Most importantly avoid doing it with cumulative ACK. However,
since we have lost_cnt_hint in the picture as well needing
adjustments, it's not as trivial as dealing with
retransmit_skb_hint (and cannot be done in the all place we
could trivially leave retransmit_skb_hint untouched).
With the previous patch, this should mostly remove O(n^2)
behavior while cumulative ACKs start flowing once rexmit
after a lossy round-trip made it through.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Most importantly avoid doing it with cumulative ACK. Not clearing
means that we no longer need n^2 processing in resolution of each
fast recovery.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Because lost counter no longer requires tuning, this is
trivial to remove (the tuning wouldn't have been too
hard either) because no "new" retransmittable skb appeared
below retransmit_skb_hint when SACKing for sure.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
I suspect it might have been related to the changed amount
of lost skbs, which was counted by retransmit_cnt_hint that
got changed.
The place for this clearing was very illogical anyway,
it should have been after the LOST-bit clearing loop to
make any sense.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Main benefit in this is that we can then freely point
the retransmit_skb_hint to anywhere we want to because
there's no longer need to know what would be the count
changes involve, and since this is really used only as a
terminator, unnecessary work is one time walk at most,
and if some retransmissions are necessary after that
point later on, the walk is not full waste of time
anyway.
Since retransmit_high must be kept valid, all lost
markers must ensure that.
Now I also have learned how those "holes" in the
rexmittable skbs can appear, mtu probe does them. So
I removed the misleading comment as well.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This useful because we'd need to verifying soon in many places
which makes things slightly more complex than it used to be.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ie., the difference between partial and all clearing doesn't
exists anymore since the SACK optimizations got dropped by
an sacktag rewrite.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch consolidates the code common to TCP and CCID-2:
* TCP uses RFC 3390 in a packet-oriented manner (tcp_input.c) and
* CCID-2 uses RFC 3390 in packet-oriented manner (RFC 4341).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Some duplicated code lying around. Located with my suffix tree
tool.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Large block of code duplication removed.
Sadly, the return value thing is a bit tricky here but it
seems the most sensible way to return positive from validator
on success rather than negative.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Removes legacy reinvent-the-wheel type thing. The generic
machinery integrates much better to automated debugging aids
such as kerneloops.org (and others), and is unambiguous due to
better naming. Non-intuively BUG_TRAP() is actually equal to
WARN_ON() rather than BUG_ON() though some might actually be
promoted to BUG_ON() but I left that to future.
I could make at least one BUILD_BUG_ON conversion.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is based upon an excellent bug report from Eric Dumazet.
tcp_ack() should clear ->icsk_probes_out even if there are packets
outstanding. Otherwise if we get a sequence of ACKs while we do have
packets outstanding over and over again, we'll never clear the
probes_out value and eventually think the connection is too sick and
we'll reset it.
This appears to be some "optimization" added to tcp_ack() in the 2.4.x
timeframe. In 2.2.x, probes_out is pretty much always cleared by
tcp_ack().
Here is Eric's original report:
----------------------------------------
Apparently, we can in some situations reset TCP connections in a couple of seconds when some frames are lost.
In order to reproduce the problem, please try the following program on linux-2.6.25.*
Setup some iptables rules to allow two frames per second sent on loopback interface to tcp destination port 12000
iptables -N SLOWLO
iptables -A SLOWLO -m hashlimit --hashlimit 2 --hashlimit-burst 1 --hashlimit-mode dstip --hashlimit-name slow2 -j ACCEPT
iptables -A SLOWLO -j DROP
iptables -A OUTPUT -o lo -p tcp --dport 12000 -j SLOWLO
Then run the attached program and see the output :
# ./loop
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,200ms,1)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,200ms,3)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,200ms,5)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,200ms,7)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,200ms,9)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,200ms,11)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,201ms,13)
State Recv-Q Send-Q Local Address:Port Peer Address:Port
ESTAB 0 40 127.0.0.1:54455 127.0.0.1:12000 timer:(persist,188ms,15)
write(): Connection timed out
wrote 890 bytes but was interrupted after 9 seconds
ESTAB 0 0 127.0.0.1:12000 127.0.0.1:54455
Exiting read() because no data available (4000 ms timeout).
read 860 bytes
While this tcp session makes progress (sending frames with 50 bytes of payload, every 500ms), linux tcp stack decides to reset it, when tcp_retries 2 is reached (default value : 15)
tcpdump :
15:30:28.856695 IP 127.0.0.1.56554 > 127.0.0.1.12000: S 33788768:33788768(0) win 32792 <mss 16396,nop,nop,sackOK,nop,wscale 7>
15:30:28.856711 IP 127.0.0.1.12000 > 127.0.0.1.56554: S 33899253:33899253(0) ack 33788769 win 32792 <mss 16396,nop,nop,sackOK,nop,wscale 7>
15:30:29.356947 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 1:61(60) ack 1 win 257
15:30:29.356966 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 61 win 257
15:30:29.866415 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 61:111(50) ack 1 win 257
15:30:29.866427 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 111 win 257
15:30:30.366516 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 111:161(50) ack 1 win 257
15:30:30.366527 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 161 win 257
15:30:30.876196 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 161:211(50) ack 1 win 257
15:30:30.876207 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 211 win 257
15:30:31.376282 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 211:261(50) ack 1 win 257
15:30:31.376290 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 261 win 257
15:30:31.885619 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 261:311(50) ack 1 win 257
15:30:31.885631 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 311 win 257
15:30:32.385705 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 311:361(50) ack 1 win 257
15:30:32.385715 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 361 win 257
15:30:32.895249 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 361:411(50) ack 1 win 257
15:30:32.895266 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 411 win 257
15:30:33.395341 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 411:461(50) ack 1 win 257
15:30:33.395351 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 461 win 257
15:30:33.918085 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 461:511(50) ack 1 win 257
15:30:33.918096 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 511 win 257
15:30:34.418163 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 511:561(50) ack 1 win 257
15:30:34.418172 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 561 win 257
15:30:34.927685 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 561:611(50) ack 1 win 257
15:30:34.927698 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 611 win 257
15:30:35.427757 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 611:661(50) ack 1 win 257
15:30:35.427766 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 661 win 257
15:30:35.937359 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 661:711(50) ack 1 win 257
15:30:35.937376 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 711 win 257
15:30:36.437451 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 711:761(50) ack 1 win 257
15:30:36.437464 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 761 win 257
15:30:36.947022 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 761:811(50) ack 1 win 257
15:30:36.947039 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 811 win 257
15:30:37.447135 IP 127.0.0.1.56554 > 127.0.0.1.12000: P 811:861(50) ack 1 win 257
15:30:37.447203 IP 127.0.0.1.12000 > 127.0.0.1.56554: . ack 861 win 257
15:30:41.448171 IP 127.0.0.1.12000 > 127.0.0.1.56554: F 1:1(0) ack 861 win 257
15:30:41.448189 IP 127.0.0.1.56554 > 127.0.0.1.12000: R 33789629:33789629(0) win 0
Source of program :
/*
* small producer/consumer program.
* setup a listener on 127.0.0.1:12000
* Forks a child
* child connect to 127.0.0.1, and sends 10 bytes on this tcp socket every 100 ms
* Father accepts connection, and read all data
*/
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <unistd.h>
#include <stdio.h>
#include <time.h>
#include <sys/poll.h>
int port = 12000;
char buffer[4096];
int main(int argc, char *argv[])
{
int lfd = socket(AF_INET, SOCK_STREAM, 0);
struct sockaddr_in socket_address;
time_t t0, t1;
int on = 1, sfd, res;
unsigned long total = 0;
socklen_t alen = sizeof(socket_address);
pid_t pid;
time(&t0);
socket_address.sin_family = AF_INET;
socket_address.sin_port = htons(port);
socket_address.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
if (lfd == -1) {
perror("socket()");
return 1;
}
setsockopt(lfd, SOL_SOCKET, SO_REUSEADDR, &on, sizeof(int));
if (bind(lfd, (struct sockaddr *)&socket_address, sizeof(socket_address)) == -1) {
perror("bind");
close(lfd);
return 1;
}
if (listen(lfd, 1) == -1) {
perror("listen()");
close(lfd);
return 1;
}
pid = fork();
if (pid == 0) {
int i, cfd = socket(AF_INET, SOCK_STREAM, 0);
close(lfd);
if (connect(cfd, (struct sockaddr *)&socket_address, sizeof(socket_address)) == -1) {
perror("connect()");
return 1;
}
for (i = 0 ; ;) {
res = write(cfd, "blablabla\n", 10);
if (res > 0) total += res;
else if (res == -1) {
perror("write()");
break;
} else break;
usleep(100000);
if (++i == 10) {
system("ss -on dst 127.0.0.1:12000");
i = 0;
}
}
time(&t1);
fprintf(stderr, "wrote %lu bytes but was interrupted after %g seconds\n", total, difftime(t1, t0));
system("ss -on | grep 127.0.0.1:12000");
close(cfd);
return 0;
}
sfd = accept(lfd, (struct sockaddr *)&socket_address, &alen);
if (sfd == -1) {
perror("accept");
return 1;
}
close(lfd);
while (1) {
struct pollfd pfd[1];
pfd[0].fd = sfd;
pfd[0].events = POLLIN;
if (poll(pfd, 1, 4000) == 0) {
fprintf(stderr, "Exiting read() because no data available (4000 ms timeout).\n");
break;
}
res = read(sfd, buffer, sizeof(buffer));
if (res > 0) total += res;
else if (res == 0) break;
else perror("read()");
}
fprintf(stderr, "read %lu bytes\n", total);
close(sfd);
return 0;
}
----------------------------------------
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove redundant checks when setting eff_sacks and make the number of SACKs a
compile time constant. Now that the options code knows how many SACK blocks can
fit in the header, we don't need to have the SACK code guessing at it.
Signed-off-by: Adam Langley <agl@imperialviolet.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some of the metrics (RTT, RTTVAR and RTAX_RTO_MIN) are stored in
kernel units (jiffies) and this leaks out through the netlink API to
user space where the units for jiffies are unknown.
This patches changes the kernel to convert to/from milliseconds. This
changes the ABI, but milliseconds seemed like the most natural unit
for these parameters. Values available via syscall in
/proc/net/rt_cache and netlink will be in milliseconds.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These places have a tcp_sock, but we'd prefer the sock itself to
get net from it. Fortunately, tcp_sk macro is just a type cast, so
this replace is really cheap.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Same as before - the sock is always there to get the net from,
but there are also some places with the net already saved on
the stack.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are some places in TCP that select one MIB index to
bump snmp statistics like this:
if (<something>)
NET_INC_STATS_BH(<some_id>);
else if (<something_else>)
NET_INC_STATS_BH(<some_other_id>);
...
else
NET_INC_STATS_BH(<default_id>);
or in a more tricky but still similar way.
On the other hand, this NET_INC_STATS_BH is a camouflaged
increment of percpu variable, which is not that small.
