wireguard-openbsd/sys/netinet/ip_input.c, branch jd/queueboostsWireGuard implementation for the OpenBSD kernelhttps://git.zx2c4.com/wireguard-openbsd/atom/sys/netinet/ip_input.c?h=jd%2Fqueueboosts2020-04-12T11:56:52ZStop processing packets under non-exclusive (read) netlock.2020-04-12T11:56:52Zmpimpi@openbsd.org2020-04-12T11:56:52Zurn:sha1:3c6c39936460d119f3d0d08761b69e36164ad920
Prevent concurrency in the socket layer which is not ready for that.
Two recent data corruptions in pfsync(4) and the socket layer pointed
out that, at least, tun(4) was incorrectly using NET_RUNLOCK(). Until
we find a way in software to avoid future mistakes and to make sure that
only the softnet thread and some ioctls are safe to use a read version
of the lock, put everything back to the exclusive version.
ok stsp@, visa@
rdr-to with loopback destination should work even though2019-12-23T22:33:57Zsashansashan@openbsd.org2019-12-23T22:33:57Zurn:sha1:fd4abc02c0bfa4e17134718bec3a3ee471075d51
IP forwarding is disabled. Issue reported by Daniel Jakots (danj@)
OK bluhm@
always pull in if_types.h, to unbreak ramdisks2019-12-09T06:48:52Zderaadtderaadt@openbsd.org2019-12-09T06:48:52Zurn:sha1:59e813f93f123809a00fd1a7eff9e9c356922aacMake sure packet destination address matches interface address,2019-12-08T11:08:22Zsashansashan@openbsd.org2019-12-08T11:08:22Zurn:sha1:144d7e8ec45c41ef58c8b309e276b5f35ab67b6e
where such packet is bound to. This check is enforced if and only
IP forwarding is disabled.
Change discussed with bluhm@, claudio@, deraadt@, markus@, tobhe@
OK bluhm@, claudio@, tobhe@
When we needed the kernel lock for local IP packet delivery, mpi@2019-08-06T22:57:54Zbluhmbluhm@openbsd.org2019-08-06T22:57:54Zurn:sha1:8f6d2e8bf0b541dc10865826bab4d248a2639690
introduced a queue to grab the lock for multiple packets. Now we
have only netlock for both IP and protocol input. So the queue is
not necessary anymore. It just switches CPU and decreases performance.
So remove the inet and inet6 ip queue for local packets.
To get TCP running on loopback, we have to queue once between TCP
input and output of the two sockets. So use the loopback queue in
looutput() unconditionally.
OK visa@
use m_microtime instead of microtime for SO_TIMESTAMP socketopt handling2019-06-10T23:48:21Zdlgdlg@openbsd.org2019-06-10T23:48:21Zurn:sha1:9f78c7ff9bbb70b9cc0719cf32120a4d62b8defc
drivers can set ph_timestamp when packets are received by the
hardware, which should be more accurate and cheaper than getting
the clock when the packet is queued on the socket.
Expose net.inet.ip.arpq.drops to help debug what's going on when a lot2018-10-13T18:36:01Zflorianflorian@openbsd.org2018-10-13T18:36:01Zurn:sha1:346625e3e9cac8662dd8a8e2c95c141a55f182fa
of packets are being dropped but non of the other counters are increasing.
From Daniel Hokka Zakrisson (daniel AT hozac DOT com), thanks!
OK florian, phessler
Convert inetctlerrmap to u_char like inet6ctlerrmap. That is also2018-09-11T21:04:03Zbluhmbluhm@openbsd.org2018-09-11T21:04:03Zurn:sha1:7017ab2caa8a94949cf2ae3d044a49408ce8d2e4
what FreeBSD does. Remove old #if 0 version of inet6ctlerrmap.
OK mpi@
Instead of calculating the mbuf packet header length here and there,2018-09-10T16:14:07Zbluhmbluhm@openbsd.org2018-09-10T16:14:07Zurn:sha1:f02cb2e83cda2e7c843af31d86dd656fd741d8ab
put the algorithm into a new function m_calchdrlen(). Also set an
uninitialized m_len to 0 in NFS code.
OK claudio@
During fragment reassembly, mbuf chains with packet headers were2018-09-10T12:47:02Zbluhmbluhm@openbsd.org2018-09-10T12:47:02Zurn:sha1:c1746f11b0c8af70eecbe03fcd247444643d7685
created. Add a new function m_removehdr() do convert packet header
mbufs within the chain to regular mbufs. Assert that the mbuf at
the beginning of the chain has a packet header.
found by Maxime Villard in NetBSD; from markus@; OK claudio@