This is actually several different bugs:
- The code is not designed to handle inpcb deletion after interface deletion
- add reference for inpcb membership
- The multicast address has to be removed from interface lists when the refcount
goes to zero OR when the interface goes away
- decouple list disconnect from refcount (v6 only for now)
- ifmultiaddr can exist past being on interface lists
- add flag for tracking whether or not it's enqueued
- deferring freeing moptions makes the incpb cleanup code simpler but opens the
door wider still to races
- call inp_gcmoptions synchronously after dropping the the inpcb lock
Fundamentally multicast needs a rewrite - but keep applying band-aids for now.
Tested by: kp
Reported by: novel, kp, lwhsu
Currently, the limits are quite high. On machines with millions of
mbuf clusters, the reassembly queue limits can also run into
the millions. Lower these values.
Also, try to ensure that no bucket will have a reassembly
queue larger than approximately 100 items. This limits the cost to
find the correct reassembly queue when processing an incoming
fragment.
Due to the low limits on each bucket's length, increase the size of
the hash table from 64 to 1024.
Reviewed by: jhb
Security: FreeBSD-SA-18:10.ip
Security: CVE-2018-6923
Currently, we process IPv6 fragments with 0 bytes of payload, add them
to the reassembly queue, and do not recognize them as duplicating or
overlapping with adjacent 0-byte fragments. An attacker can exploit this
to create long fragment queues.
There is no legitimate reason for a fragment with no payload. However,
because IPv6 packets with an empty payload are acceptable, allow an
"atomic" fragment with no payload.
Reviewed by: jhb
Security: FreeBSD-SA-18:10.ip
Security: CVE-2018-6923
There is a hashing algorithm which should distribute IPv6 reassembly
queues across the available buckets in a relatively even way. However,
if there is a flaw in the hashing algorithm which allows a large number
of IPv6 fragment reassembly queues to end up in a single bucket, a per-
bucket limit could help mitigate the performance impact of this flaw.
Implement such a limit, with a default of twice the maximum number of
reassembly queues divided by the number of buckets. Recalculate the
limit any time the maximum number of reassembly queues changes.
However, allow the user to override the value using a sysctl
(net.inet6.ip6.maxfragbucketsize).
Reviewed by: jhb
Security: FreeBSD-SA-18:10.ip
Security: CVE-2018-6923
The IPv4 fragment reassembly code supports a limit on the number of
fragments per packet. The default limit is currently 17 fragments.
Among other things, this limit serves to limit the number of fragments
the code must parse when trying to reassembly a packet.
Add a limit to the IPv6 reassembly code. By default, limit a packet
to 65 fragments (64 on the queue, plus one final fragment to complete
the packet). This allows an average fragment size of 1,008 bytes, which
should be sufficient to hold a fragment. (Recall that the IPv6 minimum
MTU is 1280 bytes. Therefore, this configuration allows a full-size
IPv6 packet to be fragmented on a link with the minimum MTU and still
carry approximately 272 bytes of headers before the fragmented portion
of the packet.)
Users can adjust this limit using the net.inet6.ip6.maxfragsperpacket
sysctl.
Reviewed by: jhb
Security: FreeBSD-SA-18:10.ip
Security: CVE-2018-6923
The IPv6 reassembly fragment limit is based on the number of mbuf clusters,
which are a global resource. However, the limit is currently applied
on a per-VNET basis. Given enough VNETs (or given sufficient customization
on enough VNETs), it is possible that the sum of all the VNET fragment
limits will exceed the number of mbuf clusters available in the system.
Given the fact that the fragment limits are intended (at least in part) to
regulate access to a global resource, the IPv6 fragment limit should
be applied on a global basis.
Note that it is still possible to disable fragmentation for a particular
VNET by setting the net.inet6.ip6.maxfragpackets sysctl to 0 for that
VNET. In addition, it is now possible to disable fragmentation globally
by setting the net.inet6.ip6.maxfrags sysctl to 0.
Reviewed by: jhb
Security: FreeBSD-SA-18:10.ip
Security: CVE-2018-6923
Currently, all IPv6 fragment reassembly queues are kept in a flat
linked list. This has a number of implications. Two significant
implications are: all reassembly operations share a common lock,
and it is possible for the linked list to grow quite large.
Improve IPv6 reassembly performance by hashing fragments into buckets,
each of which has its own lock. Calculate the hash key using a Jenkins
hash with a random seed.
Reviewed by: jhb
Security: FreeBSD-SA-18:10.ip
Security: CVE-2018-6923
It was lost when tryforward appeared. Now ip[6]_tryforward will be enabled
only when sending redirects for corresponding IP version is disabled via
sysctl. Otherwise will be used default forwarding function.
