A non-global IPv6 address can be used in more than one zone of the same
scope. This zone index is used to identify to which zone a non-global
address belongs.
Also we can have many foreign hosts with equal non-global addresses,
but from different zones. So, they can have different metrics in the
host cache.
Obtained from: Yandex LLC
Sponsored by: Yandex LLC
with no RSS hash.
When doing RSS:
* Create a new IPv4 netisr which expects the frames to have been verified;
it just directly dispatches to the IPv4 input path.
* Once IPv4 reassembly is done, re-calculate the RSS hash with the new
IP and L3 header; then reinject it as appropriate.
* Update the IPv4 netisr to be a CPU affinity netisr with the RSS hash
function (rss_soft_m2cpuid) - this will do a software hash if the
hardware doesn't provide one.
NICs that don't implement hardware RSS hashing will now benefit from RSS
distribution - it'll inject into the correct destination netisr.
Note: the netisr distribution doesn't work out of the box - netisr doesn't
query RSS for how many CPUs and the affinity setup. Yes, netisr likely
shouldn't really be doing CPU stuff anymore and should be "some kind of
'thing' that is a workqueue that may or may not have any CPU affinity";
that's for a later commit.
Differential Revision: https://reviews.freebsd.org/D527
Reviewed by: grehan
and egress.
* rss_mbuf_software_hash_v4 - look at the IPv4 mbuf to fetch the IPv4 details
+ direction to calculate a hash.
* rss_proto_software_hash_v4 - hash the given source/destination IPv4 address,
port and direction.
* rss_soft_m2cpuid - map the given mbuf to an RSS CPU ("bucket" for now)
These functions are intended to be used by the stack to support
the following:
* Not all NICs do RSS hashing, so we should support some way of doing
a hash in software;
* The NIC / driver may not hash frames the way we want (eg UDP 4-tuple
hashing when the stack is only doing 2-tuple hashing for UDP); so we
may need to re-hash frames;
* .. same with IPv4 fragments - they will need to be re-hashed after
reassembly;
* .. and same with things like IP tunneling and such;
* The transmit path for things like UDP, RAW and ICMP don't currently
have any RSS information attached to them - so they'll need an
RSS calculation performed before transmit.
TODO:
* Counters! Everywhere!
* Add a debug mode that software hashes received frames and compares them
to the hardware hash provided by the hardware to ensure they match.
The IPv6 part of this is missing - I'm going to do some re-juggling of
where various parts of the RSS framework live before I add the IPv6
code (read: the IPv6 code is going to go into netinet6/in6_rss.[ch],
rather than living here.)
Note: This API is still fluid. Please keep that in mind.
Differential Revision: https://reviews.freebsd.org/D527
Reviewed by: grehan
information as part of recvmsg().
This is primarily used for debugging/verification of the various
processing paths in the IP, PCB and driver layers.
Unfortunately the current implementation of the control message path
results in a ~10% or so drop in UDP frame throughput when it's used.
Differential Revision: https://reviews.freebsd.org/D527
Reviewed by: grehan
overriding an existing flowid/flowtype field in the outbound mbuf with
the inp_flowid/inp_flowtype details.
The upcoming RSS UDP support calculates a valid RSS value for outbound
mbufs and since it may change per send, it doesn't cache it in the inpcb.
So overriding it here would be wrong.
Differential Revision: https://reviews.freebsd.org/D527
Reviewed by: grehan
Kernel changes:
* Split kernel/userland nat structures eliminating IPFW_INTERNAL hack.
* Add IP_FW_NAT44_* codes resemblin old ones.
* Assume that instances can be named (no kernel support currently).
* Use both UH+WLOCK locks for all configuration changes.
* Provide full ABI support for old sockopts.
Userland changes:
* Use IP_FW_NAT44_* codes for nat operations.
* Remove undocumented ability to show ranges of nat "log" entries.
eliminiates some warnings when building in userland.
Thanks to Patrick Laimbock for reporting this issue.
Remove also some unnecessary casts.
There should be no functional change.
MFC after: 1 week
packets targeting a listening socket. Permit to reduce TCP input
processing starvation in context of high SYN load (e.g. short-lived TCP
connections or SYN flood).
Submitted by: Julien Charbon <jcharbon@verisign.com>
Reviewed by: adrian, hiren, jhb, Mike Bentkofsky
try to collapse adjacent pieces using m_catpkt(). In best case
scenario it copies data and frees mbufs, making mbuf exhaustion
attack harder.
Suggested by: Jonathan Looney <jonlooney gmail.com>
Security: Hardens against remote mbuf exhaustion attack.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
packets at all. Swapping byte order on SOCK_RAW was actually a bug, an
artifact from the BSD network stack, that used to convert a packet to
native byte order once it is received by kernel.
Other operating systems didn't follow this, and later other BSD
descendants fixed this, leaving us alone with the bug. Now it is
clear that we should fix the bug.
In collaboration with: Olivier Cochard-Labbé <olivier cochard.me>
See also: https://wiki.freebsd.org/SOCK_RAW
Sponsored by: Nginx, Inc.
This is the last major change in given branch.
Kernel changes:
* Use 64-bytes structures to hold multi-value variables.
* Use shared array to hold values from all tables (assume
each table algo is capable of holding 32-byte variables).
* Add some placeholders to support per-table value arrays in future.
* Use simple eventhandler-style API to ease the process of adding new
table items. Currently table addition may required multiple UH drops/
acquires which is quite tricky due to atomic table modificatio/swap
support, shared array resize, etc. Deal with it by calling special
notifier capable of rolling back state before actually performing
swap/resize operations. Original operation then restarts itself after
acquiring UH lock.
* Bump all objhash users default values to at least 64
* Fix custom hashing inside objhash.
Userland changes:
* Add support for dumping shared value array via "vlist" internal cmd.
* Some small print/fill_flags dixes to support u32 values.
* valtype is now bitmask of
<skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>.
New values can hold distinct values for each of this types.
* Provide special "legacy" type which assumes all values are the same.
* More helpers/docs following..
