changelog: start 0.6.7

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: finalise 0.6.6

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: document changes

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Add Debian upload step to release checklist

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Update my own copyright year to include 2023

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Add missing build-dependency on netbase

Closes: #1028102
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

stest: Fix debugging output about spawning secnet

The output would easily end up interleaved and nearly
incomprehensible.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

d/copyright: Link to /usr/share/common-licenses/LGPL-3

Requested-by: ftpmaster@ftp-master.debian.org
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

secnet-wireshark.lua: Fix copyright notice licence version

Reported-by: ftpmaster@ftp-master.debian.org
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

d/copyright: Alphabetise sections

Requested-by: ftpmaster@ftp-master.debian.org
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

d/copyright: Fix reference to base91-c directory.

Requested-by: ftpmaster@ftp-master.debian.org
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

polypath: Do not separately log xmit failures

udp_sock_experienced handles the necessary logging.  We don't want a
log message per packet, even if they're errors.

Empirically, using a network speed test web page, over the same
network interface, can cause this to get EAGAIN and produce logspam.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: start 0.6.6

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Tidy changelog for 0.6.5

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Bump version for re-push

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Bump version for re-push

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Finalise 0.6.3

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: Document changes

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Update my own copyright year

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Add a README.source

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

LICENCE: reorganise

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

control: Update Standards-Version

To the version prevailing when I did the updates, in December 2021.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

LICENCE: Copy the subdirmk notice too

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

LICENCE: Actually copy the notices of the included works

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

copyright: Update licence on ac_prog_cc_no_writeable_strings.m4

From the GPLv3+ upgrade in the autoconf archive.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Finalise 0.6.2

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

init script: make "restart" work if secnet is not running.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: Document changes since 0.6.1

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

legal: Add missing notice to many files

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

legal: Clean up credits for init script

Add the credits to CREDITS and LICENCE and get rid of
boilerplate/noise/version stuff from script header.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

OLD-BUGS: Rename from BUGS and don't ship in the Debian package

This is really not of any interest.  It has been unchanged since at
least 2001.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

OLD-NEWS: Rename from NEWS

This file is very old.  We use the debian/changelog file now.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

legal: Make a LICENSE symlink for Americans

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

credits: Make debian/copyright be a symlink

This reduces the number of places to edit to two.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

credits: Add missing credit re python argparse extension

See argparseactionnoyes.py.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

CREDITS, copyright: Promote Joey Hess's packaging credit

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

legal: Move legal information into its own file

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

copyright: Add notice to random-fake-userv

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

test-example/null-fake-userv: New test utility script

This doesn't get used by anything yet.  We will want it for certian
tests, eg the autopkgtest I am working on.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

example.conf: Add mtu-target

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

example.conf: Add a commented-out mobile site setting

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

example.conf: Improve syntax by deleting one of the two map examples

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

example.conf: Comment out some tuning overrides

Not setting these is better.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

test-example/common.conf: Fix a reference to the sites file

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

init script: source /lib/lsb/init-functions

As per lintian.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

debian/control: Add missing Pre-Depends

As per lintian.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

debian: Add some lintian overrides

(INSTALL contains setup instructions too.)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

debian/control: Adjust priority (as per modern policy and lintian)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Description: expand

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Description: expand

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

README, copyright: add missing credits for base91

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

site: promote creation of st->scratch

Without this, peer keyset loading does not work after restart!

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

comprehensive-test: Actually allow OLD_SECNET_DIR set to ''

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Dir.sd.mk: Prevent builtin make rule overwriting conffile.c

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

debian/rules: Use dh sequencer

Resulting changes:
 * We now provide debug symbols in the new Debian way
 * The way the prerm stops secnet accords with modern practice

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

debhelper compat: bump to 12 (stretch-backports)

According to debdiff --controlfiles ALL the only change is a change to
the postinst, which now sometimes restarts rather than starts the
daemon.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

debian/rules: Use dh_prep instead of deprecated dh_clean -k

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: start 0.6.2

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

configure: bug reporting address

The correct address is the secnet-discuss list.

Reported-by: Colin Watson <cjwatson@debian.org>
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

comm-common.h: Remove unused `notify' variable

This only needs to declare the `comm_notify_list' type, and declaring
the variable here without `extern' causes a multiple-definition link
failure on bullseye.

changelog: finalise 0.6.1

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: document changes since 0.6.0

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Use CLOCK_MONOTONIC for all our timing needs, when possible

Otherwise we can malfunction if the clock warps.

This depends on an unreleased adns feature, so the warning is
currently not actually printed.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

autoconf: Check for adns_if_monotonic and define USE_MONOTONIC

Nothing uses this yet.  Split out into its own commit for clarity.

This flag is not yet in any released version of adns but as adns
upstream I promise this is how it will be.  for my convenience I am
making secnet check for it ahead of its existence in public adns....

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

resolver: Break out adns_initflags value as variable

This provides a single place to add flags, which we'll do in a moment.

