mctp.7: Add man page for Linux MCTP support
PATCH 1/1 mctp.7: Add man page for Linux MCTP support
This change adds a brief description for the new Management Component Transport Protocol (MCTP) support added to Linux as of bc49d8169.
This is a fairly regular sockets-API implementation, so we're just describing the semantics of socket, bind, sendto and recvfrom for the new protocol.
--- man7/address_families.7 | 7 ++ man7/mctp.7 | 181 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 188 insertions(+) create mode 100644 man7/mctp.7
man7/address_families.7 +7 -0
@@ -405,6 +405,13 @@ XDP (express data path) interface (since Linux 4.18). See .I Documentation/networking/af_xdp.rst in the Linux kernel source tree for details. +.TP +.B AF_MCTP +.\" commit: bc49d8169aa72295104f1558830c568efb946315 +MCTP (Management Component Transport Protocol) interface (since Linux 5.15), +as defined by the DMTF specification DSP0236. +For further information, see +.BR mctp (7). .SH SEE ALSO .BR socket (2), .BR socket (7)
man7/mctp.7 new file +181 -0
181 lines not on this page · raw message
Signature · 1 line
2.30.2
Hello Jeremy,
Quoted 8 lines · Jeremy Kerr
This change adds a brief description for the new Management Component Transport Protocol (MCTP) support added to Linux as of
bc49d8169.This is a fairly regular sockets-API implementation, so we're just describing the semantics of socket, bind, sendto and recvfrom for the new protocol.
Signed-off-by Jeremy Kerr <[email protected]>
Thanks for the manual page!
Please, check some comments below.
Cheers,
Alex
Quoted 52 lines
--- man7/address_families.7 | 7 ++ man7/mctp.7 | 181 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 188 insertions(+)create mode 100644 man7/mctp.7
man7/address_families.7 +7
-0@@ -405,6 +405,13 @@ XDP (express data path) interface (since Linux 4.18). See .I Documentation/networking/af_xdp.rst in the Linux kernel source tree for details. +.TP +.B AF_MCTP +.\" commit: bc49d8169aa72295104f1558830c568efb946315 +MCTP (Management Component Transport Protocol) interface (since Linux 5.15), +as defined by the DMTF specification DSP0236. +For further information, see +.BR mctp (7). .SH SEE ALSO .BR socket (2), .BR socket (7)man7/mctp.7 new file +22
-0@@ -0,0 +1,181 @@ +.\" Copyright (c) 2021 Jeremy Kerr <[email protected]> +.\" +.\" %%%LICENSE_START(GPLv2+_DOC_FULL) +.\" This is free documentation; you can redistribute it and/or +.\" modify it under the terms of the GNU General Public License as +.\" published by the Free Software Foundation; either version 2 of +.\" the License, or (at your option) any later version. +.\" +.\" The GNU General Public License's references to "object code" +.\" and "executables" are to be interpreted as the output of any +.\" document formatting or typesetting system, including +.\" intermediate and printed output. +.\" +.\" This manual is distributed in the hope that it will be useful, +.\" but WITHOUT ANY WARRANTY; without even the implied warranty of +.\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +.\" GNU General Public License for more details. +.\" +.\" You should have received a copy of the GNU General Public +.\" License along with this manual; if not, see +.\" <http://www.gnu.org/licenses/>. +.\" %%%LICENSE_END
man7/mctp.7 line 32
Quoted 10 lines
man7/mctp.7 new file +10
-0+.\" commit: bc49d8169aa72295104f1558830c568efb946315 +.TH MCTP 7 2021-10-14 "Linux" "Linux Programmer's Manual" +.SH NAME +mctp \- Management Component Transport Protocol +.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.
man7/mctp.7 line 38
Quoted 6 lines
man7/mctp.7 new file +6
-0+.fi +.SH DESCRIPTION +Linux implements the Management Component Transport Protocol (MCTP), +specified by DMTF spec DSP0236, currently at version 1. +This is a connectionless protocol, typically used between devices within a +server system. Message reliability and ordering are not guaranteed, but
Please use semantic newlines. See man-pages(7):
Use semantic newlines
In the source of a manual page, new sentences should be started on new lines, and long sentences should be split into lines at clause breaks (commas, semicolons, colons, and so on). This convention, sometimes known as "seman‐ tic newlines", makes it easier to see the effect of patches, which often operate at the level of individual sentences or sentence clauses.