Some examples:
3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6
3:41 [1] zfscurr0# ipfw table mimimi info
+++ table(mimimi), set(0) +++
kindex: 2, type: addr
references: 0, valtype: skipto,limit,ipv4,ipv6
algorithm: addr:radix
items: 0, size: 296
3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1
added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1
3:42 [1] zfscurr0# ipfw table mimimi list
+++ table(mimimi), set(0) +++
10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
is found, the first usable address is returned for legacy ioctls like
SIOCGIFBRDADDR, SIOCGIFDSTADDR, SIOCGIFNETMASK and SIOCGIFADDR.
While there also fix a subtle issue that a caller from a jail asking for
INADDR_ANY may get the first IP of the host that do not belong to the jail.
Submitted by: glebius
Differential Revision: https://reviews.freebsd.org/D667
socket options. This includes managing the correspoing stat counters.
Add the SCTP_DETAILED_STR_STATS kernel option to control per policy
counters on every stream. The default is off and only an aggregated
counter is available. This is sufficient for the RTCWeb usecase.
MFC after: 1 week
Most of the tablearg-supported opcodes does not accept 0 as valid value:
O_TAG, O_TAGGED, O_PIPE, O_QUEUE, O_DIVERT, O_TEE, O_SKIPTO, O_CALLRET,
O_NETGRAPH, O_NGTEE, O_NAT treats 0 as invalid input.
The rest are O_SETDSCP and O_SETFIB.
'Fix' them by adding high-order bit (0x8000) set for non-tablearg values.
Do translation in kernel for old clients (import_rule0 / export_rule0),
teach current ipfw(8) binary to add/remove given bit.
This change does not affect handling SETDSCP values, but limit
O_SETFIB values to 32767 instead of 65k. Since currently we have either
old (16) or new (2^32) max fibs, this should not be a big deal:
we're definitely OK for former and have to add another opcode to deal
with latter, regardless of tablearg value.
* Since there seems to be lack of consensus on strict value typing,
remove non-default value types. Use userland-only "value format type"
to print values.
Kernel changes:
* Add IP_FW_XMODIFY to permit table run-time modifications.
Currently we support changing limit and value format type.
Userland changes:
* Support IP_FW_XMODIFY opcode.
* Support specifying value format type (ftype) in tablble create/modify req
* Fine-print value type/value format type.
* Implement proper checks for switching between global and set-aware tables
* Split IP_FW_DEL mess into the following opcodes:
* IP_FW_XDEL (del rules matching pattern)
* IP_FW_XMOVE (move rules matching pattern to another set)
* IP_FW_SET_SWAP (swap between 2 sets)
* IP_FW_SET_MOVE (move one set to another one)
* IP_FW_SET_ENABLE (enable/disable sets)
* Add IP_FW_XZERO / IP_FW_XRESETLOG to finish IP_FW3 migration.
* Use unified ipfw_range_tlv as range description for all of the above.
* Check dynamic states IFF there was non-zero number of deleted dyn rules,
* Del relevant dynamic states with singe traversal instead of per-rule one.
Userland changes:
* Switch ipfw(8) to use new opcodes.
Kernel changes:
* Add opcode IP_FW_TABLE_XSWAP
* Add support for swapping 2 tables with the same type/ftype/vtype.
* Make skipto cache init after ipfw locks init.
Userland changes:
* Add "table X swap Y" command.
NRSACK extension. The default will still be off, since it
it not an RFC (yet).
Changing the sysctl name will be in a separate commit.
MFC after: 1 week
option for controlling ECN on future associations and get the
status on current associations.
A simialar pattern will be used for controlling SCTP extensions in
upcoming commits.
The rss_key[] array in netinet/in_rss.c has the bytes in incorrect
order. This results in the RSS test vectors in the Microsft RSS spec
and Intel NIC specs giving incorrect results, and making it difficult
to verify correct hash operation when RSS functionality is added to
new NICs.
CR: https://phabric.freebsd.org/D516
Reviewed by: adrian
Kernel changes:
* Add TEI_FLAGS_DONTADD entry flag to indicate that insert is not possible
* Support given flag in all algorithms
* Add "limit" field to ipfw_xtable_info
* Add actual limiting code into add_table_entry()
Userland changes:
* Add "limit" option as "create" table sub-option. Limit modification
is currently impossible.
* Print human-readable errors in table enry addition/deletion code.
* Add "flow:hash" algorithm
Kernel changes:
* Add O_IP_FLOW_LOOKUP opcode to support "flow" lookups
* Add IPFW_TABLE_FLOW table type
* Add "struct tflow_entry" as strage for 6-tuple flows
* Add "flow:hash" algorithm. Basically it is auto-growing chained hash table.
Additionally, we store mask of fields we need to compare in each instance/
* Increase ipfw_obj_tentry size by adding struct tflow_entry
* Add per-algorithm stat (ifpw_ta_tinfo) to ipfw_xtable_info
* Increase algoname length: 32 -> 64 (algo options passed there as string)
* Assume every table type can be customized by flags, use u8 to store "tflags" field.
* Simplify ipfw_find_table_entry() by providing @tentry directly to algo callback.
* Fix bug in cidr:chash resize procedure.
Userland changes:
* add "flow table(NAME)" syntax to support n-tuple checking tables.
* make fill_flags() separate function to ease working with _s_x arrays
* change "table info" output to reflect longer "type" fields
Syntax:
ipfw table fl2 create type flow:[src-ip][,proto][,src-port][,dst-ip][dst-port] [algo flow:hash]
Examples:
0:02 [2] zfscurr0# ipfw table fl2 create type flow:src-ip,proto,dst-port algo flow:hash
0:02 [2] zfscurr0# ipfw table fl2 info
+++ table(fl2), set(0) +++
kindex: 0, type: flow:src-ip,proto,dst-port
valtype: number, references: 0
algorithm: flow:hash
items: 0, size: 280
0:02 [2] zfscurr0# ipfw table fl2 add 2a02:6b8::333,tcp,443 45000
0:02 [2] zfscurr0# ipfw table fl2 add 10.0.0.92,tcp,80 22000
0:02 [2] zfscurr0# ipfw table fl2 list
+++ table(fl2), set(0) +++
2a02:6b8::333,6,443 45000
10.0.0.92,6,80 22000
0:02 [2] zfscurr0# ipfw add 200 count tcp from me to 78.46.89.105 80 flow 'table(fl2)'
00200 count tcp from me to 78.46.89.105 dst-port 80 flow table(fl2)
0:03 [2] zfscurr0# ipfw show
00200 0 0 count tcp from me to 78.46.89.105 dst-port 80 flow table(fl2)
65535 617 59416 allow ip from any to any
0:03 [2] zfscurr0# telnet -s 10.0.0.92 78.46.89.105 80
Trying 78.46.89.105...
..
0:04 [2] zfscurr0# ipfw show
00200 5 272 count tcp from me to 78.46.89.105 dst-port 80 flow table(fl2)
65535 682 66733 allow ip from any to any
Previously there was a race condition between the address addition
and associating it with the CARP which resulted in the interface
MAC, instead of the CARP MAC, being used for a brief amount of time.
This caused "is using my IP address" warnings as well as data being
sent to the wrong machine due to incorrect ARP entries being recorded
by other devices on the network.
Kernel changes:
* s/IPFW_TABLE_U32/IPFW_TABLE_NUMBER/
* Force "lookup <port|uid|gid|jid>" to be IPFW_TABLE_NUMBER
* Support "lookup" method for number tables
* Add number:array algorihm (i32 as key, auto-growing).
Userland changes:
* Support named tables in "lookup <tag> Table"
* Fix handling of "table(NAME,val)" case
* Support printing "number" table data.
* Rewrite interface tables to use interface indexes
Kernel changes:
* Add generic interface tracking API:
- ipfw_iface_ref (must call unlocked, performs lazy init if needed, allocates
state & bumps ref)
- ipfw_iface_add_ntfy(UH_WLOCK+WLOCK, links comsumer & runs its callback to
update ifindex)
- ipfw_iface_del_ntfy(UH_WLOCK+WLOCK, unlinks consumer)
- ipfw_iface_unref(unlocked, drops reference)
Additionally, consumer callbacks are called in interface withdrawal/departure.
* Rewrite interface tables to use iface tracking API. Currently tables are
implemented the following way:
runtime data is stored as sorted array of {ifidx, val} for existing interfaces
full data is stored inside namedobj instance (chained hashed table).
* Add IP_FW_XIFLIST opcode to dump status of tracked interfaces
* Pass @chain ptr to most non-locked algorithm callbacks:
(prepare_add, prepare_del, flush_entry ..). This may be needed for better
interaction of given algorithm an other ipfw subsystems
* Add optional "change_ti" algorithm handler to permit updating of
cached table_info pointer (happens in case of table_max resize)
* Fix small bug in ipfw_list_tables()
* Add badd (insert into sorted array) and bdel (remove from sorted array) funcs
Userland changes:
* Add "iflist" cmd to print status of currently tracked interface
* Add stringnum_cmp for better interface/table names sorting
so it really should not be under "optional inet". The fact that uipc_accf.c
lives under kern/ lends some weight to making it a "standard" file.
Moving kern/uipc_accf.c from "optional inet" to "standard" eliminates the
need for #ifdef INET in kern/uipc_socket.c.
Also, this meant the net.inet.accf.unloadable sysctl needed to move, as
net.inet does not exist without networking compiled in (as it lives in
netinet/in_proto.c.) The new sysctl has been named net.accf.unloadable.
In order to support existing accept filter sysctls, the net.inet.accf node
has been added netinet/in_proto.c.
Submitted by: Steve Kiernan <stevek@juniper.net>
Obtained from: Juniper Networks, Inc.
PF_LINK, and multicast/broadcast flag should always be dropped because
the outer protocol uses unicast even when the inner address is not for
unicast. It had been broken since r236951 when gif_output() started to
use IFQ_HANDOFF().
by the stack.
Right now the stack isn't really setup for RSS with 4-tuple UDP hashing
for either IPv4 and IPv6.
The specifics:
* The UDP init path udp_init() and udplite_init() specify the hash as
2-tuple, so the PCBGROUPS code only tries a 2-tuple check;
* The PCBGROUPS and RSS code doesn't know about the UDP hash types
just yet, so they're never treated as valid hashes.
* For correctness, 4-tuple can't be enabled in the general case because
UDP datagrams can be more fragmented than IP datagrams may be.
Strictly speaking, TCP datagrams may also be fragmented and this could
cause issues with PCBGROUPS/RSS until the IP defragment path grows some
code to re-calculate the RSS hash.
I'll follow this commit up with awareness of the UDP 4-tuple for those
who wish to configure it, but for now it'll stay disabled.
No drivers (yet) know to use this function when RSS is enabled.
markedly better distribution of IPv6 address/ports than the previous key.
The previous key would hash large swaths of the port space for a given
source/destination IP address to the same low handful of bits, effectively
mapping them to the same queue. This made testing very .. special.
a source address was selected and cached, but it was not
stored that is was cached. This resulted in selecting
different source addresses for the INIT-ACK and COOKIE-ACK
when possible.
Thanks to Niu Zhixiong for reporting the issue.
MFC after: 1 week
awareness.
* Introduce IP_BINDMULTI - indicating that it's okay to bind multiple
sockets on the same bind details.
Although the PCB code has been taught about this (see below) this patch
doesn't introduce the rest of the PCB changes necessary to distribute
lookups among multiple PCB entries in the global wildcard table.
* Introduce IP_RSS_LISTEN_BUCKET - placing an listen socket into the
given RSS bucket (and thus a single PCBGROUP hash.)
* Modify the PCB add path to be aware of IP_BINDMULTI:
+ Only allow further PCB entries to be added if the owner credentials
and IP_BINDMULTI has been specified. Ie, only allow further
IP_BINDMULTI sockets to appear if the first bind() was IP_BINDMULTI.
* Teach the PCBGROUP code about IP_RSS_LISTE_BUCKET marked PCB entries.
