Re: [PATCH v2 1/2] MIPS: cavium-octeon: cvmx-helper-board: print unknown board warning only once

[Ralf Baechle <ralf@linux-mips.org>]

On Mon, Jun 24, 2013 at 12:38:43AM +0300, Aaro Koskinen wrote:

> When booting a new board for the first time, the console is flooded with
> "Unknown board" messages. This is not really helpful. Board type is not
> going to change after the boot, so it's sufficient to print the warning
> only once.
> 
> Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
> ---
> 
> 	v2: Adjust indentation.
> 
>  arch/mips/cavium-octeon/executive/cvmx-helper-board.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/mips/cavium-octeon/executive/cvmx-helper-board.c b/arch/mips/cavium-octeon/executive/cvmx-helper-board.c
> index 7c64977..9838c0e 100644
> --- a/arch/mips/cavium-octeon/executive/cvmx-helper-board.c
> +++ b/arch/mips/cavium-octeon/executive/cvmx-helper-board.c
> @@ -31,6 +31,8 @@
>   * network ports from the rest of the cvmx-helper files.
>   */
>  
> +#include <linux/printk.h>
> +
>  #include <asm/octeon/octeon.h>
>  #include <asm/octeon/cvmx-bootinfo.h>
>  
> @@ -184,9 +186,8 @@ int cvmx_helper_board_get_mii_address(int ipd_port)
>  	}
>  
>  	/* Some unknown board. Somebody forgot to update this function... */
> -	cvmx_dprintf
> -	    ("cvmx_helper_board_get_mii_address: Unknown board type %d\n",
> -	     cvmx_sysinfo_get()->board_type);
> +	pr_warn_once("%s: Unknown board type %d\n", __func__,
> +		     cvmx_sysinfo_get()->board_type);

David,

cvmx_dprintf() is a function frequently invoked from the OCTEON code.
I'm wondering.  Right now it's defined as:

#if CVMX_ENABLE_DEBUG_PRINTS
#define cvmx_dprintf        printk
#else
#define cvmx_dprintf(...)   {}
#endif

That is, there isn't even a severity level being used and the define
CVMX_ENABLE_DEBUG_PRINTS to control cvmx_dprintf is way to similar to
what CONFIG_DYNAMIC_DEBUG rsp. a simple #define DEBUG do.  And the
no-debug variant of cvmx_dprintf isn't correct in presence of side
effects of arguments.

So I propose to just replace cvmx_dprintf with pr_debug.  What do you
think?

  Ralf