Re: [PATCH 1/1] NTFS: Logging clean-up

[Joe Perches <joe@perches.com>]

On Sun, 2014-03-16 at 16:27 +0000, Anton Altaparmakov wrote:
> Hi,

Hello all.

> On 16 Mar 2014, at 16:12, Joe Perches <joe@perches.com> wrote:

> > This changes prefixes from "NTFS-fs" to "ntfs"
> > and should be at a minimum mentioned in the
> > changelog.
> 
> As long as it says ntfs in the string so dmesg output can be
> "grep -i ntfs"-ed I am not fussed.

Nor I really, but mentioning what output changes occur
in the changelog is helpful.

> > Converting printk(KERN_DEBUG -> pr_debug will
> > not always emit that message now.  Now, only if
> > DEBUG is #defined or dynamic_debugging is enabled
> > for the build _and_ the message is specifically
> > enabled will the message be output.
> > 
> > So, the debugging output has been silenced by default.
> > 
> > That's not necessarily good.
> 
> No that is not good at all.  I didn't know about those pr_debug
> semantics so thank you for pointing them out.  That needs fixing
> otherwise we might as well not have the messages at all...  Being able
> to enable all ntfs debug messages using a insmod option or via /proc
> as is at the moment is a very valuable debugging too and I have
> scripts that use this so am not keen on this being changed at all.

> Fabian, can you please fix the issues pointed out by Joe and also
> please make sure you actually test it!

I think a good way to enable these by default is to add
#define DEBUG before any other #include (or any include
that includes printk.h actually), but there may be other
sections of code that are #ifdef DEBUG that would need
inspecting to make sure that more code isn't compiled in
than before.

cheers, Joe