1 /proc/sys/net/ipv4/* Variables:
7 Forward Packets between interfaces.
9 This variable is special, its change resets all configuration
10 parameters to their default state (RFC1122 for hosts, RFC1812
13 ip_default_ttl - INTEGER
14 Default value of TTL field (Time To Live) for outgoing (but not
15 forwarded) IP packets. Should be between 1 and 255 inclusive.
16 Default: 64 (as recommended by RFC1700)
18 ip_no_pmtu_disc - BOOLEAN
19 Disable Path MTU Discovery.
23 default 562 - minimum discovered Path MTU
25 route/max_size - INTEGER
26 Maximum number of routes allowed in the kernel. Increase
27 this when using large numbers of interfaces and/or routes.
29 neigh/default/gc_thresh3 - INTEGER
30 Maximum number of neighbor entries allowed. Increase this
31 when using large numbers of interfaces and when communicating
32 with large numbers of directly-connected peers.
35 Time, in seconds, that cached PMTU information is kept.
38 The advertised MSS depends on the first hop route MTU, but will
39 never be lower than this setting.
41 rt_cache_rebuild_count - INTEGER
42 The per net-namespace route cache emergency rebuild threshold.
43 Any net-namespace having its route cache rebuilt due to
44 a hash bucket chain being too long more than this many times
45 will have its route caching disabled
49 ipfrag_high_thresh - INTEGER
50 Maximum memory used to reassemble IP fragments. When
51 ipfrag_high_thresh bytes of memory is allocated for this purpose,
52 the fragment handler will toss packets until ipfrag_low_thresh
55 ipfrag_low_thresh - INTEGER
56 See ipfrag_high_thresh
59 Time in seconds to keep an IP fragment in memory.
61 ipfrag_secret_interval - INTEGER
62 Regeneration interval (in seconds) of the hash secret (or lifetime
63 for the hash secret) for IP fragments.
66 ipfrag_max_dist - INTEGER
67 ipfrag_max_dist is a non-negative integer value which defines the
68 maximum "disorder" which is allowed among fragments which share a
69 common IP source address. Note that reordering of packets is
70 not unusual, but if a large number of fragments arrive from a source
71 IP address while a particular fragment queue remains incomplete, it
72 probably indicates that one or more fragments belonging to that queue
73 have been lost. When ipfrag_max_dist is positive, an additional check
74 is done on fragments before they are added to a reassembly queue - if
75 ipfrag_max_dist (or more) fragments have arrived from a particular IP
76 address between additions to any IP fragment queue using that source
77 address, it's presumed that one or more fragments in the queue are
78 lost. The existing fragment queue will be dropped, and a new one
79 started. An ipfrag_max_dist value of zero disables this check.
81 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
82 result in unnecessarily dropping fragment queues when normal
83 reordering of packets occurs, which could lead to poor application
84 performance. Using a very large value, e.g. 50000, increases the
85 likelihood of incorrectly reassembling IP fragments that originate
86 from different IP datagrams, which could result in data corruption.
91 inet_peer_threshold - INTEGER
92 The approximate size of the storage. Starting from this threshold
93 entries will be thrown aggressively. This threshold also determines
94 entries' time-to-live and time intervals between garbage collection
95 passes. More entries, less time-to-live, less GC interval.
97 inet_peer_minttl - INTEGER
98 Minimum time-to-live of entries. Should be enough to cover fragment
99 time-to-live on the reassembling side. This minimum time-to-live is
100 guaranteed if the pool size is less than inet_peer_threshold.
103 inet_peer_maxttl - INTEGER
104 Maximum time-to-live of entries. Unused entries will expire after
105 this period of time if there is no memory pressure on the pool (i.e.
106 when the number of entries in the pool is very small).
109 inet_peer_gc_mintime - INTEGER
110 Minimum interval between garbage collection passes. This interval is
111 in effect under high memory pressure on the pool.
114 inet_peer_gc_maxtime - INTEGER
115 Minimum interval between garbage collection passes. This interval is
116 in effect under low (or absent) memory pressure on the pool.
122 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
123 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
127 Controls Appropriate Byte Count (ABC) defined in RFC3465.
128 ABC is a way of increasing congestion window (cwnd) more slowly
129 in response to partial acknowledgments.
131 0 increase cwnd once per acknowledgment (no ABC)
132 1 increase cwnd once per acknowledgment of full sized segment
133 2 allow increase cwnd by two if acknowledgment is
134 of two segments to compensate for delayed acknowledgments.
137 tcp_abort_on_overflow - BOOLEAN
138 If listening service is too slow to accept new connections,
139 reset them. Default state is FALSE. It means that if overflow
140 occurred due to a burst, connection will recover. Enable this
141 option _only_ if you are really sure that listening daemon
142 cannot be tuned to accept connections faster. Enabling this
143 option can harm clients of your server.
145 tcp_adv_win_scale - INTEGER
146 Count buffering overhead as bytes/2^tcp_adv_win_scale
147 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
149 Possible values are [-31, 31], inclusive.
152 tcp_allowed_congestion_control - STRING
153 Show/set the congestion control choices available to non-privileged
154 processes. The list is a subset of those listed in
155 tcp_available_congestion_control.
156 Default is "reno" and the default setting (tcp_congestion_control).
158 tcp_app_win - INTEGER
159 Reserve max(window/2^tcp_app_win, mss) of window for application
160 buffer. Value 0 is special, it means that nothing is reserved.
163 tcp_available_congestion_control - STRING
164 Shows the available congestion control choices that are registered.
165 More congestion control algorithms may be available as modules,
168 tcp_base_mss - INTEGER
169 The initial value of search_low to be used by the packetization layer
170 Path MTU discovery (MTU probing). If MTU probing is enabled,
171 this is the initial MSS used by the connection.
173 tcp_congestion_control - STRING
174 Set the congestion control algorithm to be used for new
175 connections. The algorithm "reno" is always available, but
176 additional choices may be available based on kernel configuration.
177 Default is set as part of kernel configuration.
179 tcp_cookie_size - INTEGER
180 Default size of TCP Cookie Transactions (TCPCT) option, that may be
181 overridden on a per socket basis by the TCPCT socket option.
