mtd: spinand: Add continuous read mode support
Subsystems (2)
cover mtd: spinand: Add continuous read mode support
This series add support continuous read mode for SPI NAND.
JaimeLiao (4):
mtd: spinand: Add support continuous read mode
mtd: spinand: Add continuous read state
mtd: spinand: Add support continuous read operation
mtd: spinand: macronix: Add continuous read support for MX35LF2GE4AD
drivers/mtd/nand/spi/core.c | 118 ++++++++++++++++++++++++++++++++ drivers/mtd/nand/spi/macronix.c | 2 +- include/linux/mtd/spinand.h | 4 ++ 3 files changed, 123 insertions(+), 1 deletion(-)
Signature · 6 lines
2.17.1 ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/
PATCH 1/4 mtd: spinand: Add support continuous read mode
The patch supports setting the "CONT" bit of the configuration register and create spinand_continuous_read_enable/disable functions.
--- drivers/mtd/nand/spi/core.c | 11 +++++++++++ include/linux/mtd/spinand.h | 1 + 2 files changed, 12 insertions(+)
drivers/mtd/nand/spi/core.c +11 -0
@@ -193,6 +193,17 @@ static int spinand_init_quad_enable(struct spinand_device *spinand) enable ? CFG_QUAD_ENABLE : 0); } +static int spinand_continuous_read_enable (struct spinand_device *spinand) +{ +return spinand_upd_cfg (spinand,CFG_CONT_READ_ENABLE , +CFG_CONT_READ_ENABLE ); +} + +static int spinand_continuous_read_disable (struct spinand_device *spinand) +{ +return spinand_upd_cfg (spinand,CFG_CONT_READ_ENABLE ,0 ); +} +static int spinand_ecc_enable (struct spinand_device *spinand,bool enable) {
include/linux/mtd/spinand.h +1 -0
@@ -154,6 +154,7 @@#define REG_CFG 0xb0#define CFG_OTP_ENABLE BIT(6)#define CFG_ECC_ENABLE BIT(4) +#define CFG_CONT_READ_ENABLE BIT(2)#define CFG_QUAD_ENABLE BIT(0)/* status register */
Signature · 6 lines
2.17.1 ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/
PATCH 2/4 mtd: spinand: Add continuous read state
In reply to JaimeLiaoAdd continuous read state and initialize state to
default false.
--- drivers/mtd/nand/spi/core.c | 8 ++++++++ include/linux/mtd/spinand.h | 2 ++ 2 files changed, 10 insertions(+)
drivers/mtd/nand/spi/core.c +8 -0
@@ -193,6 +193,11 @@ static int spinand_init_quad_enable(struct spinand_device *spinand) enable ? CFG_QUAD_ENABLE : 0); } +static void spinand_init_continuous_read (struct spinand_device *spinand) +{ + spinand->use_continuous_read = false; +} +static int spinand_continuous_read_enable (struct spinand_device *spinand) { return spinand_upd_cfg(spinand, CFG_CONT_READ_ENABLE,
@@ -1261,6 +1266,9 @@ static int spinand_init(struct spinand_device *spinand) mtd->ecc_strength =nanddev_get_ecc_conf (nand)->strength; mtd->ecc_step_size =nanddev_get_ecc_conf (nand)->step_size; +/* Init continuous read */ +spinand_init_continuous_read (spinand); +return 0 ; err_cleanup_ecc_engine:
include/linux/mtd/spinand.h +2 -0
@@ -415,6 +415,7 @@ struct spinand_dirmap {* because the spi-mem interface explicitly requests that buffers * passed in spi_mem_op be DMA-able, so we can't based the bufs on * the stack +* @use_continuous_read: record the continuous read status * @manufacturer: SPI NAND manufacturer information * @priv: manufacturer private data */
@@ -443,6 +444,7 @@ struct spinand_device {u8 *databuf;u8 *oobbuf;u8 *scratchbuf; +bool use_continuous_read;const struct spinand_manufacturer *manufacturer;void *priv; };
Signature · 6 lines
2.17.1 ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/
PATCH 3/4 mtd: spinand: Add support continuous read operation
In reply to JaimeLiaoThe continuous read operation includes three phases:
Firstly, starting with the page read command and the 1st page data will be read into the cache after the read latency tRD. Secondly, Issuing the Read From Cache commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache continuously. After all the data is read out, the host should pull CS# high to terminate this continuous read operation and wait tRST for the NAND device resets read operation.
Continuous reads have a positive impact when reading more than one page and the column address is don't care in this operation, a full page data will be read out for each page.
The performance of continuous read mode is as follows. Set the flash to QUAD mode and run 25MZ direct mapping mode on the SPI bus and use the MTD test module to show the performance of continuous reads.