Factoring those cases out de-bloats 235 bytes on non-preemptible
i386 config and drives parts of the code into 80 columns.
add/remove: 0/0 grow/shrink: 0/7 up/down: 0/-235 (-235)
function old new delta
tcp_fastretrans_alert 1437 1424 -13
tcp_dsack_set 137 124 -13
tcp_xmit_retransmit_queue 690 676 -14
tcp_try_undo_recovery 283 265 -18
tcp_sacktag_write_queue 1550 1515 -35
tcp_update_reordering 162 106 -56
tcp_retransmit_timer 990 904 -86
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts two changesets, ec3c0982a2
("[TCP]: TCP_DEFER_ACCEPT updates - process as established") and
the follow-on bug fix 9ae27e0adb
("tcp: Fix slab corruption with ipv6 and tcp6fuzz").
This change causes several problems, first reported by Ingo Molnar
as a distcc-over-loopback regression where connections were getting
stuck.
Ilpo Järvinen first spotted the locking problems. The new function
added by this code, tcp_defer_accept_check(), only has the
child socket locked, yet it is modifying state of the parent
listening socket.
Fixing that is non-trivial at best, because we can't simply just grab
the parent listening socket lock at this point, because it would
create an ABBA deadlock. The normal ordering is parent listening
socket --> child socket, but this code path would require the
reverse lock ordering.
Next is a problem noticed by Vitaliy Gusev, he noted:
----------------------------------------
>--- a/net/ipv4/tcp_timer.c
>+++ b/net/ipv4/tcp_timer.c
>@@ -481,6 +481,11 @@ static void tcp_keepalive_timer (unsigned long data)
> goto death;
> }
>
>+ if (tp->defer_tcp_accept.request && sk->sk_state == TCP_ESTABLISHED) {
>+ tcp_send_active_reset(sk, GFP_ATOMIC);
>+ goto death;
Here socket sk is not attached to listening socket's request queue. tcp_done()
will not call inet_csk_destroy_sock() (and tcp_v4_destroy_sock() which should
release this sk) as socket is not DEAD. Therefore socket sk will be lost for
freeing.
----------------------------------------
Finally, Alexey Kuznetsov argues that there might not even be any
real value or advantage to these new semantics even if we fix all
of the bugs:
----------------------------------------
Hiding from accept() sockets with only out-of-order data only
is the only thing which is impossible with old approach. Is this really
so valuable? My opinion: no, this is nothing but a new loophole
to consume memory without control.
----------------------------------------
So revert this thing for now.
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch removes CVS keywords that weren't updated for a long time
from comments.
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This bug is able to corrupt fackets_out in very rare cases.
In order for this to cause corruption:
1) DSACK in the middle of previous SACK block must be generated.
2) In order to take that particular branch, part or all of the
DSACKed segment must already be SACKed so that we have that
in cache in the first place.
3) The new info must be top enough so that fackets_out will be
updated on this iteration.
...then fack_count is updated while skb wasn't, then we walk again
that particular segment thus updating fack_count twice for
a single skb and finally that value is assigned to fackets_out
by tcp_sacktag_one.
It is safe to call tcp_sacktag_one just once for a segment (at
DSACK), no need to call again for plain SACK.
Potential problem of the miscount are limited to premature entry
to recovery and to inflated reordering metric (which could even
cancel each other out in the most the luckiest scenarios :-)).
Both are quite insignificant in worst case too and there exists
also code to reset them (fackets_out once sacked_out becomes zero
and reordering metric on RTO).
This has been reported by a number of people, because it occurred
quite rarely, it has been very evasive. Andy Furniss was able to
get it to occur couple of times so that a bit more info was
collected about the problem using a debug patch, though it still
required lot of checking around. Thanks also to others who have
tried to help here.
This is listed as Bugzilla #10346. The bug was introduced by
me in commit 68f8353b48 ([TCP]: Rewrite SACK block processing &
sack_recv_cache use), I probably thought back then that there's
need to scan that entry twice or didn't dare to make it go
through it just once there. Going through twice would have
required restoring fack_count after the walk but as noted above,
I chose to drop the additional walk step altogether here.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is possible that this skip path causes TCP to end up into an
invalid state where ca_state was left to CA_Open while some
segments already came into sacked_out. If next valid ACK doesn't
contain new SACK information TCP fails to enter into
tcp_fastretrans_alert(). Thus at least high_seq is set
incorrectly to a too high seqno because some new data segments
could be sent in between (and also, limited transmit is not
being correctly invoked there). Reordering in both directions
can easily cause this situation to occur.
I guess we would want to use tcp_moderate_cwnd(tp) there as well
as it may be possible to use this to trigger oversized burst to
network by sending an old ACK with huge amount of SACK info, but
I'm a bit unsure about its effects (mainly to FlightSize), so to
be on the safe side I just currently fixed it minimally to keep
TCP's state consistent (obviously, such nasty ACKs have been
possible this far). Though it seems that FlightSize is already
underestimated by some amount, so probably on the long term we
might want to trigger recovery there too, if appropriate, to make
FlightSize calculation to resemble reality at the time when the
losses where discovered (but such change scares me too much now
and requires some more thinking anyway how to do that as it
likely involves some code shuffling).
This bug was found by Brian Vowell while running my TCP debug
patch to find cause of another TCP issue (fackets_out
miscount).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
If receiver consumes segments successfully only in-order, FRTO
fallback to conventional recovery produces RTO loop because
FRTO's forward transmissions will always get dropped and need to
be resent, yet by default they're not marked as lost (which are
the only segments we will retransmit in CA_Loss).
Price to pay about this is occassionally unnecessarily
retransmitting the forward transmission(s). SACK blocks help
a bit to avoid this, so it's mainly a concern for NewReno case
though SACK is not fully immune either.
This change has a side-effect of fixing SACKFRTO problem where
it didn't have snd_nxt of the RTO time available anymore when
fallback become necessary (this problem would have only occured
when RTO would occur for two or more segments and ECE arrives
in step 3; no need to figure out how to fix that unless the
TODO item of selective behavior is considered in future).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Damon L. Chesser <damon@damtek.com>
Tested-by: Damon L. Chesser <damon@damtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It seems that commit 009a2e3e4e ("[TCP] FRTO: Improve
interoperability with other undo_marker users") run into
another land-mine which caused fallback to conventional
recovery to break:
1. Cumulative ACK arrives after FRTO retransmission
2. tcp_try_to_open sees zero retrans_out, clears retrans_stamp
which should be kept like in CA_Loss state it would be
3. undo_marker change allowed tcp_packet_delayed to return
true because of the cleared retrans_stamp once FRTO is
terminated causing LossUndo to occur, which means all loss
markings FRTO made are reverted.
This means that the conventional recovery basically recovered
one loss per RTT, which is not that efficient. It was quite
unobvious that the undo_marker change broken something like
this, I had a quite long session to track it down because of
the non-intuitiviness of the bug (luckily I had a trivial
reproducer at hand and I was also able to learn to use kprobes
in the process as well :-)).
This together with the NewReno+FRTO fix and FRTO in-order
workaround this fixes Damon's problems, this and the first
mentioned are enough to fix Bugzilla #10063.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Damon L. Chesser <damon@damtek.com>
Tested-by: Damon L. Chesser <damon@damtek.com>
Tested-by: Sebastian Hyrwall <zibbe@cisko.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Note: there's actually another bug in FRTO's SACK variant, which
is the causing failure in NewReno case because of the error
that's fixed here. I'll fix the SACK case separately (it's
a separate bug really, though related, but in order to fix that
I need to audit tp->snd_nxt usage a bit).
There were two places where SACK variant of FRTO is getting
incorrectly used even if SACK wasn't negotiated by the TCP flow.
This leads to incorrect setting of frto_highmark with NewReno
if a previous recovery was interrupted by another RTO.
An eventual fallback to conventional recovery then incorrectly
considers one or couple of segments as forward transmissions
though they weren't, which then are not LOST marked during
fallback making them "non-retransmittable" until the next RTO.
In a bad case, those segments are really lost and are the only
one left in the window. Thus TCP needs another RTO to continue.
The next FRTO, however, could again repeat the same events
making the progress of the TCP flow extremely slow.
In order for these events to occur at all, FRTO must occur
again in FRTOs step 3 while the key segments must be lost as
well, which is not too likely in practice. It seems to most
frequently with some small devices such as network printers
that *seem* to accept TCP segments only in-order. In cases
were key segments weren't lost, things get automatically
resolved because those wrongly marked segments don't need to be
retransmitted in order to continue.
I found a reproducer after digging up relevant reports (few
reports in total, none at netdev or lkml I know of), some
cases seemed to indicate middlebox issues which seems now
to be a false assumption some people had made. Bugzilla
#10063 _might_ be related. Damon L. Chesser <damon@damtek.com>
had a reproducable case and was kind enough to tcpdump it
for me. With the tcpdump log it was quite trivial to figure
out.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are functions to refer to the value of dst->metric[THE_METRIC-1]
directly without use of a inline function "dst_metric" defined in
net/dst.h.
The following patch changes them to use the inline function
consistently.
Signed-off-by: Satoru SATOH <satoru.satoh@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
From: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
This fixes a regression added by ec3c0982a2
("[TCP]: TCP_DEFER_ACCEPT updates - process as established")
tcp_v6_do_rcv()->tcp_rcv_established(), the latter goes to step5, where
eventually skb can be freed via tcp_data_queue() (drop: label), then if
check for tcp_defer_accept_check() returns true and thus
tcp_rcv_established() returns -1, which forces tcp_v6_do_rcv() to jump
to reset: label, which in turn will pass through discard: label and free
the same skb again.
Tested by Eric Sesterhenn.
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-By: Patrick McManus <mcmanus@ducksong.com>
Returns non-zero if tp->out_of_order_queue was seen non-empty.
This allows tcp_try_rmem_schedule() to return early.
Signed-off-by: Vitaliy Gusev <vgusev@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_prune_queue() doesn't prune an out-of-order queue at all.
Therefore sk_rmem_schedule() can fail but the out-of-order queue isn't
pruned . This can lead to tcp deadlock state if the next two
conditions are held:
1. There are a sequence hole between last received in
order segment and segments enqueued to the out-of-order queue.
2. Size of all segments in the out-of-order queue is more than tcp_mem[2].
Signed-off-by: Vitaliy Gusev <vgusev@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This expresses __skb_append in terms of __skb_queue_after, exploiting that
__skb_append(old, new, list) = __skb_queue_after(list, old, new).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
MTU probe can cause some remedies for FRTO because the normal
packet ordering may be violated allowing FRTO to make a wrong
decision (it might not be that serious threat for anything
though). Thus it's safer to not run FRTO while MTU probe is
underway.
It seems that the basic FRTO variant should also look for an
skb at probe_seq.start to check if that's retransmitted one
but I didn't implement it now (plain seqno in window check
isn't robust against wraparounds).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes Bugzilla #10384
tcp_simple_retransmit does L increment without any checking
whatsoever for overflowing S+L when Reno is in use.
The simplest scenario I can currently think of is rather
complex in practice (there might be some more straightforward
cases though). Ie., if mss is reduced during mtu probing, it
may end up marking everything lost and if some duplicate ACKs
arrived prior to that sacked_out will be non-zero as well,
leading to S+L > packets_out, tcp_clean_rtx_queue on the next
cumulative ACK or tcp_fastretrans_alert on the next duplicate
ACK will fix the S counter.
More straightforward (but questionable) solution would be to
just call tcp_reset_reno_sack() in tcp_simple_retransmit but
it would negatively impact the probe's retransmission, ie.,
the retransmissions would not occur if some duplicate ACKs
had arrived.
So I had to add reno sacked_out reseting to CA_Loss state
when the first cumulative ACK arrives (this stale sacked_out
might actually be the explanation for the reports of left_out
overflows in kernel prior to 2.6.23 and S+L overflow reports
of 2.6.24). However, this alone won't be enough to fix kernel
before 2.6.24 because it is building on top of the commit
1b6d427bb7 ([TCP]: Reduce sacked_out with reno when purging
write_queue) to keep the sacked_out from overflowing.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Alessandro Suardi <alessandro.suardi@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes a long-standing bug which makes NewReno recovery crippled.
With GSO the whole head skb was marked as LOST which is in
violation of NewReno procedure that only wants to mark one packet
and ended up breaking our TCP code by causing counter overflow
because our code was built on top of assumption about valid
NewReno procedure. This manifested as triggering a WARN_ON for
the overflow in a number of places.
It seems relatively safe alternative to just do nothing if
tcp_fragment fails due to oom because another duplicate ACK is
likely to be received soon and the fragmentation will be retried.
Special thanks goes to Soeren Sonnenburg <kernel@nn7.de> who was
lucky enough to be able to reproduce this so that the warning
for the overflow was hit. It's not as easy task as it seems even
if this bug happens quite often because the amount of outstanding
data is pretty significant for the mismarkings to lead to an
overflow.
Because it's very late in 2.6.25-rc cycle (if this even makes in
time), I didn't want to touch anything with SACK enabled here.
Fragmenting might be useful for it as well but it's more or less
a policy decision rather than mandatory fix. Thus there's no need
to rush and we can postpone considering tcp_fragment with SACK
for 2.6.26.
In 2.6.24 and earlier, this very same bug existed but the effect
is slightly different because of a small changes in the if
conditions that fit to the patch's context. With them nothing
got lost marker and thus no retransmissions happened.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The fast retransmission can be forced locally to the rfc3517
branch in tcp_update_scoreboard instead of making such fragile
constructs deeper in tcp_mark_head_lost.
This is necessary for the next patch which must not have
loopholes for cnt > packets check. As one can notice,
readability got some improvements too because of this :-).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change TCP_DEFER_ACCEPT implementation so that it transitions a
connection to ESTABLISHED after handshake is complete instead of
leaving it in SYN-RECV until some data arrvies. Place connection in
accept queue when first data packet arrives from slow path.
Benefits:
- established connection is now reset if it never makes it
to the accept queue
- diagnostic state of established matches with the packet traces
showing completed handshake
- TCP_DEFER_ACCEPT timeouts are expressed in seconds and can now be
enforced with reasonable accuracy instead of rounding up to next
exponential back-off of syn-ack retry.
Signed-off-by: Patrick McManus <mcmanus@ducksong.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
__FUNCTION__ is gcc-specific, use __func__
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Updated to incorporate Eric's suggestion of using a per cpu buffer
rather than allocating on the stack. Just a two line change, but will
resend in it's entirety.
Signed-off-by: Glenn Griffin <ggriffin.kernel@gmail.com>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
It makes fackets_out to grow too slowly compared with the
real write queue.
This shouldn't cause those BUG_TRAP(packets <= tp->packets_out)
to trigger but how knows how such inconsistent fackets_out
affects here and there around TCP when everything is nowadays
assuming accurate fackets_out. So lets see if this silences
them all.
Reported by Guillaume Chazarain <guichaz@gmail.com>.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
NewReno should add cnt per skb (as with FACK) instead of depending on
SACKED_ACKED bits which won't be set with it at all. Effectively,
NewReno should always exists after the first iteration anyway (or
immediately if there's already head in lost_out.
This was fixed earlier in net-2.6.25 but got reverted among other
stuff and I didn't notice that this is still necessary (actually
wasn't even considering this case while trying to figure out the
reports because I lived with different kind of code than it in reality
was).
This should solve the WARN_ONs in TCP code that as a result of this
triggered multiple times in every place we check for this invariant.
Special thanks to Dave Young <hidave.darkstar@gmail.com> and Krishna
Kumar2 <krkumar2@in.ibm.com> for trying with my debug patches.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Tested-by: Dave Young <hidave.darkstar@gmail.com>
Tested-by: Krishna Kumar2 <krkumar2@in.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Removed case indentation level & combined some nested ifs, mostly
within 80 lines now. This is a leftover from indent patch, it
just had to be done manually to avoid messing it up completely.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
These were manually selected from indent's results which as is
are too noisy to be of any use without human reason. In addition,
some extra newlines between function and its comment were removed
too.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The snd_up check should be enough. I suspect this has been
there to provide a minor optimization in clean_rtx_queue which
used to have a small if (!->sacked) block which could skip
snd_up check among the other work.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
SACK reneging can be precalculated to a FLAG in clean_rtx_queue
which has the right skb looked up. This will help a bit in
future because skb->sacked access will be changed eventually,
changing it already won't hurt any.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Earlier resolution for NewReno's sacked_out should now keep
it small enough for this to become invariant-like check.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch introduces new memory accounting functions for each network
protocol. Most of them are renamed from memory accounting functions
for stream protocols. At the same time, some stream memory accounting
functions are removed since other functions do same thing.
Renaming:
sk_stream_free_skb() -> sk_wmem_free_skb()
__sk_stream_mem_reclaim() -> __sk_mem_reclaim()
sk_stream_mem_reclaim() -> sk_mem_reclaim()
sk_stream_mem_schedule -> __sk_mem_schedule()
sk_stream_pages() -> sk_mem_pages()
sk_stream_rmem_schedule() -> sk_rmem_schedule()
sk_stream_wmem_schedule() -> sk_wmem_schedule()
sk_charge_skb() -> sk_mem_charge()
Removeing
sk_stream_rfree(): consolidates into sock_rfree()
sk_stream_set_owner_r(): consolidates into skb_set_owner_r()
sk_stream_mem_schedule()
The following functions are added.
sk_has_account(): check if the protocol supports accounting
sk_mem_uncharge(): do the opposite of sk_mem_charge()
In addition, to achieve consolidation, updating sk_wmem_queued is
removed from sk_mem_charge().
Next, to consolidate memory accounting functions, this patch adds
memory accounting calls to network core functions. Moreover, present
memory accounting call is renamed to new accounting call.
Finally we replace present memory accounting calls with new interface
in TCP and SCTP.
Signed-off-by: Takahiro Yasui <tyasui@redhat.com>
Signed-off-by: Hideo Aoki <haoki@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While checking Gavin's patch I noticed that the returned seq_rtt
is not used by the caller.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_win_from_space() being signed, compiler might emit an integer divide
to compute tcp_win_from_space()/2 .
Using right shifts is OK here and less expensive.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pointing to the next skb is necessary to avoid referencing
already SACKed skbs which will soon be on a separate list.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Bogus seqno compares just mislead, the code is identical for
both sides of the seqno compare (and was even executed just
once because of return in between).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
To get there, highest_sack must have advanced. When it advances,
a new skb is SACKed, which already sets that FLAG. Besides, the
original purpose of it has puzzled me, never understood why
LOST bit setting of retransmitted skb is marked with
FLAG_DATA_SACKED.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Usually those skbs will have L set, not counting them as lost
retransmissions is misleading.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
1) Skip condition used to be wrong way around which made SACK
processing very broken, missed many blocks because of that.
2) Use highest_sack advancement only if some skbs are already
sacked because otherwise tcp_write_queue_next may move things
too far (occurs mainly with GSO). The other similar advancement
is not problem because highest_sack was previosly put to point
a sacked skb.
These problems were located because of problem report from Matt
Mathis <mathis@psc.edu>.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
The sock_wake_async() performs a bit different actions
depending on "how" argument. Unfortunately this argument
ony has numerical magic values.
I propose to give names to their constants to help people
reading this function callers understand what's going on
without looking into this function all the time.
I suppose this is 2.6.25 material, but if it's not (or the
naming seems poor/bad/awful), I can rework it against the
current net-2.6 tree.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously one of the in-block skip branches was missing it.
Also, drop it from tail-fully-processed case because the next
iteration will do exactly the same thing, i.e., process the
SACK block that contains the DSACK information.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Key points of this patch are:
- In case new SACK information is advance only type, no skb
processing below previously discovered highest point is done
- Optimize cases below highest point too since there's no need
to always go up to highest point (which is very likely still
present in that SACK), this is not entirely true though
because I'm dropping the fastpath_skb_hint which could
previously optimize those cases even better. Whether that's
significant, I'm not too sure.
Currently it will provide skipping by walking. Combined with
RB-tree, all skipping would become fast too regardless of window
size (can be done incrementally later).
Previously a number of cases in TCP SACK processing fails to
take advantage of costly stored information in sack_recv_cache,
most importantly, expected events such as cumulative ACK and new
hole ACKs. Processing on such ACKs result in rather long walks
building up latencies (which easily gets nasty when window is
huge). Those latencies are often completely unnecessary
compared with the amount of _new_ information received, usually
for cumulative ACK there's no new information at all, yet TCP
walks whole queue unnecessary potentially taking a number of
costly cache misses on the way, etc.!
Since the inclusion of highest_sack, there's a lot information
that is very likely redundant (SACK fastpath hint stuff,
fackets_out, highest_sack), though there's no ultimate guarantee
that they'll remain the same whole the time (in all unearthly
scenarios). Take advantage of this knowledge here and drop
fastpath hint and use direct access to highest SACKed skb as
a replacement.
Effectively "special cased" fastpath is dropped. This change
adds some complexity to introduce better coveraged "fastpath",
though the added complexity should make TCP behave more cache
friendly.
The current ACK's SACK blocks are compared against each cached
block individially and only ranges that are new are then scanned
by the high constant walk. For other parts of write queue, even
when in previously known part of the SACK blocks, a faster skip
function is used (if necessary at all). In addition, whenever
possible, TCP fast-forwards to highest_sack skb that was made
available by an earlier patch. In typical case, no other things
but this fast-forward and mandatory markings after that occur
making the access pattern quite similar to the former fastpath
"special case".
DSACKs are special case that must always be walked.
The local to recv_sack_cache copying could be more intelligent
w.r.t DSACKs which are likely to be there only once but that
is left to a separate patch.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Worker function that implements the main logic of
the inner-most loop of tcp_sacktag_write_queue().
Idea was originally presented by David S. Miller.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Highest_sack_end_seq is no longer calculated in the loop,
thus it can be pushed to the worker function altogether
making that function independent of the sacktag.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is going to replace the sack fastpath hint quite soon... :-)
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Many assumptions that are true when no reordering or other
strange events happen are not a part of the RFC3517. FACK
implementation is based on such assumptions. Previously (before
the rewrite) the non-FACK SACK was basically doing fast rexmit
and then it times out all skbs when first cumulative ACK arrives,
which cannot really be called SACK based recovery :-).