182 Values greater than the maximum (16) are interpreted as the maximum.
183 Values greater than zero and less than the minimum (8) are interpreted
184 as the minimum. Odd values are interpreted as the next even value.
188 Allows TCP to send "duplicate" SACKs.
191 Enable Explicit Congestion Notification (ECN) in TCP. ECN is only
192 used when both ends of the TCP flow support it. It is useful to
193 avoid losses due to congestion (when the bottleneck router supports
198 2 Only server-side ECN enabled. If the other end does
199 not support ECN, behavior is like with ECN disabled.
203 Enable FACK congestion avoidance and fast retransmission.
204 The value is not used, if tcp_sack is not enabled.
206 tcp_fin_timeout - INTEGER
207 Time to hold socket in state FIN-WAIT-2, if it was closed
208 by our side. Peer can be broken and never close its side,
209 or even died unexpectedly. Default value is 60sec.
210 Usual value used in 2.2 was 180 seconds, you may restore
211 it, but remember that if your machine is even underloaded WEB server,
212 you risk to overflow memory with kilotons of dead sockets,
213 FIN-WAIT-2 sockets are less dangerous than FIN-WAIT-1,
214 because they eat maximum 1.5K of memory, but they tend
215 to live longer. Cf. tcp_max_orphans.
218 Enables Forward RTO-Recovery (F-RTO) defined in RFC4138.
219 F-RTO is an enhanced recovery algorithm for TCP retransmission
220 timeouts. It is particularly beneficial in wireless environments
221 where packet loss is typically due to random radio interference
222 rather than intermediate router congestion. F-RTO is sender-side
223 only modification. Therefore it does not require any support from
226 If set to 1, basic version is enabled. 2 enables SACK enhanced
227 F-RTO if flow uses SACK. The basic version can be used also when
228 SACK is in use though scenario(s) with it exists where F-RTO
229 interacts badly with the packet counting of the SACK enabled TCP
232 tcp_frto_response - INTEGER
233 When F-RTO has detected that a TCP retransmission timeout was
234 spurious (i.e, the timeout would have been avoided had TCP set a
235 longer retransmission timeout), TCP has several options what to do
236 next. Possible values are:
237 0 Rate halving based; a smooth and conservative response,
238 results in halved cwnd and ssthresh after one RTT
239 1 Very conservative response; not recommended because even
240 though being valid, it interacts poorly with the rest of
241 Linux TCP, halves cwnd and ssthresh immediately
242 2 Aggressive response; undoes congestion control measures
243 that are now known to be unnecessary (ignoring the
244 possibility of a lost retransmission that would require
245 TCP to be more cautious), cwnd and ssthresh are restored
246 to the values prior timeout
247 Default: 0 (rate halving based)
249 tcp_keepalive_time - INTEGER
250 How often TCP sends out keepalive messages when keepalive is enabled.
253 tcp_keepalive_probes - INTEGER
254 How many keepalive probes TCP sends out, until it decides that the
255 connection is broken. Default value: 9.
257 tcp_keepalive_intvl - INTEGER
258 How frequently the probes are send out. Multiplied by
259 tcp_keepalive_probes it is time to kill not responding connection,
260 after probes started. Default value: 75sec i.e. connection
261 will be aborted after ~11 minutes of retries.
263 tcp_low_latency - BOOLEAN
264 If set, the TCP stack makes decisions that prefer lower
265 latency as opposed to higher throughput. By default, this
266 option is not set meaning that higher throughput is preferred.
267 An example of an application where this default should be
268 changed would be a Beowulf compute cluster.
271 tcp_max_orphans - INTEGER
272 Maximal number of TCP sockets not attached to any user file handle,
273 held by system. If this number is exceeded orphaned connections are
274 reset immediately and warning is printed. This limit exists
275 only to prevent simple DoS attacks, you _must_ not rely on this
276 or lower the limit artificially, but rather increase it
277 (probably, after increasing installed memory),
278 if network conditions require more than default value,
279 and tune network services to linger and kill such states
280 more aggressively. Let me to remind again: each orphan eats
281 up to ~64K of unswappable memory.
283 tcp_max_ssthresh - INTEGER
284 Limited Slow-Start for TCP with large congestion windows (cwnd) defined in
285 RFC3742. Limited slow-start is a mechanism to limit growth of the cwnd
286 on the region where cwnd is larger than tcp_max_ssthresh. TCP increases cwnd
287 by at most tcp_max_ssthresh segments, and by at least tcp_max_ssthresh/2
288 segments per RTT when the cwnd is above tcp_max_ssthresh.
289 If TCP connection increased cwnd to thousands (or tens of thousands) segments,
290 and thousands of packets were being dropped during slow-start, you can set
291 tcp_max_ssthresh to improve performance for new TCP connection.
294 tcp_max_syn_backlog - INTEGER
295 Maximal number of remembered connection requests, which are
296 still did not receive an acknowledgment from connecting client.
297 Default value is 1024 for systems with more than 128Mb of memory,
298 and 128 for low memory machines. If server suffers of overload,
299 try to increase this number.
301 tcp_max_tw_buckets - INTEGER
302 Maximal number of timewait sockets held by system simultaneously.
303 If this number is exceeded time-wait socket is immediately destroyed
304 and warning is printed. This limit exists only to prevent
305 simple DoS attacks, you _must_ not lower the limit artificially,
306 but rather increase it (probably, after increasing installed memory),
307 if network conditions require more than default value.
309 tcp_mem - vector of 3 INTEGERs: min, pressure, max
310 min: below this number of pages TCP is not bothered about its
313 pressure: when amount of memory allocated by TCP exceeds this number
314 of pages, TCP moderates its memory consumption and enters memory
315 pressure mode, which is exited when memory consumption falls
318 max: number of pages allowed for queueing by all TCP sockets.
320 Defaults are calculated at boot time from amount of available
323 tcp_moderate_rcvbuf - BOOLEAN
324 If set, TCP performs receive buffer auto-tuning, attempting to
325 automatically size the buffer (no greater than tcp_rmem[2]) to
326 match the size required by the path for full throughput. Enabled by
329 tcp_mtu_probing - INTEGER
330 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
333 1 - Disabled by default, enabled when an ICMP black hole detected
334 2 - Always enabled, use initial MSS of tcp_base_mss.
336 tcp_no_metrics_save - BOOLEAN
337 By default, TCP saves various connection metrics in the route cache
338 when the connection closes, so that connections established in the
339 near future can use these to set initial conditions. Usually, this
340 increases overall performance, but may sometimes cause performance
341 degradation. If set, TCP will not cache metrics on closing
344 tcp_orphan_retries - INTEGER
345 This value influences the timeout of a locally closed TCP connection,
346 when RTO retransmissions remain unacknowledged.
347 See tcp_retries2 for more details.
349 The default value is 8.
350 If your machine is a loaded WEB server,
351 you should think about lowering this value, such sockets
352 may consume significant resources. Cf. tcp_max_orphans.
354 tcp_reordering - INTEGER
355 Maximal reordering of packets in a TCP stream.
358 tcp_retrans_collapse - BOOLEAN
359 Bug-to-bug compatibility with some broken printers.
360 On retransmit try to send bigger packets to work around bugs in
363 tcp_retries1 - INTEGER
364 This value influences the time, after which TCP decides, that
365 something is wrong due to unacknowledged RTO retransmissions,
366 and reports this suspicion to the network layer.
367 See tcp_retries2 for more details.
369 RFC 1122 recommends at least 3 retransmissions, which is the
372 tcp_retries2 - INTEGER
373 This value influences the timeout of an alive TCP connection,
374 when RTO retransmissions remain unacknowledged.
375 Given a value of N, a hypothetical TCP connection following
376 exponential backoff with an initial RTO of TCP_RTO_MIN would
377 retransmit N times before killing the connection at the (N+1)th RTO.
379 The default value of 15 yields a hypothetical timeout of 924.6
380 seconds and is a lower bound for the effective timeout.
381 TCP will effectively time out at the first RTO which exceeds the
382 hypothetical timeout.
384 RFC 1122 recommends at least 100 seconds for the timeout,
385 which corresponds to a value of at least 8.
387 tcp_rfc1337 - BOOLEAN
388 If set, the TCP stack behaves conforming to RFC1337. If unset,
389 we are not conforming to RFC, but prevent TCP TIME_WAIT
393 tcp_rmem - vector of 3 INTEGERs: min, default, max
394 min: Minimal size of receive buffer used by TCP sockets.
395 It is guaranteed to each TCP socket, even under moderate memory
399 default: initial size of receive buffer used by TCP sockets.
400 This value overrides net.core.rmem_default used by other protocols.
401 Default: 87380 bytes. This value results in window of 65535 with
402 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
403 less for default tcp_app_win. See below about these variables.
405 max: maximal size of receive buffer allowed for automatically
406 selected receiver buffers for TCP socket. This value does not override
407 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
408 automatic tuning of that socket's receive buffer size, in which
409 case this value is ignored.
410 Default: between 87380B and 6MB, depending on RAM size.
413 Enable select acknowledgments (SACKS).
415 tcp_slow_start_after_idle - BOOLEAN
416 If set, provide RFC2861 behavior and time out the congestion
417 window after an idle period. An idle period is defined at
418 the current RTO. If unset, the congestion window will not
419 be timed out after an idle period.
423 Use the Host requirements interpretation of the TCP urgent pointer field.
424 Most hosts use the older BSD interpretation, so if you turn this on
425 Linux might not communicate correctly with them.
428 tcp_synack_retries - INTEGER
429 Number of times SYNACKs for a passive TCP connection attempt will
430 be retransmitted. Should not be higher than 255. Default value
431 is 5, which corresponds to ~180seconds.
433 tcp_syncookies - BOOLEAN
434 Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
435 Send out syncookies when the syn backlog queue of a socket
436 overflows. This is to prevent against the common 'SYN flood attack'
439 Note, that syncookies is fallback facility.
440 It MUST NOT be used to help highly loaded servers to stand
441 against legal connection rate. If you see SYN flood warnings
442 in your logs, but investigation shows that they occur
443 because of overload with legal connections, you should tune
444 another parameters until this warning disappear.
445 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
447 syncookies seriously violate TCP protocol, do not allow
448 to use TCP extensions, can result in serious degradation
449 of some services (f.e. SMTP relaying), visible not by you,
450 but your clients and relays, contacting you. While you see
451 SYN flood warnings in logs not being really flooded, your server
452 is seriously misconfigured.
454 tcp_syn_retries - INTEGER
455 Number of times initial SYNs for an active TCP connection attempt
456 will be retransmitted. Should not be higher than 255. Default value
457 is 5, which corresponds to ~180seconds.
459 tcp_timestamps - BOOLEAN
460 Enable timestamps as defined in RFC1323.
462 tcp_tso_win_divisor - INTEGER
463 This allows control over what percentage of the congestion window
464 can be consumed by a single TSO frame.
465 The setting of this parameter is a choice between burstiness and
466 building larger TSO frames.
469 tcp_tw_recycle - BOOLEAN
470 Enable fast recycling TIME-WAIT sockets. Default value is 0.
471 It should not be changed without advice/request of technical
474 tcp_tw_reuse - BOOLEAN
475 Allow to reuse TIME-WAIT sockets for new connections when it is
476 safe from protocol viewpoint. Default value is 0.
477 It should not be changed without advice/request of technical
480 tcp_window_scaling - BOOLEAN
481 Enable window scaling as defined in RFC1323.
483 tcp_wmem - vector of 3 INTEGERs: min, default, max
484 min: Amount of memory reserved for send buffers for TCP sockets.
485 Each TCP socket has rights to use it due to fact of its birth.
488 default: initial size of send buffer used by TCP sockets. This
489 value overrides net.core.wmem_default used by other protocols.
490 It is usually lower than net.core.wmem_default.
493 max: Maximal amount of memory allowed for automatically tuned
494 send buffers for TCP sockets. This value does not override
495 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
496 automatic tuning of that socket's send buffer size, in which case
497 this value is ignored.
498 Default: between 64K and 4MB, depending on RAM size.
500 tcp_workaround_signed_windows - BOOLEAN
501 If set, assume no receipt of a window scaling option means the
502 remote TCP is broken and treats the window as a signed quantity.
