Add if_requestencap() interface method which is capable of calculating
various link headers for given interface. Right now there is support
for INET/INET6/ARP llheader calculation (IFENCAP_LL type request).
Other types are planned to support more complex calculation
(L2 multipath lagg nexthops, tunnel encap nexthops, etc..).
Reshape 'struct route' to be able to pass additional data (with is length)
to prepend to mbuf.
These two changes permits routing code to pass pre-calculated nexthop data
(like L2 header for route w/gateway) down to the stack eliminating the
need for other lookups. It also brings us closer to more complex scenarios
like transparently handling MPLS nexthops and tunnel interfaces.
Last, but not least, it removes layering violation introduced by flowtable
code (ro_lle) and simplifies handling of existing if_output consumers.
ARP/ND changes:
Make arp/ndp stack pre-calculate link header upon installing/updating lle
record. Interface link address change are handled by re-calculating
headers for all lles based on if_lladdr event. After these changes,
arpresolve()/nd6_resolve() returns full pre-calculated header for
supported interfaces thus simplifying if_output().
Move these lookups to separate ether_resolve_addr() function which ether
returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr()
compat versions to return link addresses instead of pre-calculated data.
BPF changes:
Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT.
Despite the naming, both of there have ther header "complete". The only
difference is that interface source mac has to be filled by OS for
AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside
BPF and not pollute if_output() routines. Convert BPF to pass prepend data
via new 'struct route' mechanism. Note that it does not change
non-optimized if_output(): ro_prepend handling is purely optional.
Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI.
It is not needed for ethernet anymore. The only remaining FDDI user is
dev/pdq mostly untouched since 2007. FDDI support was eliminated from
OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65).
Flowtable changes:
Flowtable violates layering by saving (and not correctly managing)
rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated
header data from that lle.
Differential Revision: https://reviews.freebsd.org/D4102
initialization.
Mfp4 @180384,180385:
There is no need for a dedicated SYSINIT here. The
list can be initialized statically.
Sponsored by: CK Software GmbH
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Reviewed by: gnn
Differential Revision: https://reviews.freebsd.org/D4528
LLE structure is mostly unchanged during its lifecycle: there are only 2
things relevant for fast path lookup code:
1) link-level address change. Since r286722, these updates are performed
under AFDATA WLOCK.
2) Some sort of feedback indicating that this particular entry is used so
we send NS to perform reachability verification instead of expiring entry.
The only signal that is needed from fast path is something like binary
yes/no.
The latter is solved by the following changes:
Special r_skip_req (introduced in D3688) value is used for fast path feedback.
It is read lockless by fast path, but updated under req_mutex mutex. If this
field is non-zero, then fast path will acquire lock and set it back to 0.
After transitioning to STALE state, callout timer is armed to run each
V_nd6_delay seconds to make sure that if packet was transmitted at the start
of given interval, we would be able to switch to PROBE state in V_nd6_delay
seconds as user expects.
(in STALE state) timer is rescheduled until original V_nd6_gctimer expires
keeping lle in STALE state (remaining timer value stored in lle_remtime).
(in STALE state) timer is rescheduled if packet was transmitted less that
V_nd6_delay seconds ago to make sure we transition to PROBE state exactly
after V_n6_delay seconds.
As a result, all packets towards lle in REACHABLE/STALE/PROBE states are handled
by fast path without acquiring lle read lock.
Differential Revision: https://reviews.freebsd.org/D3780
LLE structure is mostly unchanged during its lifecycle.
To be more specific, there are 2 things relevant for fast path
lookup code:
1) link-level address change. Since r286722, these updates are performed
under AFDATA WLOCK.
2) Some sort of feedback indicating that this particular entry is used so
we re-send arp request to perform reachability verification instead of
expiring entry. The only signal that is needed from fast path is something
like binary yes/no.
The latter is solved by the following changes:
1) introduce special r_skip_req field which is read lockless by fast path,
but updated under (new) req_mutex mutex. If this field is non-zero, then
fast path will acquire lock and set it back to 0.
2) introduce simple state machine: incomplete->reachable<->verify->deleted.
Before that we implicitely had incomplete->reachable->deleted state machine,
with V_arpt_keep between "reachable" and "deleted". Verification was performed
in runtime 5 seconds before V_arpt_keep expire.
