Remove svn:mergeinfo carried over from stable/9.

Approved by: re (implicit)
Sponsored by: The FreeBSD Foundation

Copy stable/9 to releng/9.3 as part of the 9.3-RELEASE cycle.

Approved by: re (implicit)
Sponsored by: The FreeBSD Foundation

MFC r263774:

Revert r263694, and apply a better fix to squelch unnecessary warnings
from clang about possible keywords being treated as identifiers for the
remainder of the translation unit (a.k.a. -Wkeyword-compat), when using
libstdc++ in combination with -Wsystem-headers. This will not only fix
devd, but any C++ program using libstdc++.

MFC r263694:

Apply a temporary band-aid for building devd with clang 3.4, libstdc++
and -Wsystem-headers enabled (which is the default for any non-zero
WARNS level, crazily enough!). This is primarily meant to be MFC'd as
soon as possible.

MFC 259339
sbin/devd/devd.cc
Increase the size of devd's client socket's send buffer from the
default (8k) to 128k. This prevents clients from getting
POLLHUPped during event storms. For example, during zpool creation,
the kernel emits a resource.fs.zfs.statechange event for every vdev
in the pool. A 128k buffer is large enough to hold the statechange
events for a pool with nearly 800 drives.

A set of MFCs allowing stable/9 to be built again on head after r250881:

r250227:
Improve compatibility with recent flex from flex.sourceforge.net.

r250914:
Improve compatibility with old flex and fix build with GCC.

r250926 (partial):
Work around build breakages with GCC 4.2.

MFC r249951

According to devctl(4), clients must read events whole; they may not piece them
together from multiple reads(). It's as if /dev/devctl is a datagram device
instead of a stream device. However, devd's internal buffer was too small
(1025 bytes) to read an entire ereport.fs.zfs.checksum event (variable, up to
~1300 bytes). This commit enlarges the buffer to 8k.

Approved by: ken (mentor)

MFC r246134 ...

Improve devd startup time, by tweaking some string handling routines that are
heavily used when parsing config files. Mostly these changes avoid making
temporary copies of the strings, and avoid doing byte at a time append
operations, on the most-used code path.

On a 1.2 GHz ARM processor this reduces the time to parse the config files
from 13 to 6 seconds.

MFC r246361 ... bump .Dd for content change.

MFC r226775:

- Add support for a "!" character in regex matching in devd(8). It inverts
the logic (true/false) of the matching.

- Add "!usbus[0-9]+" to IFNET ATTACH notification handler in the default
devd.conf to prevent rc.d/netif from running when usbus[0-9]+ is attached.

MFC r246121 ...

Fix a descriptor leak in devd. Clients reading /var/run/devd.pipe can close
their socket connection any time, and devd only notices that when it gets an
error trying to write an event to the client. On a system with no device
change activity, clients could connect and disappear repeatedly without devd
noticing, leading to an ever-growing list of open socket descriptors in devd.

Now devd uses poll(2) looking for POLLHUP on all existing clients every time
a new client connection is established, and also periodically (once a minute)
to proactively find zombie clients and reap the socket descriptors. It also
now has a connection limit, configurable with a new -l <num> command line arg.
When the maximum number of connections is reached it stops accepting new
connections until some current clients drop off.

MFC r243930:
Constify where possible

Approved by: cperciva (mentor, implicit)

MFC r243931:
Avoid the creation of a temporary object by using the prefix operator
for non-primitive types.

Approved by: cperciva (mentor, implicit)

MFC r243932:
Prefer the use of initalizer lists to ctor assignment.

Approved by: cperciva (implicit)

MFC r243907:

Fix an old bug in devd, where it uses std::sort() to sort the various
lists it reads from its configuration files on the priority field.

Because some items in the lists have the same priority, and std::sort()
is not stable, the exact order in which the items are enumerated does
not have to correspond to the order they appear in the configuration
files.

Apparently this was never noticed with libstdc++, but with libc++ it
could cause the "uhid" entry from /etc/devd/usb.conf to be used instead
of the "ums" entry (which is earlier in the file). This caused the
problem described in the PR: the USB mouse module was never loaded, and
the other actions (such as starting moused) were not executed.

To fix the problem, make devd use std:stable_sort() instead.

Reported by: Jan Beich <jbeich@tormail.org>
PR: bin/172958

MFC r241772: document acpi_cpu devd notification about _CST change

MFC r236388:

Make devd build with libc++.

MFC r228618

In sbin/devd/devd.h, 'event_proc' and 'eps' are declared as structs. In
devd.hh, they are redeclared as classes. Fix the inconsistency.

Copy head to stable/9 as part of 9.0-RELEASE release cycle.

Approved by: re (implicit)

Typo. For USB devices, 'serial' should be 'sernum'.
See sys/dev/usb/usb_device.c for what devctl_notify() gets.

Don't generate input() since it's not used.

