All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 1/2] mtd: spinand: wait for erase completion before writing bad block maker
@ 2020-03-10 20:32 ` Mikhail Kshevetskiy
  0 siblings, 0 replies; 6+ messages in thread
From: Mikhail Kshevetskiy @ 2020-03-10 20:32 UTC (permalink / raw)
  To: miquel.raynal, richard; +Cc: linux-mtd, linux-kernel, Mikhail Kshevetskiy

SPI flash will discard any write operation while it is busy with block
erasing. As result bad block marker will not be writed to a flash.
To fix it just wait for completion of erase operation.

The erasing code is almost the same as in spinand_erase(). The only
difference is: we ignore ERASE_FAILED status.

This patch also improve error handling a bit.

Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru>
---
 drivers/mtd/nand/spi/core.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
index 89f6beefb01c..bb4eac400b0f 100644
--- a/drivers/mtd/nand/spi/core.c
+++ b/drivers/mtd/nand/spi/core.c
@@ -610,6 +610,7 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 		.oobbuf.out = spinand->oobbuf,
 	};
 	int ret;
+	u8 status;
 
 	/* Erase block before marking it bad. */
 	ret = spinand_select_target(spinand, pos->target);
@@ -620,7 +621,14 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 	if (ret)
 		return ret;
 
-	spinand_erase_op(spinand, pos);
+	ret = spinand_erase_op(spinand, pos);
+	if (ret)
+		return ret;
+
+	/* ignore status as erase may fail for bad blocks */
+	spinand_wait(spinand, &status);
+	if (ret)
+		return ret;
 
 	memset(spinand->oobbuf, 0, 2);
 	return spinand_write_page(spinand, &req);
-- 
2.25.0


^ permalink raw reply related	[flat|nested] 6+ messages in thread
* [PATCH 1/2] mtd: spinand: wait for erase completion before writing bad block maker
@ 2020-03-10 20:32 ` Mikhail Kshevetskiy
  0 siblings, 0 replies; 6+ messages in thread
From: Mikhail Kshevetskiy @ 2020-03-10 20:32 UTC (permalink / raw)
  To: miquel.raynal, richard; +Cc: linux-mtd, linux-kernel, Mikhail Kshevetskiy

SPI flash will discard any write operation while it is busy with block
erasing. As result bad block marker will not be writed to a flash.
To fix it just wait for completion of erase operation.

The erasing code is almost the same as in spinand_erase(). The only
difference is: we ignore ERASE_FAILED status.

This patch also improve error handling a bit.

Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru>
---
 drivers/mtd/nand/spi/core.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
index 89f6beefb01c..bb4eac400b0f 100644
--- a/drivers/mtd/nand/spi/core.c
+++ b/drivers/mtd/nand/spi/core.c
@@ -610,6 +610,7 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 		.oobbuf.out = spinand->oobbuf,
 	};
 	int ret;
+	u8 status;
 
 	/* Erase block before marking it bad. */
 	ret = spinand_select_target(spinand, pos->target);
@@ -620,7 +621,14 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 	if (ret)
 		return ret;
 
-	spinand_erase_op(spinand, pos);
+	ret = spinand_erase_op(spinand, pos);
+	if (ret)
+		return ret;
+
+	/* ignore status as erase may fail for bad blocks */
+	spinand_wait(spinand, &status);
+	if (ret)
+		return ret;
 
 	memset(spinand->oobbuf, 0, 2);
 	return spinand_write_page(spinand, &req);
-- 
2.25.0


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

^ permalink raw reply related	[flat|nested] 6+ messages in thread
* [PATCH 2/2] mtd: spinand: fix bad block marker writing
  2020-03-10 20:32 ` Mikhail Kshevetskiy
@ 2020-03-10 20:32   ` Mikhail Kshevetskiy
  -1 siblings, 0 replies; 6+ messages in thread
From: Mikhail Kshevetskiy @ 2020-03-10 20:32 UTC (permalink / raw)
  To: miquel.raynal, richard; +Cc: linux-mtd, linux-kernel, Mikhail Kshevetskiy

In spinand_markbad() we use spinand->oobbuf as a bad block marker, fill
it with zeroes and issue spinand_write_page() operation:

        struct nand_page_io_req req = {
                .pos = *pos,
                .ooboffs = 0,
                .ooblen = 2,
                .oobbuf.out = spinand->oobbuf,
        };
        ...
        memset(spinand->oobbuf, 0, 2);
        return spinand_write_page(spinand, &req);

spinand_write_page() will call spinand_write_to_cache_op() at some
moment. In spinand_write_to_cache_op() we have:

        nbytes = nanddev_page_size(nand) + nanddev_per_page_oobsize(nand);
        memset(spinand->databuf, 0xff, nbytes);

This will fill spinand->databuf with 0xff, but spinand->oobbuf is the
part of spinand->databuf (see spinand_init()):

        spinand->oobbuf = spinand->databuf + nanddev_page_size(nand);

As result bad block marker will be overwrited by 0xff values, hence
bad block will NOT be marked.