RFC3517 SACK disables these things:
- Per SKB timeouts & head timeout entry to recovery
- Marking at least one skb while in recovery (RFC3517 does this
only for the fast retransmission but not for the other skbs
when cumulative ACKs arrive in the recovery)
- Sacktag's loss detection flavors B and C (see comment before
tcp_sacktag_write_queue)
This does not implement the "last resort" rule 3 of NextSeg, which
allows retransmissions also when not enough SACK blocks have yet
arrived above a segment for IsLost to return true [RFC3517].
The implementation differs from RFC3517 in these points:
- Rate-halving is used instead of FlightSize / 2
- Instead of using dupACKs to trigger the recovery, the number
of SACK blocks is used as FACK does with SACK blocks+holes
(which provides more accurate number). It seems that the
difference can affect negatively only if the receiver does not
generate SACK blocks at all even though it claimed to be
SACK-capable.
- Dupthresh is not a constant one. Dynamical adjustments include
both holes and sacked segments (equal to what FACK has) due to
complexity involved in determining the number sacked blocks
between highest_sack and the reordered segment. Thus it's will
be an over-estimate.
Implementation note:
tcp_clean_rtx_queue doesn't need a lost_cnt tweak because head
skb at that point cannot be SACKED_ACKED (nor would such
situation last for long enough to cause problems).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This implements more accurately what is stated in sacktag's
overall comment:
"Both of these heuristics are not used in Loss state, when
we cannot account for retransmits accurately."
When CA_Loss state is entered, the state changer ensures that
undo_marker is only set if no TCPCB_RETRANS skbs were found,
thus having non-zero undo_marker in CA_Loss basically tells
that the R-bits still accurately reflect the current state
of TCP.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
All intermediate conditions include it already, make them
simpler as well.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a delayed ACK representing two packets arrives, there are two RTT
samples available, one for each packet. The first (in order of seq
number) will be artificially long due to the delay waiting for the
second packet, the second will trigger the ACK and so will not itself
be delayed.
According to rfc1323, the SRTT used for RTO calculation should use the
first rtt, so receivers echo the timestamp from the first packet in
the delayed ack. For congestion control however, it seems measuring
delayed ack delay is not desirable as it varies independently of
congestion.
The patch below causes seq_rtt and last_ackt to be updated with any
available later packet rtts which should have less (and hopefully
zero) delack delay. The rtt value then gets passed to
ca_ops->pkts_acked().
Where TCP_CONG_RTT_STAMP was set, effort was made to supress RTTs from
within a TSO chunk (!fully_acked), using only the final ACK (which
includes any TSO delay) to generate RTTs. This patch removes these
checks so RTTs are passed for each ACK to ca_ops->pkts_acked().
For non-delay based congestion control (cubic, h-tcp), rtt is
sometimes used for rtt-scaling. In shortening the RTT, this may make
them a little less aggressive. Delay-based schemes (eg vegas, veno,
illinois) should get a cleaner, more accurate congestion signal,
particularly for small cwnds. The congestion control module can
potentially also filter out bad RTTs due to the delayed ack alarm by
looking at the associated cnt which (where delayed acking is in use)
should probably be 1 if the alarm went off or greater if the ACK was
triggered by a packet.
Signed-off-by: Gavin McCullagh <gavin.mccullagh@nuim.ie>
Acked-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_input_metrics() refers to the built-time constant TCP_RTO_MIN
regardless of configured minimum RTO with iproute2.
Signed-off-by: Satoru SATOH <satoru.satoh@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The previous location is after sacktag processing, which affects
counters tcp_packets_in_flight depends on. This may manifest as
wrong behavior if new SACK blocks are present and all is clear
for call to tcp_cong_avoid, which in the case of
tcp_reno_cong_avoid bails out early because it thinks that
TCP is not limited by cwnd.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Though there's little need for everything that tcp_may_send_now
does (actually, even the state had to be adjusted to pass some
checks FRTO does not want to occur), it's more robust to let it
make the decision if sending is allowed. State adjustments
needed:
- Make sure snd_cwnd limit is not hit in there
- Disable nagle (if necessary) through the frto_counter == 2
The result of check for frto_counter in argument to call for
tcp_enter_frto_loss can just be open coded, therefore there
isn't need to store the previous frto_counter past
tcp_may_send_now.
In addition, returns can then be combined.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
I broke this in commit 3de96471bd:
[TCP]: Wrap-safed reordering detection FRTO check
tcp_process_frto should always see a valid frto_highmark. An invalid
frto_highmark (zero) is very likely what ultimately caused a seqno
compare in tcp_frto_enter_loss to do the wrong leading to the LOST-bit
leak.
Having LOST-bits integry ensured like done after commit
23aeeec365:
[TCP] FRTO: Plug potential LOST-bit leak
won't hurt. It may still be useful in some other, possibly legimate,
scenario.
Reported by Chazarain Guillaume <guichaz@yahoo.fr>.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
NULL ptr can be returned from tcp_write_queue_head to cached_skb
and then assigned to skb if packets_out was zero. Without this,
system is vulnerable to a carefully crafted ACKs which obviously
is remotely triggerable.
Besides, there's very little that needs to be done in sacktag
if there weren't any packets outstanding, just skipping the rest
doesn't hurt.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It might be possible that, in some extreme scenario that
I just cannot now construct in my mind, end_seq <=
frto_highmark check does not match causing the lost_out
and LOST bits become out-of-sync due to clearing and
recounting in the loop.
This may fix LOST-bit leak reported by Chazarain Guillaume
<guichaz@yahoo.fr>.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Otherwise TCP might violate packet ordering principles that FRTO
is based on. If conventional recovery path is chosen, this won't
be significant at all. In practice, any small enough value will
be sufficient to provide proper operation for FRTO, yet other
users of snd_cwnd might benefit from a "close enough" value.
FRTO's formula is now equal to what tcp_enter_cwr() uses.
FRTO used to check application limitedness a bit differently but
I changed that in commit 575ee7140d
and as a result checking for application limitedness became
completely non-existing.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case we run out of mem when fragmenting, the clearing of
FLAG_ONLY_ORIG_SACKED might get missed which then feeds FRTO
with false information. Move clearing outside skb processing
loop so that it will get executed even if the skb loop
terminates prematurely due to out-of-mem.
Besides, now the core of the loop truly deals with a single
skb only, which also enables creation a more self-contained
of tcp_sacktag_one later on.
In addition, small reorganization of if branches was made.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes subtle bug like the one with fastpath_cnt_hint happening
due to the way the GSO and hints interact. Because hints are not
reset when just a GSOed skb is partially ACKed, there's no
guarantee that the relevant part of the write queue is going to
be processed in sacktag at all (skbs below snd_una) because
fastpath hint can fast forward the entrypoint.
This was also on the way of future reductions in sacktag's skb
processing. Also future cleanups in sacktag can be made after
this (in 2.6.25).
This may make reordering update in tcp_try_undo_partial
redundant but I'm not too sure so I left it there.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reordering detection fails to take account that the reordered
skb may have pcount larger than 1. In such case the lowest of
them had the largest reordering, the old formula used the
highest of them which is pcount - 1 packets less reordered.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Similar to commit 3eec0047d9, point of this is to avoid
skipping R-bit skbs.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
DSACK inside another SACK block were missed if start_seq of DSACK
was larger than SACK block's because sorting prioritizes full
processing of the SACK block before DSACK. After SACK block
sorting situation is like this:
SSSSSSSSS
D
SSSSSS
SSSSSSS
Because write_queue is walked in-order, when the first SACK block
has been processed, TCP is already past the skb for which the
DSACK arrived and we haven't taught it to backtrack (nor should
we), so TCP just continues processing by going to the next SACK
block after the DSACK (if any).
Whenever such DSACK is present, do an embedded checking during
the previous SACK block.
If the DSACK is below snd_una, there won't be overlapping SACK
block, and thus no problem in that case. Also if start_seq of
the DSACK is equal to the actual block, it will be processed
first.
Tested this by using netem to duplicate 15% of packets, and
by printing SACK block when found_dup_sack is true and the
selected skb in the dup_sack = 1 branch (if taken):
SACK block 0: 4344-5792 (relative to snd_una 2019137317)
SACK block 1: 4344-5792 (relative to snd_una 2019137317)
equal start seqnos => next_dup = 0, dup_sack = 1 won't occur...
SACK block 0: 5792-7240 (relative to snd_una 2019214061)
SACK block 1: 2896-7240 (relative to snd_una 2019214061)
DSACK skb match 5792-7240 (relative to snd_una)
...and next_dup = 1 case (after the not shown start_seq sort),
went to dup_sack = 1 branch.
Signed-off-by: Ilpo Jrvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the current net-2.6 kernel, handling FLAG_DSACKING_ACK is broken.
The flag is cleared to 1 just after FLAG_DSACKING_ACK is set.
if (found_dup_sack)
flag |= FLAG_DSACKING_ACK;
:
flag = 1;
To fix it, this patch introduces a part of the tcp_sacktag_state patch:
http://marc.info/?l=linux-netdev&m=119210560431519&w=2
Signed-off-by: Ryousei Takano <takano-ryousei@aist.go.jp>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix inconsistency of terms:
1) D-SACK
2) F-RTO
Signed-off-by: Ryousei Takano <takano-ryousei@aist.go.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
In some places, the result of skb_headroom() is compared to an unsigned
integer, and in others, the result is compared to a signed integer. Make
the comparisons consistent and correct.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Both high-sack detection and new lowest seq variables have
unnecessary zero special case which are now removed by setting
safe initial seqnos.
This also fixes problem which caused zero received_upto being
passed to tcp_mark_lost_retrans which confused after relations
within the marker loop causing incorrect TCPCB_SACKED_RETRANS
clearing. The problem was noticed because of a performance
report from TAKANO Ryousei <takano@axe-inc.co.jp>.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Acked-by: Ryousei Takano <takano-ryousei@aist.go.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
This addition of lost_retrans_low to tcp_sock might be
unnecessary, it's not clear how often lost_retrans worker is
executed when there wasn't work to do.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Detection implemented with lost_retrans must work also when
fastpath is taken, yet most of the queue is skipped including
(very likely) those retransmitted skb's we're interested in.
This problem appeared when the hints got added, which removed
a need to always walk over the whole write queue head.
Therefore decicion for the lost_retrans worker loop entry must
be separated from the sacktag processing more than it was
necessary before.
It turns out to be problematic to optimize the worker loop
very heavily because ack_seqs of skb may have a number of
discontinuity points. Maybe similar approach as currently is
implemented could be attempted but that's becoming more and
more complex because the trend is towards less skb walking
in sacktag marker. Trying a simple work until all rexmitted
skbs heve been processed approach.
Maybe after(highest_sack_end_seq, tp->high_seq) checking is not
sufficiently accurate and causes entry too often in no-work-to-do
cases. Since that's not known, I've separated solution to that
from this patch.
