debug printk's

debug printk's

[Paul Jimenez]

  Is there stated kernel plicy on debug printk's?  The source shows
a lot of different versions of dprintk defined, some very small scripts
came up with:

# symbol
1 ASD_DPRINTK
1 bay_dprintk
1 dprintk1
1 dprintk2
1 dprintk_core
1 dprintkl
1 i2cdprintk
1 ir_dprintk
1 PDprintk
2 cmdprintk
2 ddprintk
2 DDPRINTK
2 dprintkdbg
2 fs_dprintk
2 gs_dprintk
2 P9_DPRINTK
2 PDPRINTK
2 rio_dprintk
2 SAS_DPRINTK
2 sx_dprintk
2 TOSHIBA_RBTX4927_IRQ_DPRINTK
2 TOSHIBA_RBTX4927_SETUP_DPRINTK
2 TX4927_IRQ_DPRINTK
10 Dprintk
130 DPRINTK
187 dprintk

So: 187 definitions of dprintk and 130 definitions of DPRINTK.  Plus
10 mixed-case fans and various and sundry miscellaneous versions.  Note
this is not *uses*, it's the number of *definitions*.

So, to clarify my original question a bit: is there policy we can refer
to to justify cleaning this up?

  --pj

Re: debug printk's

[Matthew Wilcox]

On Thu, Aug 16, 2007 at 02:07:19AM -0500, Paul Jimenez wrote:
> So: 187 definitions of dprintk and 130 definitions of DPRINTK.  Plus
> 10 mixed-case fans and various and sundry miscellaneous versions.  Note
> this is not *uses*, it's the number of *definitions*.
> 
> So, to clarify my original question a bit: is there policy we can refer
> to to justify cleaning this up?

pr_debug() in include/linux/kernel.h, which also refers to dev_dbg(),
neither of which you mentioned ;-)

But take it slowly, co-ordinate with the driver author, and don't try to
tell people that it's policy -- it's a best practice, sure, but people
are free to ignore it if it makes their life easier.

-- 
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."