Quoted 10 lines
man7/mctp.7 new file +10
-0+message boundaries are preserved. +.PP +The API for MCTP messaging uses a standard sockets interface, using the +.BR sendto (2) +and +.BR recvfrom (2) +classes of system calls to transfer messages. +Messages may be fragmented into packets before transmission, and reassembled at +the remote endpoint. +This fragmentation and reassembly is transparent to userspace.
man7/mctp.7 line 58
Quoted 10 lines
man7/mctp.7 new file +10
-0+.PP +.SS Address format +MCTP addresses (also referred to as EIDs, or Endpoint Identifiers) are +single-byte values, typed as +.BR mctp_eid_t . +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?
man7/mctp.7 line 70
Quoted 12 lines
man7/mctp.7 new file +12
-0+defined as: +.PP +.in +4n +.EX +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.
So, 'unsigned short'.
Quoted 4 lines
man7/mctp.7 new file +4
-0+ int smctp_network; /* local network identifier */ + struct mctp_addr smctp_addr; /* EID */ + uint8_t smctp_type; /* message type byte */ + uint8_t smctp_tag; /* tag value, including TO flag */
man7/mctp.7 line 84
Quoted 10 lines
man7/mctp.7 new file +10
-0+}; +.EE +.in +.SS Sending messages +Messages can be transmitted using the +.BR sendto (2) +and +.BR sendmsg (2) +system calls, by providing a +.B struct sockaddr_mctp
No escape here. That's correct.
Quoted 4 lines
man7/mctp.7 new file +4
-0+describing the addressing parameters. +.PP +.in +4n +.EX
man7/mctp.7 line 98
Quoted 10 lines
man7/mctp.7 new file +10
-0+struct sockaddr_mctp addr; +ssize_t len; +char *buf; + +/* set message destination */ +addr.smctp_family = AF_MCTP; +addr.smctp_network = 0; +addr.smctp_addr.s_addr = 8; /* remote EID */ +addr.smctp_tag = MCTP_TAG_OWNER; +addr.smctp_type = MCTP_TYPE_ECHO; /* fictional message type */
I don't like 'MCTP_TYPE_ECHO': It's not a flag defined by the kernel, so it shouldn't have such a name (it resembles too much other kernel identifiers, such as MCTP_TAG_OWNER). Better something like MYPROGRAM_MCTP_TYPE_ECHO, to clearly show the reader that this one is really arbitrary.
man7/mctp.7 line 103
Quoted 5 lines
man7/mctp.7 new file +5
-0+ +buf = "hello, world!" + +len = sendto(sd, buf, 13, 0, + (struct sockaddr_mctp *)&addr, sizeof(addr));
This cast is incorrect. Since the parameter is not a 'void *', that cast would give a compile error. It should be '(struct sockaddr *)'.
There are similar casts in this page which also should be fixed like this one.
Quoted 54 lines
man7/mctp.7 new file +54
-0+.EE +.in +.PP +Here, the sender owns the message tag; so +.B MCTP_TAG_OWNER +is used as the tag data. +The kernel will allocate a specific tag value for this message. +If no tag is available, +.B sendto +will return an error, with errno set to +.BR EBUSY . +This allocated tag remains associated with the socket, so that any replies to +the sent message will be received by the same socket. +.PP +Sending a MCTP message requires the +.B CAP_NET_RAW +capability. +.SS Receiving messages +Messages can be received using the +.BR recvfrom (2) +and +.BR recvmsg (2) +system calls. +.PP +.in +4n +.EX +struct sockaddr_mctp addr; +socklen_t addrlen; +char buf[13]; + +addrlen = sizeof(addr); + +len = recvfrom(sd, buf, sizeof(buf), 0, + (struct sockaddr_mctp *)&addr, &addrlen); +.EE +.in +.PP +In order to receive an incoming message, the receiver will need to either have +previously sent a message to the same endpoint, or performed a +.BR bind (2) +to receive all messages of a certain type: +.PP +.in +4n +.EX +struct sockaddr_mctp addr; + +addr.smctp_family = AF_MCTP; +addr.smctp_network = MCTP_NET_ANY; +addr.smctp_addr.s_addr = MCTP_ADDR_ANY; +addr.smctp_type = MCTP_TYPE_ECHO; /* our fictional 'echo' type */ + +int rc = bind(sd, (struct sockaddr *)&addr, sizeof(addr)); +.EE +.in
man7/mctp.7 line 167
Quoted 10 lines
man7/mctp.7 new file +10
-0+.PP +This call requires the +.B CAP_NET_BIND_SERVICE +capability, and will result in the socket receiving all messages sent to +locally-assigned EIDs, for the specified message type. +.PP +After a +.B recvfrom +or +.B recvmsg
Use man syntax:
.BR recvfrom (3)
or
.BR recvmsg (3)
Quoted 5 lines
man7/mctp.7 new file +4
-0+returns a success condition, the provided address argument will be populated +with the sender's network and EID, as well as the tag value used for the +message. +Any reply to this message should pass the same address and tag value (with the
man7/mctp.7 line 181
Quoted 10 lines
man7/mctp.7 new file +10
-0+TO bit cleared) to indicate that is is directed to the same remote endpoint. +.SH SEE ALSO +.BR socket (7) +.PP +The kernel source file +.IR Documentation/networking/mctp.rst . +.PP +The DSP0236 specification, at +.UR https://www.dmtf.org/standards/pmci +.UE .