A separate buffer for bad block marker used to fix this issue.

Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru>
---
 drivers/mtd/nand/spi/core.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
index bb4eac400b0f..d1355773d484 100644
--- a/drivers/mtd/nand/spi/core.c
+++ b/drivers/mtd/nand/spi/core.c
@@ -603,11 +603,12 @@ static int spinand_mtd_block_isbad(struct mtd_info *mtd, loff_t offs)
 static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 {
 	struct spinand_device *spinand = nand_to_spinand(nand);
+	char bad_block_marker[2] = {0, 0};
 	struct nand_page_io_req req = {
 		.pos = *pos,
 		.ooboffs = 0,
-		.ooblen = 2,
-		.oobbuf.out = spinand->oobbuf,
+		.ooblen = sizeof(bad_block_marker),
+		.oobbuf.out = bad_block_marker,
 	};
 	int ret;
 	u8 status;
@@ -630,7 +631,6 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 	if (ret)
 		return ret;
 
-	memset(spinand->oobbuf, 0, 2);
 	return spinand_write_page(spinand, &req);
 }
 
-- 
2.25.0


^ permalink raw reply related	[flat|nested] 6+ messages in thread
* [PATCH 2/2] mtd: spinand: fix bad block marker writing
@ 2020-03-10 20:32   ` Mikhail Kshevetskiy
  0 siblings, 0 replies; 6+ messages in thread
From: Mikhail Kshevetskiy @ 2020-03-10 20:32 UTC (permalink / raw)
  To: miquel.raynal, richard; +Cc: linux-mtd, linux-kernel, Mikhail Kshevetskiy

In spinand_markbad() we use spinand->oobbuf as a bad block marker, fill
it with zeroes and issue spinand_write_page() operation:

        struct nand_page_io_req req = {
                .pos = *pos,
                .ooboffs = 0,
                .ooblen = 2,
                .oobbuf.out = spinand->oobbuf,
        };
        ...
        memset(spinand->oobbuf, 0, 2);
        return spinand_write_page(spinand, &req);

spinand_write_page() will call spinand_write_to_cache_op() at some
moment. In spinand_write_to_cache_op() we have:

        nbytes = nanddev_page_size(nand) + nanddev_per_page_oobsize(nand);
        memset(spinand->databuf, 0xff, nbytes);

This will fill spinand->databuf with 0xff, but spinand->oobbuf is the
part of spinand->databuf (see spinand_init()):

        spinand->oobbuf = spinand->databuf + nanddev_page_size(nand);

As result bad block marker will be overwrited by 0xff values, hence
bad block will NOT be marked.

A separate buffer for bad block marker used to fix this issue.

Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru>
---
 drivers/mtd/nand/spi/core.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
index bb4eac400b0f..d1355773d484 100644
--- a/drivers/mtd/nand/spi/core.c
+++ b/drivers/mtd/nand/spi/core.c
@@ -603,11 +603,12 @@ static int spinand_mtd_block_isbad(struct mtd_info *mtd, loff_t offs)
 static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 {
 	struct spinand_device *spinand = nand_to_spinand(nand);
+	char bad_block_marker[2] = {0, 0};
 	struct nand_page_io_req req = {
 		.pos = *pos,
 		.ooboffs = 0,
-		.ooblen = 2,
-		.oobbuf.out = spinand->oobbuf,
+		.ooblen = sizeof(bad_block_marker),
+		.oobbuf.out = bad_block_marker,
 	};
 	int ret;
 	u8 status;
@@ -630,7 +631,6 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
 	if (ret)
 		return ret;
 
-	memset(spinand->oobbuf, 0, 2);
 	return spinand_write_page(spinand, &req);
 }
 
-- 
2.25.0


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

^ permalink raw reply related	[flat|nested] 6+ messages in thread
* Re: [PATCH 1/2] mtd: spinand: wait for erase completion before writing bad block maker
  2020-03-10 20:32 ` Mikhail Kshevetskiy
@ 2020-03-11 10:13   ` Miquel Raynal
  -1 siblings, 0 replies; 6+ messages in thread
From: Miquel Raynal @ 2020-03-11 10:13 UTC (permalink / raw)
  To: Mikhail Kshevetskiy; +Cc: richard, linux-mtd, linux-kernel

Hi Mikhail,

Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru> wrote on Tue, 10
Mar 2020 23:32:23 +0300:

> SPI flash will discard any write operation while it is busy with block
> erasing. As result bad block marker will not be writed to a flash.
> To fix it just wait for completion of erase operation.
> 
> The erasing code is almost the same as in spinand_erase(). The only
> difference is: we ignore ERASE_FAILED status.
> 
> This patch also improve error handling a bit.
> 
> Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru>

Thanks a lot for sharing this!

