public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Chris Wright <chrisw@sous-sol.org>
To: linux-kernel@vger.kernel.org, stable@kernel.org
Cc: Justin Forbes <jmforbes@linuxtx.org>,
	Zwane Mwaikambo <zwane@arm.linux.org.uk>,
	"Theodore Ts'o" <tytso@mit.edu>,
	Randy Dunlap <rdunlap@xenotime.net>,
	Dave Jones <davej@redhat.com>,
	Chuck Wolber <chuckw@quantumlinux.com>,
	Chris Wedgwood <reviews@ml.cw.f00f.org>,
	Michael Krufky <mkrufky@linuxtv.org>,
	Chuck Ebbert <cebbert@redhat.com>,
	Domenico Andreoli <cavokz@gmail.com>,
	torvalds@linux-foundation.org, akpm@linux-foundation.org,
	alan@lxorguk.ukuu.org.uk, Patrick McHardy <kaber@trash.net>,
	David S Miller <davem@davemloft.net>
Subject: TCP: Fix shrinking windows with window scaling
Date: Wed, 16 Apr 2008 18:02:02 -0700	[thread overview]
Message-ID: <20080417010348.192847503@sous-sol.org> (raw)
In-Reply-To: 20080417010122.148289106@sous-sol.org

[-- Attachment #1: tcp-fix-shrinking-windows-with-window-scaling.patch --]
[-- Type: text/plain, Size: 2309 bytes --]

-stable review patch.  If anyone has any objections, please let us know.
---------------------

From: Patrick McHardy <kaber@trash.net>

Upstream commit: 607bfbf2d55dd1cfe5368b41c2a81a8c9ccf4723

When selecting a new window, tcp_select_window() tries not to shrink
the offered window by using the maximum of the remaining offered window
size and the newly calculated window size. The newly calculated window
size is always a multiple of the window scaling factor, the remaining
window size however might not be since it depends on rcv_wup/rcv_nxt.
This means we're effectively shrinking the window when scaling it down.


The dump below shows the problem (scaling factor 2^7):

- Window size of 557 (71296) is advertised, up to 3111907257:

IP 172.2.2.3.33000 > 172.2.2.2.33000: . ack 3111835961 win 557 <...>

- New window size of 514 (65792) is advertised, up to 3111907217, 40 bytes
  below the last end:

IP 172.2.2.3.33000 > 172.2.2.2.33000: . 3113575668:3113577116(1448) ack 3111841425 win 514 <...>

The number 40 results from downscaling the remaining window:

3111907257 - 3111841425 = 65832
65832 / 2^7 = 514
65832 % 2^7 = 40

If the sender uses up the entire window before it is shrunk, this can have
chaotic effects on the connection. When sending ACKs, tcp_acceptable_seq()
will notice that the window has been shrunk since tcp_wnd_end() is before
tp->snd_nxt, which makes it choose tcp_wnd_end() as sequence number.
This will fail the receivers checks in tcp_sequence() however since it
is before it's tp->rcv_wup, making it respond with a dupack.

If both sides are in this condition, this leads to a constant flood of
ACKs until the connection times out.

Make sure the window is never shrunk by aligning the remaining window to
the window scaling factor.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
---
 net/ipv4/tcp_output.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -258,7 +258,7 @@ static u16 tcp_select_window(struct sock
 		 *
 		 * Relax Will Robinson.
 		 */
-		new_win = cur_win;
+		new_win = ALIGN(cur_win, 1 << tp->rx_opt.rcv_wscale);
 	}
 	tp->rcv_wnd = new_win;
 	tp->rcv_wup = tp->rcv_nxt;

-- 

  parent reply	other threads:[~2008-04-17  1:16 UTC|newest]