Instead of using the wildcard logic and hashing, these sockets are
simply placed into the PCBGROUP and _not_ in the wildcard hash.
* When doing a PCBGROUP lookup, also do a wildcard match as well.
This allows for an RSS bucket PCB entry to appear in a PCBGROUP
rather than having to exist in the wildcard list.
Tested:
* TCP IPv4 server testing with igb(4)
* TCP IPv4 server testing with ix(4)
TODO:
* The pcbgroup lookup code duplicated the wildcard and wildcard-PCB
logic. This could be refactored into a single function.
* This doesn't yet work for IPv6 (The PCBGROUP code in netinet6/ doesn't
yet know about this); nor does it yet fully work for UDP.
* Switch kernel to use per-cpu counters for rules.
* Keep ABI/API.
Kernel changes:
* Each rules is now exported as TLV with optional extenable
counter block (ip_fW_bcounter for base one) and
ip_fw_rule for rule&cmd data.
* Counters needs to be explicitly requested by IPFW_CFG_GET_COUNTERS flag.
* Separate counters from rules in kernel and clean up ip_fw a bit.
* Pack each rule in IPFW_TLV_RULE_ENT tlv to ease parsing.
* Introduce versioning in container TLV (may be needed in future).
* Fix ipfw_cfg_lheader broken u64 alignment.
Userland changes:
* Use set_mask from cfg header when requesting config
* Fix incorrect read accouting in ipfw_show_config()
* Use IPFW_RULE_NOOPT flag instead of playing with _pad
* Fix "ipfw -d list": do not print counters for dynamic states
* Some small fixes
Kernel changes:
* Change dump format for dynamic states:
each state is now stored inside ipfw_obj_dyntlv
last dynamic state is indicated by IPFW_DF_LAST flag
* Do not perform sooptcopyout() for !SOPT_GET requests.
Userland changes:
* Introduce foreach_state() function handler to ease work
with different states passed by ipfw_dump_config().
* Bump table dump format preserving old ABI.
Kernel size:
* Add IP_FW_TABLE_XFIND to handle "lookup" request from userland.
* Add ta_find_tentry() algorithm callbacks/handlers to support lookups.
* Fully switch to ipfw_obj_tentry for various table dumps:
algorithms are now required to support the latest (ipfw_obj_tentry) entry
dump format, the rest is handled by generic dump code.
IP_FW_TABLE_XLIST opcode version bumped (0 -> 1).
* Eliminate legacy ta_dump_entry algo handler:
dump_table_entry() converts data from current to legacy format.
Userland side:
* Add "lookup" table parameter.
* Change the way table type is guessed: call table_get_info() first,
and check value for IPv4/IPv6 type IFF table does not exist.
* Fix table_get_list(): do more tries if supplied buffer is not enough.
* Sparate table_show_entry() from table_show_list().
Kernel changes:
* Introduce ipfw_obj_tentry table entry structure to force u64 alignment.
* Support "update-on-existing-key" "add" bahavior (TEI_FLAGS_UPDATED).
* Use "subtype" field to distingush between IPv4 and IPv6 table records
instead of previous hack.
* Add value type (vtype) field for kernel tables. Current types are
number,ip and dscp
* Fix sets mask retrieval for old binaries
* Fix crash while using interface tables
Userland changes:
* Switch ipfw_table_handler() to use named-only tables.
* Add "table NAME create [type {cidr|iface|u32} [valtype {number|ip|dscp}] ..."
* Switch ipfw_table_handler to match_token()-based parser.
* Switch ipfw_sets_handler to use new ipfw_get_config() for mask retrieval.
* Allow ipfw set X table ... syntax to permit using per-set table namespaces.
Kernel changes:
* change base TLV header to be u64 (so size can be u32).
* Introduce ipfw_obj_ctlv generc container TLV.
* Add IP_FW_XGET opcode which is now used for atomic configuration
retrieval. One can specify needed configuration pieces to retrieve
via flags field. Currently supported are
IPFW_CFG_GET_STATIC (static rules) and
IPFW_CFG_GET_STATES (dynamic states).
Other configuration pieces (tables, pipes, etc..) support is planned.
Userland changes:
* Switch ipfw(8) to use new IP_FW_XGET for rule listing.
* Split rule listing code get and show pieces.
* Make several steps forward towards libipfw:
permit printing states and rules(paritally) to supplied buffer.
do not die on malloc/kernel failure inside given printing functions.
stop assuming cmdline_opts is global symbol.
These changes prevent sysctl(8) from returning proper output,
such as:
1) no output from sysctl(8)
2) erroneously returning ENOMEM with tools like truss(1)
or uname(1)
truss: can not get etype: Cannot allocate memory
there is an environment variable which shall initialize the SYSCTL
during early boot. This works for all SYSCTL types both statically and
dynamically created ones, except for the SYSCTL NODE type and SYSCTLs
which belong to VNETs. A new flag, CTLFLAG_NOFETCH, has been added to
be used in the case a tunable sysctl has a custom initialisation
function allowing the sysctl to still be marked as a tunable. The
kernel SYSCTL API is mostly the same, with a few exceptions for some
special operations like iterating childrens of a static/extern SYSCTL
node. This operation should probably be made into a factored out
common macro, hence some device drivers use this. The reason for
changing the SYSCTL API was the need for a SYSCTL parent OID pointer
and not only the SYSCTL parent OID list pointer in order to quickly
generate the sysctl path. The motivation behind this patch is to avoid
parameter loading cludges inside the OFED driver subsystem. Instead of
adding special code to the OFED driver subsystem to post-load tunables
into dynamically created sysctls, we generalize this in the kernel.
Other changes:
- Corrected a possibly incorrect sysctl name from "hw.cbb.intr_mask"
to "hw.pcic.intr_mask".
- Removed redundant TUNABLE statements throughout the kernel.
- Some minor code rewrites in connection to removing not needed
TUNABLE statements.
- Added a missing SYSCTL_DECL().
- Wrapped two very long lines.
- Avoid malloc()/free() inside sysctl string handling, in case it is
called to initialize a sysctl from a tunable, hence malloc()/free() is
not ready when sysctls from the sysctl dataset are registered.
