]>
git.ipfire.org Git - thirdparty/dhcpcd.git/log
Roy Marples [Tue, 6 Oct 2020 06:10:41 +0000 (07:10 +0100)]
dhcpcd: Simplify carrier handling more by using IS_LINK_UP macro
Removes the need for the LINK_DOWN_IFFUP state.
While here, remove the check for IFF_RUNNING when LINK_UNKNOWN
because that is OS specific.
Roy Marples [Mon, 5 Oct 2020 15:33:51 +0000 (16:33 +0100)]
Release dhcpcd-9.3.0
Roy Marples [Sun, 4 Oct 2020 21:04:49 +0000 (22:04 +0100)]
privsep: Remove capsicum specific hooks from BPF
We no longer change the filter as it's locked if the OS supports.
Roy Marples [Sat, 3 Oct 2020 16:17:45 +0000 (17:17 +0100)]
Don't log backticks.
Roy Marples [Sat, 3 Oct 2020 16:00:56 +0000 (17:00 +0100)]
privsep: We need getsockopt as well as setsockopt on the link socket
So we can report receive buffer size.
Important for route(4) overflow so we can try and set a bigger buffer.
Roy Marples [Fri, 2 Oct 2020 14:57:01 +0000 (15:57 +0100)]
privsep: allow CAP_SETSOCKOPT for route(4) fd.
If FreeBSD ever implements RO_MISSFILTER it will need special
rights to work over Capsium.
Roy Marples [Wed, 30 Sep 2020 16:25:32 +0000 (17:25 +0100)]
privsep: We now need to carry ifa_data for BSD
Roy Marples [Wed, 30 Sep 2020 15:14:39 +0000 (16:14 +0100)]
BSD: Fix compile for non NetBSD
Roy Marples [Tue, 29 Sep 2020 07:00:24 +0000 (08:00 +0100)]
Add a comment to prior incase I do something dumb like removing it
in the future if I forgot why it was there.
SAITOH Masanobu [Tue, 29 Sep 2020 06:30:27 +0000 (07:30 +0100)]
Fix a typo in a comment
Roy Marples [Mon, 28 Sep 2020 20:10:21 +0000 (21:10 +0100)]
privsep: fix crash when interface departs before bpf returns for it
Roy Marples [Mon, 28 Sep 2020 16:09:38 +0000 (17:09 +0100)]
BSD: struct if_data->ifi_link_state is the single source of truth
Vastly improve and simplify link detection on BSD.
dhcpcd either examines the whole system via getifaddrs(3) or
reacts to events via route(4).
In both cases we have struct if_data which has ifi_link_state.
Armed with this knowledge, we no longer need SIOCGIFDATA or
SIOCGIFMEDIA.
To solve the issue of newly attached interfaces having
LINK_STATE_UNKNOWN or some interfaces not even changing it,
we only change the local knowledge of interface flags when
reports them by getifaddrs(3) or route(4) when we change them.
For example, if we set IFF_UP and it succeeds we don't set this
internally until reported by the kernel as above.
This keeps flags and link state in sync with each other.
The hope is that the kernel can set the real link state before
it reports IFF_UP.
As such, we no longer require the poll option or need to enter a
tight loop for old interfaces.
Roy Marples [Sun, 27 Sep 2020 10:28:03 +0000 (11:28 +0100)]
BSD: Ignore vether(4) devices by default.
Sascha Wildner [Thu, 24 Sep 2020 19:53:56 +0000 (20:53 +0100)]
Clean up some warnings.
Roy Marples [Thu, 24 Sep 2020 02:49:30 +0000 (03:49 +0100)]
FreeBSD: Anticipate SIOCGIFDATA not working in Capsicum
Roy Marples [Thu, 24 Sep 2020 02:31:43 +0000 (03:31 +0100)]
BSD: NetBSD is the odd man out with SIOCGIFDATA
So setup the #defines like so.
On OpenBSD, pledge blocks it and there is no escape.
Luckily we already allow indirect ioctls via privsep so it works fine.
Roy Marples [Wed, 23 Sep 2020 12:14:06 +0000 (13:14 +0100)]
if: Always warn about ignored interfaces.
Roy Marples [Wed, 23 Sep 2020 08:59:47 +0000 (09:59 +0100)]
BSD: Of course DragonFly does SIOCGIFDATA differently..
Roy Marples [Wed, 23 Sep 2020 08:31:24 +0000 (09:31 +0100)]
Add a warning about prior
Roy Marples [Wed, 23 Sep 2020 08:27:58 +0000 (09:27 +0100)]
Revert double variable define
Roy Marples [Wed, 23 Sep 2020 08:26:16 +0000 (09:26 +0100)]
FreeBSD lacks SIOCGIFDATA it seems.
Roy Marples [Tue, 22 Sep 2020 20:53:13 +0000 (21:53 +0100)]
BSD: Use SIOCGIFDATA if no media support and no ifa_data
Hopefully this nails link state once and for all on BSD.
Roy Marples [Tue, 22 Sep 2020 12:09:03 +0000 (13:09 +0100)]
BSD: Detect initial link state in ifa_data
Not all interfaces report media state to get the link state.
However, link state is available from getifaddrs(3) ifa_data
for AF_LINK addresses.
Testing shows that link state is also sent correctly via
route(4) messages for the same interface.
This makes pppoe(4) interfaces more reliable on FreeBSD and OpenBSD.
Roy Marples [Tue, 22 Sep 2020 12:08:32 +0000 (13:08 +0100)]
dhcpcd: log carrier lost once more
Roy Marples [Mon, 21 Sep 2020 21:24:18 +0000 (22:24 +0100)]
Try and track link state better, regardless of if we're using
the interface or not.
Roy Marples [Mon, 21 Sep 2020 16:40:28 +0000 (17:40 +0100)]
privsep: Don't log sandbox type twice
Roy Marples [Mon, 21 Sep 2020 10:20:46 +0000 (11:20 +0100)]
privsep: Allow __NR_mmap for seccomp
Some malloc implementations might use it rather than brk.
Roy Marples [Sun, 20 Sep 2020 20:22:49 +0000 (21:22 +0100)]
privsep: Don't remove pidfile at exit
The privileged actioneer will remove it.
Roy Marples [Sun, 20 Sep 2020 19:22:40 +0000 (20:22 +0100)]
privsep: Add more platforms for seccomp
Really nutty. Why cannot a system header define a default?
Roy Marples [Sun, 20 Sep 2020 18:44:23 +0000 (19:44 +0100)]
privsep: Allow more syscalls for seccomp
dumplease now works over the control socket and stdin.
Roy Marples [Sun, 20 Sep 2020 18:24:26 +0000 (19:24 +0100)]
privsep: Send signal from launcher to master over the socket
rather than using kill which is not permitted in capsicum.
This also allows us to drop the proc pledge.
Roy Marples [Sun, 20 Sep 2020 18:09:08 +0000 (19:09 +0100)]
privsep: sandbox the launcher process
Roy Marples [Sun, 20 Sep 2020 16:43:31 +0000 (17:43 +0100)]
FreeBSD: Fix build
Roy Marples [Sat, 19 Sep 2020 23:53:47 +0000 (00:53 +0100)]
privsep: Avoid "unconfined" seccomp warnings
Unsure why these are needed atm....
Roy Marples [Sat, 19 Sep 2020 23:43:36 +0000 (00:43 +0100)]
privsep: Don't be noisy about the sandbox
Roy Marples [Sat, 19 Sep 2020 23:35:08 +0000 (00:35 +0100)]
privsep: Log if the platform sandbox is unavailable or available
This is kinda important.
Roy Marples [Sat, 19 Sep 2020 19:53:23 +0000 (20:53 +0100)]
privsep: Add the SECCOMP BPF sandbox for Linux
Now we have capsicum, pledge and the POSIX resource limited sandboxes
this was quite easy really.
Roy Marples [Sat, 19 Sep 2020 19:36:07 +0000 (20:36 +0100)]
Linux: Use SIOCSIFVLAN rather than SIOCSIFVLAN
Roy Marples [Sat, 19 Sep 2020 17:58:52 +0000 (18:58 +0100)]
privsep: Fold capsicum and pledge entry points into ps_entersandbox
Roy Marples [Sat, 19 Sep 2020 14:05:51 +0000 (15:05 +0100)]
Adjust prior so that message is logged before starting dev
Also add os_init incase other os need similar in the future.
