Re: [PATCH] mtd: spi-nor: stop printing superfluous debug info

[Michael Walle]

Hi,

> > > The mtd data can be obtain with the mtd ioctls and the SPI NOR
> > > flash name can be determined interrogating the sysfs entries.
> > > Stop polluting the kernel log.
> > >
> > > Signed-off-by: Tudor Ambarus <tudor.ambarus@xxxxxxxxxx>
> > >
> > > ---
> > >  drivers/mtd/spi-nor/core.c | 19 -------------------
> > >  1 file changed, 19 deletions(-)
> > >
> > > diff --git a/drivers/mtd/spi-nor/core.c b/drivers/mtd/spi-nor/core.c
> > > index 25a64c65717d..6de76fd009d1 100644
> > > --- a/drivers/mtd/spi-nor/core.c
> > > +++ b/drivers/mtd/spi-nor/core.c
> > > @@ -3517,25 +3517,6 @@ int spi_nor_scan(struct spi_nor *nor, const char > *name,
> > >  	/* No mtd_info fields should be used up to this point. */
> > >  	spi_nor_set_mtd_info(nor);
> > >
> > > -	dev_info(dev, "%s (%lld Kbytes)\n", info->name,
> > > -			(long long)mtd->size >> 10);
> >
> > I'd lower this to dev_dbg() and print the jedec id. It might come in
> > handy for a quick glance during bootup if debug is enabled.
>
> Ack. Although, your boot time will almost be unaffected if you don't
> print the info messages to the console. What takes the most time is not
> writing to the kernel buffer, it's to display the lines on a serial
> console, and dev_info() are by default discarded, you need to select a
> lower log level manually, and if you do that it means you're not
> looking for quick boot times but rather more for additional 
> information.

Also with recent (or planned, dunno the current state) printk won't wait
anymore for the slowest console. I really don't have a strong opinion,
either dev_info(jedec_id) or dev_dbg(jedec_id). Whatever Tudor prefers.

> > > -	dev_dbg(dev,
> > > -		"mtd .name = %s, .size = 0x%llx (%lldMiB), "
> > > -		".erasesize = 0x%.8x (%uKiB) .numeraseregions = %d\n",
> > > -		mtd->name, (long long)mtd->size, (long long)(mtd->size >> 20),
> > > -		mtd->erasesize, mtd->erasesize / 1024, mtd->numeraseregions);
> > > -
> > > -	if (mtd->numeraseregions)
> > > -		for (i = 0; i < mtd->numeraseregions; i++)
> > > -			dev_dbg(dev,
> > > -				"mtd.eraseregions[%d] = { .offset = 0x%llx, "
> > > -				".erasesize = 0x%.8x (%uKiB), "
> > > -				".numblocks = %d }\n",
> > > -				i, (long long)mtd->eraseregions[i].offset,
> > > -				mtd->eraseregions[i].erasesize,
> > > -				mtd->eraseregions[i].erasesize / 1024,
> > > -				mtd->eraseregions[i].numblocks);
> > >  	return 0;
> >
> > Part of this is already available through the spi-nor debugfs, 
> > although not
> > the actual mtd properties. These I think, should go into the mtdcore
> > itself if really needed. Either through dev_dbg() or debugfs.
>
> Maybe we don't need this at all, as long as one message remains about
> the JEDEC ID, but keep in mind that spi-nors are commonly storing the
> rootfs and if your spi-nor does not boot you don't have a userspace yet
> and all the debugfs entries are purely useless.

Good point.

Just curious, do you know any boards which has the rootfs writable on
the spi-nor flash?

-michael