Re: port_t again, sorry didn't type the whole thing.

Janos Farkas (Janos.Farkas-nouce/priv-#DyQtpWAZZ3xFtWelno54PCx.kde@lk9qw.mail.eon.ml.org)
Fri, 22 May 1998 10:54:32 +0200

On 1998-05-22 at 08:15:15, Tigran Aivazian wrote:
> Let's have an architecture-specific type called port_t that is
> typedefed to the appropriate type (unsigned short on i386, unsigned
> long on alpha etc) and redeclare all the in/out stuff in io.h to use
> port_t (as well as check all the (million?) lines of .c files to make
> sure they use the proper declaration. This looks a tiny bit cleaner
> than the current state.

The I/O port number range, thus the C type needed to represent it is not
just a function of the CPU architecture, but more so of the bus type
(think PCI). Also (IMHO), the in/out routines/macros are not quite
useful for multiple (I/O) buses reachable through bridges. Although
handling these "extra" situations seems to be uninteresting for almost
everyone but me.. :)

[Ah, no more "out of control commercial b.s."? Thank you! :)]

-- 
Janos - Don't worry, my address is real.  I'm just bored of spam.




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu