Re: [PATCH] mctp.7: Add man page for Linux MCTP support

[Alejandro Colomar (man-pages)]

[CC += checkpatch.pl maintainers (see reason below)]


Hi Jeremy,

On 10/18/21 7:05 AM, Jeremy Kerr wrote:
> Hi Alex,
>
> > Thanks for the manual page!
>
> And thanks for the review! In general, I've updated to suit your
> comments, just a couple of queries inline.
>
> > > +.SH SYNOPSIS
> > > +.nf
> > > +.B #include <sys/socket.h>
> > > +.B #include <linux/mctp.h>
> > > +.PP
> > > +.B mctp_socket = socket(AF_MCTP, SOCK_DGRAM, 0);
> >
> > mctp_socket is a variable name.  See socket.7 for an example.
> > It should be in italics.
>
> This was based on udp.7; want me to send a patch for that too?

Sure. Thanks!

> > > +Packets between a local and remote endpoint are identified by the
> > > source
> > > +and destination EIDs, plus a three-bit tag value.
> > > +.PP
> > > +Addressing data is passed in socket system calls through
> > > +.B struct sockaddr\_mctp
> >
> > That escape is unnecessary.  Did you see it in another page perhaps?
>
> I thought I'd seen some odd line-breaks at the underscore, but can't
> replicate that now. Will remove.
>
> > > +typedef uint8_t        mctp_eid_t;
> > > +
> > > +struct mctp_addr {
> > > +    mctp_eid_t         s_addr;
> > > +};
> > > +
> > > +struct sockaddr_mctp {
> > > +    unsigned short int smctp_family;  /* = AF_MCTP */
> >
> > We only use 'int' in 'unsigned int', as the kernel does (or attempts
> > to do).  checkpatch.pl warns about 'unsigned short int', IIRC.
>
> No, there are no warnings from checkpatch there; that's just copied from
> the current kernel header.

Huh!  That's weird; 'unsigned long int' does, so I expected the same 
with 'short'.  Maybe a bug in checkpatch?


WARNING:UNNECESSARY_INT: Prefer 'unsigned long' over 'unsigned long int' 
as the int is unnecessary
#42: FILE: /home/user/src/alx/test/unsigned_short_int.c:42:
+	unsigned long int a;

total: 0 errors, 1 warnings, 0 checks, 65 lines checked


> However, I have just sent a separate patch to change that to
> __kernel_sa_family_t. Should I use that here (keeping this an exact
> match of the kernel header), or stick to the more familiar unsigned
> short?


I prefer 'unsigned short' for consistency with 'unsigned long'.

> Cheers,
>
>
> Jeremy

Cheers,

Alex


--
Alejandro Colomar
Linux man-pages comaintainer; https://www.kernel.org/doc/man-pages/
http://www.alejandro-colomar.es/