=================================================mtd_speedtest: MTD device: 0 count: 100mtd_speedtest: MTD device size 268435456, eraseblock size 131072,
page size 2048, count of eraseblocks 2048, pages per eraseblock 64, OOB size 64mtd_test: scanning for bad eraseblocksmtd_test: scanned 100 eraseblocks, 0 are badmtd_speedtest: testing eraseblock write speedmtd_speedtest: eraseblock write speed is 1298 KiB/smtd_speedtest: testing eraseblock read speedmtd_speedtest: eraseblock read speed is 11053 KiB/smtd_speedtest: testing page write speedmtd_speedtest: page write speed is 1291 KiB/smtd_speedtest: testing page read speedmtd_speedtest: page read speed is 3240 KiB/smtd_speedtest: testing 2 page write speedmtd_speedtest: 2 page write speed is 1289 KiB/smtd_speedtest: testing 2 page read speedmtd_speedtest: 2 page read speed is 2909 KiB/smtd_speedtest: Testing erase speedmtd_speedtest: erase speed is 45229 KiB/smtd_speedtest: Testing 2x multi-block erase speedmtd_speedtest: 2x multi-block erase speed is 62135 KiB/smtd_speedtest: Testing 4x multi-block erase speedmtd_speedtest: 4x multi-block erase speed is 60093 KiB/smtd_speedtest: Testing 8x multi-block erase speedmtd_speedtest: 8x multi-block erase speed is 61244 KiB/smtd_speedtest: Testing 16x multi-block erase speedmtd_speedtest: 16x multi-block erase speed is 61538 KiB/smtd_speedtest: Testing 32x multi-block erase speedmtd_speedtest: 32x multi-block erase speed is 61835 KiB/smtd_speedtest: Testing 64x multi-block erase speedmtd_speedtest: 64x multi-block erase speed is 60663 KiB/smtd_speedtest: finished
=================================================
--- drivers/mtd/nand/spi/core.c | 99 +++++++++++++++++++++++++++++++++++++ include/linux/mtd/spinand.h | 1 + 2 files changed, 100 insertions(+)
drivers/mtd/nand/spi/core.c +99 -0
@@ -386,6 +386,10 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand,if (req->datalen) { buf = spinand->databuf; nbytes =nanddev_page_size (nand); +if (spinand->use_continuous_read) { + buf = req->databuf.in; + nbytes = req->datalen; + } column =0 ; }
@@ -412,6 +416,9 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand, buf += ret; } +if (spinand->use_continuous_read) +return 0 ; +if (req->datalen)memcpy (req->databuf.in, spinand->databuf + req->dataoffs, req->datalen);
@@ -640,6 +647,80 @@ static int spinand_write_page(struct spinand_device *spinand,return nand_ecc_finish_io_req (nand, (struct nand_page_io_req *)req); } +static int spinand_mtd_continuous_read (struct mtd_info *mtd,loff_t from, +struct mtd_oob_ops *ops, +struct nand_io_iter *iter) +{ +struct spinand_device *spinand =mtd_to_spinand (mtd); +struct nand_device *nand =mtd_to_nanddev (mtd); +int ret =0 ; + +/* +* Continuous read mode could reduce some operation in On-die ECC free +* flash when read page sequentially. +*/ + iter->req.type =NAND_PAGE_READ ; + iter->req.mode =MTD_OPS_RAW ; + iter->req.dataoffs =nanddev_offs_to_pos (nand, from, &iter->req.pos); + iter->req.databuf.in = ops->datbuf; + iter->req.datalen = ops->len; + +if (from & (nanddev_page_size (nand) -1 )) { +pr_debug ("%s: unaligned address\n" , __func__); +return -EINVAL ; + } + + ret =spinand_continuous_read_enable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = true; + + ret =spinand_select_target (spinand, iter->req.pos.target); +if (ret) +return ret; + +/* +* The continuous read operation including: firstly, starting with the +* page read command and the 1 st page data will be read into the cache +* after the read latency tRD. Secondly, Issuing the Read From Cache +* commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache +* continuously. +* +* The cache is divided into two halves, while one half of the cache is +* outputting the data, the other half will be loaded for the new data; +* therefore, the host can read out the data continuously from page to +* page. Multiple of Read From Cache commands can be issued in one +* continuous read operation, each Read From Cache command is required +* to read multiple 4-byte data exactly; otherwise, the data output will +* be out of sequence from one Read From Cache command to another Read +* From Cache command. +* +* After all the data is read out, the host should pull CS# high to +* terminate this continuous read operation and wait a 6us of tRST for +* the NAND device resets read operation. The data output for each page +* will always start from byte 0 and a full page data should be read out +* for each page. +*/ + ret =spinand_read_page (spinand, &iter->req); +if (ret) +return ret; + + ret =spinand_reset_op (spinand); +if (ret) +return ret; + + ret =spinand_continuous_read_disable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = false; + + ops->retlen = iter->req.datalen; + +return ret; +} +static int spinand_mtd_read (struct mtd_info *mtd,loff_t from,struct mtd_oob_ops *ops) {
@@ -656,6 +737,24 @@ static int spinand_mtd_read(struct mtd_info *mtd, loff_t from,mutex_lock (&spinand->lock); +/* +* If the device support continuous read mode and read length larger +* than one page size will enter the continuous read mode. This mode +* helps avoid issuing a page read command and read from cache command +* again, and improves read performance for continuous addresses. +*/ +if ((spinand->flags &SPINAND_HAS_CONT_READ_BIT ) && + (ops->len >nanddev_page_size (nand))) { + ret =spinand_mtd_continuous_read (mtd, from, ops, &iter); + +mutex_unlock (&spinand->lock); + +if (ecc_failed && !ret) + ret = -EBADMSG ; + +return ret ? ret : max_bitflips; + } +nanddev_io_for_each_page (nand,NAND_PAGE_READ , from, ops, &iter) {if (disable_ecc) iter.req.mode =MTD_OPS_RAW ;
include/linux/mtd/spinand.h +1 -0
@@ -308,6 +308,7 @@ struct spinand_ecc_info {#define SPINAND_HAS_QE_BIT BIT(0)#define SPINAND_HAS_CR_FEAT_BIT BIT(1) +#define SPINAND_HAS_CONT_READ_BIT BIT(2)/** * struct spinand_ondie_ecc_conf - private SPI-NAND on-die ECC engine structure
Signature · 6 lines
2.17.1 ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/
Miquel Raynal
maintainerRe: [PATCH v1 3/4] mtd: spinand: Add support continuous read operation
Hi JaimeLiao,
[email protected] wrote on Thu, 11 Aug 2022 17:45:35 +0800:
Quoted 9 lines
The continuous read operation includes three phases:
Firstly, starting with the page read command and the 1st page data will be read into the cache after the read latency tRD. Secondly, Issuing the Read From Cache commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache continuously. After all the data is read out, the host should pull CS# high to terminate this continuous read operation and wait tRST for the NAND device resets read operation.
