[PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[Alexander Stein]

If there is an unallocated memory area before the last, filling parting
the size calculation for MTD_SIZE_REMAINING does not take this hole
into account.
Fix this by calculating the remaining size just based on total size
and partition offset.

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
---
 drivers/mtd/mtdpart.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/mtdpart.c b/drivers/mtd/mtdpart.c
index 4886392a1cf..4bc0960bb3c 100644
--- a/drivers/mtd/mtdpart.c
+++ b/drivers/mtd/mtdpart.c
@@ -207,7 +207,7 @@ int mtd_parse_partitions(struct mtd_info *parent, const char **_mtdparts,
 {
 	struct mtd_partition partition = {}, *parts;
 	const char *mtdparts = *_mtdparts;
-	uint64_t cur_off = 0, cur_sz = 0;
+	uint64_t cur_off = 0;
 	int nparts = 0;
 	int ret, idx;
 	u64 sz;
@@ -236,8 +236,7 @@ int mtd_parse_partitions(struct mtd_info *parent, const char **_mtdparts,
 			return ret;
 
 		if (parts[idx].size == MTD_SIZE_REMAINING)
-			parts[idx].size = parent->size - cur_sz;
-		cur_sz += parts[idx].size;
+			parts[idx].size = parent->size - parts[idx].offset;
 
 		sz = parts[idx].size;
 		if (sz < parent->writesize || do_div(sz, parent->writesize)) {
-- 
2.43.0

Re: [PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[Tom Rini]

On Thu, 20 Feb 2025 15:58:07 +0100, Alexander Stein wrote:

> If there is an unallocated memory area before the last, filling parting
> the size calculation for MTD_SIZE_REMAINING does not take this hole
> into account.
> Fix this by calculating the remaining size just based on total size
> and partition offset.
> 
> 
> [...]

Applied to u-boot/next, thanks!

-- 
Tom

[REGRESSION] [PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[Francesco Dolcini]

Hello Tom, Alexander

On Thu, Feb 20, 2025 at 03:58:07PM +0100, Alexander Stein wrote:
> If there is an unallocated memory area before the last, filling parting
> the size calculation for MTD_SIZE_REMAINING does not take this hole
> into account.
> Fix this by calculating the remaining size just based on total size
> and partition offset.
> 
> Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>

This change introduced a regression on colibri imx7 and imx6ull.

```
Booting from NAND...
Partition size must be a multiple of 2048
Could not parse device nand0
Failed parsing MTD partitions from mtdparts!
Partition ubi not found!
```

Reverting commit 1ca97ee90392 ("mtd: mtdpart: Support MTD_SIZE_REMAINING
with unallocated memory area") on top of current master fixes the issue.

Any suggestion? I have not looked at the code, I just did a git bisect.

Francesco

Re: [REGRESSION] [PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[Alexander Stein]

Hi Francesco,

Am Dienstag, 22. April 2025, 12:37:32 CEST schrieb Francesco Dolcini:
> Hello Tom, Alexander
> 
> On Thu, Feb 20, 2025 at 03:58:07PM +0100, Alexander Stein wrote:
> > If there is an unallocated memory area before the last, filling parting
> > the size calculation for MTD_SIZE_REMAINING does not take this hole
> > into account.
> > Fix this by calculating the remaining size just based on total size
> > and partition offset.
> > 
> > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
> 
> This change introduced a regression on colibri imx7 and imx6ull.
> 
> ```
> Booting from NAND...
> Partition size must be a multiple of 2048
> Could not parse device nand0
> Failed parsing MTD partitions from mtdparts!
> Partition ubi not found!
> ```
> 
> Reverting commit 1ca97ee90392 ("mtd: mtdpart: Support MTD_SIZE_REMAINING
> with unallocated memory area") on top of current master fixes the issue.
> 
> Any suggestion? I have not looked at the code, I just did a git bisect.

How does your mtdparts look like?