- Bumped FreeBSD version to indicate SYSCTL API change.
MFC after: 2 weeks
Sponsored by: Mellanox Technologies
map the bucket to an RSS queue, then map the queue to a CPU ID.
This way the bucket->queue and queue->CPU mapping can change
over time.
Introduce IP_RSSBUCKETID - which instead looks up the RSS bucket.
User applications can then map the RSS bucket to a CPU.
There's 128 indirection table entries which correspond to the
low 7 bits of the 32 bit RSS hash. Each value will correspond
to an RSS bucket. (Then each RSS bucket currently will map
to a CPU.)
This is a more explicit way of figuring out which RSS bucket
is in each RSS indirection slot. It can be inferred by the other
methods but I'd rather drivers use something more simplified and
explicit.
reporting IP-addresses to the peer during the handshake, adding
addresses to the host, reporting the addresses via the sysctl
interface (used by netstat, for example) and reporting the
addresses to the application via socket options.
This issue was reported by Bernd Walter.
MFC after: 3 days
* Add 'algoname' string to ipfw_xtable_info permitting to specify lookup
algoritm with parameters.
* Rework part of ipfw_rewrite_table_uidx()
Sponsored by: Yandex LLC
* Use one u16 from op3 header to implement opcode versioning.
* IP_FW_TABLE_XLIST has now 2 handlers, for ver.0 (old) and ver.1 (current).
* Every getsockopt request is now handled in ip_fw_table.c
* Rename new opcodes:
IP_FW_OBJ_DEL -> IP_FW_TABLE_XDESTROY
IP_FW_OBJ_LISTSIZE -> IP_FW_TABLES_XGETSIZE
IP_FW_OBJ_LIST -> IP_FW_TABLES_XLIST
IP_FW_OBJ_INFO -> IP_FW_TABLE_XINFO
IP_FW_OBJ_INFO -> IP_FW_TABLE_XFLUSH
* Add some docs about using given opcodes.
* Group some legacy opcode/handlers.
Kernel changes:
* Add IP_FW_OBJ_FLUSH opcode (flush table based on its name/set)
* Add IP_FW_OBJ_DUMP opcode (dumps table data based on its names/set)
* Add IP_FW_OBJ_LISTSIZE / IP_FW_OBJ_LIST opcodes (get list of kernel tables)
Userland changes:
* move tables code to separate tables.c file
* get rid of tables_max
* switch "all"/list handling to new opcodes
Kernel-side changelog:
* Split general tables code and algorithm-specific table data.
Current algorithms (IPv4/IPv6 radix and interface tables radix) moved to
new ip_fw_table_algo.c file.
Tables code now supports any algorithm implementing the following callbacks:
+struct table_algo {
+ char name[64];
+ int idx;
+ ta_init *init;
+ ta_destroy *destroy;
+ table_lookup_t *lookup;
+ ta_prepare_add *prepare_add;
+ ta_prepare_del *prepare_del;
+ ta_add *add;
+ ta_del *del;
+ ta_flush_entry *flush_entry;
+ ta_foreach *foreach;
+ ta_dump_entry *dump_entry;
+ ta_dump_xentry *dump_xentry;
+};
* Change ->state, ->xstate, ->tabletype fields of ip_fw_chain to
->tablestate pointer (array of 32 bytes structures necessary for
runtime lookups (can be probably shrinked to 16 bytes later):
+struct table_info {
+ table_lookup_t *lookup; /* Lookup function */
+ void *state; /* Lookup radix/other structure */
+ void *xstate; /* eXtended state */
+ u_long data; /* Hints for given func */
+};
* Add count method for namedobj instance to ease size calculations
* Bump ip_fw3 buffer in ipfw_clt 128->256 bytes.
* Improve bitmask resizing on tables_max change.
* Remove table numbers checking from most places.
* Fix wrong nesting in ipfw_rewrite_table_uidx().
* Add IP_FW_OBJ_LIST opcode (list all objects of given type, currently
implemented for IPFW_OBJTYPE_TABLE).
* Add IP_FW_OBJ_LISTSIZE (get buffer size to hold IP_FW_OBJ_LIST data,
currenly implemented for IPFW_OBJTYPE_TABLE).
* Add IP_FW_OBJ_INFO (requests info for one object of given type).
Some name changes:
s/ipfw_xtable_tlv/ipfw_obj_tlv/ (no table specifics)
s/ipfw_xtable_ntlv/ipfw_obj_ntlv/ (no table specifics)
Userland changes:
* Add do_set3() cmd to ipfw2 to ease dealing with op3-embeded opcodes.
* Add/improve support for destroy/info cmds.
* Add namedobject set-aware api capable of searching/allocation objects by their name/idx.
* Switch tables code to use string ids for configuration tasks.
* Change locking model: most configuration changes are protected with UH lock, runtime-visible are protected with both locks.
* Reduce number of arguments passed to ipfw_table_add/del by using separate structure.
* Add internal V_fw_tables_sets tunable (set to 0) to prepare for set-aware tables (requires opcodes/client support)
* Implement typed table referencing (and tables are implicitly allocated with all state like radix ptrs on reference)
* Add "destroy" ipfw(8) using new IP_FW_DELOBJ opcode
Namedobj more detailed:
* Blackbox api providing methods to add/del/search/enumerate objects
* Statically-sized hashes for names/indexes
* Per-set bitmask to indicate free indexes
* Separate methods for index alloc/delete/resize
Basically, there should not be any user-visible changes except the following:
* reducing table_max is not supported
* flush & add change table type won't work if table is referenced
Sponsored by: Yandex LLC
ifa_ifwithnet() and ifa_ifwithdstaddr() The legacy functions will call the
_fib() versions with RT_ALL_FIBS, preserving legacy behavior.
sys/net/if_var.h
sys/net/if.c
Add legacy-compatible functions as described above. Ensure legacy
behavior when RT_ALL_FIBS is passed as fibnum.
sys/netinet/in_pcb.c
sys/netinet/ip_output.c
sys/netinet/ip_options.c
sys/net/route.c
sys/net/rtsock.c
sys/netinet6/nd6.c
Call with _fib() functions if we must use a specific fib, or the
legacy functions otherwise.
tests/sys/netinet/fibs_test.sh
tests/sys/netinet/udp_dontroute.c
Improve the udp_dontroute test. The bug that this test exercises is
that ifa_ifwithnet() will return the wrong address, if multiple
interfaces have addresses on the same subnet but with different
fibs. The previous version of the test only considered one possible
failure mode: that ifa_ifwithnet_fib() might fail to find any
suitable address at all. The new version also checks whether
ifa_ifwithnet_fib() finds the correct address by checking where the
ARP request goes.