Quoted 6 lines
Continuous reads have a positive impact when reading more than one page and the column address is don't care in this operation, a full page data will be read out for each page.
The performance of continuous read mode is as follows. Set the flash to QUAD mode and run 25MZ direct mapping mode on the SPI
MHz
Quoted 2 lines
bus and use the MTD test module to show the performance of continuous reads.
Please show before and after changes and just give us the lines regarding the read speeds rather than all the output of the tool.
Quoted 48 lines
=================================================
mtd_speedtest: MTD device: 0 count: 100mtd_speedtest: MTD device size 268435456, eraseblock size 131072,
page size 2048, count of eraseblocks 2048, pages per eraseblock 64, OOB size 64mtd_test: scanning for bad eraseblocksmtd_test: scanned 100 eraseblocks, 0 are badmtd_speedtest: testing eraseblock write speedmtd_speedtest: eraseblock write speed is 1298 KiB/smtd_speedtest: testing eraseblock read speedmtd_speedtest: eraseblock read speed is 11053 KiB/smtd_speedtest: testing page write speedmtd_speedtest: page write speed is 1291 KiB/smtd_speedtest: testing page read speedmtd_speedtest: page read speed is 3240 KiB/smtd_speedtest: testing 2 page write speedmtd_speedtest: 2 page write speed is 1289 KiB/smtd_speedtest: testing 2 page read speedmtd_speedtest: 2 page read speed is 2909 KiB/smtd_speedtest: Testing erase speedmtd_speedtest: erase speed is 45229 KiB/smtd_speedtest: Testing 2x multi-block erase speedmtd_speedtest: 2x multi-block erase speed is 62135 KiB/smtd_speedtest: Testing 4x multi-block erase speedmtd_speedtest: 4x multi-block erase speed is 60093 KiB/smtd_speedtest: Testing 8x multi-block erase speedmtd_speedtest: 8x multi-block erase speed is 61244 KiB/smtd_speedtest: Testing 16x multi-block erase speedmtd_speedtest: 16x multi-block erase speed is 61538 KiB/smtd_speedtest: Testing 32x multi-block erase speedmtd_speedtest: 32x multi-block erase speed is 61835 KiB/smtd_speedtest: Testing 64x multi-block erase speedmtd_speedtest: 64x multi-block erase speed is 60663 KiB/smtd_speedtest: finished
=================================================Signed-off-by JaimeLiao <[email protected]>--- drivers/mtd/nand/spi/core.c | 99 +++++++++++++++++++++++++++++++++++++ include/linux/mtd/spinand.h | 1 + 2 files changed, 100 insertions(+)drivers/mtd/nand/spi/core.c +0
-0
drivers/mtd/nand/spi/core.c static int spinand_read_from_cache_op(struct spinand_device *spinand, · line 392
Quoted 7 lines
drivers/mtd/nand/spi/core.c +4
-0@@ -386,6 +386,10 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand,if (req->datalen) { buf = spinand->databuf; nbytes =nanddev_page_size (nand); +if (spinand->use_continuous_read) { + buf = req->databuf.in; + nbytes = req->datalen; + }
Does not look right. A cache read is still one page, and I believe the caller should have taken care of that?
drivers/mtd/nand/spi/core.c static int spinand_read_from_cache_op(struct spinand_device *spinand, · line 420
Quoted 9 lines
drivers/mtd/nand/spi/core.c +2
-0column =0 ; }@@ -412,6 +416,9 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand, buf += ret; } +if (spinand->use_continuous_read) +return 0 ;
For me this is not the right place.