Best regards
Alexander
-- 
TQ-Systems GmbH | Mühlstraße 2, Gut Delling | 82229 Seefeld, Germany
Amtsgericht München, HRB 105018
Geschäftsführer: Detlef Schneider, Rüdiger Stahl, Stefan Schneider
http://www.tq-group.com/

Re: [REGRESSION] [PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[Francesco Dolcini]

On Tue, Apr 22, 2025 at 03:05:58PM +0200, Alexander Stein wrote:
> Am Dienstag, 22. April 2025, 12:37:32 CEST schrieb Francesco Dolcini:
> > On Thu, Feb 20, 2025 at 03:58:07PM +0100, Alexander Stein wrote:
> > > If there is an unallocated memory area before the last, filling parting
> > > the size calculation for MTD_SIZE_REMAINING does not take this hole
> > > into account.
> > > Fix this by calculating the remaining size just based on total size
> > > and partition offset.
> > > 
> > > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
> > 
> > This change introduced a regression on colibri imx7 and imx6ull.
> > 
> > ```
> > Booting from NAND...
> > Partition size must be a multiple of 2048
> > Could not parse device nand0
> > Failed parsing MTD partitions from mtdparts!
> > Partition ubi not found!
> > ```
> > 
> > Reverting commit 1ca97ee90392 ("mtd: mtdpart: Support MTD_SIZE_REMAINING
> > with unallocated memory area") on top of current master fixes the issue.
> > 
> > Any suggestion? I have not looked at the code, I just did a git bisect.
> 
> How does your mtdparts look like?

$ grep mtdparts configs/colibri_imx7_defconfig
CONFIG_MTDPARTS_DEFAULT="mtdparts=gpmi-nand:512k(mx7-bcb),1536k(u-boot1)ro,1536k(u-boot2)ro,512k(u-boot-env),-(ubi)"

Re: [REGRESSION] [PATCH 1/1] mtd: mtdpart: Support MTD_SIZE_REMAINING with unallocated memory area

[Weijie Gao (高惟杰)]

On Tue, 2025-04-22 at 15:07 +0200, Francesco Dolcini wrote:
> External email : Please do not click links or open attachments until
> you have verified the sender or the content.
> 
> 
> On Tue, Apr 22, 2025 at 03:05:58PM +0200, Alexander Stein wrote:
> > Am Dienstag, 22. April 2025, 12:37:32 CEST schrieb Francesco
> > Dolcini:
> > > On Thu, Feb 20, 2025 at 03:58:07PM +0100, Alexander Stein wrote:
> > > > If there is an unallocated memory area before the last, filling
> > > > parting
> > > > the size calculation for MTD_SIZE_REMAINING does not take this
> > > > hole
> > > > into account.
> > > > Fix this by calculating the remaining size just based on total
> > > > size
> > > > and partition offset.
> > > > 
> > > > Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com
> > > > >
> > > 
> > > This change introduced a regression on colibri imx7 and imx6ull.
> > > 
> > > ```
> > > Booting from NAND...
> > > Partition size must be a multiple of 2048
> > > Could not parse device nand0
> > > Failed parsing MTD partitions from mtdparts!
> > > Partition ubi not found!
> > > ```
> > > 
> > > Reverting commit 1ca97ee90392 ("mtd: mtdpart: Support
> > > MTD_SIZE_REMAINING
> > > with unallocated memory area") on top of current master fixes the
> > > issue.
> > > 
> > > Any suggestion? I have not looked at the code, I just did a git
> > > bisect.
> > 
> > How does your mtdparts look like?
> 
> $ grep mtdparts configs/colibri_imx7_defconfig
> CONFIG_MTDPARTS_DEFAULT="mtdparts=gpmi-nand:512k(mx7-bcb),1536k(u-
> boot1)ro,1536k(u-boot2)ro,512k(u-boot-env),-(ubi)"
> 

I just reproduced this issue with a rather simple mtdparts:
mtdparts=2048k(bl2),-(ubi)

After printing some log, I beliebve this is caused by the mtd partition
offset value:
```
spi-nand: spi_nand spi_nand@0: 128 MiB, block size: 128 KiB, page size:
2048, OOB size: 64
parent->writesize = 0x800
parent->size = 0x8000000
parts[0]: name = bl2, offset = 0xffffffffffffffff, size = 0x200000
parts[1]: name = ubi, offset = 0xffffffffffffffff, size = 0x8000001
Partition size must be a multiple of 2048
Could not parse device spi-nand0
Failed parsing MTD partitions from mtdparts!
Partition ubi not found!
```

The mtd partition offset has not been updated when calculating the mtd
partition size:
```
		if (parts[idx].size == MTD_SIZE_REMAINING)
			parts[idx].size = parent->size -
parts[idx].offset;
```

It's also easy to solve: calculate the offset before calculating the
size