Re: [PATCH v4] mtd: ubi: Extend UBI layer debug/messaging capabilities

[Artem Bityutskiy <dedekind1@gmail.com>]

On Tue, 2014-10-14 at 07:33 -0700, Joe Perches wrote:
> If you are going to change all the ubi_<level> calls,
> can you also please add a terminating newline to all
> the uses for consistency with all the other
> pr_<level>/dev_<level>/<etc>_<level> calls?

I get the consistency argument.

On the other hand, this is about printing a single line. It is gets
prefixed (with "UBI: ") automatically, why wouldn't we append the
newline character automatically too?

In the generic functions this is for flexibility: rarely, people to want
to print a multi-line message with those. The first line will be
prefixed, the following line won't be prefixed.

We do not need that flexibility. And I think that adding hundreds of
'\n's just for the sake of consistency to be not very attractive option.

IOW, I do not support this suggestion.

> > >   /* UBI error messages */
> > > -#define ubi_err(fmt, ...) pr_err("UBI error: %s: " fmt "\n",      \
> > > -				 __func__, ##__VA_ARGS__)
> > > +#define ubi_err(ubi, fmt, ...) pr_err("UBI-%d error: %s: " fmt "\n",      \
> > > +				 ubi->ubi_num, __func__, ##__VA_ARGS__)
> 
> Converting these macros to functions using "%pV"
> will save quite a bit of text space by removing
> a lot of "UBI-%d <foo>: " duplication.

These were added before '%pV' existed, I think. I never used this printk
extension, but if it results in a more concise code, sounds like a good
idea. But I'd do this separately.

> Using ubi_notice instead of ubi_msg would be a
> lot more standard too.

Yes, this could be an OK separate nicification, I think, if someone is
willing to do this work. I would not put this item to my TODO list,
since this is a lot of changes for with little gain. But I would accept
such a patch, sure.

Thanks!

--
Artem.