This is changed to "change state to verify 5 seconds before V_arpt_keep,
set r_skip_req to non-zero value and check it every second". If the value
is zero - then send arp verification probe.
These changes do not introduce any signifficant control plane overhead:
typically lle callout timer would fire 1 time more each V_arpt_keep (1200s)
for used lles and up to arp_maxtries (5) for dead lles.
As a result, all packets towards "reachable" lle are handled by fast path without
acquiring lle read lock.
Additional "req_mutex" is needed because callout / arpresolve_slow() or eventhandler
might keep LLE lock for signifficant amount of time, which might not be feasible
for fast path locking (e.g. having rmlock as ether AFDATA or lltable own lock).
Differential Revision: https://reviews.freebsd.org/D3688
new return codes of -1 were mistakenly being considered "true". Callout_stop
now returns -1 to indicate the callout had either already completed or
was not running and 0 to indicate it could not be stopped. Also update
the manual page to make it more consistent no non-zero in the callout_stop
or callout_reset descriptions.
MFC after: 1 Month with associated callout change.
Initially function was introduced in r53541 (KAME initial commit) to
"provide hints from upper layer protocols that indicate a connection
is making "forward progress"" (quote from RFC 2461 7.3.1 Reachability
Confirmation).
However, it was converted to do nothing (e.g. just return) in r122922
(tcp_hostcache implementation) back in 2003. Some defines were moved
to tcp_var.h in r169541. Then, it was broken (for non-corner cases)
by r186119 (L2<>L3 split) in 2008 (NULL ifp in nd6_lookup). So,
right now this code is broken and has no "real" base users.
Differential Revision: https://reviews.freebsd.org/D3699
without holding afdata wlock
* convert per-af delete_address callback to global lltable_delete_entry() and
more low-level "delete this lle" per-af callback
* fix some bugs/inconsistencies in IPv4/IPv6 ifscrub procedures
Sponsored by: Yandex LLC
Differential Revision: https://reviews.freebsd.org/D3573
Before that, the logic besides lle_create() was the following:
return existing if found, create if not. This behaviour was error-prone
since we had to deal with 'sudden' static<>dynamic lle changes.
This commit fixes bunch of different issues like:
- refcount leak when lle is converted to static.
Simple check case:
console 1:
while true;
do for i in `arp -an|awk '$4~/incomp/{print$2}'|tr -d '()'`;
do arp -s $i 00:22:44:66:88:00 ; arp -d $i;
done;
done
console 2:
ping -f any-dead-host-in-L2
console 3:
# watch for memory consumption:
vmstat -m | awk '$1~/lltable/{print$2}'
- possible problems in arptimer() / nd6_timer() when dropping/reacquiring
lock.
New logic explicitly handles use-or-create cases in every lla_create
user. Basically, most of the changes are purely mechanical. However,
we explicitly avoid using existing lle's for interface/static LLE records.
* While here, call lle_event handlers on all real table lle change.
* Create lltable_free_entry() calling existing per-lltable
lle_free_t callback for entry deletion
* Split lltable_init() into lltable_allocate_htbl() (alloc
hash table with default callbacks) and lltable_link() (
links any lltable to the list).
* Switch from LLTBL_HASHTBL_SIZE to per-lltable hash size field.
* Move lltable setup to separate functions in in[6]_domifattach.
differences between projects/routing and HEAD.
This commit tries to keep code logic the same while changing underlying
code to use unified callbacks.
* Add llt_foreach_entry method to traverse all entries in given llt
* Add llt_dump_entry method to export particular lle entry in sysctl/rtsock
format (code is not indented properly to minimize diff). Will be fixed
in the next commits.
* Add llt_link_entry/llt_unlink_entry methods to link/unlink particular lle.
* Add llt_fill_sa_entry method to export address in the lle to sockaddr
format.
* Add llt_hash method to use in generic hash table support code.
* Add llt_free_entry method which is used in llt_prefix_free code.
* Prepare for fine-grained locking by separating lle unlink and deletion in
lltable_free() and lltable_prefix_free().
* Provide lltable_get<ifp|af>() functions to reduce direct 'struct lltable'
access by external callers.
* Remove @llt agrument from lle_free() lle callback since it was unused.
* Temporarily add L3_CADDR() macro for 'const' sockaddr typecasting.
* Switch to per-af hashing code.