Noticed because of report against a related problem from TAKANO
Ryousei <takano@axe-inc.co.jp>. He also provided a patch to
that part of the problem. This patch includes solution to it
(though this patch has to use somewhat different placement).
TAKANO's description and patch is available here:
http://marc.info/?l=linux-netdev&m=119149311913288&w=2
...In short, TAKANO's problem is that end_seq the loop is using
not necessarily the largest SACK block's end_seq because the
current ACK may still have higher SACK blocks which are later
by the loop.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Both sacked_out and fackets_out are directly known from how
parameter. Since fackets_out is accurate, there's no need for
recounting (sacked_out was previously unnecessarily counted
in the loop anyway).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is necessary for upcoming DSACK bugfix. Reduces sacktag
length which is not very sad thing at all... :-)
Notice that there's a need to handle out-of-mem at caller's
place.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's on the way for future cutting of that function.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This condition (plain R) can arise at least in recovery that
is triggered after tcp_undo_loss. There isn't any reason why
they should not be marked as lost, not marking makes in_flight
estimator to return too large values.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
I was reading tcp_enter_loss while looking for Cedric's bug and
noticed bytes_acked adjustment is missing from FRTO side.
Since bytes_acked will only be used in tcp_cong_avoid, I think
it's safe to assume RTO would be spurious. During FRTO cwnd
will be not controlled by tcp_cong_avoid and if FRTO calls for
conventional recovery, cwnd is adjusted and the result of wrong
assumption is cleared from bytes_acked. If RTO was in fact
spurious, we did normal ABC already and can continue without
any additional adjustments.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Follows own function for each task principle, this is really
somewhat separate task being done in sacktag. Also reduces
indentation.
In addition, added ack_seq local var to break some long
lines & fixed coding style things.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix a bunch of sparse warnings. Mostly about 0 used as
NULL pointer, and shadowed variable declarations.
One notable case was that hash size should have been unsigned.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This should no longer be necessary because fackets_out is
accurate. It indicates bugs elsewhere, thus report it.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case somebody has a suggestion about a better place for this
check, which must guarantee execution "early enough" (i.e,
before the wrap can occur), I'm very open to them.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Just came across what RFC2018 states about generation of valid
SACK blocks in case of reneging. Alter comment a bit to point
out clearly.
IMHO, there isn't any reason to change code because the
validation is there for a purpose (counters will inform user
about decision TCP made if this case ever surfaces).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case of ACK reordering, the SACK block might be valid in it's
time but is already obsoleted since we've received another kind
of confirmation about arrival of the segments through snd_una
advancement of an earlier packet.
I didn't bother to build distinguishing of valid and invalid
SACK blocks but simply made reordered SACK blocks that are too
old always not counted regardless of their "real" validity which
could be determined by using the ack field of the reordered
packet (won't be significant IMHO).
DSACKs can very well be considered useful even in this situation,
so won't do any of this for them.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's no reason to clear the sacktag skb hint when small part
of the rexmit queue changes. Account changes (if any) instead when
fragmenting/collapsing. RTO/FRTO do not touch SACKED_ACKED bits so
no need to discard SACK tag hint at all.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Most of the description that follows comes from my mail to
netdev (some editing done):
Main obstacle to FRTO use is its deployment as it has to be on
the sender side where as wireless link is often the receiver's
access link. Take initiative on behalf of unlucky receivers and
enable it by default in future Linux TCP senders. Also IETF
seems to interested in advancing FRTO from experimental [1].
How does FRTO help?
===================
FRTO detects spurious RTOs and avoids a number of unnecessary
retransmissions and a couple of other problems that can arise
due to incorrect guess made at RTO (i.e., that segments were
lost when they actually got delayed which is likely to occur
e.g. in wireless environments with link-layer retransmission).
Though FRTO cannot prevent the first (potentially unnecessary)
retransmission at RTO, I suspect that it won't cost that much
even if you have to pay for each bit (won't be that high
percentage out of all packets after all :-)). However, usually
when you have a spurious RTO, not only the first segment
unnecessarily retransmitted but the *whole window*. It goes like
this: all cumulative ACKs got delayed due to in-order delivery,
then TCP will actually send 1.5*original cwnd worth of data in
the RTO's slow-start when the delayed ACKs arrive (basically the
original cwnd worth of it unnecessarily). In case one is
interested in minimizing unnecessary retransmissions e.g. due to
cost, those rexmissions must never see daylight. Besides, in the
worst case the generated burst overloads the bottleneck buffers
which is likely to significantly delay the further progress of
the flow. In case of ll rexmissions, ACK compression often
occurs at the same time making the burst very "sharp edged" (in
that case TCP often loses most of the segments above high_seq
=> very bad performance too). When FRTO is enabled, those
unnecessary retransmissions are fully avoided except for the
first segment and the cwnd behavior after detected spurious RTO
is determined by the response (one can tune that by sysctl).
Basic version (non-SACK enhanced one), FRTO can fail to detect
spurious RTO as spurious and falls back to conservative
behavior. ACK lossage is much less significant than reordering,
usually the FRTO can detect spurious RTO if at least 2
cumulative ACKs from original window are preserved (excluding
the ACK that advances to high_seq). With SACK-enhanced version,
the detection is quite robust.
FRTO should remove the need to set a high lower bound for the
RTO estimator due to delay spikes that occur relatively common
in some environments (esp. in wireless/cellular ones).
[1] http://www1.ietf.org/mail-archive/web/tcpm/current/msg02862.html
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Basically this change enables it, previously other undo_marker
users were left with nothing. Reverse undo_marker logic
completely to get it set right in CA_Loss. On the other hand,
when spurious RTO is detected, clear it. Clearing might be too
heavy for some scenarios but seems safe enough starting point
for now and shouldn't have much effect except in majority of
cases (if in any).
By adding a new FLAG_ we avoid looping through write_queue when
RTO occurs.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implements following cleanups:
- Comment re-placement (CodingStyle)
- tcp_tso_acked() local (wrapper-like) variable removal
(readability)
- __-types removed (IMHO they make local variables jumpy looking
and just was space)
- acked -> flag (naming conventions elsewhere in TCP code)
- linebreak adjustments (readability)
- nested if()s combined (reduced indentation)
- clarifying newlines added
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The accounting code is pretty much the same, so it's a shame
we do it in two places.
I'm not too sure if added fully_acked check in MTU probing is
really what we want perhaps the added end_seq could be used in
the after() comparison.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Substraction for fackets_out is unconditional when snd_una
advances, thus there's no need to do it inside the loop. Just
make sure correct bounds are honored.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In DSACK case, some events are not extraordinary, such as packet
duplication generated DSACK. They can arrive easily below
snd_una when undo_marker is not set (TCP being in CA_Open),
counting such DSACKs amoung SACK discards will likely just
mislead if they occur in some scenario when there are other
problems as well. Similarly, excessively delayed packets could
cause "normal" DSACKs. Therefore, separate counters are
allocated for DSACK events.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
SACK processing code has been a sort of russian roulette as no
validation of SACK blocks is previously attempted. Besides, it
is not very clear what all kinds of broken SACK blocks really
mean (e.g., one that has start and end sequence numbers
reversed). So now close the roulette once and for all.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Only thing that tiny function does is rearming the RTO (if
necessary), name it accordingly.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Makes caller side more obvious, there's no need to have
a wrapper for this oneliner!
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously code had IsReno/IsFack defined as macros that were
local to tcp_input.c though sack_ok field has user elsewhere too
for the same purpose. This changes them to static inlines as
preferred according the current coding style and unifies the
access to sack_ok across multiple files. Magic bitops of sack_ok
for FACK and DSACK are also abstracted to functions with
appropriate names.
Note:
- One sack_ok = 1 remains but that's self explanary, i.e., it
enables sack
- Couple of !IsReno cases are changed to tcp_is_sack
- There were no users for IsDSack => I dropped it
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously TCP had a transitional state during which reno
counted segments that are already below the current window into
sacked_out, which is now prevented. In addition, re-try now
the unconditional S+L skb catching.
This approach conservatively calls just remove_sack and leaves
reset_sack() calls alone. The best solution to the whole problem
would be to first calculate the new sacked_out fully (this patch
does not move reno_sack_reset calls from original sites and thus
does not implement this). However, that would require very
invasive change to fastretrans_alert (perhaps even slicing it to
two halves). Alternatively, all callers of tcp_packets_in_flight
(i.e., users that depend on sacked_out) should be postponed
until the new sacked_out has been calculated but it isn't any
simpler alternative.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This happens rather infrequently and is only possible during
FRTO. We must not allow TCP to slip to Open state because
tcp_fastretrans_alert might then not be called on it's time
when FRTO has exited. This become a problem when left_out
got removed and was replaced by just sacked_out.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_verify_left_out is useful for verifying S+L condition, so
add it back to couple of places in where the code was not
calling to tcp_sync_left_out but used own ad-hoc solution
(before the tcp_sync_left_out got removed).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Left_out was dropped a while ago, thus leaving verifying
consistency of the "left out" as only task for the function in
question. Thus make it's name more appropriate.
In addition, it is intentionally converted to #define instead
of static inline because the location of the invariant failure
is the most important thing to have if this ever triggers. I
think it would have been helpful e.g. in this case where the
location of the failure point had to be based on some quesswork:
http://lkml.org/lkml/2007/5/2/464
...Luckily the guesswork seems to have proved to be correct.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
tp->left_out got removed but nothing came to replace it back
then (users just did addition by themselves), so add function
for users now.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is easily calculable when needed and user are not that many
after all.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
No other users exist for tcp_ecn.h. Very few things remain in
tcp.h, for most TCP ECN functions callers reside within a
single .c file and can be placed there.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
F-RTO does not touch SACKED_ACKED bits at all, so there is no
need to recount them in tcp_enter_frto_loss. After removal of
the else branch, nested ifs can be combined.
This must also reset sacked_out when SACK is not in use as TCP
could have received some duplicate ACKs prior RTO. To achieve
that in a sane manner, tcp_reset_reno_sack was re-placed by the
previous patch.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stupid error from my side. Even though now that I noticed this,
I hoped it would have been an optimization but no, the counter
hint is then incorrect. Thus clearing is necessary for now (I
still suspect though that this path is never executed).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is guaranteed to be valid only when !tp->sacked_out. In most
cases this seqno is available in the last ACK but there is no
guarantee for that. The new fast recovery loss marking algorithm
needs this as entry point.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
When only GSO skb was partially ACKed, no hints are reset,
therefore fastpath_cnt_hint must be tweaked too or else it can
corrupt fackets_out. The corruption to occur, one must have
non-trivial ACK/SACK sequence, so this bug is not very often
that harmful. There's a fackets_out state reset in TCP because
fackets_out is known to be inaccurate and that fixes the issue
eventually anyway.
In case there was also at least one skb that got fully ACKed,
the fastpath_skb_hint is set to NULL which causes a recount for
fastpath_cnt_hint (the old value won't be accessed anymore),
thus it can safely be decremented without additional checking.
Reported by Cedric Le Goater <clg@fr.ibm.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cell phone networks do link layer retransmissions and other
things that cause unnecessary timeout retransmits. So allow
the minimum RTO to be inflated per-route to deal with this.