503 If unset, assume the remote TCP is not broken even if we do
504 not receive a window scaling option from them.
507 tcp_dma_copybreak - INTEGER
508 Lower limit, in bytes, of the size of socket reads that will be
509 offloaded to a DMA copy engine, if one is present in the system
510 and CONFIG_NET_DMA is enabled.
513 tcp_thin_linear_timeouts - BOOLEAN
514 Enable dynamic triggering of linear timeouts for thin streams.
515 If set, a check is performed upon retransmission by timeout to
516 determine if the stream is thin (less than 4 packets in flight).
517 As long as the stream is found to be thin, up to 6 linear
518 timeouts may be performed before exponential backoff mode is
519 initiated. This improves retransmission latency for
520 non-aggressive thin streams, often found to be time-dependent.
521 For more information on thin streams, see
522 Documentation/networking/tcp-thin.txt
525 tcp_thin_dupack - BOOLEAN
526 Enable dynamic triggering of retransmissions after one dupACK
527 for thin streams. If set, a check is performed upon reception
528 of a dupACK to determine if the stream is thin (less than 4
529 packets in flight). As long as the stream is found to be thin,
530 data is retransmitted on the first received dupACK. This
531 improves retransmission latency for non-aggressive thin
532 streams, often found to be time-dependent.
533 For more information on thin streams, see
534 Documentation/networking/tcp-thin.txt
537 tcp_challenge_ack_limit - INTEGER
538 Limits number of Challenge ACK sent per second, as recommended
539 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
544 udp_mem - vector of 3 INTEGERs: min, pressure, max
545 Number of pages allowed for queueing by all UDP sockets.
547 min: Below this number of pages UDP is not bothered about its
548 memory appetite. When amount of memory allocated by UDP exceeds
549 this number, UDP starts to moderate memory usage.
551 pressure: This value was introduced to follow format of tcp_mem.
553 max: Number of pages allowed for queueing by all UDP sockets.
555 Default is calculated at boot time from amount of available memory.
557 udp_rmem_min - INTEGER
558 Minimal size of receive buffer used by UDP sockets in moderation.
559 Each UDP socket is able to use the size for receiving data, even if
560 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
563 udp_wmem_min - INTEGER
564 Minimal size of send buffer used by UDP sockets in moderation.
565 Each UDP socket is able to use the size for sending data, even if
566 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
571 cipso_cache_enable - BOOLEAN
572 If set, enable additions to and lookups from the CIPSO label mapping
573 cache. If unset, additions are ignored and lookups always result in a
574 miss. However, regardless of the setting the cache is still
575 invalidated when required when means you can safely toggle this on and
576 off and the cache will always be "safe".
579 cipso_cache_bucket_size - INTEGER
580 The CIPSO label cache consists of a fixed size hash table with each
581 hash bucket containing a number of cache entries. This variable limits
582 the number of entries in each hash bucket; the larger the value the
583 more CIPSO label mappings that can be cached. When the number of
584 entries in a given hash bucket reaches this limit adding new entries
585 causes the oldest entry in the bucket to be removed to make room.
588 cipso_rbm_optfmt - BOOLEAN
589 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
590 the CIPSO draft specification (see Documentation/netlabel for details).
591 This means that when set the CIPSO tag will be padded with empty
592 categories in order to make the packet data 32-bit aligned.
595 cipso_rbm_structvalid - BOOLEAN
596 If set, do a very strict check of the CIPSO option when
597 ip_options_compile() is called. If unset, relax the checks done during
598 ip_options_compile(). Either way is "safe" as errors are caught else
599 where in the CIPSO processing code but setting this to 0 (False) should
600 result in less work (i.e. it should be faster) but could cause problems
601 with other implementations that require strict checking.
606 ip_local_port_range - 2 INTEGERS
607 Defines the local port range that is used by TCP and UDP to
608 choose the local port. The first number is the first, the
609 second the last local port number. Default value depends on
610 amount of memory available on the system:
612 < 128Mb 1024-4999 or even less.
613 This number defines number of active connections, which this
614 system can issue simultaneously to systems not supporting
615 TCP extensions (timestamps). With tcp_tw_recycle enabled
616 (i.e. by default) range 1024-4999 is enough to issue up to
617 2000 connections per second to systems supporting timestamps.
619 ip_local_reserved_ports - list of comma separated ranges
620 Specify the ports which are reserved for known third-party
621 applications. These ports will not be used by automatic port
622 assignments (e.g. when calling connect() or bind() with port
623 number 0). Explicit port allocation behavior is unchanged.
625 The format used for both input and output is a comma separated
626 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
627 10). Writing to the file will clear all previously reserved
628 ports and update the current list with the one given in the
631 Note that ip_local_port_range and ip_local_reserved_ports
632 settings are independent and both are considered by the kernel
633 when determining which ports are available for automatic port
636 You can reserve ports which are not in the current
637 ip_local_port_range, e.g.:
639 $ cat /proc/sys/net/ipv4/ip_local_port_range
641 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
644 although this is redundant. However such a setting is useful
645 if later the port range is changed to a value that will
646 include the reserved ports.
650 ip_nonlocal_bind - BOOLEAN
651 If set, allows processes to bind() to non-local IP addresses,
652 which can be quite useful - but may break some applications.
656 If set non-zero, enables support for dynamic addresses.
657 If set to a non-zero value larger than 1, a kernel log
658 message will be printed when dynamic address rewriting
662 icmp_echo_ignore_all - BOOLEAN
663 If set non-zero, then the kernel will ignore all ICMP ECHO
667 icmp_echo_ignore_broadcasts - BOOLEAN
668 If set non-zero, then the kernel will ignore all ICMP ECHO and
669 TIMESTAMP requests sent to it via broadcast/multicast.
672 icmp_ratelimit - INTEGER
673 Limit the maximal rates for sending ICMP packets whose type matches
674 icmp_ratemask (see below) to specific targets.
675 0 to disable any limiting,
676 otherwise the minimal space between responses in milliseconds.