* Rename LLE_FREE_LOCKED() callback from in[6]_lltable_free() to
in_[6]lltable_destroy() to avoid clashing with llt_free_entry() method.
Update description from these functions.
* Use unified lltable_free_entry() function instead of per-af one.
Reviewed by: ae
* Make most of lltable_* methods 'normal' functions instead of inline
* Add lltable_get_<af|ifp>() functions to access given lltable fields
* Temporarily resurrect nd6_lookup() function
and explicit calls to RTENTRY_FREE_LOCKED()
* Use lltable_prefix_free() in arp_ifscrub to be consistent with nd6.
* Rename <lltable_|llt>_delete function to _delete_addr() to note that
this function is used to external callers. Make this function maintain
its own locking.
* Use lookup/unlink/clear call chain from internal callers instead of
delete_addr.
* Fix LLE_DELETED flag handling
cleanup including unlinking/freeing
* Relax locking in lltable_prefix_free_af/lltable_free
* Do not pass @llt to lle free callback: it is always NULL now.
* Unify arptimer/nd6_llinfo_timer: explicitly unlock lle avoiding
unlock/lock sequinces
* Do not pass unlocked lle to nd6_ns_output(): add nd6_llinfo_get_holdsrc()
to retrieve preferred source address from lle hold queue and pass it
instead of lle.
* Finally, make nd6_create() create and return unlocked lle
* Separate defrtr handling code from nd6_free():
use nd6_check_del_defrtr() to check if we need to keep entry instead of
performing GC,
use nd6_check_recalc_defrtr() to perform actual recalc on lle removal.
* Move isRouter handling from nd6_cache_lladdr() to separate
nd6_check_router()
* Add initial code to maintain lle runtime flags in sync.
* struct llentry is now basically split into 2 pieces:
all fields within 64 bytes (amd64) are now protected by both
ifdata lock AND lle lock, e.g. you require both locks to be held
exclusively for modification. All data necessary for fast path
operations is kept here. Some fields were added:
- r_l3addr - makes lookup key liev within first 64 bytes.
- r_flags - flags, containing pre-compiled decision whether given
lle contains usable data or not. Current the only flag is RLLE_VALID.
- r_len - prepend data len, currently unused
- r_kick - used to provide feedback to control plane (see below).
All other fields are protected by lle lock.
* Add simple state machine for ARP to handle "about to expire" case:
Current model (for the fast path) is the following:
- rlock afdata
- find / rlock rte
- runlock afdata
- see if "expire time" is approaching
(time_uptime + la->la_preempt > la->la_expire)
- if true, call arprequest() and decrease la_preempt
- store MAC and runlock rte
New model (data plane):
- rlock afdata
- find rte
- check if it can be used using r_* fields only
- if true, store MAC
- if r_kick field != 0 set it to 0.
- runlock afdata
New mode (control plane):
- schedule arptimer to be called in (V_arpt_keep - V_arp_maxtries)
seconds instead of V_arpt_keep.
- on first timer invocation change state from ARP_LLINFO_REACHABLE
to ARP_LLINFO_VERIFY, sets r_kick to 1 and shedules next call in
V_arpt_rexmit (default to 1 sec).
- on subsequent timer invocations in ARP_LLINFO_VERIFY state, checks
for r_kick value: reschedule if not changed, and send arprequest()
if set to zero (e.g. entry was used).
* Convert IPv4 path to use new single-lock approach. IPv6 bits to follow.
* Slow down in_arpinput(): now valid reply will (in most cases) require
acquiring afdata WLOCK twice. This is requirement for storing changed
lle data. This change will be slightly optimized in future.
* Provide explicit hash link/unlink functions for both ipv4/ipv6 code.
This will probably be moved to generic lle code once we have per-AF
hashing callback inside lltable.
* Perform lle unlink on deletion immediately instead of delaying it to
the timer routine.
* Make r244183 more explicit: use new LLE_CALLOUTREF flag to indicate the
presence of lle reference used for safe callout calls.
lla_lookup(LLE_CREATE) -> lla_create
lla_lookup(LLE_DELETE) -> lla_delete
Assume lla_create to return LLE_EXCLUSIVE lock for lle.
* Rework lla_rt_output to perform all lle changes under afdata WLOCK.
