Re: [PATCH 1/1] mtd/intel-spi: Support cmdline-based partition

[Miquel Raynal <miquel.raynal@xxxxxxxxxxx>]

Hi Ron,

Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx> wrote on Mon, 27 Apr
2020 11:49:54 +0200:

> On Mon, 27 Apr 2020 11:16:23 +0200
> Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx> wrote:
> 
> > Hi all,
> > 
> > On Wed, 1 Apr 2020 09:41:48 +0200
> > Miquel Raynal <miquel.raynal@xxxxxxxxxxx> wrote:
> >   
> > > Hi ron,
> > > 
> > > ron minnich <rminnich@xxxxxxxxx> wrote on Mon, 30 Mar 2020 08:53:22
> > > -0700:
> > >     
> > > > On Mon, Mar 30, 2020 at 12:27 AM Miquel Raynal
> > > > <miquel.raynal@xxxxxxxxxxx> wrote:
> > > >       
> > > > > Would it be hard to support an extra ':' after the MTD device name?
> > > > > This way would would allow anything inside the optional '(' ')' but
> > > > > would keep the trailing ':'.
> > > > >
> > > > > toTay:
> > > > >         mtdparts=name:part1,part2
> > > > >
> > > > > So:
> > > > >         mtdparts=(0000:00:1f.5):25165824(BIOS),-(squashfs)        
> > > > 
> > > > 
> > > > I thought about that ':' too. It does add a bit more to the code, and
> > > > a bit more in the way of error cases. I always worry, when code is
> > > > going into flash,
> > > > about errors where something looks close to right but is wrong. (says
> > > > the person who just typed it instead of is a few times :-)
> > > > 
> > > > What if we did this:
> > > > mtdparts=[0000:00:1f.5]25165824(BIOS),-(squashfs)
> > > > 
> > > > Is the "]" 'enough different' that we do not need the ':'?
> > > > 
> > > > I kind of like the [] better anyway as it makes the mtdid stand out a
> > > > bit more from the part names? But is it enough that we don't need the
> > > > ':'? Would you still prefer the () as opposed to the []?      
> > > 
> > > I like the [] as well, maybe more than () because at least it does not
> > > conflict with the partition names. But I really prefer keeping the : if
> > > the code is still readable.
> > > 
> > > It is much easier to explain to people : "if you have a : in the name,
> > > enclose it with []".    
> > 
> > Sorry to chime in so late in the discussion, but I wonder if any of
> > that is necessary. Can't we just split the string per device (split
> > strings every time we see a ';'), and then find the last ':' in each of
> > those strings and consider everything before that last ':' to be the MTD
> > name. That should work even if the MTD name contains one or more ':'.
> > 
> > Don't get me wrong, I'm perfectly fine with intel enclosing the PCI
> > address in [] to make it clearer, but I see that other drivers use ':'
> > in their MTD device names (the atmel raw NAND controller driver to name
> > one), so I think it'd be good to make the mtd part parsing robust to
> > this use case.  
> 
> I just gave it a try and the following patch should solve the problem
> (only compile-tested). As I said previously, it doesn't prevent you from
> enclosing the PCI address in [] if you think it's clearer, but I think
> the problem should be addressed in the cmdline parser anyway.
> 
> --->8---  
> From 08b30597dd73efd9c4c8d1906ab02a9540875419 Mon Sep 17 00:00:00 2001
> From: Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx>
> Date: Mon, 27 Apr 2020 11:44:50 +0200
> Subject: [PATCH] mtd: parser: cmdline: Support MTD names containing one or
>  more colons
> 
> Looks like some drivers define MTD names with a colon in it, thus
> making mtdpart= parsing impossible. Let's fix the parser to gracefully
> handle that case: the last ':' in a partition definition sequence is
> considered instead of the first one.
> 
> Signed-off-by: Boris Brezillon <boris.brezillon@xxxxxxxxxxxxx>
> ---
>  drivers/mtd/parsers/cmdlinepart.c | 23 ++++++++++++++++++++---
>  1 file changed, 20 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/mtd/parsers/cmdlinepart.c b/drivers/mtd/parsers/cmdlinepart.c
> index c86f2db8c882..0625b25620ca 100644
> --- a/drivers/mtd/parsers/cmdlinepart.c
> +++ b/drivers/mtd/parsers/cmdlinepart.c
> @@ -218,12 +218,29 @@ static int mtdpart_setup_real(char *s)
>  		struct cmdline_mtd_partition *this_mtd;
>  		struct mtd_partition *parts;
>  		int mtd_id_len, num_parts;
> -		char *p, *mtd_id;
> +		char *p, *mtd_id, *semicol;
> +
> +		/*
> +		 * Replace the first ';' by a NULL char so strrchr can work
> +		 * properly.
> +		 */
> +		semicol = strchr(s, ';');
> +		if (semicol)
> +			*semicol = '\0';
>  
>  		mtd_id = s;
>  
> -		/* fetch <mtd-id> */
> -		p = strchr(s, ':');
> +		/*
> +		 * fetch <mtd-id>. We use strrchr to ignore all ':' that could
> +		 * be present in the MTD name, only the last one is interpreted
> +		 * as an <mtd-id>/<part-definition> separator.
> +		 */
> +		p = strrchr(s, ':');
> +
> +		/* Restore the ';' now. */
> +		if (semicol)
> +			*semicol = ';';
> +
>  		if (!p) {
>  			pr_err("no mtd-id\n");
>  			return -EINVAL;

This is also the approach I like the most. It avoids modifying the
syntax on the cmdline (no change in Bootloaders needed) and we don't
have to change the parser of a driver every time a colon is in the
name.

I would like to drop "mtd: parsers: support '[]' fir ud ub mtdparts" as
welle as "mtd: spi-nor: controllers: intel-spi: Add support for command
line partitions", what do you think?


Thanks,
Miquèl

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