Roy Marples [Sat, 19 Sep 2020 13:40:50 +0000 (14:40 +0100)]
Linux: detect network namespace and deny udev in one
udev says whether an interface name is stable or not.
In a network namespace, udev claims the interface does not exist.
This makes sense because udev only operates in the root namespace.
As such disable udev in a network namespace.
While here correctly spell initialise.
Roy Marples [Wed, 16 Sep 2020 14:55:11 +0000 (15:55 +0100)]
duid: Allow optional argument to specify ll or llt alongside uuid
Roy Marples [Tue, 15 Sep 2020 10:45:21 +0000 (11:45 +0100)]
install: Create DBDIR with mode 0750 by default
It shouldn't be 0755 as we might not want everyone to read
DHCP leases.
Roy Marples [Sat, 12 Sep 2020 19:14:47 +0000 (20:14 +0100)]
dhcpcd: Only manipulate stdin, stdout and stderr when valid
UNIX application expect these to exist even if pointed at /dev/null.
We cannot change which fd they use, it's always 0, 1 and 2.
But if these fd's are not open when dhcpcd is called, they could
be assigned to dhcpcd internals.
In this instance we should not use the streams in anyway or form.
Roy Marples [Sat, 12 Sep 2020 14:58:03 +0000 (15:58 +0100)]
logerr: Don't leak the logfile stream to scripts
Roy Marples [Fri, 11 Sep 2020 14:07:19 +0000 (15:07 +0100)]
README: Remove old reference to db dir needing user permissions
We've been using the original dirs since dhcpcd-9.1.
Roy Marples [Fri, 11 Sep 2020 11:21:07 +0000 (12:21 +0100)]
dhcpcd: avoid privsep errors if we error locking the pidfile
Should not happen on the happy path, but if it does, tidy the
resultant errors.
Roy Marples [Sun, 6 Sep 2020 14:27:28 +0000 (15:27 +0100)]
Release dhcpcd-9.2.0
Roy Marples [Sun, 6 Sep 2020 14:26:36 +0000 (15:26 +0100)]
DHCP: Fix compile without IPv4LL
Roy Marples [Sun, 6 Sep 2020 14:25:59 +0000 (15:25 +0100)]
configure: Remove PPOLL from config.mk
Roy Marples [Sun, 6 Sep 2020 12:53:08 +0000 (13:53 +0100)]
privsep: dump leases in a sandbox
Roy Marples [Sun, 6 Sep 2020 12:27:31 +0000 (13:27 +0100)]
dhcpcd: stderr callback fd may the reset by peer
When the process exits which is normal.
Roy Marples [Sun, 6 Sep 2020 11:20:40 +0000 (12:20 +0100)]
privsep: Dump leases from stdin in a limited sandbox
Roy Marples [Sun, 6 Sep 2020 10:58:29 +0000 (11:58 +0100)]
privsep: dropprivs can be static
Roy Marples [Sun, 6 Sep 2020 10:57:19 +0000 (11:57 +0100)]
privsep: limit rights for stdout/stderr/stdin using capsicum
Roy Marples [Sun, 6 Sep 2020 02:01:49 +0000 (03:01 +0100)]
FreeBSD: Don't rights limit stderr
Otherwise some utils in scripts will error with insufficient
capabilities.
Roy Marples [Sun, 6 Sep 2020 01:41:08 +0000 (02:41 +0100)]
dhcpcd: Redirect stdout/stderr to the launcher stderr descriptor
This actually make life really simple!
We no longer need to redirect stdout/stderr to /dev/null for privsep
and any script output is now captured again - and it all goes to stderr
as it should even if a script wants it to go to stdout.
On the happy path, only the master process will actually log anything
to stderr so we turn that off after we "fork".
On the unhappy path, logging to stderr/stdout *may* fail because
the launcher process *may* have exited.
We *could* have the master process as an intermediary but that's
just excess code to avoid errors which *should* not happen.
Regardless, any errror should still hit syslog.
Roy Marples [Sat, 5 Sep 2020 16:01:59 +0000 (17:01 +0100)]
privsep: Fix prior for FreeBSD.
Roy Marples [Sat, 5 Sep 2020 15:16:22 +0000 (16:16 +0100)]
dhcpcd: Setup a socketpair in the launcher to write to stderr
Rather than duping stderr down to the processes.