No functional change.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

config parsing: When closure is of wrong type, report how

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Closures: Provide closure_type_name

This will be used in error reporting.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

config parsing: site: Use cfgfatal_cl_type for transform

Minor improvement to the message.  We are going to improve
cfgfatal_cl_type.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

config parsing: Break out cfgfatal_cl_type

No functional change.  We are going to reuse and improve this.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

example.conf: Fix "transform" syntax

These global closure names need to be invoked.

This was broken in 5b5f297f9a9d47ee
  site: support multiple transforms
(first released in 0.3.0)

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

rsa: Print errno value if we fail to open key file

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

rsa: Do not crash with -j if key file does not exist

This was messed up by 58913a3b93a6
  rsa1: Break rsa_loadpriv_core out of rsapriv_apply
where the early exit path should have been handled in both places but
wasn't.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

log: Set log level for early logging

Unless ->level is set, logging is just a no-op.  We must set this
twice, because the system log is used to report command line parsing
errors, but the command line might increase the log level.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

example.conf: Use new name-prefixed format

Otherwise it won't work at all.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

stest: No longer use localhost addresses

This will make testing polypath stuff more sensible.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

mtest/t-userv: Check lack of optional group in fragment

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

make-secnet-sites: Tolerate missing group in userv sites file

When processing a sites file fragment via userv, the group in the
provided file is supposed to be optional.  This was accidentally
regressed in 19482a2958fa
   make-secnet-sites: Do not write out unchecked output in sites

The additional call to w[2].groupname ought to have been conditional.
Now we have precisely the right code in OpBase, so simply call it.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

mtest/t-userv: Test that output conf file is correct

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

mtest/t-userv: Test that updated sites file is readable.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

make-secnet-sites: Tolerate unused group in additions to location

When processing a sites file other than via userv, the group (if
specified) is not of any interest.

But since tainting, we need to bless it for re-output.  (This is
necessary even in modes where we don't actually write anything out,
like .conf generation.)

The correct place to do this is in OpConf's base class.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

mtest: Break out diff-mss-sites-conf

No functional change.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

mtest/t-userv: Check that user in wrong group is rejected

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

make-secnet-sites: Fix error handling if caller is in wrong group

We would crash with an unbound variable error, instead of the right
error message.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Makefiles: Use Final.sd.mk to implementing RECHECK_RM

This is now read by make after all the other makefiles.  This allows
us to move the addition of {stest,mtest}/d-* to RECHECK_RM from
Dir.sd.mk into test-common.sd.mk, where it belongs.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Merge subdirmk 0.4

git subtree pull.  Fix up test-example/Dir.sd.mk for the incompatible
change (`&${' needs to become `&{').

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

README: Update copyright date

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

README: Fix documentation errors relating to &{ etc.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

logging: site: Log state on PHASE_RUN entry instead of initially

site_startup calls enter_state_run which would print a message, but
logging is not set up that early.  The result is a message printed to
stderr before daemonisation.

We can distinguish this situation from other calls to enter_state_run
because the old state is SITE_STOP, which only occurs between config
reading (closure invocation) and site_startup being called.
So we can suppress this message.

But it did serve a purpose: it would only be printed if the site was
listed in `sites'; otherwise site_startup wouldn't be called and the
`entering state RUN' message would be absent.

So instead we provide a more explicit way to tell: on entering
PHASE_RUN, site_startup has either been called, or not.  And logging
is set up.  state is then STOP or RUN.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

site: Rename site_phase_hook to site_phase_shutdown_hook

This is misnamed.  And we are going to add yet another phase hook.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

logging: Move "starting" message earlier

We are going to add some log messages to PHASE_RUN hooks.  We want the
overall startup message to come first.  Doing this right after
PHASE_DAEMONIZE makes sense as PHASE_DAEMONIZE now sets up logging,
too.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

site: Change site->control(bool_t) to site->startup()

This is only ever called with run=True.  We are going to want to rely
on this property.

If we ever do more runtime reconfig that will be done differently.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

logging; site: Do not log transport_peers_clear if already clear

This suppresses a pointles message at startup.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

logging: site: Introduce transport_peers_init which doesn't log

This eliminates some spurious startup messages.  It also allows
transport_peers_clear to read *peers, since it doesn't have to do
initialisation.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

logging: start rather earlier

This moves some messages printed by by early netlink and polypath
setup to the proper logfile / syslog directly, rather than having them
captured by the stderr capture from daemonize.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

startup: Break out start_sites

We were going to want to change when this happens.  But actually it
seems that is not right.  Keep the commit anyway, as I think it
improves clarity a bit.

No functional change.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

changelog: Start 0.6.1

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Finalise 0.6.0

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

configure[.ac]: Arrange to cope with jessie's gcc

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>

Revert "site, pubkeys: Avoid for (int a=..."

This reverts commit 983e0900816ece898f3d53a530fe0f2c73932bbc.

site, pubkeys: Avoid for (int a=...

In jessie, gcc hates this unless you pass -std=gnu11 or something.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>