Thread overview: 86+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-04-17  1:01 [patch 00/66] 2.6.24-stable review Chris Wright
2008-04-17  1:01 ` time: prevent the loop in timespec_add_ns() from being optimised away Chris Wright
2008-04-17  1:01 ` kbuild: soften modpost checks when doing cross builds Chris Wright
2008-04-17  1:01 ` mtd: memory corruption in block2mtd.c Chris Wright
2008-04-17  1:01 ` md: remove the super sysfs attribute from devices in an md array Chris Wright
2008-04-17  1:01 ` V4L: ivtv: Add missing sg_init_table() Chris Wright
2008-04-17  1:01 ` UIO: add pgprot_noncached() to UIO mmap code Chris Wright
2008-04-17  1:01 ` USB: add support for Motorola ROKR Z6 cellphone in mass storage mode Chris Wright
2008-04-17  1:01 ` USB: new quirk flag to avoid Set-Interface Chris Wright
2008-04-17  1:01 ` inotify: fix race Chris Wright
2008-04-17  1:01 ` inotify: remove debug code Chris Wright
2008-04-17  1:01 ` NOHZ: reevaluate idle sleep length after add_timer_on() Chris Wright
2008-04-17  1:01 ` slab: fix cache_cache bootstrap in kmem_cache_init() Chris Wright
2008-04-17  1:01 ` xen: fix RMW when unmasking events Chris Wright
2008-04-17  1:01 ` xen: mask out SEP from CPUID Chris Wright
2008-04-17  1:01 ` xen: fix UP setup of shared_info Chris Wright
2008-04-17  1:01 ` PERCPU : __percpu_alloc_mask() can dynamically size percpu_data storage Chris Wright
2008-04-17  1:01 ` alloc_percpu() fails to allocate percpu data Chris Wright
2008-04-17  1:01 ` vfs: fix data leak in nobh_write_end() Chris Wright
2008-04-17  1:01 ` pci: revert SMBus unhide on HP Compaq nx6110 Chris Wright
2008-04-17  1:01 ` hwmon: (w83781d) Fix I/O resource conflict with PNP Chris Wright
2008-04-17  1:01 ` vmcoreinfo: add the symbol "phys_base" Chris Wright
2008-04-17  9:24   ` Eric W. Biederman
2008-04-17 17:16     ` Chris Wright
2008-04-17 17:29       ` Vivek Goyal
2008-04-18 10:17         ` Ken'ichi Ohmichi
2008-04-17 23:31       ` Eric W. Biederman
2008-04-17  1:01 ` USB: Allow initialization of broken keyspan serial adapters Chris Wright
2008-04-17  1:01 ` USB: serial: fix regression in Visor/Palm OS module for kernels >= 2.6.24 Chris Wright
2008-04-17  1:01 ` USB: serial: ti_usb_3410_5052: Correct TUSB3410 endpoint requirements Chris Wright
2008-04-17  8:01   ` Oliver Neukum
2008-04-17 17:02     ` Greg KH
2008-04-17  1:01 ` CRYPTO xcbc: Fix crash when ipsec uses xcbc-mac with big data chunk Chris Wright
2008-04-17 11:26   ` S.Çağlar Onur
2008-04-17 14:22     ` Herbert Xu
2008-04-17 23:33       ` Chris Wright
2008-04-17  1:01 ` mtd: fix broken state in CFI driver caused by FL_SHUTDOWN Chris Wright
2008-04-17  1:01 ` ipmi: change device node ordering to reflect probe order Chris Wright
2008-04-17  1:01 ` AX25 ax25_out: check skb for NULL in ax25_kick() Chris Wright
2008-04-17  1:01 ` NET: include <linux/types.h> into linux/ethtool.h for __u* typedef Chris Wright
2008-04-17  1:01 ` SUNGEM: Fix NAPI assertion failure Chris Wright
2008-04-17  1:01 ` INET: inet_frag_evictor() must run with BH disabled Chris Wright
2008-04-17  1:01 ` LLC: Restrict LLC sockets to root Chris Wright
2008-04-17  1:01 ` netpoll: zap_completion_queue: adjust skb->users counter Chris Wright
2008-04-17  1:01 ` PPPOL2TP: Make locking calls softirq-safe Chris Wright
2008-04-17  1:01 ` PPPOL2TP: Fix SMP issues in skb reorder queue handling Chris Wright
2008-04-17  1:01 ` NET: Add preemption point in qdisc_run Chris Wright
2008-04-17  1:01 ` sch_htb: fix "too many events" situation Chris Wright
2008-04-17  1:02 ` SCTP: Fix local_addr deletions during list traversals Chris Wright