PR: 221137
Submitted by: mckay@
MFC after: 2 weeks
On PowerPC (and possibly other architectures), that doesn't use
EARLY_AP_STARTUP, the config task queue may be used initialized.
This was observed while trying to mount the root fs from NFS, as
reported here: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=230168.
This patch has 2 main changes:
1- Perform a basic initialization of qgroup_config, similar to
what is done in taskqgroup_adjust, but simpler.
This makes qgroup_config ready to be used during NFS root mount.
2- When EARLY_AP_STARTUP is not used, call inm_init() and
in6m_init() right before SI_SUB_ROOT_CONF, because bootp needs
to send multicast packages to request an IP.
PR: Bug 230168
Reported by: sbruno
Reviewed by: jhibbits, mmacy, sbruno
Approved by: jhibbits
Differential Revision: D16633
The dtrace provider for UDP-Lite is modeled after the UDP provider.
This fixes the bug that UDP-Lite packets were triggering the UDP
provider.
Thanks to dteske@ for providing the dwatch module.
Reviewed by: dteske@, markj@, rrs@
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D16377
TCP/IPv4 allows an implicit connection setup using sendto(), which
is used for TTCP and TCP fast open. This patch adds support for
TCP/IPv6.
While there, improve some tests for detecting multicast addresses,
which are mapped.
Reviewed by: bz@, kbowling@, rrs@
Sponsored by: Netflix, Inc.
Differential Revision: https://reviews.freebsd.org/D16458
The timespecadd(3) family of macros were imported from NetBSD back in
r35029. However, they were initially guarded by #ifdef _KERNEL. In the
meantime, we have grown at least 28 syscalls that use timespecs in some
way, leading many programs both inside and outside of the base system to
redefine those macros. It's better just to make the definitions public.
Our kernel currently defines two-argument versions of timespecadd and
timespecsub. NetBSD, OpenBSD, and FreeDesktop.org's libbsd, however, define
three-argument versions. Solaris also defines a three-argument version, but
only in its kernel. This revision changes our definition to match the
common three-argument version.
Bump _FreeBSD_version due to the breaking KPI change.
Discussed with: cem, jilles, ian, bde
Differential Revision: https://reviews.freebsd.org/D14725
Fire UDP receive probes when a packet is received and there is no
endpoint consuming it. Fire the probe also if the TTL of the
received packet is smaller than the minimum required by the endpoint.
Clarify also in the man page, when the probe fires.
Reviewed by: dteske@, markj@, rrs@
Sponsored by: Netflix, Inc.
Differential Revision: https://reviews.freebsd.org/D16046
This deduplicates the code a bit, and also implicitly adds missing
callout_stop() to in[6]_lltable_delete_entry() functions.
PR: 209682, 225927
Submitted by: hselasky (previous version)
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D4605
Simple fix to address panics relating to setting IPV6_TCLASS
with setsockopt(). The premise of this change is that it is
ok to call malloc with M_NOWAIT while holding a lock on the
in6p.
If it later turns out that it is not ok, then major surgery
will be required, as ip6_setpktopt() will have to be fixed
(as it also calls malloc with M_NOWAIT) which pulls in the
ip6_pcbopts(), ip6_setpktopts(), ip6_setpktopt() call chain.
Submitted by: Jason Eggnet
Reviewed by: rrs, transport, sbruno
Sponsored by: Limelight Networks
Differential Revision: https://reviews.freebsd.org/D16201
- Add tracker argument to preemptible epochs
- Inline epoch read path in kernel and tied modules
- Change in_epoch to take an epoch as argument
- Simplify tfb_tcp_do_segment to not take a ti_locked argument,
there's no longer any benefit to dropping the pcbinfo lock
and trying to do so just adds an error prone branchfest to
these functions
- Remove cases of same function recursion on the epoch as
recursing is no longer free.
- Remove the the TAILQ_ENTRY and epoch_section from struct
thread as the tracker field is now stack or heap allocated
as appropriate.
Tested by: pho and Limelight Networks
Reviewed by: kbowling at llnw dot com
Sponsored by: Limelight Networks
Differential Revision: https://reviews.freebsd.org/D16066
encap_lookup_t method can be invoked by IP encap subsytem even if none
of gif/gre/me interfaces are exist. Hash tables are allocated on demand,
when first interface is created. So, make NULL pointer check before
doing access to hash table.
PR: 229378
Using of rwlock with multiqueue NICs for IP forwarding on high pps
produces high lock contention and inefficient. Rmlock fits better for
such workloads.
Reviewed by: melifaro, olivier
Obtained from: Yandex LLC
Sponsored by: Yandex LLC
Differential Revision: https://reviews.freebsd.org/D15789
It is possible that ifma_protospec becomes NULL in this function for
some entry, but it is still referenced and thus it will not unlinked
from the list. Then "restart" condition triggers and this entry with
NULL ifma_protospec will lead to page fault.
PR: 228982
of needed interface when many gre interfaces are present.
Remove rmlock from gre_softc, use epoch(9) and CK_LIST instead.
Move more AF-related code into AF-related locations. Use hash table to
speedup lookup of needed softc.
This patch adds a new socket option, SO_REUSEPORT_LB, which allow multiple
programs or threads to bind to the same port and incoming connections will be
load balanced using a hash function.
Most of the code was copied from a similar patch for DragonflyBSD.
However, in DragonflyBSD, load balancing is a global on/off setting and can not
be set per socket. This patch allows for simultaneous use of both the current
SO_REUSEPORT and the new SO_REUSEPORT_LB options on the same system.
Required changes to structures:
Globally change so_options from 16 to 32 bit value to allow for more options.
Add hashtable in pcbinfo to hold all SO_REUSEPORT_LB sockets.
Limitations:
As DragonflyBSD, a load balance group is limited to 256 pcbs (256 programs or
threads sharing the same socket).
This is a substantially different contribution as compared to its original
incarnation at svn r332894 and reverted at svn r332967. Thanks to rwatson@
for the substantive feedback that is included in this commit.
Submitted by: Johannes Lundberg <johalun0@gmail.com>
Obtained from: DragonflyBSD
Relnotes: Yes
Sponsored by: Limelight Networks
Differential Revision: https://reviews.freebsd.org/D11003
of needed interface when many gif interfaces are present.
Remove rmlock from gif_softc, use epoch(9) and CK_LIST instead.
Move more AF-related code into AF-related locations.
Use hash table to speedup lookup of needed softc. Interfaces
with GIF_IGNORE_SOURCE flag are stored in plain CK_LIST.
Sysctl net.link.gif.parallel_tunnels is removed. The removal was planed
16 years ago, and actually it could work only for outbound direction.
Each protocol, that can be handled by if_gif(4) interface is registered
by separate encap handler, this helps avoid invoking the handler
for unrelated protocols (GRE, PIM, etc.).
This change allows dramatically improve performance when many gif(4)
interfaces are used.
Sponsored by: Yandex LLC
Currently it has several disadvantages:
- it uses single mutex to protect internal structures. It is used by
data- and control- path, thus there are no parallelism at all.
- it uses single list to keep encap handlers for both INET and INET6
families.
- struct encaptab keeps unneeded information (src, dst, masks, protosw),
that isn't used by code in the source tree.
- matches are prioritized and when many tunneling interfaces are
registered, encapcheck handler of each interface is invoked for each
packet. The search takes O(n) for n interfaces. All this work is done
with exclusive lock held.
What this patch includes:
- the datapath is converted to be lockless using epoch(9) KPI.
- struct encaptab now linked using CK_LIST.
- all unused fields removed from struct encaptab. Several new fields
addedr: min_length is the minimum packet length, that encapsulation
handler expects to see; exact_match is maximum number of bits, that
can return an encapsulation handler, when it wants to consume a packet.
- IPv6 and IPv4 handlers are stored in separate lists;
- added new "encap_lookup_t" method, that will be used later. It is
targeted to speedup lookup of needed interface, when gif(4)/gre(4) have
many interfaces.
- the need to use protosw structure is eliminated. The only pr_input
method was used from this structure, so I don't see the need to keep
using it.
- encap_input_t method changed to avoid using mbuf tags to store softc
pointer. Now it is passed directly trough encap_input_t method.
encap_getarg() funtions is removed.
- all sockaddr structures and code that uses them removed. We don't have
any code in the tree that uses them. All consumers use encap_attach_func()
method, that relies on invoking of encapcheck() to determine the needed
handler.
- introduced struct encap_config, it contains parameters of encap handler
that is going to be registered by encap_attach() function.
- encap handlers are stored in lists ordered by exact_match value, thus
handlers that need more bits to match will be checked first, and if
encapcheck method returns exact_match value, the search will be stopped.
- all current consumers changed to use new KPI.
Reviewed by: mmacy
Sponsored by: Yandex LLC
Differential Revision: https://reviews.freebsd.org/D15617
Per vnet(9), CURVNET_SET and CURVNET_RESTORE cannot be used as a single
statement for a conditional and CURVNET_RESTORE must be in the same
block as CURVNET_SET (or a subblock).
Reviewed by: andrew
Sponsored by: The FreeBSD Foundation
Avoid the ugly unlock / lock of the inpcbinfo where we need to
figure out what kind of lock we hold by simply deferring the
operation to another context. (Also a small dependency for
converting the pcbinfo read lock to epoch)
r333175 updated the join_group functions, but not the leave_group ones.
Reviewed by: sbruno
Sponsored by: Limelight Networks
Differential Revision: https://reviews.freebsd.org/D15393
