public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Neal Cardwell <ncardwell@google.com>,
	Eric Dumazet <edumazet@google.com>,
	Yuchung Cheng <ycheng@google.com>,
	"David S. Miller" <davem@davemloft.net>
Subject: [ 76/86] ipv6: use a stronger hash for tcp
Date: Tue, 26 Feb 2013 16:08:24 -0800	[thread overview]
Message-ID: <20130226235920.825016157@linuxfoundation.org> (raw)
In-Reply-To: <20130226235912.881663118@linuxfoundation.org>

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

------------------


From: Eric Dumazet <edumazet@google.com>

[ Upstream commit 08dcdbf6a7b9d14c2302c5bd0c5390ddf122f664 ]

It looks like its possible to open thousands of TCP IPv6
sessions on a server, all landing in a single slot of TCP hash
table. Incoming packets have to lookup sockets in a very
long list.

We should hash all bits from foreign IPv6 addresses, using
a salt and hash mix, not a simple XOR.

inet6_ehashfn() can also separately use the ports, instead
of xoring them.

Reported-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 include/net/inet6_hashtables.h |    8 ++++----
 include/net/inet_sock.h        |    1 +
 include/net/ipv6.h             |   12 ++++++++++++
 net/ipv4/af_inet.c             |    9 +++++++--
 4 files changed, 24 insertions(+), 6 deletions(-)

--- a/include/net/inet6_hashtables.h
+++ b/include/net/inet6_hashtables.h
@@ -28,16 +28,16 @@
 
 struct inet_hashinfo;
 
-/* I have no idea if this is a good hash for v6 or not. -DaveM */
 static inline unsigned int inet6_ehashfn(struct net *net,
 				const struct in6_addr *laddr, const u16 lport,
 				const struct in6_addr *faddr, const __be16 fport)
 {
-	u32 ports = (lport ^ (__force u16)fport);
+	u32 ports = (((u32)lport) << 16) | (__force u32)fport;
 
 	return jhash_3words((__force u32)laddr->s6_addr32[3],
-			    (__force u32)faddr->s6_addr32[3],
-			    ports, inet_ehash_secret + net_hash_mix(net));
+			    ipv6_addr_jhash(faddr),
+			    ports,
+			    inet_ehash_secret + net_hash_mix(net));
 }
 
 static inline int inet6_sk_ehashfn(const struct sock *sk)
--- a/include/net/inet_sock.h
+++ b/include/net/inet_sock.h
@@ -202,6 +202,7 @@ static inline void inet_sk_copy_descenda
 extern int inet_sk_rebuild_header(struct sock *sk);
 
 extern u32 inet_ehash_secret;
+extern u32 ipv6_hash_secret;
 extern void build_ehash_secret(void);
 
 static inline unsigned int inet_ehashfn(struct net *net,
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -15,6 +15,7 @@
 
 #include <linux/ipv6.h>
 #include <linux/hardirq.h>
+#include <linux/jhash.h>
 #include <net/if_inet6.h>
 #include <net/ndisc.h>
 #include <net/flow.h>
@@ -390,6 +391,17 @@ struct ip6_create_arg {
 void ip6_frag_init(struct inet_frag_queue *q, void *a);
 int ip6_frag_match(struct inet_frag_queue *q, void *a);
 
+/* more secured version of ipv6_addr_hash() */
+static inline u32 ipv6_addr_jhash(const struct in6_addr *a)
+{
+	u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
+
+	return jhash_3words(v,
+			    (__force u32)a->s6_addr32[2],
+			    (__force u32)a->s6_addr32[3],
+			    ipv6_hash_secret);
+}
+
 static inline int ipv6_addr_any(const struct in6_addr *a)
 {
 	return (a->s6_addr32[0] | a->s6_addr32[1] |
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -227,8 +227,12 @@ EXPORT_SYMBOL(inet_listen);
 u32 inet_ehash_secret __read_mostly;
 EXPORT_SYMBOL(inet_ehash_secret);
 
+u32 ipv6_hash_secret __read_mostly;
+EXPORT_SYMBOL(ipv6_hash_secret);
+
 /*
- * inet_ehash_secret must be set exactly once
+ * inet_ehash_secret must be set exactly once, and to a non nul value
+ * ipv6_hash_secret must be set exactly once.
  */
 void build_ehash_secret(void)
 {
@@ -238,7 +242,8 @@ void build_ehash_secret(void)
 		get_random_bytes(&rnd, sizeof(rnd));
 	} while (rnd == 0);
 