Signed-off-by: David S. Miller <davem@davemloft.net>
People often get tripped up by this function and think that
it does not implemented the prescribed algorithms from
RFC2414 and RFC3390, even though it does.
So add a comment to head off such misunderstandings in the
future.
Signed-off-by: David S. Miller <davem@davemloft.net>
In case a DSACK is received, it's better to lower cwnd as it's
a sign of data receival.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_cwnd_down must check for it too as it should be conservative
in case of collapse stuff and also when receiver is trying to
lie (though that wouldn't be very successful/useful anyway).
Note:
- Separated also is_dupack and do_lost in fast_retransalert
* Much cleaner look-and-feel now
* This time it really fixes cumulative ACK with many new
SACK blocks recovery entry (I claimed this fixes with
last patch but it wasn't). TCP will now call
tcp_update_scoreboard regardless of is_dupack when
in recovery as long as there is enough fackets_out.
- Introduce FLAG_SND_UNA_ADVANCED
* Some prior_snd_una arguments are unnecessary after it
- Added helper FLAG_ANY_PROGRESS to avoid long FLAG...|FLAG...
constructs
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's possible that new SACK blocks that should trigger new LOST
markings arrive with new data (which previously made is_dupack
false). In addition, I think this fixes a case where we get
a cumulative ACK with enough SACK blocks to trigger the fast
recovery (is_dupack would be false there too).
I'm not completely pleased with this solution because readability
of the code is somewhat questionable as 'is_dupack' in SACK case
is no longer about dupacks only but would mean something like
'lost_marker_work_todo' too... But because of Eifel stuff done
in CA_Recovery, the FLAG_DATA_SACKED check cannot be placed to
the if statement which seems attractive solution. Nevertheless,
I didn't like adding another variable just for that either... :-)
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Actually, the ratehalving seems to work too well, as cwnd is
reduced on every second ACK even though the packets in flight
remains unchanged. Recoveries in a bidirectional flows suffer
quite badly because of this, both NewReno and SACK are affected.
After this patch, rate halving is performed for ACK only if
packets in flight was supposedly changed too.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch changes the API for the callback that is done after an ACK is
received. It solves a couple of issues:
* Some congestion controls want higher resolution value of RTT
(controlled by TCP_CONG_RTT_SAMPLE flag). These don't really want a ktime, but
all compute a RTT in microseconds.
* Other congestion control could use RTT at jiffies resolution.
To keep API consistent the units should be the same for both cases, just the
resolution should change.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
None of the existing TCP congestion controls use the rtt value pased
in the ca_ops->cong_avoid interface. Which is lucky because seq_rtt
could have been -1 when handling a duplicate ack.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
For yet unknown reason, something cleared SACKED_RETRANS bit
underneath FRTO.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 6f74651ae6 is found guilty
of breaking DSACK counting, which should be done only for the
SACK block reported by the DSACK instead of every SACK block
that is received along with DSACK information.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 164891aadf broke RTT
sampling of congestion control modules. Inaccurate timestamps
could be fed to them without providing any way for them to
identify such cases. Previously RTT sampler was called only if
FLAG_RETRANS_DATA_ACKED was not set filtering inaccurate
timestamps nicely. In addition, the new behavior could give an
invalid timestamp (zero) to RTT sampler if only skbs with
TCPCB_RETRANS were ACKed. This solves both problems.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This flaw does not affect any behavior (currently).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Without FRTO, the tcp_try_to_open is never called with
lost_out > 0 (see tcp_time_to_recover). However, when FRTO is
enabled, the !tp->lost condition is not used until end of FRTO
because that way TCP avoids premature entry to fast recovery
during FRTO.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The code used to ignore GSO completely, passing either way too
small or zero pkts_acked when GSO skb or part of it got ACKed.
In addition, there is no need to calculate the value in the loop
but simple arithmetics after the loop is sufficient. There is
no need to handle SYN case specially because congestion control
modules are not yet initialized when FLAG_SYN_ACKED is set.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
State could become inconsistent in two cases:
1) Userspace disabled FRTO by tuning sysctl when one of the TCP
flows was in the middle of FRTO algorithm (and then RTO is
again triggered)
2) SACK reneging occurs during FRTO algorithm
A simple solution is just to abort the previous FRTO when such
obscure condition occurs...
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The conservative spurious RTO response did not queue CWR even
though the sending rate was lowered. Whenever reduction happens
regardless of reason, CWR should be sent (forgetting to send it
is not very fatal though).
A better approach would be to queue CWR when one of the sending
rate reducing responses (rate-halving one or this conservative
response) is used already at RTO. Doing that would allow CWR to
be sent along with the two new data segments that are sent
during FRTO. However, it's a bit "racy" because userland could
tune the response sysctl to a more aggressive one in between.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is a corner case where less than MSS sized new data thingie
is awaiting in the send queue. For F-RTO to work correctly, a
new data segment must be sent at certain point or F-RTO cannot
be used at all. RFC4138 allows overriding of Nagle at that
point.
Implementation uses frto_counter states 2 and 3 to distinguish
when Nagle override is needed.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
No new data is needed until the first ACK comes, so no need to check
for application limitedness until then.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Do some simple changes to make congestion control API faster/cleaner.
* use ktime_t rather than timeval
* merge rtt sampling into existing ack callback
this means one indirect call versus two per ack.
* use flags bits to store options/settings
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Spring cleaning time...
There seems to be a lot of places in the network code that have
extra bogus semicolons after conditionals. Most commonly is a
bogus semicolon after: switch() { }
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a transmitted packet is looped back directly, CHECKSUM_PARTIAL
maps to the semantics of CHECKSUM_UNNECESSARY. Therefore we should
treat it as such in the stack.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the places where we need a pointer to the transport header, it is
still legal to touch skb->h.raw directly if just adding to,
subtracting from or setting it to another layer header.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the cases where the transport header is being set to a offset from
skb->data.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the cases where the network header is being set to a offset from skb->data.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the places where we need a pointer to the network header, it is still legal
to touch skb->nh.raw directly if just adding to, subtracting from or setting it
to another layer header.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the places where we need a pointer to the mac header, it is still legal to
touch skb->mac.raw directly if just adding to, subtracting from or setting it
to another layer header.
This one also converts some more cases to skb_reset_mac_header() that my
regex missed as it had no spaces before nor after '=', ugh.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
That is equal to skb->head before skb_reserve, to help in the layer header
changes.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add whitespace around keywords.
Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This allows the write queue implementation to be changed,
for example, to one which allows fast interval searching.
Signed-off-by: David S. Miller <davem@davemloft.net>
Where appropriate, convert references to xtime.tv_sec to the
get_seconds() helper function.
Signed-off-by: James Morris <jmorris@namei.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Undoing ssthresh is disabled in fastretrans_alert whenever
FLAG_ECE is set by clearing prior_ssthresh. The clearing does
not protect FRTO because FRTO operates before fastretrans_alert.
Moving the clearing of prior_ssthresh earlier seems to be a
suboptimal solution to the FRTO case because then FLAG_ECE will
cause a second ssthresh reduction in try_to_open (the first
occurred when FRTO was entered). So instead, FRTO falls back
immediately to the rate halving response, which switches TCP to
CA_CWR state preventing the latter reduction of ssthresh.
If the first ECE arrived before the ACK after which FRTO is able
to decide RTO as spurious, prior_ssthresh is already cleared.
Thus no undoing for ssthresh occurs. Besides, FLAG_ECE should be
set also in the following ACKs resulting in rate halving response
that sees TCP is already in CA_CWR, which again prevents an extra
ssthresh reduction on that round-trip.
If the first ECE arrived before RTO, ssthresh has already been
adapted and prior_ssthresh remains cleared on entry because TCP
is in CA_CWR (the same applies also to a case where FRTO is
entered more than once and ECE comes in the middle).
High_seq must not be touched after tcp_enter_cwr because CWR
round-trip calculation depends on it.
I believe that after this patch, FRTO should be ECN-safe and
even able to take advantage of synergy benefits.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
A local variable for icsk was created but this change was
missing. Spotted by Jarek Poplawski.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
New sysctl tcp_frto_response is added to select amongst these
responses:
- Rate halving based; reuses CA_CWR state (default)
- Very conservative; used to be the only one available (=1)
- Undo cwr; undoes ssthresh and cwnd reductions (=2)
The response with rate halving requires a new parameter to
tcp_enter_cwr because FRTO has already reduced ssthresh and
doing a second reduction there has to be prevented. In addition,
to keep things nice on 80 cols screen, a local variable was
added.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The reordering detection must work also when FRTO has not been
used at all which was the original intention of mine, just the
expression of the idea was flawed.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implements the SACK-enhanced FRTO given in RFC4138 using the
variant given in Appendix B.
RFC4138, Appendix B:
"This means that in order to declare timeout spurious, the TCP
sender must receive an acknowledgment for non-retransmitted
segment between SND.UNA and RecoveryPoint in algorithm step 3.
RecoveryPoint is defined in conservative SACK-recovery
algorithm [RFC3517]"
The basic version of the FRTO algorithm can still be used also
when SACK is enabled. To enabled SACK-enhanced version, tcp_frto
sysctl is set to 2.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
To be honest, I'm not too sure how the reord stuff works in the
first place but this seems necessary.
When FRTO has been active, the one and only retransmission could
be unnecessary but the state and sending order might not be what
the sacktag code expects it to be (to work correctly).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP without FRTO would be in Loss state with small cwnd. FRTO,
however, leaves cwnd (typically) to a larger value which causes
ssthresh to become too large in case RTO is triggered again
compared to what conventional recovery would do. Because
consecutive RTOs result in only a single ssthresh reduction,
RTO+cumulative ACK+RTO pattern is required to trigger this
event.
A large comment is included for congestion control module writers
trying to figure out what CA_EVENT_FRTO handler should do because
there exists a remote possibility of incompatibility between
FRTO and module defined ssthresh functions.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously RETRANS bits were cleared on the entry to FRTO. We
postpone that into tcp_enter_frto_loss, which is really the
place were the clearing should be done anyway. This allows
simplification of the logic from a clearing loop to the head skb
clearing only.
Besides, the other changes made in the previous patches to
tcp_use_frto made it impossible for the non-SACKed FRTO to be
entered if other than the head has been rexmitted.
With SACK-enhanced FRTO (and Appendix B), however, there can be
a number retransmissions in flight when RTO expires (same thing
could happen before this patchset also with non-SACK FRTO). To
not introduce any jumpiness into the packet counting during FRTO,
instead of clearing RETRANS bits from skbs during entry, do it
later on.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This interpretation comes from RFC4138:
"If the sender implements some loss recovery algorithm other
than Reno or NewReno [FHG04], the F-RTO algorithm SHOULD
NOT be entered when earlier fast recovery is underway."
I think the RFC means to say (especially in the light of
Appendix B) that ...recovery is underway (not just fast recovery)
or was underway when it was interrupted by an earlier (F-)RTO
that hasn't yet been resolved (snd_una has not advanced enough).