Quoted 12 lines
drivers/mtd/nand/spi/core.c +5
-0+if (req->datalen)memcpy (req->databuf.in, spinand->databuf + req->dataoffs, req->datalen);@@ -640,6 +647,80 @@ static int spinand_write_page(struct spinand_device *spinand,return nand_ecc_finish_io_req (nand, (struct nand_page_io_req *)req); } +static int spinand_mtd_continuous_read (struct mtd_info *mtd,loff_t from, +struct mtd_oob_ops *ops, +struct nand_io_iter *iter) +{
drivers/mtd/nand/spi/core.c static int spinand_write_page(struct spinand_device *spinand, · line 663
Quoted 10 lines
drivers/mtd/nand/spi/core.c +10
-0+struct spinand_device *spinand =mtd_to_spinand (mtd); +struct nand_device *nand =mtd_to_nanddev (mtd); +int ret =0 ; + +/* +* Continuous read mode could reduce some operation in On-die ECC free +* flash when read page sequentially. +*/ + iter->req.type =NAND_PAGE_READ ; + iter->req.mode =MTD_OPS_RAW ;
I don't get what you do here.
Continuous read shall either work with ECC engines or just not be implemented at all. You cannot blindly disable ECC support.
Quoted 44 lines
drivers/mtd/nand/spi/core.c +44
-0+ iter->req.dataoffs =nanddev_offs_to_pos (nand, from, &iter->req.pos); + iter->req.databuf.in = ops->datbuf; + iter->req.datalen = ops->len; + +if (from & (nanddev_page_size (nand) -1 )) { +pr_debug ("%s: unaligned address\n" , __func__); +return -EINVAL ; + } + + ret =spinand_continuous_read_enable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = true; + + ret =spinand_select_target (spinand, iter->req.pos.target); +if (ret) +return ret; + +/* +* The continuous read operation including: firstly, starting with the +* page read command and the 1 st page data will be read into the cache +* after the read latency tRD. Secondly, Issuing the Read From Cache +* commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache +* continuously. +* +* The cache is divided into two halves, while one half of the cache is +* outputting the data, the other half will be loaded for the new data; +* therefore, the host can read out the data continuously from page to +* page. Multiple of Read From Cache commands can be issued in one +* continuous read operation, each Read From Cache command is required +* to read multiple 4-byte data exactly; otherwise, the data output will +* be out of sequence from one Read From Cache command to another Read +* From Cache command. +* +* After all the data is read out, the host should pull CS# high to +* terminate this continuous read operation and wait a 6us of tRST for +* the NAND device resets read operation. The data output for each page +* will always start from byte 0 and a full page data should be read out +* for each page. +*/ + ret =spinand_read_page (spinand, &iter->req); +if (ret) +return ret;
drivers/mtd/nand/spi/core.c static int spinand_write_page(struct spinand_device *spinand, · line 717
Quoted 10 lines
drivers/mtd/nand/spi/core.c +10
-0+ + ret =spinand_reset_op (spinand); +if (ret) +return ret; + + ret =spinand_continuous_read_disable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = false;
Any error in this function should lead to resetting use_continuous_read to false and disabling continuous read on the flash side.
Quoted 39 lines
drivers/mtd/nand/spi/core.c +24
-0+ + ops->retlen = iter->req.datalen; + +return ret; +} +static int spinand_mtd_read (struct mtd_info *mtd,loff_t from,struct mtd_oob_ops *ops) {@@ -656,6 +737,24 @@ static int spinand_mtd_read(struct mtd_info *mtd, loff_t from,mutex_lock (&spinand->lock); +/* +* If the device support continuous read mode and read length larger +* than one page size will enter the continuous read mode. This mode +* helps avoid issuing a page read command and read from cache command +* again, and improves read performance for continuous addresses. +*/ +if ((spinand->flags &SPINAND_HAS_CONT_READ_BIT ) && + (ops->len >nanddev_page_size (nand))) { + ret =spinand_mtd_continuous_read (mtd, from, ops, &iter); + +mutex_unlock (&spinand->lock); + +if (ecc_failed && !ret) + ret = -EBADMSG ; + +return ret ? ret : max_bitflips; + } +nanddev_io_for_each_page (nand,NAND_PAGE_READ , from, ops, &iter) {if (disable_ecc) iter.req.mode =MTD_OPS_RAW ;include/linux/mtd/spinand.h +0
-0
include/linux/mtd/spinand.h struct spinand_ecc_info { · line 311
Quoted 4 lines
include/linux/mtd/spinand.h +1
-0@@ -308,6 +308,7 @@ struct spinand_ecc_info {#define SPINAND_HAS_QE_BIT BIT(0)#define SPINAND_HAS_CR_FEAT_BIT BIT(1) +#define SPINAND_HAS_CONT_READ_BIT BIT(2)
Is this bit standard?
include/linux/mtd/spinand.h struct spinand_ecc_info { · line 314
Quoted 3 lines
include/linux/mtd/spinand.h +0
-0/** * struct spinand_ondie_ecc_conf - private SPI-NAND on-die ECC engine structure
Thanks,
Miquèl
______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
Re: [PATCH v1 3/4] mtd: spinand: Add support continuous read operation
Hi Miquel
Quoted 14 lines
Hi JaimeLiao,
[email protected] wrote on Thu, 11 Aug 2022 17:45:35 +0800:
Quoted 9 lines
The continuous read operation includes three phases:
Firstly, starting with the page read command and the 1st page data will be read into the cache after the read latency tRD. Secondly, Issuing the Read From Cache commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache continuously. After all the data is read out, the host should pull CS# high to terminate this continuous read operation and wait tRST for the NAND device resets read operation.