679 icmp_ratemask - INTEGER
680 Mask made of ICMP types for which rates are being limited.
681 Significant bits: IHGFEDCBA9876543210
682 Default mask: 0000001100000011000 (6168)
684 Bit definitions (see include/linux/icmp.h):
686 3 Destination Unreachable *
691 C Parameter Problem *
696 H Address Mask Request
699 * These are rate limited by default (see default mask above)
701 icmp_ignore_bogus_error_responses - BOOLEAN
702 Some routers violate RFC1122 by sending bogus responses to broadcast
703 frames. Such violations are normally logged via a kernel warning.
704 If this is set to TRUE, the kernel will not give such warnings, which
705 will avoid log file clutter.
708 icmp_errors_use_inbound_ifaddr - BOOLEAN
710 If zero, icmp error messages are sent with the primary address of
711 the exiting interface.
713 If non-zero, the message will be sent with the primary address of
714 the interface that received the packet that caused the icmp error.
715 This is the behaviour network many administrators will expect from
716 a router. And it can make debugging complicated network layouts
719 Note that if no primary address exists for the interface selected,
720 then the primary address of the first non-loopback interface that
721 has one will be used regardless of this setting.
725 igmp_max_memberships - INTEGER
726 Change the maximum number of multicast groups we can subscribe to.
729 Theoretical maximum value is bounded by having to send a membership
730 report in a single datagram (i.e. the report can't span multiple
731 datagrams, or risk confusing the switch and leaving groups you don't
734 The number of supported groups 'M' is bounded by the number of group
735 report entries you can fit into a single datagram of 65535 bytes.
737 M = 65536-sizeof (ip header)/(sizeof(Group record))
739 Group records are variable length, with a minimum of 12 bytes.
740 So net.ipv4.igmp_max_memberships should not be set higher than:
742 (65536-24) / 12 = 5459
744 The value 5459 assumes no IP header options, so in practice
745 this number may be lower.
747 conf/interface/* changes special settings per interface (where
748 "interface" is the name of your network interface)
750 conf/all/* is special, changes the settings for all interfaces
752 log_martians - BOOLEAN
753 Log packets with impossible addresses to kernel log.
754 log_martians for the interface will be enabled if at least one of
755 conf/{all,interface}/log_martians is set to TRUE,
756 it will be disabled otherwise
758 accept_redirects - BOOLEAN
759 Accept ICMP redirect messages.
760 accept_redirects for the interface will be enabled if:
761 - both conf/{all,interface}/accept_redirects are TRUE in the case
762 forwarding for the interface is enabled
764 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
765 case forwarding for the interface is disabled
766 accept_redirects for the interface will be disabled otherwise
771 Enable IP forwarding on this interface.
773 mc_forwarding - BOOLEAN
774 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
775 and a multicast routing daemon is required.
776 conf/all/mc_forwarding must also be set to TRUE to enable multicast
777 routing for the interface
780 Integer value used to differentiate the devices by the medium they
781 are attached to. Two devices can have different id values when
782 the broadcast packets are received only on one of them.
783 The default value 0 means that the device is the only interface
784 to its medium, value of -1 means that medium is not known.
786 Currently, it is used to change the proxy_arp behavior:
787 the proxy_arp feature is enabled for packets forwarded between
788 two devices attached to different media.
792 proxy_arp for the interface will be enabled if at least one of
793 conf/{all,interface}/proxy_arp is set to TRUE,
794 it will be disabled otherwise
796 proxy_arp_pvlan - BOOLEAN
797 Private VLAN proxy arp.
798 Basically allow proxy arp replies back to the same interface
799 (from which the ARP request/solicitation was received).
801 This is done to support (ethernet) switch features, like RFC
802 3069, where the individual ports are NOT allowed to
803 communicate with each other, but they are allowed to talk to
804 the upstream router. As described in RFC 3069, it is possible
805 to allow these hosts to communicate through the upstream
806 router by proxy_arp'ing. Don't need to be used together with
809 This technology is known by different names:
810 In RFC 3069 it is called VLAN Aggregation.
811 Cisco and Allied Telesyn call it Private VLAN.
812 Hewlett-Packard call it Source-Port filtering or port-isolation.
813 Ericsson call it MAC-Forced Forwarding (RFC Draft).
815 shared_media - BOOLEAN
816 Send(router) or accept(host) RFC1620 shared media redirects.
817 Overrides ip_secure_redirects.
818 shared_media for the interface will be enabled if at least one of
819 conf/{all,interface}/shared_media is set to TRUE,
820 it will be disabled otherwise
823 secure_redirects - BOOLEAN
824 Accept ICMP redirect messages only for gateways,
825 listed in default gateway list.
826 secure_redirects for the interface will be enabled if at least one of
827 conf/{all,interface}/secure_redirects is set to TRUE,
828 it will be disabled otherwise
831 send_redirects - BOOLEAN
832 Send redirects, if router.
833 send_redirects for the interface will be enabled if at least one of
834 conf/{all,interface}/send_redirects is set to TRUE,
835 it will be disabled otherwise
838 bootp_relay - BOOLEAN
839 Accept packets with source address 0.b.c.d destined
840 not to this host as local ones. It is supposed, that
841 BOOTP relay daemon will catch and forward such packets.
842 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
847 accept_source_route - BOOLEAN
848 Accept packets with SRR option.
849 conf/all/accept_source_route must also be set to TRUE to accept packets
850 with SRR option on the interface
851 default TRUE (router)
854 accept_local - BOOLEAN
855 Accept packets with local source addresses. In combination with
856 suitable routing, this can be used to direct packets between two
857 local interfaces over the wire and have them accepted properly.
861 0 - No source validation.
862 1 - Strict mode as defined in RFC3704 Strict Reverse Path
863 Each incoming packet is tested against the FIB and if the interface
864 is not the best reverse path the packet check will fail.
865 By default failed packets are discarded.
866 2 - Loose mode as defined in RFC3704 Loose Reverse Path
867 Each incoming packet's source address is also tested against the FIB
868 and if the source address is not reachable via any interface
869 the packet check will fail.
871 Current recommended practice in RFC3704 is to enable strict mode
872 to prevent IP spoofing from DDos attacks. If using asymmetric routing
873 or other complicated routing, then loose mode is recommended.