Thus, my interpretation is that whenever TCP has ever
retransmitted other than head, basic version cannot be used
because then the order assumptions which are used as FRTO basis
do not hold.
NewReno has only the head segment retransmitted at a time.
Therefore, walk up to the segment that has not been SACKed, if
that segment is not retransmitted nor anything before it, we know
for sure, that nothing after the non-SACKed segment should be
either. This assumption is valid because TCPCB_EVER_RETRANS does
not leave holes but each non-SACKed segment is rexmitted
in-order.
Check for retrans_out > 1 avoids more expensive walk through the
skb list, as we can know the result beforehand: F-RTO will not be
allowed.
SACKed skb can turn into non-SACked only in the extremely rare
case of SACK reneging, in this case we might fail to detect
retransmissions if there were them for any other than head. To
get rid of that feature, whole rexmit queue would have to be
walked (always) or FRTO should be prevented when SACK reneging
happens. Of course RTO should still trigger after reneging which
makes this issue even less likely to show up. And as long as the
response is as conservative as it's now, nothing bad happens even
then.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
FRTO controls cwnd when it still processes the ACK input or it
has just reverted back to conventional RTO recovery; the normal
rules apply when FRTO has reverted to standard congestion
control.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Because TCP is not in Loss state during FRTO recovery, fast
recovery could be triggered by accident. Non-SACK FRTO is more
robust than not yet included SACK-enhanced version (that can
receiver high number of duplicate ACKs with SACK blocks during
FRTO), at least with unidirectional transfers, but under
extraordinary patterns fast recovery can be incorrectly
triggered, e.g., Data loss+ACK losses => cumulative ACK with
enough SACK blocks to meet sacked_out >= dupthresh condition).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since purpose is to reduce CWND, we prevent immediate growth. This
is not a major issue nor is "the correct way" specified anywhere.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The FRTO detection did not care how ACK pattern affects to cwnd
calculation of the conventional recovery. This caused incorrect
setting of cwnd when the fallback becames necessary. The
knowledge tcp_process_frto() has about the incoming ACK is now
passed on to tcp_enter_frto_loss() in allowed_segments parameter
that gives the number of segments that must be added to
packets-in-flight while calculating the new cwnd.
Instead of snd_una we use FLAG_DATA_ACKED in duplicate ACK
detection because RFC4138 states (in Section 2.2):
If the first acknowledgment after the RTO retransmission
does not acknowledge all of the data that was retransmitted
in step 1, the TCP sender reverts to the conventional RTO
recovery. Otherwise, a malicious receiver acknowledging
partial segments could cause the sender to declare the
timeout spurious in a case where data was lost.
If the next ACK after RTO is duplicate, we do not retransmit
anything, which is equal to what conservative conventional
recovery does in such case.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Handles RFC4138 shortcoming (in step 2); it should also have case
c) which ignores ACKs that are not duplicates nor advance window
(opposite dir data, winupdate).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Retransmission counter assumptions are to be changed. Forcing
reason to do this exist: Using sysctl in check would be racy
as soon as FRTO starts to ignore some ACKs (doing that in the
following patches). Userspace may disable it at any moment
giving nice oops if timing is right. frto_counter would be
inaccessible from userspace, but with SACK enhanced FRTO
retrans_out can include other than head, and possibly leaving
it non-zero after spurious RTO, boom again.
Luckily, solution seems rather simple: never go directly to Open
state but use Disorder instead. This does not really change much,
since TCP could anyway change its state to Disorder during FRTO
using path tcp_fastretrans_alert -> tcp_try_to_open (e.g., when
a SACK block makes ACK dubious). Besides, Disorder seems to be
the state where TCP should be if not recovering (in Recovery or
Loss state) while having some retransmissions in-flight (see
tcp_try_to_open), which is exactly what happens with FRTO.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case a latency spike causes more than one RTO, the later should not
cause the already reduced ssthresh to propagate into the prior_ssthresh
since FRTO declares all such RTOs spurious at once or none of them. In
treating of ssthresh, we mimic what tcp_enter_loss() does.
The previous state (in frto_counter) must be available until we have
checked it in tcp_enter_frto(), and also ACK information flag in
process_frto().
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Moved comments out from the body of process_frto() to the head
(preferred way; see Documentation/CodingStyle). Bonus: it's much
easier to read in this compacted form.
FRTO algorithm and implementation is described in greater detail.
For interested reader, more information is available in RFC4138.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
FRTO spurious RTO detection algorithm (RFC4138) does not include response
to a detected spurious RTO but can use different response algorithms.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
FRTO was slightly too brave... Should only clear
TCPCB_SACKED_RETRANS bit.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
We clear the unused parts of the SACK cache, This prevents us from mistakenly
taking the cache data if the old data in the SACK cache is the same as the data
in the SACK block. This assumes that we never receive an empty SACK block with
start and end both at zero.
Signed-off-by: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move DSACK code outside the SACK fast-path checking code. If the DSACK
determined that the information was too old we stayed with a partial cache
copied. Most likely this matters very little since the next packet will not be
DSACK and we will find it in the cache. but it's still not good form and there
is little reason to couple the two checks.
Since the SACK receive cache doesn't need the data to be in host order we also
remove the ntohl in the checking loop.
Signed-off-by: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Only advance the SACK fast-path pointer for the first block, the
fast-path assumes that only the first block advances next time so we
should not move the cached skb for the next sack blocks.
Signed-off-by: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The sorting of SACK blocks actually munges them rather than sort,
causing the TCP stack to ignore some SACK information and breaking the
assumption of ordered SACK blocks after sorting.
The sort takes the data from a second buffer which isn't moved causing
subsequent data moves to occur from the wrong location. The fix is to
use a temporary buffer as a normal sort does.
Signed-off-By: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
I encountered a kernel panic with my test program, which is a very
simple IPv6 client-server program.
The server side sets IPV6_RECVPKTINFO on a listening socket, and the
client side just sends a message to the server. Then the kernel panic
occurs on the server. (If you need the test program, please let me
know. I can provide it.)
This problem happens because a skb is forcibly freed in
tcp_rcv_state_process().
When a socket in listening state(TCP_LISTEN) receives a syn packet,
then tcp_v6_conn_request() will be called from
tcp_rcv_state_process(). If the tcp_v6_conn_request() successfully
returns, the skb would be discarded by __kfree_skb().
However, in case of a listening socket which was already set
IPV6_RECVPKTINFO, an address of the skb will be stored in
treq->pktopts and a ref count of the skb will be incremented in
tcp_v6_conn_request(). But, even if the skb is still in use, the skb
will be freed. Then someone still using the freed skb will cause the
kernel panic.
I suggest to use kfree_skb() instead of __kfree_skb().
Signed-off-by: Masayuki Nakagawa <nakagawa.msy@ncos.nec.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
I believe all the below memory barriers only matter on SMP so
therefore the smp_* variant of the barrier should be used.
I'm wondering if the barrier in net/ipv4/inet_timewait_sock.c should be
dropped entirely. schedule_work's implementation currently implies a
memory barrier and I think sane semantics of schedule_work() should imply
a memory barrier, as needed so the caller shouldn't have to worry.
It's not quite obvious why the barrier in net/packet/af_packet.c is
needed; maybe it should be implied through flush_dcache_page?
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix SO_PEERSEC for tcp sockets to return the security context of
the peer (as represented by the SA from the peer) as opposed to the
SA used by the local/source socket.
Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com>
Signed-off-by: James Morris <jmorris@namei.org>
GCC can't tell we always initialize 'tv' in all the cases
we actually use it, so explicitly set it up with zeros.
Signed-off-by: David S. Miller <davem@davemloft.net>
This changes the microsecond RTT sampling so that samples are taken in
the same way that RTT samples are taken for the RTO calculator: on the
last segment acknowledged, and only when the segment hasn't been
retransmitted.
Signed-off-by: John Heffner <jheffner@psc.edu>
Acked-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some of the instances of tcp_sack_block are host-endian, some - net-endian.
Define struct tcp_sack_block_wire identical to struct tcp_sack_block
with u32 replaced with __be32; annotate uses of tcp_sack_block replacing
net-endian ones with tcp_sack_block_wire. Change is obviously safe since
for cc(1) __be32 is typedefed to u32.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
It does not affect either mss-sized connections (obviously) or
connections controlled by Nagle (because there is only one small
segment in flight).
The idea is to record the fact that a small segment arrives on a
connection, where one small segment has already been received and
still not-ACKed. In this case ACK is forced after tcp_recvmsg() drains
receive buffer.
In other words, it is a "soft" each-2nd-segment ACK, which is enough
to preserve ACK clock even when ABC is enabled.
Signed-off-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
By passing a Linux-generated TSO packet straight back into Linux, Xen
becomes our first LRO user :) Unfortunately, there is at least one spot
in our stack that needs to be changed to cope with this.
The receive MSS estimate is computed from the raw packet size. This is
broken if the packet is GSO/LRO. Fortunately the real MSS can be found
in gso_size so we simply need to use that if it is non-zero.
Real LRO NICs should of course set the gso_size field in future.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change net/core, ipv4 and ipv6 sysctl variables to __read_mostly.
Couldn't actually measure any performance increase while testing (.3%
I consider noise), but seems like the right thing to do.
Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Turn Appropriate Byte Count off by default because it unfairly
penalizes applications that do small writes. Add better documentation
to describe what it is so users will understand why they might want to
turn it on.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
1) fix slow start after retransmit timeout
2) fix case of L=2*SMSS acked bytes comparison
Signed-off-by: Daikichi Osuga <osugad@s1.nttdocomo.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
Whenever a transfer is application limited, we are allowed at least
initial window worth of data per window unless cwnd is previously
less than that.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the current TSO implementation, NETIF_F_TSO and ECN cannot be
turned on together in a TCP connection. The problem is that most
hardware that supports TSO does not handle CWR correctly if it is set
in the TSO packet. Correct handling requires CWR to be set in the
first packet only if it is set in the TSO header.
This patch adds the ability to turn on NETIF_F_TSO and ECN using
GSO if necessary to handle TSO packets with CWR set. Hardware
that handles CWR correctly can turn on NETIF_F_TSO_ECN in the dev->
features flag.
All TSO packets with CWR set will have the SKB_GSO_TCPV4_ECN set. If
the output device does not have the NETIF_F_TSO_ECN feature set, GSO
will split the packet up correctly with CWR only set in the first
segment.
With help from Herbert Xu <herbert@gondor.apana.org.au>.
Since ECN can always be enabled with TSO, the SOCK_NO_LARGESEND sock
flag is completely removed.
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Having separate fields in sk_buff for TSO/UFO (tso_size/ufo_size) is not
going to scale if we add any more segmentation methods (e.g., DCCP). So
let's merge them.
They were used to tell the protocol of a packet. This function has been
subsumed by the new gso_type field. This is essentially a set of netdev
feature bits (shifted by 16 bits) that are required to process a specific
skb. As such it's easy to tell whether a given device can process a GSO
skb: you just have to and the gso_type field and the netdev's features
field.
I've made gso_type a conjunction. The idea is that you have a base type
(e.g., SKB_GSO_TCPV4) that can be modified further to support new features.