Quoted 8 lines
Quoted 6 lines
Continuous reads have a positive impact when reading more than one page and the column address is don't care in this operation, a full page data will be read out for each page.
The performance of continuous read mode is as follows. Set the flash to QUAD mode and run 25MZ direct mapping mode on the SPI
MHz
Thanks, this will be correct next patch.
Quoted 6 lines
Quoted 3 lines
bus and use the MTD test module to show the performance of continuous reads.
Please show before and after changes and just give us the lines regarding the read speeds rather than all the output of the tool.
Ok, I will attach test report for two cases.
Quoted 44 lines
Quoted 44 lines
=================================================
mtd_speedtest: MTD device: 0 count: 100mtd_speedtest: MTD device size 268435456, eraseblock size 131072,
page size 2048, count of eraseblocks 2048, pages per eraseblock 64, OOB size 64mtd_test: scanning for bad eraseblocksmtd_test: scanned 100 eraseblocks, 0 are badmtd_speedtest: testing eraseblock write speedmtd_speedtest: eraseblock write speed is 1298 KiB/smtd_speedtest: testing eraseblock read speedmtd_speedtest: eraseblock read speed is 11053 KiB/smtd_speedtest: testing page write speedmtd_speedtest: page write speed is 1291 KiB/smtd_speedtest: testing page read speedmtd_speedtest: page read speed is 3240 KiB/smtd_speedtest: testing 2 page write speedmtd_speedtest: 2 page write speed is 1289 KiB/smtd_speedtest: testing 2 page read speedmtd_speedtest: 2 page read speed is 2909 KiB/smtd_speedtest: Testing erase speedmtd_speedtest: erase speed is 45229 KiB/smtd_speedtest: Testing 2x multi-block erase speedmtd_speedtest: 2x multi-block erase speed is 62135 KiB/smtd_speedtest: Testing 4x multi-block erase speedmtd_speedtest: 4x multi-block erase speed is 60093 KiB/smtd_speedtest: Testing 8x multi-block erase speedmtd_speedtest: 8x multi-block erase speed is 61244 KiB/smtd_speedtest: Testing 16x multi-block erase speedmtd_speedtest: 16x multi-block erase speed is 61538 KiB/smtd_speedtest: Testing 32x multi-block erase speedmtd_speedtest: 32x multi-block erase speed is 61835 KiB/smtd_speedtest: Testing 64x multi-block erase speedmtd_speedtest: 64x multi-block erase speed is 60663 KiB/smtd_speedtest: finished
=================================================Signed-off-by JaimeLiao <[email protected]>--- drivers/mtd/nand/spi/core.c | 99 +++++++++++++++++++++++++++++++++++++ include/linux/mtd/spinand.h | 1 + 2 files changed, 100 insertions(+)
Quoted 15 lines
drivers/mtd/nand/spi/core.c static int spinand_read_from_cache_op(struct spinand_device *spinand, · line 392
Quoted 12 lines
drivers/mtd/nand/spi/core.c +4
-0@@ -386,6 +386,10 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand,if (req->datalen) { buf = spinand->databuf; nbytes =nanddev_page_size (nand); +if (spinand->use_continuous_read) { + buf = req->databuf.in; + nbytes = req->datalen; + }Does not look right. A cache read is still one page, and I believe the caller should have taken care of that?
For this part, buf is req->databuf.in.
Multi page could be receive at this operation.
Quoted 4 lines
Quoted 4 lines
drivers/mtd/nand/spi/core.c +0
-0column =0 ; }
Quoted 8 lines
drivers/mtd/nand/spi/core.c static int spinand_read_from_cache_op(struct spinand_device *spinand, · line 420
Quoted 6 lines
drivers/mtd/nand/spi/core.c +2
-0@@ -412,6 +416,9 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand, buf += ret; } +if (spinand->use_continuous_read) +return 0 ;For me this is not the right place.
Would it be better If using "goto"?
Quoted 26 lines
Quoted 13 lines
drivers/mtd/nand/spi/core.c +5
-0+if (req->datalen)memcpy (req->databuf.in, spinand->databuf + req->dataoffs, req->datalen);@@ -640,6 +647,80 @@ static int spinand_write_page(struct spinand_device *spinand,return nand_ecc_finish_io_req (nand, (struct nand_page_io_req *)req); } +static int spinand_mtd_continuous_read (struct mtd_info *mtd,loff_t from, +struct mtd_oob_ops *ops, +struct nand_io_iter *iter) +{drivers/mtd/nand/spi/core.c static int spinand_write_page(struct spinand_device *spinand, · line 663
Quoted 10 lines
drivers/mtd/nand/spi/core.c +10
-0+struct spinand_device *spinand =mtd_to_spinand (mtd); +struct nand_device *nand =mtd_to_nanddev (mtd); +int ret =0 ; + +/* +* Continuous read mode could reduce some operation in On-die ECC free +* flash when read page sequentially. +*/ + iter->req.type =NAND_PAGE_READ ; + iter->req.mode =MTD_OPS_RAW ;I don't get what you do here.
Quoted 2 lines
Continuous read shall either work with ECC engines or just not be implemented at all. You cannot blindly disable ECC support.