* change arp_ifscrub() ackquire afdata WLOCK, the same as arp_ifinit().
reside on their own cache line to prevent false sharing with other
nearby structures, especially for those in the .bss segment.
NB: Those mutexes and rwlocks with variables next to them that get
changed on every invocation do not benefit from their own cache line.
Actually it may be net negative because two cache misses would be
incurred in those cases.
Since ARP and routing are separated, "proxy only" entries
don't have any meaning, thus we don't need additional field
in sockaddr to pass SIN_PROXY flag.
New kernel is binary compatible with old tools, since sizes
of sockaddr_inarp and sockaddr_in match, and sa_family are
filled with same value.
The structure declaration is left for compatibility with
third party software, but in tree code no longer use it.
Reviewed by: ru, andre, net@
entering llentry_free(), and in case if we lose the race, we should simply
perform LLE_FREE_LOCKED(). Otherwise, if the race is lost by the thread
performing arptimer(), it will remove two references from the lle instead
of one.
Reported by: Ian FREISLICH <ianf clue.co.za>
llentry_free() and arptimer():
o Use callout_init_rw() for lle timeout, this allows us safely
disestablish them.
- This allows us to simplify the arptimer() and make it
race safe.
o Consistently use ifp->if_afdata_lock to lock access to
linked lists in the lle hashes.
o Introduce new lle flag LLE_LINKED, which marks an entry that
is attached to the hash.
- Use LLE_LINKED to avoid double unlinking via consequent
calls to llentry_free().
- Mark lle with LLE_DELETED via |= operation istead of =,
so that other flags won't be lost.
o Make LLE_ADDREF(), LLE_REMREF() and LLE_FREE_LOCKED() more
consistent and provide more informative KASSERTs.
The patch is a collaborative work of all submitters and myself.
PR: kern/165863
Submitted by: Andrey Zonov <andrey zonov.org>
Submitted by: Ryan Stone <rysto32 gmail.com>
Submitted by: Eric van Gyzen <eric_van_gyzen dell.com>
referenced within its timeout window. This change clears the LLE_VALID flag when an llentry
is removed from an interface's hash table and adds an extra check to the flowtable code
for the LLE_VALID flag in llentry to avoid retaining and using a stale reference.
Reviewed by: qingli@
MFC after: 2 weeks
interface is brought down, even though the interface address is still
valid. This patch maintains the permanent ARP entries as long as the
interface address (having the same prefix as that of the ARP entries)
is valid.
Reviewed by: delphij
MFC after: 5 days
DPCPU_DEFINE and VNET_DEFINE macros, as these cause problems for various
people working on the affected files. A better long-term solution is
still being considered. This reversal may give some modules empty
set_pcpu or set_vnet sections, but these are harmless.
Changes reverted:
------------------------------------------------------------------------
r215318 | dim | 2010-11-14 21:40:55 +0100 (Sun, 14 Nov 2010) | 4 lines
Instead of unconditionally emitting .globl's for the __start_set_xxx and
__stop_set_xxx symbols, only emit them when the set_vnet or set_pcpu
sections are actually defined.
------------------------------------------------------------------------
r215317 | dim | 2010-11-14 21:38:11 +0100 (Sun, 14 Nov 2010) | 3 lines
Apply the STATIC_VNET_DEFINE and STATIC_DPCPU_DEFINE macros throughout
the tree.
------------------------------------------------------------------------
r215316 | dim | 2010-11-14 21:23:02 +0100 (Sun, 14 Nov 2010) | 2 lines
Add macros to define static instances of VNET_DEFINE and DPCPU_DEFINE.
When a fast machine first brings up some non TCP networking program
it is quite possible that we will drop packets due to the fact that
only one packet can be held per ARP entry. This leads to packets
being missed when a program starts or restarts if the ARP data is
not currently in the ARP cache.
This code adds a new sysctl, net.link.ether.inet.maxhold, which defines
a system wide maximum number of packets to be held in each ARP entry.
Up to maxhold packets are queued until an ARP reply is received or
the ARP times out. The default setting is the old value of 1
which has been part of the BSD networking code since time
immemorial.
Expose the time we hold an incomplete ARP entry by adding
the sysctl net.link.ether.inet.wait, which defaults to 20
seconds, the value used when the new ARP code was added..
Reviewed by: bz, rpaulo
MFC after: 3 weeks
