The commit 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
adds helper function, try_write_vid_and_data(), to simplify the code, but this
helper function has bug, it will return 0 (success) when ubi_io_write_vid_hdr()
or the ubi_io_write_data() return error number (-EIO, etc), because the return
value of ubi_wl_put_peb() will overwrite the original return value.

This issue will cause unexpected data loss issue, because the caller of this
function and UBIFS willn't know the data is lost.

Fixes: 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")

Signed-off-by: Wang YanQing <udknight@gmail.com>
---
 drivers/mtd/ubi/eba.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/drivers/mtd/ubi/eba.c b/drivers/mtd/ubi/eba.c
index 09c408c..4e32b25 100644
--- a/drivers/mtd/ubi/eba.c
+++ b/drivers/mtd/ubi/eba.c
@@ -981,10 +981,17 @@ static int try_write_vid_and_data(struct ubi_volume *vol, int lnum,
 out_put:
 	up_read(&ubi->fm_eba_sem);
 
-	if (err && pnum >= 0)
-		err = ubi_wl_put_peb(ubi, vol_id, lnum, pnum, 1);
-	else if (!err && opnum >= 0)
-		err = ubi_wl_put_peb(ubi, vol_id, lnum, opnum, 0);
+	if (err && pnum >= 0) {
+		if (ubi_wl_put_peb(ubi, vol_id, lnum, pnum, 1) != 0) {
+			ubi_warn(ubi, "failed to return physical eraseblock %d",
+				 pnum);
+		}
+	} else if (!err && opnum >= 0) {
+		if (ubi_wl_put_peb(ubi, vol_id, lnum, opnum, 0) != 0) {
+			ubi_warn(ubi, "failed to return physical eraseblock %d",
+				 opnum);
+		}
+	}
 
 	return err;
 }
-- 
1.8.5.6.2.g3d8a54e.dirty

Good catch. Maybe many of history happened unconsistent ubifs images 
have related to this point. One small nit below.

> The commit 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
> adds helper function, try_write_vid_and_data(), to simplify the code, but this
> helper function has bug, it will return 0 (success) when ubi_io_write_vid_hdr()
> or the ubi_io_write_data() return error number (-EIO, etc), because the return
> value of ubi_wl_put_peb() will overwrite the original return value.
> 
> This issue will cause unexpected data loss issue, because the caller of this
> function and UBIFS willn't know the data is lost.
> 
> Fixes: 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
> 
> Signed-off-by: Wang YanQing <udknight@gmail.com>
> ---
>   drivers/mtd/ubi/eba.c | 15 +++++++++++----
>   1 file changed, 11 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/mtd/ubi/eba.c b/drivers/mtd/ubi/eba.c
> index 09c408c..4e32b25 100644
> --- a/drivers/mtd/ubi/eba.c
> +++ b/drivers/mtd/ubi/eba.c
> @@ -981,10 +981,17 @@ static int try_write_vid_and_data(struct ubi_volume *vol, int lnum,
>   out_put:
>   	up_read(&ubi->fm_eba_sem);
>   
> -	if (err && pnum >= 0)
> -		err = ubi_wl_put_peb(ubi, vol_id, lnum, pnum, 1);
> -	else if (!err && opnum >= 0)
> -		err = ubi_wl_put_peb(ubi, vol_id, lnum, opnum, 0);
> +	if (err && pnum >= 0) {
> +		if (ubi_wl_put_peb(ubi, vol_id, lnum, pnum, 1) != 0) {

Could we also add error code returned from ubi_wl_put_peb() in warning 
message? Same with next ubifs_warn.
> +			ubi_warn(ubi, "failed to return physical eraseblock %d",
> +				 pnum);
> +		}
> +	} else if (!err && opnum >= 0) {
> +		if (ubi_wl_put_peb(ubi, vol_id, lnum, opnum, 0) != 0) {
> +			ubi_warn(ubi, "failed to return physical eraseblock %d",
> +				 opnum);
> +		}
> +	}
>   
>   	return err;
>   }
>

On Sat, Feb 18, 2023 at 12:58:08AM +0800, Wang YanQing wrote:
> The commit 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
> adds helper function, try_write_vid_and_data(), to simplify the code, but this
> helper function has bug, it will return 0 (success) when ubi_io_write_vid_hdr()
> or the ubi_io_write_data() return error number (-EIO, etc), because the return
> value of ubi_wl_put_peb() will overwrite the original return value.
> 
> This issue will cause unexpected data loss issue, because the caller of this
> function and UBIFS willn't know the data is lost.
> 
> Fixes: 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
> 
> Signed-off-by: Wang YanQing <udknight@gmail.com>

Hi! Miquel Raynal

What is the status about this patch?

Thanks!

----- Ursprüngliche Mail -----
> Von: "Wang YanQing" <udknight@gmail.com>
> An: "Miquel Raynal" <miquel.raynal@bootlin.com>, "Boris Brezillon" <bbrezillon@kernel.org>, "richard" <richard@nod.at>,
> "Vignesh Raghavendra" <vigneshr@ti.com>, "linux-mtd" <linux-mtd@lists.infradead.org>, "linux-kernel"
> <linux-kernel@vger.kernel.org>
> Gesendet: Sonntag, 26. Februar 2023 15:14:06
> Betreff: Re: [PATCH] mtd: ubi: eba.c: fix return value overwrite issue in try_write_vid_and_data()

> On Sat, Feb 18, 2023 at 12:58:08AM +0800, Wang YanQing wrote:
>> The commit 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
>> adds helper function, try_write_vid_and_data(), to simplify the code, but this
>> helper function has bug, it will return 0 (success) when ubi_io_write_vid_hdr()
>> or the ubi_io_write_data() return error number (-EIO, etc), because the return
>> value of ubi_wl_put_peb() will overwrite the original return value.
>> 
>> This issue will cause unexpected data loss issue, because the caller of this
>> function and UBIFS willn't know the data is lost.
>> 
>> Fixes: 2d78aee426d8 ("UBI: simplify LEB write and atomic LEB change code")
>> 
>> Signed-off-by: Wang YanQing <udknight@gmail.com>
> 
> Hi! Miquel Raynal
> 
> What is the status about this patch?

We're in the middle of the merge window. It will be applied to the fixes
branch after rc1.

Thanks,
//richard