For Macronix continuous read feature is for "On-die ECC" flash only. So that I think ecc have been handled in flash.
Quoted 50 lines
Quoted 50 lines
drivers/mtd/nand/spi/core.c +49
-0+ iter->req.dataoffs =nanddev_offs_to_pos (nand, from, &iter->req.pos); + iter->req.databuf.in = ops->datbuf; + iter->req.datalen = ops->len; + +if (from & (nanddev_page_size (nand) -1 )) { +pr_debug ("%s: unaligned address\n" , __func__); +return -EINVAL ; + } + + ret =spinand_continuous_read_enable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = true; + + ret =spinand_select_target (spinand, iter->req.pos.target); +if (ret) +return ret; + +/* +* The continuous read operation including: firstly, starting with the +* page read command and the 1 st page data will be read into the cache +* after the read latency tRD. Secondly, Issuing the Read From Cache +* commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache +* continuously. +* +* The cache is divided into two halves, while one half of the cache is +* outputting the data, the other half will be loaded for the new data; +* therefore, the host can read out the data continuously from page to +* page. Multiple of Read From Cache commands can be issued in one +* continuous read operation, each Read From Cache command is required +* to read multiple 4-byte data exactly; otherwise, the data output will +* be out of sequence from one Read From Cache command to another Read +* From Cache command. +* +* After all the data is read out, the host should pull CS# high to +* terminate this continuous read operation and wait a 6us of tRST for +* the NAND device resets read operation. The data output for each page +* will always start from byte 0 and a full page data should be read out +* for each page. +*/ + ret =spinand_read_page (spinand, &iter->req); +if (ret) +return ret; + + ret =spinand_reset_op (spinand); +if (ret) +return ret; +
Quoted 8 lines
drivers/mtd/nand/spi/core.c static int spinand_write_page(struct spinand_device *spinand, · line 717
Quoted 5 lines
drivers/mtd/nand/spi/core.c +5
-0+ ret =spinand_continuous_read_disable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = false;Any error in this function should lead to resetting
use_continuous_readto false and disabling continuous read on the flash side.
OK, this will be correct next version.
Quoted 41 lines
Quoted 41 lines
drivers/mtd/nand/spi/core.c +24
-0+ + ops->retlen = iter->req.datalen; + +return ret; +} +static int spinand_mtd_read (struct mtd_info *mtd,loff_t from,struct mtd_oob_ops *ops) {@@ -656,6 +737,24 @@ static int spinand_mtd_read(struct mtd_info *mtd, loff_t from,mutex_lock (&spinand->lock); +/* +* If the device support continuous read mode and read length larger +* than one page size will enter the continuous read mode. This mode +* helps avoid issuing a page read command and read from cache command +* again, and improves read performance for continuous addresses. +*/ +if ((spinand->flags &SPINAND_HAS_CONT_READ_BIT ) && + (ops->len >nanddev_page_size (nand))) { + ret =spinand_mtd_continuous_read (mtd, from, ops, &iter); + +mutex_unlock (&spinand->lock); + +if (ecc_failed && !ret) + ret = -EBADMSG ; + +return ret ? ret : max_bitflips; + } +nanddev_io_for_each_page (nand,NAND_PAGE_READ , from, ops, &iter) {if (disable_ecc) iter.req.mode =MTD_OPS_RAW ;include/linux/mtd/spinand.h +0
-0@@ -308,6 +308,7 @@ struct spinand_ecc_info {
Quoted 5 lines
include/linux/mtd/spinand.h struct spinand_ecc_info { · line 311
Quoted 3 lines
include/linux/mtd/spinand.h +1
-0#define SPINAND_HAS_QE_BIT BIT(0)#define SPINAND_HAS_CR_FEAT_BIT BIT(1) +#define SPINAND_HAS_CONT_READ_BIT BIT(2)Is this bit standard?
It is using to add into flash id table and judge whether continuous read feature is support or not.
Quoted 8 lines
include/linux/mtd/spinand.h struct spinand_ecc_info { · line 314
Quoted 4 lines
include/linux/mtd/spinand.h +0
-0/** * struct spinand_ondie_ecc_conf - private SPI-NAND on-die ECC engine structure Thanks,
Miquèl
Thanks
Jaime
______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
Miquel Raynal
maintainerRe: [PATCH v1 3/4] mtd: spinand: Add support continuous read operation
Hi Jaime,
[email protected] wrote on Wed, 8 Feb 2023 19:14:55 +0800:
Quoted 93 lines
Hi Miquel
Quoted 14 lines
Hi JaimeLiao,
[email protected] wrote on Thu, 11 Aug 2022 17:45:35 +0800:
Quoted 9 lines
The continuous read operation includes three phases:
Firstly, starting with the page read command and the 1st page data will be read into the cache after the read latency tRD. Secondly, Issuing the Read From Cache commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache continuously. After all the data is read out, the host should pull CS# high to terminate this continuous read operation and wait tRST for the NAND device resets read operation.Quoted 8 lines
Quoted 6 lines
Continuous reads have a positive impact when reading more than one page and the column address is don't care in this operation, a full page data will be read out for each page.
