba.c static void toshiba_nand_decode_id(struct · line 245

Quoted 2 lines
drivers/mtd/nand/raw/nand_toshiba.c +1 -0
 *chip) +{
+	chip->data_interface = th58nvg2s3hbai4_timings;

this data interface should be tested against the controller's capabilities and return an error otherwise.

Please check that, in case of error (simulate it) it fallbacks to mode 0 and does not fail silently.

Quoted 22 lines
drivers/mtd/nand/raw/nand_toshiba.c +10 -0
+
+	return 0;
+}
+
 static int tc58teg5dclta00_init(struct nand_chip *chip)
 {
 	struct mtd_info *mtd = nand_to_mtd(chip);
@@ -205,6 +258,12 @@ static int tc58teg5dclta00_init(struct nand_chip
 *chip) return 0;
 }
 
+static int th58nvg2s3hbai4_init(struct nand_chip *chip)
+{
+	chip->ops.init_data_interface =
 th58nvg2s3hbai4_init_data_interface;
+	return 0;
+}
+
 static int toshiba_nand_init(struct nand_chip *chip)
 {
drivers/mtd/nand/raw/nand_toshiba.c +0 -0
 	if (nand_is_slc(chip))

drivers/mtd/nand/raw/nand_toshiba.c static int toshiba_nand_init(struct nand_chip · line 283

Quoted 8 lines
drivers/mtd/nand/raw/nand_toshiba.c +2 -0
@@ -217,6 +276,8 @@ static int toshiba_nand_init(struct nand_chip
 *chip) 
 	if (!strcmp("TC58TEG5DCLTA00", chip->parameters.model))
 		tc58teg5dclta00_init(chip);
+	if (!strncmp("TH58NVG2S3HBAI4", chip->parameters.model, 15))
+		th58nvg2s3hbai4_init(chip);
 
 	return 0;
 }

Thanks,
Miquèl

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

maintainer
#4 lore raw
In reply to Rickard Andersson

Hi Rickard,

Rickard Andersson <[email protected]> wrote on Thu, 14 May 2020 11:13:41 +0200:

Quoted 20 lines

From: Rickard x Andersson <[email protected]>

Check if the NAND vendor provided hook works and check that the NAND controller can handle the timings.

---
 drivers/mtd/nand/raw/nand_base.c | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)
drivers/mtd/nand/raw/nand_base.c +2 -0
index 8744f0033f78..974050a04f23 100644
@@ -955,6 +955,30 @@ static int nand_setup_data_interface(struct nand_chip *chip, int chipnr)
 	return ret;
 }
 
+static int nand_try_init_data_interface(struct nand_chip *chip)
+{

drivers/mtd/nand/raw/nand_base.c static int nand_setup_data_interface(struct nand_chip *chip, int chipnr) · line 969

Quoted 10 lines
drivers/mtd/nand/raw/nand_base.c +10 -0
+	int ret = chip->ops.init_data_interface(chip);
+
+	if (!ret) {
+		/*
+		 * Pass NAND_DATA_IFACE_CHECK_ONLY to only check if the
+		 * controller supports the requested timings.
+		 */
+		ret = chip->controller->ops->setup_data_interface(chip,
+						 NAND_DATA_IFACE_CHECK_ONLY,
+						 &chip->data_interface);

I would like this to happen in Toshiba's code. Imagine you can try several timings, the vendor code should be able to check by itself the controller supports the timings.

I think you can drop this "try init data interface" and move the content to Toshiba's driver.

Quoted 16 lines
drivers/mtd/nand/raw/nand_base.c +12 -0
+	}
+
+	if (ret) {
+		/* The provided data interface timings did not work */
+		chip->ops.init_data_interface = NULL;
+		memset(&chip->data_interface, 0,
+		       sizeof(struct nand_data_interface));
+	}
+
+	return ret;
+}
+
 /**
  * nand_choose_data_interface - find the best data interface and timings
  * @chip: The NAND chip

drivers/mtd/nand/raw/nand_base.c static int nand_choose_data_interface(struct nand_chip *chip) · line 984

Quoted 5 lines
drivers/mtd/nand/raw/nand_base.c +0 -2
@@ -980,8 +1004,11 @@ static int nand_choose_data_interface(struct nand_chip *chip)
 	 * ->init_data_interface() is expected to update the entire chip's
 	 * nand_data_interface structure.
 	 */
-	if (nand_has_init_data_interface(chip))
-		return chip->ops.init_data_interface(chip);

I renamed it "choose_data_interface" now, you can pull changes from Github.

drivers/mtd/nand/raw/nand_base.c static int nand_choose_data_interface(struct nand_chip *chip) · line 1010

Quoted 4 lines
drivers/mtd/nand/raw/nand_base.c +4 -0
+	if (nand_has_init_data_interface(chip)) {
+		ret = nand_try_init_data_interface(chip);
+		if (!ret)
+			return 0;

I'm fine with a fallback on the regular ONFI research though, which in this case would choose timings mode 0 I guess. Can you validate?

On my side I changed the code on Github and now a negative return code just silently fails so that we fallback on mode 0.

drivers/mtd/nand/raw/nand_base.c static int nand_choose_data_interface(struct nand_chip *chip) · line 1014

Quoted 4 lines
drivers/mtd/nand/raw/nand_base.c +1 -0
+	}
 
 	/*
 	 * First try to identify the best timings from ONFI parameters and

Thanks,
Miquèl

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/