Please, use break points for both URIs above. See the source code of uri.7, which has plenty of examples.
See also "the Chicago Manual of Style" <https://libguides.lorainccc.edu/c.php?g=29361&p=183502> for guidelines on that.
Signature · 3 lines
Alejandro Colomar Linux man-pages comaintainer; https://www.kernel.org/doc/man-pages/ http://www.alejandro-colomar.es/
Hi Alex,
Quoted 1 line
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.
Quoted 9 lines
Quoted 6 lines
+.SH SYNOPSIS
+.nf
+.B #include <sys/socket.h>
+.B #include <linux/mctp.h>
+.PP
+.Bmctp_socket=socket(AF_MCTP, SOCK_DGRAM, 0);
mctp_socketis 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?
Quoted 8 lines
man7/mctp.7 line 58
Quoted 6 lines
+Packets between a local and remote endpoint are identified by the source
man7/mctp.7 new file +4
-0+and destination EIDs, plus a three-bit tag value. +.PP +Addressing data is passed in socket system calls through +.B struct sockaddr\_mctpThat 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.
Quoted 11 lines
man7/mctp.7 line 70
Quoted 8 lines
man7/mctp.7 new file +8
-0+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.
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?
Cheers,
Jeremy
Hi Jeremy,
Quoted 7 lines · Jeremy Kerr
Quoted 4 lines
Quoted 2 lines
+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.
My groff experiments don't reveal _ or \_ as being permissible break points[1]. However, the structure tag _could_ break like this:
sock‐addr_mctp
In other words (if my UTF-8 gets mangled), after "sock".
To prevent that, you can prefix the word with the hyphenation control escape sequence, "\%". This escape sequence is extremely portable; it dates back to 1970s AT&T troff.
Further, if you wanted to prevent breaking between the "struct" keyword and the structure tag, you could use a nonbreaking adjustable space escape sequence, "\~". While this was a groff innovation (about 30 years ago), it's been adopted by Heirloom Doctools troff and mandoc(1), so it's pretty portable to systems likely to install the Linux man-pages.
So we might write
.B struct\~\%sockaddr_mctp
for instance in running text (that is, in sentences). (When filling is off, as in code examples and the synopses of most pages documenting C interfaces, neither of these escape sequences is necessary; unfilled lines are neither automatically hyphenated nor adjusted.)
Just a typographical FYI. I know these issues sometimes frustrate people.
Regards,
Branden
[1] I'm using groff's defaults. It's possible to change the groff "hyphenation code" of any character, but few do so.
Hi Branden,
Quoted 14 lines · G. Branden Robinson
Hi Jeremy,
Quoted 7 lines · Jeremy Kerr
Quoted 4 lines
Quoted 2 lines
+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.
My groff experiments don't reveal _ or \_ as being permissible break points[1]. However, the structure tag _could_ break like this:
Quoted 7 lines
sock‐addr_mctp
In other words (if my UTF-8 gets mangled), after "sock".
To prevent that, you can prefix the word with the hyphenation control escape sequence, "\%". This escape sequence is extremely portable; it dates back to 1970s AT&T troff.
Hmm, this is one of the few points we disagree, it seems :)
As I said previously, I think this uglyfies code too much, for something that I think should be default (it is rarer to want break points at highlighted words than not wanting them). But, to try to confirm my thoughts, I'll accept using this in the man-pages and see what happens (maybe some day, a winter day it shall be, you'll wake up and see a cold commit reaping them all :p)
Quoted 7 lines
Further, if you wanted to prevent breaking between the "struct" keyword and the structure tag, you could use a nonbreaking adjustable space escape sequence, "\~". While this was a groff innovation (about 30 years ago), it's been adopted by Heirloom Doctools troff and mandoc(1), so it's pretty portable to systems likely to install the Linux man-pages.
Ah yes, I should have spotted that one :p
Quoted 4 lines
So we might write
.B struct\~\%sockaddr_mctp
Okay.
Quoted 10 lines
for instance in running text (that is, in sentences). (When filling is off, as in code examples and the synopses of most pages documenting C interfaces, neither of these escape sequences is necessary; unfilled lines are neither automatically hyphenated nor adjusted.)
Just a typographical FYI. I know these issues sometimes frustrate people.
Regards, > Branden
Thanks!
Alex
Quoted 4 lines
[1] I'm using groff's defaults. It's possible to change the groff "hyphenation code" of any character, but few do so.
Signature · 3 lines
Alejandro Colomar Linux man-pages comaintainer; https://www.kernel.org/doc/man-pages/ http://www.alejandro-colomar.es/
Quoted 5 lines · Alejandro Colomar (man-pages)
Quoted 3 lines
So we might write
.B struct\~\%sockaddr_mctp
Okay.
Actually, wouldn't it be better to just write?:
.B \%struct\~sockaddr_mctp
This way \% applies to the whole (even if it was unnecessary for 'struct\~').
Signature · 3 lines
Alejandro Colomar Linux man-pages comaintainer; https://www.kernel.org/doc/man-pages/ http://www.alejandro-colomar.es/
Suppressing hyphenation (was: [PATCH] mctp.7: Add man page for Linux MCTP support)
[Jeremy Kerr dropped from CC--I hope that's okay]
Hi Alex,
Getting back to this after a month...
Quoted 5 lines · Alejandro Colomar (man-pages)
Quoted 4 lines · Alejandro Colomar (man-pages)
Quoted 3 lines
So we might write
.B struct\~\%sockaddr_mctp
Quoted 8 lines
Quoted 1 line
Okay.
Actually, wouldn't it be better to just write?:
.B \%struct\~sockaddr_mctp
This way \% applies to the whole (even if it was unnecessary for 'struct\~').
In fact it does not apply to the whole; '\~' still counts as a word delimiter to groff even if it is not a permissible location for a "break" (line break).
Before I bust out the long explanation, I'll try to present some short advice for man page writers.
* If you wish to suppress hyphenation with the '\%' escape sequence, place it at the _beginning_ of each such word. Except for special character escape sequences like '\-', '\(ha', and '\[aq]', most groff escape sequences act as word boundaries, so you may need to specify '\%' before each word in a series, as in '\%typedef\~int\~\%strsize'.
Now for the deeper dive.
As strange as it may seem, this is consistent with the behavior of hyphenation when it encounters most other escape sequences[1] (most of which a portable man page should not attempt to use). The key factor to consider in matters of hyphenation suppression is where the _word boundaries_ are, not where white space appears.
By contrast, anything that formats a glyph for output generally _is_ part of a word. But only glyphs that not part of natural language words (in English, [A-Za-z]) are eligible for adjacent hyphenation.
Here's the documentation of '\%' (and '\:') from the Info documentation of the forthcoming groff 1.23.0 release.
[[
-- Escape: \%
-- Escape: \:
To tell GNU 'troff' how to hyphenate words as they occur in input, use the '\%' escape, also known as the "hyphenation character". Each instance within a word indicates to GNU 'troff' that the word may be hyphenated at that point, while prefixing a word with this escape prevents it from being otherwise hyphenated. This mechanism affects only that occurrence of the word; to change the hyphenation of a word for the remainder of input processing, use the 'hw' request.
GNU 'troff' regards the escapes '\X' and '\Y' as starting a word; that is, the '\%' escape in, say, '\X'...'\%foobar' or '\Y'...'\%foobar' no longer prevents hyphenation of 'foobar' but inserts a hyphenation point just prior to it; most likely this isn't what you want. *Note Postprocessor Access::.
The '\:' escape inserts a non-printing break point; that is, the word can break there, but the soft hyphen glyph (see below) is not written to the output if it does. This escape is an input word boundary, so the remainder of the word is subject to hyphenation as normal.
You can use '\:' and '\%' in combination to control breaking of a file name or URL or to permit hyphenation only after certain explicit hyphens within a word.
The \%Lethbridge-Stewart-\:\%Sackville-Baggins divorce was, in retrospect, inevitable once the contents of \%/var/log/\:\%httpd/\:\%access_log on the family web server came to light, revealing visitors from Hogwarts.