The performance of continuous read mode is as follows. Set the flash to QUAD mode and run 25MZ direct mapping mode on the SPI
MHz
Thanks, this will be correct next patch.
Quoted 6 lines
Quoted 3 lines
bus and use the MTD test module to show the performance of continuous reads.
Please show before and after changes and just give us the lines regarding the read speeds rather than all the output of the tool.
Ok, I will attach test report for two cases.
Quoted 44 lines
Quoted 44 lines
=================================================
mtd_speedtest: MTD device: 0 count: 100mtd_speedtest: MTD device size 268435456, eraseblock size 131072,
page size 2048, count of eraseblocks 2048, pages per eraseblock 64, OOB size 64mtd_test: scanning for bad eraseblocksmtd_test: scanned 100 eraseblocks, 0 are badmtd_speedtest: testing eraseblock write speedmtd_speedtest: eraseblock write speed is 1298 KiB/smtd_speedtest: testing eraseblock read speedmtd_speedtest: eraseblock read speed is 11053 KiB/smtd_speedtest: testing page write speedmtd_speedtest: page write speed is 1291 KiB/smtd_speedtest: testing page read speedmtd_speedtest: page read speed is 3240 KiB/smtd_speedtest: testing 2 page write speedmtd_speedtest: 2 page write speed is 1289 KiB/smtd_speedtest: testing 2 page read speedmtd_speedtest: 2 page read speed is 2909 KiB/smtd_speedtest: Testing erase speedmtd_speedtest: erase speed is 45229 KiB/smtd_speedtest: Testing 2x multi-block erase speedmtd_speedtest: 2x multi-block erase speed is 62135 KiB/smtd_speedtest: Testing 4x multi-block erase speedmtd_speedtest: 4x multi-block erase speed is 60093 KiB/smtd_speedtest: Testing 8x multi-block erase speedmtd_speedtest: 8x multi-block erase speed is 61244 KiB/smtd_speedtest: Testing 16x multi-block erase speedmtd_speedtest: 16x multi-block erase speed is 61538 KiB/smtd_speedtest: Testing 32x multi-block erase speedmtd_speedtest: 32x multi-block erase speed is 61835 KiB/smtd_speedtest: Testing 64x multi-block erase speedmtd_speedtest: 64x multi-block erase speed is 60663 KiB/smtd_speedtest: finished
=================================================Signed-off-by JaimeLiao <[email protected]>--- drivers/mtd/nand/spi/core.c | 99 +++++++++++++++++++++++++++++++++++++ include/linux/mtd/spinand.h | 1 + 2 files changed, 100 insertions(+)Quoted 15 lines
drivers/mtd/nand/spi/core.c static int spinand_read_from_cache_op(struct spinand_device *spinand, · line 392
Quoted 12 lines
drivers/mtd/nand/spi/core.c +4
-0@@ -386,6 +386,10 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand,if (req->datalen) { buf = spinand->databuf; nbytes =nanddev_page_size (nand); +if (spinand->use_continuous_read) { + buf = req->databuf.in; + nbytes = req->datalen; + }Does not look right. A cache read is still one page, and I believe the caller should have taken care of that?
Quoted 2 lines
For this part, buf is
req->databuf.in.
Multi page could be receive at this operation.
That is not what we expect from spinand_read_from_cache_op(). If you change the semantics, please make your own function.
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/mtd/nand/spi/core.c +0
-0column =0 ; }@@ -412,6 +416,9 @@ static int spinand_read_from_cache_op(struct spinand_device *spinand, buf += ret; }
Quoted 5 lines
Quoted 4 lines
drivers/mtd/nand/spi/core.c static int spinand_read_from_cache_op(struct spinand_device *spinand, · line 420
Quoted 2 lines
drivers/mtd/nand/spi/core.c +2
-0+if (spinand->use_continuous_read) +return 0 ;For me this is not the right place.
Would it be better If using "goto"?
I think I meant: "this is not the right function" to do that.
Quoted 27 lines
Quoted 27 lines
Quoted 14 lines
drivers/mtd/nand/spi/core.c +5
-0+if (req->datalen)memcpy (req->databuf.in, spinand->databuf + req->dataoffs, req->datalen);@@ -640,6 +647,80 @@ static int spinand_write_page(struct spinand_device *spinand,return nand_ecc_finish_io_req (nand, (struct nand_page_io_req *)req); } +static int spinand_mtd_continuous_read (struct mtd_info *mtd,loff_t from, +struct mtd_oob_ops *ops, +struct nand_io_iter *iter) +{drivers/mtd/nand/spi/core.c static int spinand_write_page(struct spinand_device *spinand, · line 663
Quoted 10 lines
drivers/mtd/nand/spi/core.c +10
-0+struct spinand_device *spinand =mtd_to_spinand (mtd); +struct nand_device *nand =mtd_to_nanddev (mtd); +int ret =0 ; + +/* +* Continuous read mode could reduce some operation in On-die ECC free +* flash when read page sequentially. +*/ + iter->req.type =NAND_PAGE_READ ; + iter->req.mode =MTD_OPS_RAW ;I don't get what you do here.
Quoted 4 lines
Quoted 2 lines
Continuous read shall either work with ECC engines or just not be implemented at all. You cannot blindly disable ECC support.