This allows us to reopen stdout and stderr onto /dev/null right
away and means only the launcher process writes to stderr.
The downside is that any stdout from the script is now lost.
If that's needed, we could setup a stdout socketpair as well.
Roy Marples [Sat, 5 Sep 2020 15:13:36 +0000 (16:13 +0100)]
logerr: Ensure that the fd err log is line buffered
Roy Marples [Sat, 5 Sep 2020 15:12:30 +0000 (16:12 +0100)]
privsep: Use xsocketpair
Roy Marples [Sat, 5 Sep 2020 15:10:30 +0000 (16:10 +0100)]
if: introduce xsocketpair, similar to xsocket
Old systems don't have SOCK_CLOEXEC, etc, this makes it easy.
While here, right limit the sockets.
Roy Marples [Wed, 2 Sep 2020 08:56:32 +0000 (09:56 +0100)]
DHCP: allow leasetime -1 to represent infinity
Easier to enter than
4294967295 seconds which is the real
representation of infinity.
Roy Marples [Tue, 1 Sep 2020 11:24:42 +0000 (12:24 +0100)]
DHCP: optimize prior for systems without IPv4 address lifetimes
Roy Marples [Tue, 1 Sep 2020 11:15:00 +0000 (12:15 +0100)]
DHCP: Adjust prior to be more generic
Roy Marples [Tue, 1 Sep 2020 10:56:53 +0000 (11:56 +0100)]
DHCP: Don't set address lifetimes when extending leases
Otherwise the kernel WILL remove them.
dhcpcd already manages address removal when needed because
some OS's do not support address lifetimes even for IPv6.
While here apply the same logic to IPv6.
Roy Marples [Mon, 31 Aug 2020 05:51:27 +0000 (06:51 +0100)]
dhcpcd: Rename function for prior now it no longer inits
Roy Marples [Mon, 31 Aug 2020 05:48:59 +0000 (06:48 +0100)]
dhcpcd: Adjust prior to ensure options exist for active interfaces before timeouts
Roy Marples [Sun, 30 Aug 2020 08:41:47 +0000 (09:41 +0100)]
DHCP: reason is BOUND when recovering from an expired extended lease
Roy Marples [Sun, 30 Aug 2020 08:07:35 +0000 (09:07 +0100)]
DHCP: When extending the lease, re-enter the DISCOVER phase.
Roy Marples [Sun, 30 Aug 2020 07:28:44 +0000 (08:28 +0100)]
dhcpcd: init new interfaces from route(4) overflow
This ensures that any newly active ones have an options structure.
Roy Marples [Sun, 30 Aug 2020 07:03:01 +0000 (08:03 +0100)]
DHCP: Request offered IP when using lastlease
Also, set expired state on using last lease rather than last lease extension.
Roy Marples [Thu, 20 Aug 2020 15:28:47 +0000 (16:28 +0100)]
privsep: Only the master process accepts signals
The master process dictates when processes should stop, which
allows for a clean shutdown when the admin issues `pkill dhcpcd`.
Roy Marples [Thu, 20 Aug 2020 15:09:53 +0000 (16:09 +0100)]
DHCP: Mark the state as expired on lase lease extension
Expired states do not claim any client IP to request from.
Roy Marples [Sun, 16 Aug 2020 18:21:48 +0000 (19:21 +0100)]
Use a minimum bufsize of 1 as 0 doesn't work on some OS.
Roy Marples [Sun, 16 Aug 2020 17:52:17 +0000 (18:52 +0100)]
privsep: Set a zero length receive buffer for write only sockets
We cannot use shutdown(2) because they are not connected.
Constantly draining would be a waste of CPU time, so just let
the buffer overflow. To ease the kernel as much as we can, set
a zero length buffer.
The kernel may still allocate a small buffer, but this is kernel
dependant and we're just trying to be helpful.
Roy Marples [Sat, 8 Aug 2020 19:27:34 +0000 (20:27 +0100)]
privsep: Improve some errors
Roy Marples [Sat, 8 Aug 2020 19:06:55 +0000 (20:06 +0100)]
dhcpcd: Fix logging we forked at exit.
Roy Marples [Tue, 4 Aug 2020 13:26:18 +0000 (14:26 +0100)]
Revert "Don't install /var/db/dhcpcd in the Makefile"
This was when the privsep code had the database directory in the
chroot, which it not longer is.
This reverts commit
5f275b7bd1ed4d1f830b7a60ba253a98f7ef6127 .
Roy Marples [Mon, 3 Aug 2020 16:04:51 +0000 (17:04 +0100)]
Revert "privsep: shutdown read end of the write only sockets"
Fails noisly on Linux.
This reverts commit
5fc20ae1aa368e36f9cabcbc047cd7d8e468c5b7 .
Roy Marples [Mon, 3 Aug 2020 15:56:14 +0000 (16:56 +0100)]
dhcpcd: If using -4 or -6, check we have support for it
Better to error early sensibly.
Roy Marples [Mon, 3 Aug 2020 15:55:51 +0000 (16:55 +0100)]
inet6: Linux will convert NULL to all, so don't specify all
Roy Marples [Mon, 3 Aug 2020 15:54:59 +0000 (16:54 +0100)]
Linux: fix reading the ipv6 forwarding proc entry
We need to ensure that the carriage return doesn't give issue.
Roy Marples [Mon, 3 Aug 2020 15:15:39 +0000 (16:15 +0100)]
Linux: IP6 forwaring only applies to "all" interface
Per interface forwarding is apparently only for setting things
like IsRouter in NA messages.
Roy Marples [Sat, 1 Aug 2020 19:56:13 +0000 (20:56 +0100)]
options: Let clientid override a prior duid
So you can do
duid
ssid foo
clientid
Roy Marples [Sat, 25 Jul 2020 13:22:18 +0000 (14:22 +0100)]
options: open an address socket to detect if inet6 is available
This should silence dhcpcd warnings if IPv6 is compiled out.
Do the same for IPv4, although that's highly unlikely as
it's kind of required for some interface ioctls.
Fabrice Fontaine [Mon, 13 Jul 2020 08:43:51 +0000 (10:43 +0200)]
src/dhcpcd.c: fix build without fork
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Sat, 1 Aug 2020 19:32:34 +0000 (20:32 +0100)]
privsep: Don't read control_group in privsep
There is no need as it's a global option and it will have already
been read in by this point.
Also means we don't need to pledge getpw.
Roy Marples [Sat, 18 Jul 2020 11:40:30 +0000 (12:40 +0100)]
privsep: shutdown read end of the write only sockets
Clearly shows our intent and hopefully is an optimisation
within the kernel.
Roy Marples [Wed, 8 Jul 2020 11:36:36 +0000 (12:36 +0100)]
Fix prior
Roy Marples [Wed, 8 Jul 2020 11:03:15 +0000 (12:03 +0100)]
route: Ensure IPv4LL routes always come last in priority
Roy Marples [Sat, 4 Jul 2020 12:21:09 +0000 (13:21 +0100)]
Fix compile without IPv6 on systems that do not define ALIGN
Roy Marples [Fri, 3 Jul 2020 10:41:23 +0000 (11:41 +0100)]
Make DBDIR permissions same as the files it creates
Roy Marples [Fri, 3 Jul 2020 10:33:39 +0000 (11:33 +0100)]
Release dhcpcd-9.1.4
Roy Marples [Fri, 3 Jul 2020 10:30:25 +0000 (11:30 +0100)]
dhcpcd: Fix creating DBDIR
As privsep now uses files in / and not the dhcpcd users directory
we need to ensure the mkdir is back with the rundir.
Roy Marples [Thu, 2 Jul 2020 17:14:01 +0000 (18:14 +0100)]
Fix SMALL build
Roy Marples [Thu, 2 Jul 2020 13:42:57 +0000 (14:42 +0100)]
Release dhcpcd-9.1.3
Roy Marples [Thu, 2 Jul 2020 13:05:19 +0000 (14:05 +0100)]
make: clean before import-src
Roy Marples [Wed, 1 Jul 2020 19:09:42 +0000 (20:09 +0100)]
DHCP6: Restore lease timings
Erroneously changed in
060f5a9e93 , only seems to affect
INFORM getting into a loop.
Roy Marples [Wed, 1 Jul 2020 10:46:01 +0000 (11:46 +0100)]
BSD: Setup sysctls before redirecting stderr
Need to do this as early as.