Actually Frieder sent a fix for that I already added to nand/next,
please have a look at:
https://lore.kernel.org/linux-mtd/20200218100432.32433-1-frieder.schrempf@kontron.de/

If you think I'm missing something else, please tell me!

> ---
>  drivers/mtd/nand/spi/core.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
> index 89f6beefb01c..bb4eac400b0f 100644
> --- a/drivers/mtd/nand/spi/core.c
> +++ b/drivers/mtd/nand/spi/core.c
> @@ -610,6 +610,7 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
>  		.oobbuf.out = spinand->oobbuf,
>  	};
>  	int ret;
> +	u8 status;
>  
>  	/* Erase block before marking it bad. */
>  	ret = spinand_select_target(spinand, pos->target);
> @@ -620,7 +621,14 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
>  	if (ret)
>  		return ret;
>  
> -	spinand_erase_op(spinand, pos);
> +	ret = spinand_erase_op(spinand, pos);
> +	if (ret)
> +		return ret;
> +
> +	/* ignore status as erase may fail for bad blocks */
> +	spinand_wait(spinand, &status);
> +	if (ret)
> +		return ret;
>  
>  	memset(spinand->oobbuf, 0, 2);
>  	return spinand_write_page(spinand, &req);

Thanks,
Miquèl

^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [PATCH 1/2] mtd: spinand: wait for erase completion before writing bad block maker
@ 2020-03-11 10:13   ` Miquel Raynal
  0 siblings, 0 replies; 6+ messages in thread
From: Miquel Raynal @ 2020-03-11 10:13 UTC (permalink / raw)
  To: Mikhail Kshevetskiy; +Cc: richard, linux-mtd, linux-kernel

Hi Mikhail,

Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru> wrote on Tue, 10
Mar 2020 23:32:23 +0300:

> SPI flash will discard any write operation while it is busy with block
> erasing. As result bad block marker will not be writed to a flash.
> To fix it just wait for completion of erase operation.
> 
> The erasing code is almost the same as in spinand_erase(). The only
> difference is: we ignore ERASE_FAILED status.
> 
> This patch also improve error handling a bit.
> 
> Signed-off-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@oktetlabs.ru>

Thanks a lot for sharing this!

Actually Frieder sent a fix for that I already added to nand/next,
please have a look at:
https://lore.kernel.org/linux-mtd/20200218100432.32433-1-frieder.schrempf@kontron.de/

If you think I'm missing something else, please tell me!

> ---
>  drivers/mtd/nand/spi/core.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mtd/nand/spi/core.c b/drivers/mtd/nand/spi/core.c
> index 89f6beefb01c..bb4eac400b0f 100644
> --- a/drivers/mtd/nand/spi/core.c
> +++ b/drivers/mtd/nand/spi/core.c
> @@ -610,6 +610,7 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
>  		.oobbuf.out = spinand->oobbuf,
>  	};
>  	int ret;
> +	u8 status;
>  
>  	/* Erase block before marking it bad. */
>  	ret = spinand_select_target(spinand, pos->target);
> @@ -620,7 +621,14 @@ static int spinand_markbad(struct nand_device *nand, const struct nand_pos *pos)
>  	if (ret)
>  		return ret;
>  
> -	spinand_erase_op(spinand, pos);
> +	ret = spinand_erase_op(spinand, pos);
> +	if (ret)
> +		return ret;
> +
> +	/* ignore status as erase may fail for bad blocks */
> +	spinand_wait(spinand, &status);
> +	if (ret)
> +		return ret;
>  
>  	memset(spinand->oobbuf, 0, 2);
>  	return spinand_write_page(spinand, &req);

Thanks,
Miquèl

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

^ permalink raw reply	[flat|nested] 6+ messages in thread
end of thread, other threads:[~2020-03-11 10:28 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-03-10 20:32 [PATCH 1/2] mtd: spinand: wait for erase completion before writing bad block maker Mikhail Kshevetskiy
2020-03-10 20:32 ` Mikhail Kshevetskiy
2020-03-10 20:32 ` [PATCH 2/2] mtd: spinand: fix bad block marker writing Mikhail Kshevetskiy
2020-03-10 20:32   ` Mikhail Kshevetskiy
2020-03-11 10:13 ` [PATCH 1/2] mtd: spinand: wait for erase completion before writing bad block maker Miquel Raynal
2020-03-11 10:13   ` Miquel Raynal

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.