For Macronix continuous read feature is for "On-die ECC" flash only. So that I think ecc have been handled in flash.
Then just prevent continuous read from being enabled on !on-die ECC flash devices, and please do not perform anything like changing the mode of operation.
Quoted 100 lines
Quoted 49 lines
Quoted 49 lines
drivers/mtd/nand/spi/core.c +49
-0+ iter->req.dataoffs =nanddev_offs_to_pos (nand, from, &iter->req.pos); + iter->req.databuf.in = ops->datbuf; + iter->req.datalen = ops->len; + +if (from & (nanddev_page_size (nand) -1 )) { +pr_debug ("%s: unaligned address\n" , __func__); +return -EINVAL ; + } + + ret =spinand_continuous_read_enable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = true; + + ret =spinand_select_target (spinand, iter->req.pos.target); +if (ret) +return ret; + +/* +* The continuous read operation including: firstly, starting with the +* page read command and the 1 st page data will be read into the cache +* after the read latency tRD. Secondly, Issuing the Read From Cache +* commands (03h/0Bh/3Bh/6Bh/BBh/EBh) to read out the data from cache +* continuously. +* +* The cache is divided into two halves, while one half of the cache is +* outputting the data, the other half will be loaded for the new data; +* therefore, the host can read out the data continuously from page to +* page. Multiple of Read From Cache commands can be issued in one +* continuous read operation, each Read From Cache command is required +* to read multiple 4-byte data exactly; otherwise, the data output will +* be out of sequence from one Read From Cache command to another Read +* From Cache command. +* +* After all the data is read out, the host should pull CS# high to +* terminate this continuous read operation and wait a 6us of tRST for +* the NAND device resets read operation. The data output for each page +* will always start from byte 0 and a full page data should be read out +* for each page. +*/ + ret =spinand_read_page (spinand, &iter->req); +if (ret) +return ret; + + ret =spinand_reset_op (spinand); +if (ret) +return ret; +Quoted 8 lines
drivers/mtd/nand/spi/core.c static int spinand_write_page(struct spinand_device *spinand, · line 717
Quoted 5 lines
drivers/mtd/nand/spi/core.c +5
-0+ ret =spinand_continuous_read_disable (spinand); +if (ret) +return ret; + + spinand->use_continuous_read = false;Any error in this function should lead to resetting
use_continuous_readto false and disabling continuous read on the flash side.OK, this will be correct next version.
Quoted 41 lines
Quoted 41 lines
drivers/mtd/nand/spi/core.c +24
-0+ + ops->retlen = iter->req.datalen; + +return ret; +} +static int spinand_mtd_read (struct mtd_info *mtd,loff_t from,struct mtd_oob_ops *ops) {@@ -656,6 +737,24 @@ static int spinand_mtd_read(struct mtd_info *mtd, loff_t from,mutex_lock (&spinand->lock); +/* +* If the device support continuous read mode and read length larger +* than one page size will enter the continuous read mode. This mode +* helps avoid issuing a page read command and read from cache command +* again, and improves read performance for continuous addresses. +*/ +if ((spinand->flags &SPINAND_HAS_CONT_READ_BIT ) && + (ops->len >nanddev_page_size (nand))) { + ret =spinand_mtd_continuous_read (mtd, from, ops, &iter); + +mutex_unlock (&spinand->lock); + +if (ecc_failed && !ret) + ret = -EBADMSG ; + +return ret ? ret : max_bitflips; + } +nanddev_io_for_each_page (nand,NAND_PAGE_READ , from, ops, &iter) {if (disable_ecc) iter.req.mode =MTD_OPS_RAW ;include/linux/mtd/spinand.h +0
-0@@ -308,6 +308,7 @@ struct spinand_ecc_info {
Quoted 7 lines
Quoted 5 lines
include/linux/mtd/spinand.h struct spinand_ecc_info { · line 311
Quoted 3 lines
include/linux/mtd/spinand.h +1
-0#define SPINAND_HAS_QE_BIT BIT(0)#define SPINAND_HAS_CR_FEAT_BIT BIT(1) +#define SPINAND_HAS_CONT_READ_BIT BIT(2)Is this bit standard?
It is using to add into flash id table and judge whether continuous read feature is support or not.
I understand what it is meant for, but my question is, shall we consider this bit standard across manufacturers? Or is it specific to Macronix?
Thanks,
Miquèl
______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
PATCH 4/4 mtd: spinand: macronix: Add continuous read support for MX35LF2GE4AD
In reply to JaimeLiaoAdd continuous read flag for MX35LF2GE4AD
--- drivers/mtd/nand/spi/macronix.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/mtd/nand/spi/macronix.c +1 -1
@@ -126,7 +126,7 @@ static const struct spinand_info macronix_spinand_table[] = { SPINAND_INFO_OP_VARIANTS(&read_cache_variants, &write_cache_variants, &update_cache_variants), - SPINAND_HAS_QE_BIT, + SPINAND_HAS_QE_BIT | SPINAND_HAS_CONT_READ_BIT, SPINAND_ECCINFO(&mx35lfxge4ab_ooblayout, mx35lf1ge4ab_ecc_get_status)), SPINAND_INFO("MX35LF4GE4AD",
Signature · 6 lines
2.17.1 ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/