]]
Here's a short shell script to tell you where your installed version of groff will hyphenate words: it forces hyphenation to occur at every possible location.
$ cat ~/bin/hyphen #!/bin/sh
for W
do
printf ".hy 4\n.ll 1u\n%s\n" "$W" | nroff -Wbreak | sed '/^$/d' \
| tr -d '\n'
echodone
$ LC_ALL=C hyphen antidisestablishmentarianism 'struct\\~sockaddr' an-tidis-es-tab-lish-men-tar-i-an-ism struct\~sock-addr $ LC_ALL=C hyphen sockaddr \\%sockaddr \\%sock\\%addr sock_addr sock^addr sock-addr sockaddr sock-addr sock_addr sock^addr
(I set the locale so as to keep this email strictly "basic Latin", groff will happily emit proper Unicode hyphens U+2010 to a supporting output device.)
You can see from the above that we can't recklessly sprinkle '\%': apart from looking ugly, '\%' at the beginning of a word suppresses only _automatic_ hyphenation. If you specify it both at the beginning _and_ within a word, its other meaning of marking a hyphenation point is still honored.
Regards,
Branden
[1] There are a few exceptions, like those which "don't produce an input token" as the groff Texinfo manual puts it, a construction that is more intelligible to the groff developer than the groff user. These have to do with escape sequences that change the way glyphs are rendered, such as changes to the font style or family, type size, or stroke or fill colors. Most of these should never occur in portable man pages and even '\f' is, in my view, better handled with man(7) font style macros and the '\c' escape sequence if required for break suppression.
[CC += checkpatch.pl maintainers (see reason below)]
Hi Jeremy,
Quoted 14 lines · Jeremy Kerr
Hi Alex,
Quoted 1 line
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.
Quoted 7 lines
Quoted 6 lines
+.SH SYNOPSIS
+.nf
+.B #include <sys/socket.h>
+.B #include <linux/mctp.h>
+.PP
+.Bmctp_socket=socket(AF_MCTP, SOCK_DGRAM, 0);
Quoted 4 lines
Quoted 2 lines
mctp_socketis 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!
Quoted 25 lines
Quoted 9 lines
man7/mctp.7 line 58
Quoted 7 lines
+Packets between a local and remote endpoint are identified by the source
man7/mctp.7 new file +4
-0+and destination EIDs, plus a three-bit tag value. +.PP +Addressing data is passed in socket system calls through +.B struct sockaddr\_mctpThat 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.
Quoted 11 lines
man7/mctp.7 line 70
Quoted 8 lines
man7/mctp.7 new file +8
-0+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.
Quoted 2 lines
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
Quoted 5 lines
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'.
Quoted 6 lines
Cheers,
Jeremy
Cheers,
Alex
Signature · 3 lines
Alejandro Colomar Linux man-pages comaintainer; https://www.kernel.org/doc/man-pages/ http://www.alejandro-colomar.es/
Hello Jeremy,
Quoted 15 lines · Alejandro Colomar (man-pages)
man7/mctp.7 line 181
Quoted 12 lines
+TO bit cleared) to indicate that is is directed to the same remote endpoint.
+.SH SEE ALSO
+.BR socket (7)
+.PPman7/mctp.7 new file +6
-0+The kernel source file +.IR Documentation/networking/mctp.rst . +.PP +The DSP0236 specification, at +.UR https://www.dmtf.org/standards/pmci +.UE .Please, use break points for both URIs above. See the source code of uri.7, which has plenty of examples.
Quoted 1 line
See also "the Chicago Manual of Style" <https://libguides.lorainccc.edu/c.php?g=29361&p=183502> for guidelines on that.
Please, don't read those. As Branden and I discussed in another thread today, we prefer not following that style.
So instead of those sources, I'll just write it for you.
.IR Documentation/\:networking/\:mctp.rst .
.UR https://\:www.dmtf.org/\:standards/\:pmci
We could also break before dots, but since the names are short, I don't think it's needed. It could be:
.IR Documentation/\:networking/\:mctp\:.rst .
.UR https://\:www\:.dmtf\:.org/\:standards/\:pmci
If you are curious about the discussion we had about this, it's here:
<https://lists.gnu.org/archive/html/groff/2021-10/msg00045.html>.
Thanks,
Alex
Signature · 3 lines
Alejandro Colomar Linux man-pages comaintainer; https://www.kernel.org/doc/man-pages/ http://www.alejandro-colomar.es/