875 The max value from conf/{all,interface}/rp_filter is used
876 when doing source validation on the {interface}.
878 Default value is 0. Note that some distributions enable it
882 1 - Allows you to have multiple network interfaces on the same
883 subnet, and have the ARPs for each interface be answered
884 based on whether or not the kernel would route a packet from
885 the ARP'd IP out that interface (therefore you must use source
886 based routing for this to work). In other words it allows control
887 of which cards (usually 1) will respond to an arp request.
889 0 - (default) The kernel can respond to arp requests with addresses
890 from other interfaces. This may seem wrong but it usually makes
891 sense, because it increases the chance of successful communication.
892 IP addresses are owned by the complete host on Linux, not by
893 particular interfaces. Only for more complex setups like load-
894 balancing, does this behaviour cause problems.
896 arp_filter for the interface will be enabled if at least one of
897 conf/{all,interface}/arp_filter is set to TRUE,
898 it will be disabled otherwise
900 arp_announce - INTEGER
901 Define different restriction levels for announcing the local
902 source IP address from IP packets in ARP requests sent on
904 0 - (default) Use any local address, configured on any interface
905 1 - Try to avoid local addresses that are not in the target's
906 subnet for this interface. This mode is useful when target
907 hosts reachable via this interface require the source IP
908 address in ARP requests to be part of their logical network
909 configured on the receiving interface. When we generate the
910 request we will check all our subnets that include the
911 target IP and will preserve the source address if it is from
912 such subnet. If there is no such subnet we select source
913 address according to the rules for level 2.
914 2 - Always use the best local address for this target.
915 In this mode we ignore the source address in the IP packet
916 and try to select local address that we prefer for talks with
917 the target host. Such local address is selected by looking
918 for primary IP addresses on all our subnets on the outgoing
919 interface that include the target IP address. If no suitable
920 local address is found we select the first local address
921 we have on the outgoing interface or on all other interfaces,
922 with the hope we will receive reply for our request and
923 even sometimes no matter the source IP address we announce.
925 The max value from conf/{all,interface}/arp_announce is used.
927 Increasing the restriction level gives more chance for
928 receiving answer from the resolved target while decreasing
929 the level announces more valid sender's information.
932 Define different modes for sending replies in response to
933 received ARP requests that resolve local target IP addresses:
934 0 - (default): reply for any local target IP address, configured
936 1 - reply only if the target IP address is local address
937 configured on the incoming interface
938 2 - reply only if the target IP address is local address
939 configured on the incoming interface and both with the
940 sender's IP address are part from same subnet on this interface
941 3 - do not reply for local addresses configured with scope host,
942 only resolutions for global and link addresses are replied
944 8 - do not reply for all local addresses
946 The max value from conf/{all,interface}/arp_ignore is used
947 when ARP request is received on the {interface}
950 Define mode for notification of address and device changes.
951 0 - (default): do nothing
952 1 - Generate gratuitous arp requests when device is brought up
953 or hardware address changes.
956 Define behavior for gratuitous ARP frames who's IP is not
957 already present in the ARP table:
958 0 - don't create new entries in the ARP table
959 1 - create new entries in the ARP table
961 Both replies and requests type gratuitous arp will trigger the
962 ARP table to be updated, if this setting is on.
964 If the ARP table already contains the IP address of the
965 gratuitous arp frame, the arp table will be updated regardless
966 if this setting is on or off.
969 app_solicit - INTEGER
970 The maximum number of probes to send to the user space ARP daemon
971 via netlink before dropping back to multicast probes (see
972 mcast_solicit). Defaults to 0.
974 disable_policy - BOOLEAN
975 Disable IPSEC policy (SPD) for this interface
977 disable_xfrm - BOOLEAN
978 Disable IPSEC encryption on this interface, whatever the policy
983 Allows you to write a number, which can be used as required.
993 delon.nicolas@wanadoo.fr
998 /proc/sys/net/ipv6/* Variables:
1000 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1001 apply to IPv6 [XXX?].
1003 bindv6only - BOOLEAN
1004 Default value for IPV6_V6ONLY socket option,
1005 which restricts use of the IPv6 socket to IPv6 communication
1007 TRUE: disable IPv4-mapped address feature
1008 FALSE: enable IPv4-mapped address feature
1010 Default: FALSE (as specified in RFC2553bis)
1014 ip6frag_high_thresh - INTEGER
1015 Maximum memory used to reassemble IPv6 fragments. When
1016 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1017 the fragment handler will toss packets until ip6frag_low_thresh
1020 ip6frag_low_thresh - INTEGER
1021 See ip6frag_high_thresh
1023 ip6frag_time - INTEGER
1024 Time in seconds to keep an IPv6 fragment in memory.
1026 ip6frag_secret_interval - INTEGER
1027 Regeneration interval (in seconds) of the hash secret (or lifetime
1028 for the hash secret) for IPv6 fragments.
1032 Change the interface-specific default settings.
1036 Change all the interface-specific settings.
1038 [XXX: Other special features than forwarding?]
1040 conf/all/forwarding - BOOLEAN
1041 Enable global IPv6 forwarding between all interfaces.
1043 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1044 to control which interfaces may forward packets and which not.
1046 This also sets all interfaces' Host/Router setting
1047 'forwarding' to the specified value. See below for details.
1049 This referred to as global forwarding.
1055 Change special settings per interface.
1057 The functional behaviour for certain settings is different
1058 depending on whether local forwarding is enabled or not.
1061 Accept Router Advertisements; autoconfigure using them.
1063 Possible values are:
1064 0 Do not accept Router Advertisements.
1065 1 Accept Router Advertisements if forwarding is disabled.
1066 2 Overrule forwarding behaviour. Accept Router Advertisements
1067 even if forwarding is enabled.
1069 Functional default: enabled if local forwarding is disabled.
1070 disabled if local forwarding is enabled.
1072 accept_ra_defrtr - BOOLEAN
1073 Learn default router in Router Advertisement.
1075 Functional default: enabled if accept_ra is enabled.
1076 disabled if accept_ra is disabled.