For example, if we add a hardware TSO type that supports ECN, they would
declare NETIF_F_TSO | NETIF_F_TSO_ECN. All TSO packets with CWR set would
have a gso_type of SKB_GSO_TCPV4 | SKB_GSO_TCPV4_ECN while all other TSO
packets would be SKB_GSO_TCPV4. This means that only the CWR packets need
to be emulated in software.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Many of the TCP congestion methods all just use ssthresh
as the minimum congestion window on decrease. Rather than
duplicating the code, just have that be the default if that
handle in the ops structure is not set.
Minor behaviour change to TCP compound. It probably wants
to use this (ssthresh) as lower bound, rather than ssthresh/2
because the latter causes undershoot on loss.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
We only want to take receive RTT mesaurements for data
bearing frames, here in the header prediction fast path
for a pure-sender, we know that we have a pure-ACK and
thus the checks in tcp_rcv_rtt_mesaure_ts() will not pass.
Signed-off-by: David S. Miller <davem@davemloft.net>
Locks down user pages and sets up for DMA in tcp_recvmsg, then calls
dma_async_try_early_copy in tcp_v4_do_rcv
Signed-off-by: Chris Leech <christopher.leech@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
From: Aki M Nyrhinen <anyrhine@cs.helsinki.fi>
IMHO the current fix to the problem (in_flight underflow in reno)
is incorrect. it treats the symptons but ignores the problem. the
problem is timing out packets other than the head packet when we
don't have sack. i try to explain (sorry if explaining the obvious).
with sack, scanning the retransmit queue for timed out packets is
fine because we know which packets in our retransmit queue have been
acked by the receiver.
without sack, we know only how many packets in our retransmit queue the
receiver has acknowledged, but no idea which packets.
think of a "typical" slow-start overshoot case, where for example
every third packet in a window get lost because a router buffer gets
full.
with sack, we check for timeouts on those every third packet (as the
rest have been sacked). the packet counting works out and if there
is no reordering, we'll retransmit exactly the packets that were
lost.
without sack, however, we check for timeout on every packet and end up
retransmitting consecutive packets in the retransmit queue. in our
slow-start example, 2/3 of those retransmissions are unnecessary. these
unnecessary retransmissions eat the congestion window and evetually
prevent fast recovery from continuing, if enough packets were lost.
Signed-off-by: David S. Miller <davem@davemloft.net>
From: "Angelo P. Castellani" <angelo.castellani+lkml@gmail.com>
Using NewReno, if a sk_buff is timed out and is accounted as lost_out,
it should also be removed from the sacked_out.
This is necessary because recovery using NewReno fast retransmit could
take up to a lot RTTs and the sk_buff RTO can expire without actually
being really lost.
left_out = sacked_out + lost_out
in_flight = packets_out - left_out + retrans_out
Using NewReno without this patch, on very large network losses,
left_out becames bigger than packets_out + retrans_out (!!).
For this reason unsigned integer in_flight overflows to 2^32 - something.
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch contains the following possible cleanups:
- make the following needlessly global function static:
- arp.c: arp_rcv()
- remove the following unused EXPORT_SYMBOL's:
- devinet.c: devinet_ioctl
- fib_frontend.c: ip_rt_ioctl
- inet_hashtables.c: inet_bind_bucket_create
- inet_hashtables.c: inet_bind_hash
- tcp_input.c: sysctl_tcp_abc
- tcp_ipv4.c: sysctl_tcp_tw_reuse
- tcp_output.c: sysctl_tcp_mtu_probing
- tcp_output.c: sysctl_tcp_base_mss
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This moves some TCP-specific MTU probing state out of
inet_connection_sock back to tcp_sock.
Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Implementation of packetization layer path mtu discovery for TCP, based on
the internet-draft currently found at
<http://www.ietf.org/internet-drafts/draft-ietf-pmtud-method-05.txt>.
Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
This changes some simple "if (x) BUG();" statements to "BUG_ON(x);"
Signed-off-by: Kris Katterjohn <kjak@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP inline usage cleanup:
* get rid of inline in several places
* replace __inline__ with inline where possible
* move functions used in one file out of tcp.h
* let compiler decide on used once cases
On x86_64:
text data bss dec hex filename
3594701 648348 567400 4810449 4966d1 vmlinux.orig
3593133 648580 567400 4809113 496199 vmlinux
On sparc64:
text data bss dec hex filename
2538278 406152 530392 3474822 350586 vmlinux.ORIG
2536382 406384 530392 3473158 34ff06 vmlinux
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
As DCCP needs to be called in the same spots.
Now we have a member in inet_sock (is_icsk), set at sock creation time from
struct inet_protosw->flags (if INET_PROTOSW_ICSK is set, like for TCP and
DCCP) to see if a struct sock instance is a inet_connection_sock for places
like the ones in ip_sockglue.c (v4 and v6) where we previously were looking if
sk_type was SOCK_STREAM, that is insufficient because we now use the same code
for DCCP, that has sk_type SOCK_DCCP.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
And move it to struct inet_connection_sock. DCCP will use it in the
upcoming changesets.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use "hints" to speed up the SACK processing. Various forms
of this have been used by TCP developers (Web100, STCP, BIC)
to avoid the 2x linear search of outstanding segments.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is a patch for discussion addressing some receive buffer growing issues.
This is partially related to the thread "Possible BUG in IPv4 TCP window
handling..." last week.
Specifically it addresses the problem of an interaction between rcvbuf
moderation (receiver autotuning) and rcv_ssthresh. The problem occurs when
sending small packets to a receiver with a larger MTU. (A very common case I
have is a host with a 1500 byte MTU sending to a host with a 9k MTU.) In
such a case, the rcv_ssthresh code is targeting a window size corresponding
to filling up the current rcvbuf, not taking into account that the new rcvbuf
moderation may increase the rcvbuf size.
One hunk makes rcv_ssthresh use tcp_rmem[2] as the size target rather than
rcvbuf. The other changes the behavior when it overflows its memory bounds
with in-order data so that it tries to grow rcvbuf (the same as with
out-of-order data).
These changes should help my problem of mixed MTUs, and should also help the
case from last week's thread I think. (In both cases though you still need
tcp_rmem[2] to be set much larger than the TCP window.) One question is if
this is too aggressive at trying to increase rcvbuf if it's under memory
stress.
Orignally-from: John Heffner <jheffner@psc.edu>
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is an updated version of the RFC3465 ABC patch originally
for Linux 2.6.11-rc4 by Yee-Ting Li. ABC is a way of counting
bytes ack'd rather than packets when updating congestion control.
The orignal ABC described in the RFC applied to a Reno style
algorithm. For advanced congestion control there is little
change after leaving slow start.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simplify the code that comuputes microsecond rtt estimate used
by TCP Vegas. Move the callback out of the RTT sampler and into
the end of the ack cleanup.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This bug is responsible for causing the infamous "Treason uncloaked"
messages that's been popping up everywhere since the printk was added.
It has usually been blamed on foreign operating systems. However,
some of those reports implicate Linux as both systems are running
Linux or the TCP connection is going across the loopback interface.
In fact, there really is a bug in the Linux TCP header prediction code
that's been there since at least 2.1.8. This bug was tracked down with
help from Dale Blount.
The effect of this bug ranges from harmless "Treason uncloaked"
messages to hung/aborted TCP connections. The details of the bug
and fix is as follows.
When snd_wnd is updated, we only update pred_flags if
tcp_fast_path_check succeeds. When it fails (for example,
when our rcvbuf is used up), we will leave pred_flags with
an out-of-date snd_wnd value.
When the out-of-date pred_flags happens to match the next incoming
packet we will again hit the fast path and use the current snd_wnd
which will be wrong.
In the case of the treason messages, it just happens that the snd_wnd
cached in pred_flags is zero while tp->snd_wnd is non-zero. Therefore
when a zero-window packet comes in we incorrectly conclude that the
window is non-zero.
In fact if the peer continues to send us zero-window pure ACKs we
will continue making the same mistake. It's only when the peer
transmits a zero-window packet with data attached that we get a
chance to snap out of it. This is what triggers the treason
message at the next retransmit timeout.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
From: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Handle better the case where the sender sends full sized
frames initially, then moves to a mode where it trickles
out small amounts of data at a time.
This known problem is even mentioned in the comments
above tcp_grow_window() in tcp_input.c, specifically:
...
* The scheme does not work when sender sends good segments opening
* window and then starts to feed us spagetti. But it should work
* in common situations. Otherwise, we have to rely on queue collapsing.
...
When the sender gives full sized frames, the "struct sk_buff" overhead
from each packet is small. So we'll advertize a larger window.
If the sender moves to a mode where small segments are sent, this
ratio becomes tilted to the other extreme and we start overrunning
the socket buffer space.
tcp_clamp_window() tries to address this, but it's clamping of
tp->window_clamp is a wee bit too aggressive for this particular case.
Fix confirmed by Ion Badulescu.
Signed-off-by: David S. Miller <davem@davemloft.net>
The problem is that the SACK fragmenting code may incorrectly call
tcp_fragment() with a length larger than the skb->len. This happens
when the skb on the transmit queue completely falls to the LHS of the
SACK.
And add a BUG() check to tcp_fragment() so we can spot this kind of
error more quickly in the future.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
All we need to do is resegment the queue so that
we record SACK information accurately. The edges
of the SACK blocks guide our resegmenting decisions.
With help from Herbert Xu.
Signed-off-by: David S. Miller <davem@davemloft.net>
Of this type, mostly:
CHECK net/ipv6/netfilter.c
net/ipv6/netfilter.c:96:12: warning: symbol 'ipv6_netfilter_init' was not declared. Should it be static?
net/ipv6/netfilter.c:101:6: warning: symbol 'ipv6_netfilter_fini' was not declared. Should it be static?
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This changeset basically moves tcp_sk()->{ca_ops,ca_state,etc} to inet_csk(),
minimal renaming/moving done in this changeset to ease review.
Most of it is just changes of struct tcp_sock * to struct sock * parameters.
With this we move to a state closer to two interesting goals:
1. Generalisation of net/ipv4/tcp_diag.c, becoming inet_diag.c, being used
for any INET transport protocol that has struct inet_hashinfo and are
derived from struct inet_connection_sock. Keeps the userspace API, that will
just not display DCCP sockets, while newer versions of tools can support
DCCP.
2. INET generic transport pluggable Congestion Avoidance infrastructure, using
the current TCP CA infrastructure with DCCP.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With this we're very close to getting all of the current TCP
refactorings in my dccp-2.6 tree merged, next changeset will export
some functions needed by the current DCCP code and then dccp-2.6.git
will be born!
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Completing the previous changeset, this also generalises tcp_v4_synq_add,
renaming it to inet_csk_reqsk_queue_hash_add, already geing used in the
DCCP tree, which I plan to merge RSN.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
This creates struct inet_connection_sock, moving members out of struct
tcp_sock that are shareable with other INET connection oriented
protocols, such as DCCP, that in my private tree already uses most of
these members.
The functions that operate on these members were renamed, using a
inet_csk_ prefix while not being moved yet to a new file, so as to
ease the review of these changes.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