Reported by: bz, hrs
Reviewed by: hrs
MFC after: 1 week
X-MFC-with: 264905
Sponsored by: Spectra Logic
mode.
Put the htonl(), htons(), ntohl() and ntohs() declarations under
__POSIX_VISIBLE >= 200112. POSIX.1-2001 and newer require these to be
exposed from <netinet/in.h> (as well as <arpa/inet.h>).
Note that it may be unnecessary to check __POSIX_VISIBLE >= 200112 because
older versions of POSIX and the C standard do not define this header.
However, other places in the same file already perform the check.
PR: 188316
Submitted by: Christian Neukirchen
mappings. Instead, they should be first mapping to an RSS bucket and
then querying the RSS bucket -> CPU ID mapping to figure out the target
CPU.
When (if?) RSS rebalancing is implemented or some other (non round-robin)
distribution of work from buckets to CPU IDs, various bits of code - both
userland and kernel - will need to know how this mapping works.
So, to support this:
* Add a new function rss_m2bucket() - this maps an mbuf to a given bucket.
Anything which is currently doing hash -> CPU work may instead wish to
do hash -> bucket, and then query the bucket->cpuid map for which
CPU it belongs on. Or, map it to a bucket, then re-pin that bucket ->
CPU during a rebalance operation.
* For userland applications which wish to exploit affinity to RSS buckets,
the bucket -> CPU ID mapping is now available via a sysctl.
net.inet.rss.bucket_mapping lists the bucket to CPU ID mapping via
a list of bucket:cpu pairs.
lookup for the inp flowid/flowtype to destination CPU.
This only modifies the case where RSS is enabled and the per-cpu tcp
timer option is enabled. Otherwise the behaviour should be the same
as before.
This is intended to be used by various places that wish to hash some
information about a TCP/UDP/IP flow but don't necessarily have a
live mbuf to do it with.
Refactor rss_m2cpuid() to use the refactored function.
Add `flags` u16 field to the hole in ipfw_table_xentry structure.
Kernel has been guessing address family for supplied record based
on xent length size.
Userland, however, has been getting fixed-size ipfw_table_xentry structures
guessing address family by checking address by IN6_IS_ADDR_V4COMPAT().
Fix this behavior by providing specific IPFW_TCF_INET flag for IPv4 records.
PR: bin/189471
Submitted by: Dennis Yusupoff <dyr@smartspb.net>
MFC after: 2 weeks
near-term future use.
These are intended to fetch the current flow id, flow hash type
(M_HASHTYPE_* from the sys/mbuf.h) and if RSS is enabled, the
RSS destined CPU ID for the receive path.
eight years. The original concept was to improve the
corner case where you run out of ephemeral ports, but it
was causing performance problems and the mechanism
of limiting the number of time_wait sockets serves
the same purpose in the end.
Reviewed by: bz
same events that tcpstat's tcps_rcvmemdrop counter counts.
- Rename tcps_rcvmemdrop to tcps_rcvreassfull and improve its
description in netstat(1) output.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
mixing on stack memory and UMA memory in one linked list.
Thus, rewrite TCP reassembly code in terms of memory usage. The
algorithm remains unchanged.
We actually do not need extra memory to build a reassembly queue.
Arriving mbufs are always packet header mbufs. So we got the length
of data as pkthdr.len. We got m_nextpkt for linkage. And we need
only one pointer to point at the tcphdr, use PH_loc for that.
In tcpcb the t_segq fields becomes mbuf pointer. The t_segqlen
field now counts not packets, but bytes in the queue. This gives
us more precision when comparing to socket buffer limits.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
exists on another interface. The panic was introduced by change 264887, which
changed the fibnum parameter in the call to rtalloc1_fib() in
ifa_switch_loopback_route() from RT_DEFAULT_FIB to RT_ALL_FIBS. The solution
is to use the interface fib in that call. For the majority of users, that will
be equivalent to the legacy behavior.
PR: kern/189089
Reported by: neel
Reviewed by: neel
MFC after: 3 weeks
X-MFC with: 264887
Sponsored by: Spectra Logic
These two bugs are closely related. The root cause is that ifa_ifwithnet
does not consider FIBs when searching for an interface address.
sys/net/if_var.h
sys/net/if.c
Add a fib argument to ifa_ifwithnet and ifa_ifwithdstadddr. Those
functions will only return an address whose interface fib equals the
argument.
sys/net/route.c
Update calls to ifa_ifwithnet and ifa_ifwithdstaddr with fib
arguments.
sys/netinet/in.c
Update in_addprefix to consider the interface fib when adding
prefixes. This will prevent it from not adding a subnet route when
one already exists on a different fib.
sys/net/rtsock.c
sys/netinet/in_pcb.c
sys/netinet/ip_output.c
sys/netinet/ip_options.c
sys/netinet6/nd6.c
Add RT_DEFAULT_FIB arguments to ifa_ifwithdstaddr and ifa_ifwithnet.
In some cases it there wasn't a clear specific fib number to use.
In others, I was unable to test those functions so I chose
RT_DEFAULT_FIB to minimize divergence from current behavior. I will
fix some of the latter changes along with PR kern/187553.
tests/sys/netinet/fibs_test.sh
tests/sys/netinet/udp_dontroute.c
tests/sys/netinet/Makefile
Revert r263738. The udp_dontroute test was right all along.
However, bugs kern/187550 and kern/187553 cancelled each other out
when it came to this test. Because of kern/187553, ifa_ifwithnet
searched the default fib instead of the requested one, but because
of kern/187550, there was an applicable subnet route on the default
fib. The new test added in r263738 doesn't work right, however. I
can verify with dtrace that ifa_ifwithnet returned the wrong address
before I applied this commit, but route(8) miraculously found the
correct interface to use anyway. I don't know how.
Clear expected failure messages for kern/187550 and kern/187552.
PR: kern/187550
PR: kern/187552
Reviewed by: melifaro
MFC after: 3 weeks
Sponsored by: Spectra Logic
sys/net/route.c
In rtinit1, use the interface fib instead of the process fib. The
latter wasn't very useful because ifconfig(8) is usually invoked
with the default process fib. Changing ifconfig(8) to use setfib(2)
would be redundant, because it already sets the interface fib.
tests/sys/netinet/fibs_test.sh
Clear the expected ATF failure
sys/net/if.c
Pass the interface fib in calls to rtrequest1_fib and rtalloc1_fib
sys/netinet/in.c
sys/net/if_var.h
Add a fibnum argument to ifa_switch_loopback_route, a subroutine of
in_scrubprefix. Pass it the interface fib.
PR: kern/187549
Reviewed by: melifaro
MFC after: 3 weeks
Sponsored by: Spectra Logic Corporation
walks the list of connections in TIME_WAIT closing expired connections
due to contention on the global TCP pcbinfo lock.
To remediate, introduce a new global lock to protect the list of
connections in TIME_WAIT. Only acquire the TCP pcbinfo lock when
closing an expired connection. This limits the window of time when
TCP input processing is stopped to the amount of time needed to close
a single connection.
Submitted by: Julien Charbon <jcharbon@verisign.com>
Reviewed by: rwatson, rrs, adrian
MFC after: 2 months
restricted to a single FIB in a multifib system.
Restricting an interface's routes to the FIB to which it is assigned (by
setting net.add_addr_allfibs=0) causes ARP updates to fail with "arpresolve:
can't allocate llinfo for x.x.x.x". This is due to the ARP update code hard
coding it's lookup for existing routing entries to FIB 0.
sys/netinet/in.c:
When dealing with RTM_ADD (add route) requests for an interface, use
the interface's assigned FIB instead of the default (FIB 0).
sys/netinet/if_ether.c:
In arpresolve(), enhance error message generated when an
lla_lookup() fails so that the interface causing the error is
visible in logs.
tests/sys/netinet/fibs_test.sh
Clear ATF expected error.
PR: kern/167947
Submitted by: Nikolay Denev <ndenev@gmail.com> (previous version)
Reviewed by: melifaro
MFC after: 3 weeks
Sponsored by: Spectra Logic Corporation
proprietary binary format.
* Add support for a diagnostic information error cause.
The code is sysctlable and the default is 0, which
means it is not sent.
This is joint work with rrs@.
MFC after: 1 week
linking NIC Receive Side Scaling (RSS) to the network stack's
connection-group implementation. This prototype (and derived patches)
are in use at Juniper and several other FreeBSD-using companies, so
despite some reservations about its maturity, merge the patch to the
base tree so that it can be iteratively refined in collaboration rather
than maintained as a set of gradually diverging patch sets.
(1) Merge a software implementation of the Toeplitz hash specified in
RSS implemented by David Malone. This is used to allow suitable
pcbgroup placement of connections before the first packet is
received from the NIC. Software hashing is generally avoided,
however, due to high cost of the hash on general-purpose CPUs.
(2) In in_rss.c, maintain authoritative versions of RSS state intended
to be pushed to each NIC, including keying material, hash
algorithm/ configuration, and buckets. Provide software-facing
interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both
the RSS standardised Toeplitz and a 'naive' variation with a hash
efficient in software but with poor distribution properties.
Implement rss_m2cpuid()to be used by netisr and other load
balancing code to look up the CPU on which an mbuf should be
processed.
(3) In the Ethernet link layer, allow netisr distribution using RSS as
a source of policy as an alternative to source ordering; continue
to default to direct dispatch (i.e., don't try and requeue packets
for processing on the 'right' CPU if they arrive in a directly
dispatchable context).
(4) Allow RSS to control tuning of connection groups in order to align
groups with RSS buckets. If a packet arrives on a protocol using
connection groups, and contains a suitable hardware-generated
hash, use that hash value to select the connection group for pcb
lookup for both IPv4 and IPv6. If no hardware-generated Toeplitz
hash is available, we fall back on regular PCB lookup risking
contention rather than pay the cost of Toeplitz in software --
this is a less scalable but, at my last measurement, faster
approach. As core counts go up, we may want to revise this
strategy despite CPU overhead.
Where device drivers suitably configure NICs, and connection groups /
RSS are enabled, this should avoid both lock and line contention during
connection lookup for TCP. This commit does not modify any device
drivers to tune device RSS configuration to the global RSS
configuration; patches are in circulation to do this for at least
Chelsio T3 and Intel 1G/10G drivers. Currently, the KPI for device
drivers is not particularly robust, nor aware of more advanced features
such as runtime reconfiguration/rebalancing. This will hopefully prove
a useful starting point for refinement.
No MFC is scheduled as we will first want to nail down a more mature
and maintainable KPI/KBI for device drivers.
Sponsored by: Juniper Networks (original work)
Sponsored by: EMC/Isilon (patch update and merge)
AppleTalk was a network transport protocol for Apple Macintosh devices
in 80s and then 90s. Starting with Mac OS X in 2000 the AppleTalk was
a legacy protocol and primary networking protocol is TCP/IP. The last
Mac OS X release to support AppleTalk happened in 2009. The same year
routing equipment vendors (namely Cisco) end their support.
Thus, AppleTalk won't be supported in FreeBSD 11.0-RELEASE.
IPX was a network transport protocol in Novell's NetWare network operating
system from late 80s and then 90s. The NetWare itself switched to TCP/IP
as default transport in 1998. Later, in this century the Novell Open
Enterprise Server became successor of Novell NetWare. The last release
that claimed to still support IPX was OES 2 in 2007. Routing equipment
vendors (e.g. Cisco) discontinued support for IPX in 2011.
Thus, IPX won't be supported in FreeBSD 11.0-RELEASE.
The virtio driver needs the offset to be stored in csum_data,
like in the case for UDP and TCP.
The virtio problem was reported by
Niu Zhixiong <kaiaixi@gmail.com>, who helped in debugging
and testing the patch.
MFC after: 3 days
the protocol specific mbuf flags are shared between them.
- Move all M_FOO definitions into a single place: netinet/in6.h, to
avoid future clashes.
- Resolve clash between M_DECRYPTED and M_SKIP_FIREWALL which resulted
in a failure of operation of IPSEC and packet filters.
Thanks to Nicolas and Georgios for all the hard work on bisecting,
testing and finally finding the root of the problem.
PR: kern/186755
PR: kern/185876
In collaboration with: Georgios Amanakis <gamanakis gmail.com>
In collaboration with: Nicolas DEFFAYET <nicolas-ml deffayet.com>
Sponsored by: Nginx, Inc.
- Use counter(9) for rt_pksent (former rt_rmx.rmx_pksent). This
removes another cache trashing ++ from packet forwarding path.
- Create zini/fini methods for the rtentry UMA zone. Via initialize
mutex and counter in them.
- Fix reporting of rmx_pksent to routing socket.
- Fix netstat(1) to report "Use" both in kvm(3) and sysctl(3) mode.
The change is mostly targeted for stable/10 merge. For head,
rt_pksent is expected to just disappear.
Discussed with: melifaro
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
The ifaddr is already referenced by the rtentry, and we are holding
reference on the rtentry throughout the function execution.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
were primarily used to size the sysctl name list macros that were removed
in r254295. A few other constants either did not have an associated
sysctl node, or the associated node used OID_AUTO instead.
PR: ports/184525 (exp-run)
tested and is unfinished. However, I've tested my version,
it works okay. As before it is unfinished: timeout aren't
driven by TCP session state. To enable the HASH_ALL mode,
one needs in kernel config:
options FLOWTABLE_HASH_ALL
o Reduce the alignment on flentry to 64 bytes. Without
the FLOWTABLE_HASH_ALL option, twice less memory would
be consumed by flows.
o API to ip_output()/ip6_output() got even more thin: 1 liner.
o Remove unused unions. Simply use fle->f_key[].
o Merge all IPv4 code into flowtable_lookup_ipv4(), and do same
flowtable_lookup_ipv6(). Stop copying data to on stack
sockaddr structures, simply use key[] on stack.
o Move code from flowtable_lookup_common() that actually works
on insertion into flowtable_insert().
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
- ip_output() and ip_output6() simply call flowtable_lookup(),
passing mbuf and address family. That's the only code under
#ifdef FLOWTABLE in the protocols code now.
o Revamp statistics gathering and export.
- Remove hand made pcpu stats, and utilize counter(9).
- Snapshot of statistics is available via 'netstat -rs'.
- All sysctls are moved into net.flowtable namespace, since
spreading them over net.inet isn't correct.
o Properly separate at compile time INET and INET6 parts.
o General cleanup.
- Remove chain of multiple flowtables. We simply have one for
IPv4 and one for IPv6.
- Flowtables are allocated in flowtable.c, symbols are static.
- With proper argument to SYSINIT() we no longer need flowtable_ready.
- Hash salt doesn't need to be per-VNET.
- Removed rudimentary debugging, which use quite useless in dtrace era.
The runtime behavior of flowtable shouldn't be changed by this commit.
Sponsored by: Netflix
Sponsored by: Nginx, Inc.
By definition, the very first FIN is not a duplicate. Process it normally
and don't feed it to congestion control as though it were a dupe. Don't
prevent CC from seeing later dupe acks while in a half close state.
the INP_INFO lock from tcp_usr_accept. As the PR/patch states
this was following the advice already in the code.
See the PR below for a full disucssion of this change and its
measured effects.
PR: 183659
Submitted by: Julian Charbon
Reviewed by: jhb
pim_input() properly.
While here, remove extra variable and incorrect condition
before m_pullup().
Reported by: Olivier Cochard-Labbé <olivier cochard.me>
Sponsored by: Nginx, Inc.
* Set ia address/mask values BEFORE attaching to address lists.
Inet6 address assignment is not atomic, so the simplest way to
do this atomically is to fill in ia before attach.
* Validate irfa->ia_addr field before use (we permit ANY sockaddr in old code).
* Do some renamings:
in6_ifinit -> in6_notify_ifa (interaction with other subsystems is here)
in6_setup_ifa -> in6_broadcast_ifa (LLE/Multicast/DaD code)
in6_ifaddloop -> nd6_add_ifa_lle
in6_ifremloop -> nd6_rem_ifa_lle
* Split working with LLE and route announce code for last two.
Add temporary in6_newaddrmsg() function to mimic current rtsock behaviour.
* Call device SIOCSIFADDR handler IFF we're adding first address.
In IPv4 we have to call it on every address change since ARP record
is installed by arp_ifinit() which is called by given handler.
IPv6 stack, on the opposite is responsible to call nd6_add_ifa_lle() so
there is no reason to call SIOCSIFADDR often.
of a syncache connection, copy it into the inp_flowid field.
Without this, an incoming TCP connection won't have an inp_flowid marked
until some data comes in, and this means that things like the per-CPU
TCP timer option will choose a different CPU for the timer work.
(It also means that if one grabbed the flowid via an ioctl from userland,
it won't be available until some data has been received.)
Sponsored by: Netflix, Inc.
Problem case:
Original lookup returns route with GW set, so gw points to
rte->rt_gateway.
After that we're changing dst and performing lookup another time.
Since fwd host is most probably directly reachable, resulting
rte does not contain rt_gateway, so gw is not set. Finally, we
end with packet transmitted to proper interface but wrong
link-layer address.
Found by: lstewart
Discussed with: ae,lstewart
MFC after: 2 weeks
Sponsored by: Yandex LLC