Fix missing closedir()

Reviewed by: imp

Allow the kernel to generate more spacy things and still have devd
cope. Skip multiple spaces in a few contexts.

PR: 96854
Submitted by: Shin'ya Kumabuchi
MFC after: 1 week

mdoc: drop redundant .Pp and .LP calls

They have no effect when coming in pairs, or before .Bl/.Bd

Fix typos, spelling, formatting and mdoc mistakes found by Nobuyuki while
translating these manual pages. Minor corrections by me.

Submitted by: Nobuyuki Koganemaru <n-kogane@syd.odn.ne.jp>

- Avoid calling the copy constructor when it is not necessary.

- Remove unused instance of string.

Expand system into my_system, and add the necessary tidyness that we
need. Close the pidfile. Then close all descriptors >= 3 to avoid
information leakage to children.

This solves the problem of not being able to restart devd when you
have, for example, a dhclient forked to configure your network...

MFC after: 3 days

Document the new USB notification types.

Bump WARNS where possible.

Checked by: make universe
Approved by: ed (co-mentor)

o Spell. Sort .Xrs.

Document several notifications, among them are DEVFS, update to ifneti,
coretemp and kern.

The asmc(4) and zfs(5) are still not documented.

Based on the patch by Roland Smith <rsmith xs4all nl>.
MFC after: 1 week

Send all debug to stderr.

Clarify that configuration files must be named '*.conf'

No need to be gratuitously style(9) non-compliant here, even though
C++ lets me get away with it.

Minor initialization change to not trigger bogus gcc warning about
passing literal strings to the constructors of string.

Reduce the WARNS level across the board due to a warning that occurs
on at least arm and sparc64. i386/pc98, ia64 and amd64 are all OK.

Fix the description of "media-type".

Revise the markup, sort sections, fix some grammar bugs.
Not fixed: the description of media-type is completely bogus.

flesh out the devd.conf man page... Mostly from the PR, but did a couple
gramatical tweaks along w/ sorting the list, and adding that serial is
available for USB....

PR: 85097
Submitted by: Fredrik Lindberg
MFC after: 1 week

Add -f option to program's usage(), fix manpage's SYNOPSIS.

o Yacc's lineno variable holds a total number of parsed lines.
Reset it to 1 for each devd config file so if the parser finds
a syntax error devd(8) will report a correct line number.

Submitted by: Niki Denev
MFC after: 2 weeks

Use pidfile(3).

OK'ed by: imp

It was always intended that regular expression matching be case
insensitive. Make it so.

Rev 1.15 should have incremented the date on this manual page.

Add a -f configfile option to devd(8), based on a patch submitted by
Wojciech A. Koszek.

Submitted by: Wojciech A. Koszek <dunstan@freebsd.czest.pl>

Reuse delete_and_clear() template helper.

Approved by: imp

Make devd WARNS=4 clean and bump WARNS accordingly. This will insure
that future variable shadowing bugs don't compile.

Reviewed by: imp
Compiled on: alpha i386 sparc64

As pointed out on current@, we don't want to declare a variable in a scope
that just uses the variable and throws it away.

This should fix the subsystem keyword wrt media-type.

MFC After: 2 days

media-type predicate is used in contexts where device-name is undefined.
In those contexted, use subsystem instead.

# This causes dhclient to run again when I plug in my ethernet cable to
# my fxp card in my laptop.

* Open permissions on /var/run/devd.pipe so that any user can read devd events
from this socket
* Enable non-blocking I/O on devd.pipe to keep clients from wedging devd.
If a write(2) on devd.pipe would block, the client in question will be
removed [1]

Requested by: rwatson [1]
Approved by: imp

Replace fch{mod,own} with straight ch{mod,own} as the former cannot be used
on socket file descriptors.

Reviewed by: secteam (cperciva)
Approved by: imp

Add the ability to match the on a media type of the device in question.

Submitted by: sam
Approved by: re (scottl)

Ignore sigpipe so we can properly detach clients to the pipe.

Submitted by: Fredrik Lindberg

Fix parsing of nomatch events.
minor debug fix.

Sync program's usage() with manpage's SYNOPSIS.

Scheduled mdoc(7) sweep.

NOSHARED -> NO_SHARED

Clarify that devd will only parse files whose names end in .conf.
Prompted by

PR: docs/73668
Submitted by: Dan Pelleg
MFC After: 3 days

mdoc(7) fixes.

Create a unix domain socket and write to it all the events that come
in over the socket.

Parse the ! lines that will soon be coming from the kernel. These are
a generalized notification mechanism for subsystems wishing to report
events.

Revieded by: njl

# The kernel side seems like it might be causing panics for me, but should
# be forthcoming shortly.

fix typo: s/Instaed/Instead/

At imp's request, force devd to be statically compiled. This avoids the
need for libstdc++ in /lib, and the generated binary is actually smaller
statically linked than dynamically + sizeof(libstdc++). Additionally,
devd doesn't use get*by*() which is one of the main motivations for
dynamically linking your root partition anyway.

