On Thu, Nov 10, 2022 at 09:12:35PM +0100, Richard Weinberger wrote:
> ----- Ursprüngliche Mail -----
> > On Thu, Nov 10, 2022 at 08:44:53PM +0100, Richard Weinberger wrote:
> >> ----- Ursprüngliche Mail -----
> >> > Von: "Daniel Golle" <daniel@xxxxxxxxxxxxxx>
> >> > +
> >> > + if (!IS_ENABLED(CONFIG_MTD_BLOCK_PARTITIONS) || mtd_type_is_nand(new->mtd))
> >> > + gd->flags |= GENHD_FL_NO_PART;
> >>
> >> I know that NAND should not get used with mtdblock because lack of wearleveling
> >> and
> >> in general too many writes. But what exactly is the rationale to deny part
> >> scanning for NAND?
> >
> > As UBI should be used on NAND, partition scanning should be enabled for
> > ubiblock devices to have uImage.FIT filesystem subimages mapped by the
> > partition parser.
> >
> > If not skipping partition scanning on NAND-backed mtdblock devices the
> > scanning itself will already trigger multiple warnings which now happen
> > every time when a NAND-backed mtdblock device is being opened since
> > commit 96a3295c ("mtdblock: warn if opened on NAND").
>
> I see, you want to promote UBI. Makes sense.
> In case you do a v5 series, please add a comment to the code.
Will do, I'm planning to send v5 early next week.
If we are going to have only CONFIG_MTD_BLOCK_PARTITIONS and no other
config symbol for ubiblock devices I'd also merge the two patches for
mtdblock and ubiblock partition scanning into a single one.
Thank you for reviewing!
Daniel