1078 accept_ra_pinfo - BOOLEAN
1079 Learn Prefix Information in Router Advertisement.
1081 Functional default: enabled if accept_ra is enabled.
1082 disabled if accept_ra is disabled.
1084 accept_ra_rt_info_max_plen - INTEGER
1085 Maximum prefix length of Route Information in RA.
1087 Route Information w/ prefix larger than or equal to this
1088 variable shall be ignored.
1090 Functional default: 0 if accept_ra_rtr_pref is enabled.
1091 -1 if accept_ra_rtr_pref is disabled.
1093 accept_ra_rtr_pref - BOOLEAN
1094 Accept Router Preference in RA.
1096 Functional default: enabled if accept_ra is enabled.
1097 disabled if accept_ra is disabled.
1099 accept_redirects - BOOLEAN
1102 Functional default: enabled if local forwarding is disabled.
1103 disabled if local forwarding is enabled.
1105 accept_source_route - INTEGER
1106 Accept source routing (routing extension header).
1108 >= 0: Accept only routing header type 2.
1109 < 0: Do not accept routing header.
1114 Autoconfigure addresses using Prefix Information in Router
1117 Functional default: enabled if accept_ra_pinfo is enabled.
1118 disabled if accept_ra_pinfo is disabled.
1120 dad_transmits - INTEGER
1121 The amount of Duplicate Address Detection probes to send.
1124 forwarding - BOOLEAN
1125 Configure interface-specific Host/Router behaviour.
1127 Note: It is recommended to have the same setting on all
1128 interfaces; mixed router/host scenarios are rather uncommon.
1130 Possible values are:
1131 0 Forwarding disabled
1132 1 Forwarding enabled
1133 2 Forwarding enabled (Hybrid Mode)
1137 By default, Host behaviour is assumed. This means:
1139 1. IsRouter flag is not set in Neighbour Advertisements.
1140 2. Router Solicitations are being sent when necessary.
1141 3. If accept_ra is TRUE (default), accept Router
1142 Advertisements (and do autoconfiguration).
1143 4. If accept_redirects is TRUE (default), accept Redirects.
1147 If local forwarding is enabled, Router behaviour is assumed.
1148 This means exactly the reverse from the above:
1150 1. IsRouter flag is set in Neighbour Advertisements.
1151 2. Router Solicitations are not sent.
1152 3. Router Advertisements are ignored unless accept_ra is 2.
1153 4. Redirects are ignored.
1157 Hybrid mode. Same behaviour as TRUE, except for:
1159 2. Router Solicitations are being sent when necessary.
1161 Default: 0 (disabled) if global forwarding is disabled (default),
1162 otherwise 1 (enabled).
1165 Default Hop Limit to set.
1169 Default Maximum Transfer Unit
1170 Default: 1280 (IPv6 required minimum)
1172 router_probe_interval - INTEGER
1173 Minimum interval (in seconds) between Router Probing described
1178 router_solicitation_delay - INTEGER
1179 Number of seconds to wait after interface is brought up
1180 before sending Router Solicitations.
1183 router_solicitation_interval - INTEGER
1184 Number of seconds to wait between Router Solicitations.
1187 router_solicitations - INTEGER
1188 Number of Router Solicitations to send until assuming no
1189 routers are present.
1192 use_tempaddr - INTEGER
1193 Preference for Privacy Extensions (RFC3041).
1194 <= 0 : disable Privacy Extensions
1195 == 1 : enable Privacy Extensions, but prefer public
1196 addresses over temporary addresses.
1197 > 1 : enable Privacy Extensions and prefer temporary
1198 addresses over public addresses.
1199 Default: 0 (for most devices)
1200 -1 (for point-to-point devices and loopback devices)
1202 temp_valid_lft - INTEGER
1203 valid lifetime (in seconds) for temporary addresses.
1204 Default: 604800 (7 days)
1206 temp_prefered_lft - INTEGER
1207 Preferred lifetime (in seconds) for temporary addresses.
1208 Default: 86400 (1 day)
1210 max_desync_factor - INTEGER
1211 Maximum value for DESYNC_FACTOR, which is a random value
1212 that ensures that clients don't synchronize with each
1213 other and generate new addresses at exactly the same time.
1214 value is in seconds.
1217 regen_max_retry - INTEGER
1218 Number of attempts before give up attempting to generate
1219 valid temporary addresses.
1222 max_addresses - INTEGER
1223 Maximum number of autoconfigured addresses per interface. Setting
1224 to zero disables the limitation. It is not recommended to set this
1225 value too large (or to zero) because it would be an easy way to
1226 crash the kernel by allowing too many addresses to be created.
1229 disable_ipv6 - BOOLEAN
1230 Disable IPv6 operation. If accept_dad is set to 2, this value
1231 will be dynamically set to TRUE if DAD fails for the link-local
1233 Default: FALSE (enable IPv6 operation)
1235 When this value is changed from 1 to 0 (IPv6 is being enabled),
1236 it will dynamically create a link-local address on the given
1237 interface and start Duplicate Address Detection, if necessary.
1239 When this value is changed from 0 to 1 (IPv6 is being disabled),
1240 it will dynamically delete all address on the given interface.
1242 accept_dad - INTEGER
1243 Whether to accept DAD (Duplicate Address Detection).
1245 1: Enable DAD (default)
1246 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1247 link-local address has been found.
1249 force_tllao - BOOLEAN
1250 Enable sending the target link-layer address option even when
1251 responding to a unicast neighbor solicitation.
1254 Quoting from RFC 2461, section 4.4, Target link-layer address:
1256 "The option MUST be included for multicast solicitations in order to
1257 avoid infinite Neighbor Solicitation "recursion" when the peer node
1258 does not have a cache entry to return a Neighbor Advertisements
1259 message. When responding to unicast solicitations, the option can be
1260 omitted since the sender of the solicitation has the correct link-
1261 layer address; otherwise it would not have be able to send the unicast
1262 solicitation in the first place. However, including the link-layer
1263 address in this case adds little overhead and eliminates a potential
1264 race condition where the sender deletes the cached link-layer address
1265 prior to receiving a response to a previous solicitation."
1269 Limit the maximal rates for sending ICMPv6 packets.
1270 0 to disable any limiting,
1271 otherwise the minimal space between responses in milliseconds.
1276 Pekka Savola <pekkas@netcore.fi>
1277 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1280 /proc/sys/net/bridge/* Variables:
1282 bridge-nf-call-arptables - BOOLEAN
1283 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1287 bridge-nf-call-iptables - BOOLEAN
1288 1 : pass bridged IPv4 traffic to iptables' chains.
1292 bridge-nf-call-ip6tables - BOOLEAN
1293 1 : pass bridged IPv6 traffic to ip6tables' chains.
1297 bridge-nf-filter-vlan-tagged - BOOLEAN
1298 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1302 bridge-nf-filter-pppoe-tagged - BOOLEAN
1303 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1308 proc/sys/net/sctp/* Variables:
1310 addip_enable - BOOLEAN
1311 Enable or disable extension of Dynamic Address Reconfiguration
1312 (ADD-IP) functionality specified in RFC5061. This extension provides
1313 the ability to dynamically add and remove new addresses for the SCTP
1316 1: Enable extension.
1318 0: Disable extension.
1322 addip_noauth_enable - BOOLEAN
1323 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1324 authentication to protect the operations of adding or removing new
1325 addresses. This requirement is mandated so that unauthorized hosts
1326 would not be able to hijack associations. However, older
1327 implementations may not have implemented this requirement while
1328 allowing the ADD-IP extension. For reasons of interoperability,
1329 we provide this variable to control the enforcement of the
1330 authentication requirement.
1332 1: Allow ADD-IP extension to be used without authentication. This
1333 should only be set in a closed environment for interoperability
1334 with older implementations.
1336 0: Enforce the authentication requirement
1340 auth_enable - BOOLEAN
1341 Enable or disable Authenticated Chunks extension. This extension
1342 provides the ability to send and receive authenticated chunks and is
1343 required for secure operation of Dynamic Address Reconfiguration
1346 1: Enable this extension.
1347 0: Disable this extension.
1351 prsctp_enable - BOOLEAN
1352 Enable or disable the Partial Reliability extension (RFC3758) which
1353 is used to notify peers that a given DATA should no longer be expected.
1361 The limit of the number of new packets that can be initially sent. It
1362 controls how bursty the generated traffic can be.
1366 association_max_retrans - INTEGER
1367 Set the maximum number for retransmissions that an association can
1368 attempt deciding that the remote end is unreachable. If this value
1369 is exceeded, the association is terminated.
1373 max_init_retransmits - INTEGER
1374 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1375 that an association will attempt before declaring the destination
1376 unreachable and terminating.
1380 path_max_retrans - INTEGER
1381 The maximum number of retransmissions that will be attempted on a given
1382 path. Once this threshold is exceeded, the path is considered
1383 unreachable, and new traffic will use a different path when the
1384 association is multihomed.
1388 rto_initial - INTEGER
1389 The initial round trip timeout value in milliseconds that will be used
1390 in calculating round trip times. This is the initial time interval
1391 for retransmissions.
1396 The maximum value (in milliseconds) of the round trip timeout. This
1397 is the largest time interval that can elapse between retransmissions.
1402 The minimum value (in milliseconds) of the round trip timeout. This
1403 is the smallest time interval the can elapse between retransmissions.
1407 hb_interval - INTEGER
1408 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1409 are sent at the specified interval on idle paths to probe the state of
1410 a given path between 2 associations.
1414 sack_timeout - INTEGER
1415 The amount of time (in milliseconds) that the implementation will wait
1420 valid_cookie_life - INTEGER
1421 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1422 is used during association establishment.
1426 cookie_preserve_enable - BOOLEAN
1427 Enable or disable the ability to extend the lifetime of the SCTP cookie
1428 that is used during the establishment phase of SCTP association
1430 1: Enable cookie lifetime extension.
1435 rcvbuf_policy - INTEGER
1436 Determines if the receive buffer is attributed to the socket or to
1437 association. SCTP supports the capability to create multiple
1438 associations on a single socket. When using this capability, it is
1439 possible that a single stalled association that's buffering a lot
1440 of data may block other associations from delivering their data by
1441 consuming all of the receive buffer space. To work around this,
1442 the rcvbuf_policy could be set to attribute the receiver buffer space
1443 to each association instead of the socket. This prevents the described
1446 1: rcvbuf space is per association
1447 0: recbuf space is per socket
1451 sndbuf_policy - INTEGER
1452 Similar to rcvbuf_policy above, this applies to send buffer space.
1454 1: Send buffer is tracked per association
1455 0: Send buffer is tracked per socket.
1459 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1460 Number of pages allowed for queueing by all SCTP sockets.
1462 min: Below this number of pages SCTP is not bothered about its
1463 memory appetite. When amount of memory allocated by SCTP exceeds
1464 this number, SCTP starts to moderate memory usage.
1466 pressure: This value was introduced to follow format of tcp_mem.
1468 max: Number of pages allowed for queueing by all SCTP sockets.
1470 Default is calculated at boot time from amount of available memory.
1472 sctp_rmem - vector of 3 INTEGERs: min, default, max
1473 See tcp_rmem for a description.
1475 sctp_wmem - vector of 3 INTEGERs: min, default, max
1476 See tcp_wmem for a description.
1478 addr_scope_policy - INTEGER
1479 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1481 0 - Disable IPv4 address scoping
1482 1 - Enable IPv4 address scoping
1483 2 - Follow draft but allow IPv4 private addresses
1484 3 - Follow draft but allow IPv4 link local addresses
1489 /proc/sys/net/core/*
1490 dev_weight - INTEGER
1491 The maximum number of packets that kernel can handle on a NAPI
1492 interrupt, it's a Per-CPU variable.
1496 /proc/sys/net/unix/*
1497 max_dgram_qlen - INTEGER
1498 The maximum length of dgram socket receive queue
1505 /proc/sys/net/irda/*
1506 fast_poll_increase FIXME
1507 warn_noreply_time FIXME
1508 discovery_slots FIXME
1511 discovery_timeout FIXME
1512 lap_keepalive_time FIXME
1513 max_noreply_time FIXME
1514 max_tx_data_size FIXME
1516 min_tx_turn_time FIXME