Simplistic C comment re is wrong, use more correct one

Remove old workaround

Drop the pid file after we call the final daemon call. w/o -n would
give the wrong pid.

Submitted by: ru and Lukas Ertl
PR: 54113

mdoc(7) fixes: Expand contraction.

Allow zero or more actions in an action list, rather than requiring
one or more actions in the list. This makes constructs like:

attach 10 {
// echo "Driver $device_name attached"
};

to be accepted by the parser. It will be treated as if the user had
entered:

// attach 10 {
// echo "Driver $device_name attached"
// };

(eg totally ignored).

Approved by: re@ (rwatson)

sysctlbyname takes a size_t as its 5th argument, not a pointer.
However, since NULL was used, and this is C++, this error went
undetected until g++ 3.3 somehow managed to whine about it.

Reported by: obrien

Create a new C++ header file called devd.hh which has all the class
definitions in it. Begin to document the classes that we use, and how
they interrelate (using comments that I can use with doxygen to
automatically generate docs with).

Update to use is_id_char() in the last patch. I should have done this
before the last commit.

Make better use of the string type when parsing things. This switches
expand one from using a fixed buffer to using a string which
dynamically allocates these things.

Submitted by: green@ (against an earlier version of devd)
Ignored for too long by: imp

Change default to be 'process all events before calling daemon.'

Also, put a small work around into devd to prevent a hang on boot this
would cause because select used to return 2 rather than 0 for no
evetnts due to a bug I fixed a few days ago in subr_bus.c. I'll
remove this workaround May 7th. You have until then to upgrade your
kernel if you want to run a new devd with an older kernel.

Kill debug printf that crept in my last commit

Implement a two new flags, and change the meaning of another one.
'd': now means don't do daemon().
'D': Debug
'n': Don't wait to process all pending events before calling daemon.

In the past, devd would call daemon immediately. However, this causes
a race. If anything in the boot process depends on configuring the
devices configured by devd, maybe they would be configured in time,
maybe not. Now we don't call daemon until all pending events are
processed, unless -n is specified.

# -n is actually the default for a while due to the select(2) bug in devctl
# that I just fixed to give people a chance to upgrade.

Put an anti-foot-shooting measure in place. Set hw.bus.devctl_disable
to 0 when we startup. Print a warning in this case. This allows
people that are playing with devd by hand to have something happen.
Otherwise, it appears that devd isn't working because /dev/devctl is
disabled and producing no events.

Suggested by: peter on irc a long time ago.

o Skip white space between variables better. This makes
'a="b" c="d" at loc=1 on busN' properly set 'c' and process the rest of
the stirng. Before it would ignore everything after variable 'a'.
o Parse nomatch and other events differently. They are more different
than the code allowed for, so we weren't properly parsing nomatch
events. It appears this fixes some of the demand loading issues that
I was having with devd.

Noticed by: Gary Palmer

devd.conf already has the examples in the todo list

This can't obviously be cleaner than system headers.

mdoc(7) police: Revision.

The bugs section hasn't been true since before 5.0-R

devd bugs section is now OBE

These are OBE. devd.conf now lives in /etc.

Terminate the usage message with \n.

Submitted by: joe

MFp4: make it work

o Expand variables correctly.
o Set variables for each event.
o rewrite event loop to execute the commands in the config file, rather
than the hard wired generic command
o better(?) debug when running -d
o sort vectors of actions so that we just have to search for the first
one to match rather than the best one that matches.
o better attempts to clear all resources used on 'restart'
o Remove now bogus comments

MFC After: 1 centiyear

devd not devdd

submitted by: joe

english(4) police.

Latest snapshot of devd. This one adds re-matching and string
expansion.

MFp4 (imp_freebsd branch): snapshot of devd work:
o improve parsing and lexing
o create data structures based on the parsed file now.
o Still need to rewrite main loop and add regex (still uses hard coded
devd-generic)
o minor man page updates.

# There should be one more commit before rc2

Approved by: re (blanket)

Fixup FILES section to use proper width and to use a cross-reference to
devd(8).

Remove leading ^ from example of match line in vendor-supplied rules since
the ^ is implicit at the beginning of the expressions.

Add BUGS section and note departure of actual implementation from what
is documented so the pioneers will know why it works the way it does.

Install devd.conf(5).

Try to reword things a bit more to make it clearer.

Reviewed by: imp

- Fix some minor grammar and spelling nits.
- Massage the wording in a few places.
- Put .Nm on lines by itself so it renders correctly.

Reviewed by: imp

devd. A daemon that hooks into the kernel's /dev/devctl to produce
arbitrary commands when devices come and go in the device tree (which is
different than the /dev directory).

This is an initial version. Much of the planned power isn't here.
Instead of doing the full matching, we always run /etc/devd-generic.
/etc/devd.generic will go away at some point, I think.

I'm committing it in this early state so I can start getting feedback
from early adapters.

Approved by: re