Re: [PATCH v2 3/3] mtd: cmdlinepart: check the partitions

[Artem Bityutskiy <dedekind1@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 955 bytes --]

On Tue, 2012-09-04 at 16:25 +0800, Huang Shijie wrote:
> 于 2012年09月04日 16:19, Artem Bityutskiy 写道:
> > Just do
> >
> > if (ret)
> > 	return ret;
> >
> >
> > and kill the huge 'switch' that you added.
> If the cmdline have two mtd-ids such as:
>    
> gpmi-nand:100m(root),100m@100m(kernel),1g@200m(rootfs);edb7212-nand:-(home)
> 
> The `return` will also bypass the `edb7212-nand`.
> 
> Is it acceptable?

I thought the whole idea is to refuse command lines which are ambiguous
or make no sense, no?

You could actually do this in two step and split your 3rd patch:

1. Add 'check_partitions' which prints an error messages and returns
-EINVAL if the command line has overlaps or gaps or other bad things.
But ignore the return code in 'mtdpart_setup_real()'

2. Make 'mtdpart_setup_real()' check for the error code and propagate it
up, so we'll have a prink + real refusal.

-- 
Best Regards,
Artem Bityutskiy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]