2008-04-17  1:02 ` NET: Fix multicast device ioctl checks Chris Wright
2008-04-17  1:02 ` Chris Wright [this message]
2008-04-17  1:02 ` TCP: Let skbs grow over a page on fast peers Chris Wright
2008-04-17  1:02 ` VLAN: Dont copy ALLMULTI/PROMISC flags from underlying device Chris Wright
2008-04-17  1:02 ` SPARC64: Fix atomic backoff limit Chris Wright
2008-04-17  1:02 ` SPARC64: Fix __get_cpu_var in preemption-enabled area Chris Wright
2008-04-17  1:02 ` SPARC64: flush_ptrace_access() needs preemption disable Chris Wright
2008-04-17  1:02 ` libata: assume no device is attached if both IDENTIFYs are aborted Chris Wright
2008-04-17  1:02 ` sis190: read the mac address from the eeprom first Chris Wright
2008-04-17  1:02 ` bluetooth: hci_core: defer hci_unregister_sysfs() Chris Wright
2008-04-17  1:02 ` SPARC64: Fix FPU saving in 64-bit signal handling Chris Wright
2008-04-17  1:02 ` DVB: tda10086: make the 22kHz tone for DISEQC a config option Chris Wright
2008-04-17  1:02 ` SUNRPC: Fix a memory leak in rpc_create() Chris Wright
2008-04-17 21:25   ` Stefan Lippers-Hollmann
2008-04-17 22:06     ` Trond Myklebust
2008-04-17 22:09       ` Chris Wright
2008-04-18 14:42       ` Chuck Lever
2008-04-17  1:02 ` HFS+: fix unlink of links Chris Wright
2008-04-17  1:02 ` acpi: fix "buggy BIOS check" when CPUs are hot removed Chris Wright
2008-04-17  1:02 ` plip: replace spin_lock_irq with spin_lock_irqsave in irq context Chris Wright
2008-04-17  1:02 ` signalfd: fix for incorrect SI_QUEUE user data reporting Chris Wright
2008-04-17  1:02 ` md: close a livelock window in handle_parity_checks5 Chris Wright
2008-04-17  1:02 ` POWERPC: Fix build of modular drivers/macintosh/apm_emu.c Chris Wright
2008-04-17  1:02 ` pnpacpi: reduce printk severity for "pnpacpi: exceeded the max number of ..." Chris Wright
2008-04-17 15:24   ` Nick Andrew
2008-04-17 17:09     ` Chris Wright
2008-04-18 21:48   ` Bjorn Helgaas
2008-04-23  4:09     ` [stable PATCH for 2.6.24.5 and 2.6.25] pnpacpi: fix potential corruption on "pnpacpi: exceeded the max number of IRQ resources 2" Len Brown
2008-04-17  1:02 ` PARISC futex: special case cmpxchg NULL in kernel space Chris Wright
2008-04-17  1:02 ` PARISC pdc_console: fix bizarre panic on boot Chris Wright
2008-04-17  1:02 ` PARISC fix signal trampoline cache flushing Chris Wright
2008-04-17  1:02 ` acpi: bus: check once more for an empty list after locking it Chris Wright
2008-04-17  1:02 ` fbdev: fix /proc/fb oops after module removal Chris Wright
2008-04-17  1:02 ` macb: Call phy_disconnect on removing Chris Wright
2008-04-17  1:02 ` file capabilities: remove cap_task_kill() Chris Wright
2008-04-17  1:02 ` locks: fix possible infinite loop in fcntl(F_SETLKW) over nfs Chris Wright
2008-04-18  7:50 ` [stable] [patch 00/66] 2.6.24-stable review Chris Wright

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080417010348.192847503@sous-sol.org \
    --to=chrisw@sous-sol.org \
    --cc=akpm@linux-foundation.org \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=cavokz@gmail.com \
    --cc=cebbert@redhat.com \
    --cc=chuckw@quantumlinux.com \
    --cc=davej@redhat.com \
    --cc=davem@davemloft.net \
    --cc=jmforbes@linuxtx.org \
    --cc=kaber@trash.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mkrufky@linuxtv.org \
    --cc=rdunlap@xenotime.net \
    --cc=reviews@ml.cw.f00f.org \
    --cc=stable@kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=tytso@mit.edu \
    --cc=zwane@arm.linux.org.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox