Apparently Haiku has a misbehaving /dev/urandom.
While we're at it, simplify the function signature to completely succeed
or completely fail and make sure the caller checks the result.
Reported-by: Alexander von Gluck IV <kallisti5@unixzen.com>
Nitpicked-by: Aaron Jones <aaronmdjones@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The commit 7c833642 ("wg-quick: freebsd: allow loopback to work") was
supposed to make things better, but actually it just started sending
legitimate localhost traffic over the WireGuard interface, which is
really quite bad.
This reverts commit 7c833642dfa342218602ab18e7091e86408d2982.
Reported-by: Matt Smith <matt.xtaz@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since wg-quick(8) calls wg(8) which does hostname lookups, we should
probably only run this after we're allowed to look up hostnames.
Reported-by: Anton Castelli <anton.c42@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
FreeBSD adds a route for point-to-point destination addresses. We don't
really want to specify any destination address, but unfortunately we
have to. Before we tried to cheat by giving our own address as the
destination, but this had the unfortunate effect of preventing
loopback from working on our local ip address. We work around this with
yet another kludge: we set the destination address to 127.0.0.1. Since
127.0.0.1 is already assigned to an interface, this has the same effect
of not specifying a destination address, and therefore we accomplish the
intended behavior.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This avoids another ip(8) invocation for little benefit.
Confirmed to work with iproute2 and busybox.
Signed-off-by: Aaron Jones <aaronmdjones@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
An empty allowed IPs is totally valid, for folks wishing to move IP
addresses between multiple peers atomically.
Suggested-by: Comex <comexk@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Netlink returns NLM_F_DUMP_INTR if the set of all tunnels changed
during the dump. That's unfortunate, but is pretty common on busy
systems that are adding and removing tunnels all the time. Rather
than retrying, potentially indefinitely, we just work with the
partial results.
Reported-by: Robert Gerus <ar@is-a.cat>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
We don't actually use any C11 features, so we can at least compile with
ancient gcc.
Reported-by: Aaron M. D. Jones <aaronmdjones@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
It's not used for anything, and LKML doesn't like the type being used as
an index value.
Suggested-by: Eugene Syromiatnikov <esyr@redhat.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Suggested-by: Toke Høiland-Jørgensen <toke@toke.dk>
Suggested-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The kernel has very specific rules correlating file type with comment
type, and also SPDX identifiers can't be merged with other comments.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This is not useful for WireGuard, but for the general use case we
probably want it this way, and the speed difference is mostly lost in
the noise.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The only things wg-quick(8) needs from Homebrew are bash(1) and wg(8).
Other than that, it's explicitly coded against the native system
utilities. Since wg-quick(8) and bash(1) are invoked in auto_su by their
full absolute path (via $SELF and $BASH, respectively), we can simply
set the $PATH to be prefixed by the default system binary paths. This
way, if users install tools that conflict with system tools -- such as
GNU coreutils -- we won't accidently call those.
Reported-by: Deirdre Connolly <durumcrustulum@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
It's unclear why it was like this in the first place, but it apparently
broke certain IPv6 setups.
Reported-by: Jonas Blahut <j@die-blahuts.de>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This eliminates a few style warnings from "mandoc -T lint src/tools/wg*.8".
Signed-off-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
`ndc users add` eventually invokes SOCK_DESTROY on user sockets, causing
them to reconnect. By delaying this until after routes are set, we
ensure that the sockets reconnect using the tunnel, rather than the old
route.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This works around a race condition in macOS's network daemons, while
also adding one in the form of possibly calling kill -ALRM on a stale
PID; unfortunately bash can't wait from a trap.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This knob might disappear at some point, and we don't want to encourage
its use, so it's not being documented, but this should help with
development of new implementations.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
For properly configured Homebrew installations /usr/local/bin should be
before /bin, so this should still work. This allows the script to be
used in more than one setting.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
If we're doing automatic routing with default routes, but the config has
also specified an explicit fwmark, then use that explicit fwmark, even
if it's conflicting, since the administrator has explicitly opted into
using it. Also, when shutting down the interface, we only now remove the
fancy rules if we're in automatic routing mode with default routes.
Suggested-by: Luis Ressel <aranea@aixah.de>
Reported-by: Saeid Akbari <saeidscorp@yahoo.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Some older broken resolvconfs don't support resolvconf -l, but do have a
file in a standard location, so use it.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Some older broken resolvconf implementations ignore -m, but do have an
interface-order list. It's better to use this list dynamically, in case
it changes, or in case it's not used by the OS's resolvconf
implementation, such as in the case of systemd or openresolv.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Before, this meant that it simply took the last 15 characters, instead
of erroring out when there's more than 15 chars.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This reverts commit da4ff396cc5d5e0ff21f9ecbc2f951c048c63fff and adds
some optimizations to hacl64.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
For now, it's faster:
hacl64: 109782 cycles per call
fiat64: 108984 cycles per call
It's quite possible this commit will be reverted with nice changes from
INRIA, though.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
While this has a negative performance impact on x86_64, it has a
positive performance impact on smaller machines, which is where we're
actually using this code. For example, an A53:
Before: fiat32: 228605 cycles per call
After: fiat32: 188307 cycles per call
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Alexander von Gluck IV
Jason A. Donenfeld
Aaron Jones
Jonathan Neuschäfer
Filippo Valsorda