Re: [PATCH 3/6] firewire: char device interface

["Kristian Høgsberg" <krh@redhat.com>]

John Stoffel wrote:
>>>>>> "Stefan" == Stefan Richter <stefanr@s5r6.in-berlin.de> writes:
> 
> Stefan> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
> Stefan> ---
> Stefan>  drivers/firewire/fw-cdev.c    |  954 ++++++++++++++++++++++++++++++++++
> Stefan>  include/linux/firewire-cdev.h |  268 +++++++++
> Stefan>  2 files changed, 1222 insertions(+)
> 
> Stefan> Index: linux_juju/include/linux/firewire-cdev.h
...
> Stefan> +#define RCODE_SEND_ERROR		0x10
> Stefan> +#define RCODE_CANCELLED			0x11
> Stefan> +#define RCODE_BUSY			0x12
> Stefan> +#define RCODE_GENERATION		0x13
> Stefan> +#define RCODE_NO_ACK			0x14
> Stefan> +
> Stefan> +#define SCODE_100			0x0
> Stefan> +#define SCODE_200			0x1
> Stefan> +#define SCODE_400			0x2
> Stefan> +#define SCODE_800			0x3
> Stefan> +#define SCODE_1600			0x4
> Stefan> +#define SCODE_3200			0x5
> 
> These are also defined in fw-transaction.h, though that file doesn't
> have all the values.  Can these just be combined into a single
> fw-constants.h file instead?  
> 
> I honestly haven't checked all your defines....

The firewire-cdev.h file is meant to be a self-contained userspace header file 
and shouldn't include other kernel header files.  All duplicated values are 
standardized ieee1394 values and won't ever change.  I should put a #ifndef 
__FW_COMMON_DEFINES protection around the duplicate values, I guess, but I'm 
just wondering why I never saw a "symbol redefined" warning...

Kristian