-	cmpxchg(&inet_ehash_secret, 0, rnd);
+	if (cmpxchg(&inet_ehash_secret, 0, rnd) == 0)
+		get_random_bytes(&ipv6_hash_secret, sizeof(ipv6_hash_secret));
 }
 EXPORT_SYMBOL(build_ehash_secret);
 



  parent reply	other threads:[~2013-02-27  0:12 UTC|newest]

Thread overview: 98+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-02-27  0:07 [ 00/86] 3.4.34-stable review Greg Kroah-Hartman
2013-02-27  0:07 ` [ 01/86] x86-32, mm: Rip out x86_32 NUMA remapping code Greg Kroah-Hartman
2013-02-27  0:07 ` [ 02/86] x86-32, mm: Remove reference to resume_map_numa_kva() Greg Kroah-Hartman
2013-02-27  0:07 ` [ 03/86] x86-32, mm: Remove reference to alloc_remap() Greg Kroah-Hartman
2013-02-27  0:07 ` [ 04/86] mm: fix pageblock bitmap allocation Greg Kroah-Hartman
2013-02-27  0:07 ` [ 05/86] timeconst.pl: Eliminate Perl warning Greg Kroah-Hartman
2013-02-27  0:07 ` [ 06/86] genirq: Avoid deadlock in spurious handling Greg Kroah-Hartman
2013-02-27  0:07 ` [ 07/86] posix-cpu-timers: Fix nanosleep task_struct leak Greg Kroah-Hartman
2013-02-27  0:07 ` [ 08/86] hrtimer: Prevent hrtimer_enqueue_reprogram race Greg Kroah-Hartman
2013-02-27  0:07 ` [ 09/86] x86: Hyper-V: register clocksource only if its advertised Greg Kroah-Hartman
2013-02-27  0:07 ` [ 10/86] ALSA: ali5451: remove irq enabling in pointer callback Greg Kroah-Hartman
2013-02-27  0:07 ` [ 11/86] ALSA: rme32.c irq enabling after spin_lock_irq Greg Kroah-Hartman
2013-02-27  0:07 ` [ 12/86] tty: Prevent deadlock in n_gsm driver Greg Kroah-Hartman
2013-02-27  0:07 ` [ 13/86] tty: set_termios/set_termiox should not return -EINTR Greg Kroah-Hartman
2013-02-27  0:07 ` [ 14/86] USB: serial: fix null-pointer dereferences on disconnect Greg Kroah-Hartman
2013-02-27  0:07 ` [ 15/86] b43: Increase number of RX DMA slots Greg Kroah-Hartman
2013-02-27  0:07 ` [ 16/86] rtlwifi: rtl8192cu: Add new USB ID Greg Kroah-Hartman
2013-02-27  0:07 ` [ 17/86] rtlwifi: usb: allocate URB control message setup_packet and data buffer separately Greg Kroah-Hartman
2013-02-27  0:07 ` [ 18/86] xen: Send spinlock IPI to all waiters Greg Kroah-Hartman
2013-02-27  0:07 ` [ 19/86] xen: close evtchn port if binding to irq fails Greg Kroah-Hartman
2013-02-27  0:07 ` [ 20/86] Driver core: treat unregistered bus_types as having no devices Greg Kroah-Hartman
2013-02-27  0:07 ` [ 21/86] mm: mmu_notifier: have mmu_notifiers use a global SRCU so they may safely schedule Greg Kroah-Hartman
2013-02-27  0:07 ` [ 22/86] mm: mmu_notifier: make the mmu_notifier srcu static Greg Kroah-Hartman
2013-02-27  0:07 ` [ 23/86] mmu_notifier_unregister NULL Pointer deref and multiple ->release() callouts Greg Kroah-Hartman
2013-02-27  0:07 ` [ 24/86] KVM: s390: Handle hosts not supporting s390-virtio Greg Kroah-Hartman
2013-02-27  0:07 ` [ 25/86] s390/kvm: Fix store status for ACRS/FPRS Greg Kroah-Hartman
2013-02-27  0:07 ` [ 26/86] futex: Revert "futex: Mark get_robust_list as deprecated" Greg Kroah-Hartman
2013-02-27  0:07 ` [ 27/86] inotify: remove broken mask checks causing unmount to be EINVAL Greg Kroah-Hartman
2013-02-27  0:07 ` [ 28/86] fs/block_dev.c: page cache wrongly left invalidated after revalidate_disk() Greg Kroah-Hartman
2013-02-27  0:07 ` [ 29/86] ocfs2: unlock super lock if lockres refresh failed Greg Kroah-Hartman
2013-02-27  0:07 ` [ 30/86] drivers/video/backlight/adp88?0_bl.c: fix resume Greg Kroah-Hartman
2013-02-27  0:07 ` [ 31/86] tmpfs: fix use-after-free of mempolicy object Greg Kroah-Hartman
2013-02-27  0:07 ` [ 32/86] mm/fadvise.c: drain all pagevecs if POSIX_FADV_DONTNEED fails to discard all pages Greg Kroah-Hartman
2013-02-27  0:07 ` [ 33/86] drivercore: Fix ordering between deferred_probe and exiting initcalls Greg Kroah-Hartman
2013-02-27  0:07 ` [ 34/86] umount oops when remove blocklayoutdriver first Greg Kroah-Hartman
2013-02-27  0:07 ` [ 35/86] NLM: Ensure that we resend all pending blocking locks after a reclaim Greg Kroah-Hartman
2013-02-27  0:07 ` [ 36/86] p54usb: corrected USB ID for T-Com Sinus 154 data II Greg Kroah-Hartman
2013-02-27  0:07 ` [ 37/86] ALSA: usb-audio: fix Roland A-PRO support Greg Kroah-Hartman
2013-02-27  0:07 ` [ 38/86] ALSA: usb: Fix Processing Unit Descriptor parsers Greg Kroah-Hartman
2013-02-27  0:07 ` [ 39/86] ALSA: hda - Release assigned pin/cvt at error path of hdmi_pcm_open() Greg Kroah-Hartman
2013-02-27  0:07 ` [ 40/86] ALSA: hda - Workaround for silent output on Sony Vaio VGC-LN51JGB with ALC889 Greg Kroah-Hartman
2013-02-27  0:07 ` [ 41/86] ALSA: hda - hdmi: ELD shouldnt be valid after unplug Greg Kroah-Hartman
2013-02-27  0:07 ` [ 42/86] sunvdc: Fix off-by-one in generic_request() Greg Kroah-Hartman
2013-02-27  0:07 ` [ 43/86] drm/radeon/dce6: fix display powergating Greg Kroah-Hartman
2013-02-27  0:07 ` [ 44/86] drm/udl: make usage as a console safer Greg Kroah-Hartman
2013-02-27  0:07 ` [ 45/86] drm/udl: disable fb_defio by default Greg Kroah-Hartman
2013-02-27  0:07 ` [ 46/86] vgacon/vt: clear buffer attributes when we load a 512 character font (v2) Greg Kroah-Hartman
2013-02-27  0:07 ` [ 47/86] drm: dont add inferred modes for monitors that dont support them Greg Kroah-Hartman
2013-02-27  0:07 ` [ 48/86] drm: Fill depth/bits_per_pixel for C8 format Greg Kroah-Hartman
2013-02-27  0:07 ` [ 49/86] drm: Use C8 instead of RGB332 when determining the format from depth/bpp Greg Kroah-Hartman
2013-02-27  0:07 ` [ 50/86] drm/usb: bind driver to correct device Greg Kroah-Hartman
2013-02-27  0:07 ` [ 51/86] target: Fix divide by zero bug in fabric_max_sectors for unconfigured devices Greg Kroah-Hartman
2013-02-27  0:08 ` [ 52/86] intel/iommu: force writebuffer-flush quirk on Gen 4 Chipsets Greg Kroah-Hartman
2013-02-27  0:08 ` [ 53/86] drm/i915: disable shared panel fitter for pipe Greg Kroah-Hartman
2013-02-27  0:08 ` [ 54/86] drm/i915: Set i9xx sdvo clock limits according to specifications Greg Kroah-Hartman
2013-02-27 10:11   ` Patrik Jakobsson
2013-02-27 10:19     ` Chris Wilson
2013-02-27 10:23       ` Patrik Jakobsson
2013-02-27  0:08 ` [ 55/86] staging: comedi: disallow COMEDI_DEVCONFIG on non-board minors Greg Kroah-Hartman
2013-02-27  0:08 ` [ 56/86] staging: vt6656: Fix URB submitted while active warning Greg Kroah-Hartman
2013-02-27  0:08 ` [ 57/86] ASoC: wm2200: correct IN2L and IN3L digital mute Greg Kroah-Hartman
2013-02-27  0:08 ` [ 58/86] ARM: PXA3xx: program the CSMSADRCFG register Greg Kroah-Hartman
2013-02-27  0:08 ` [ 59/86] ARM: samsung: fix assembly syntax for new gas Greg Kroah-Hartman
2013-02-27  0:08 ` [ 60/86] ARM: 7643/1: sched: correct update_sched_clock() Greg Kroah-Hartman
2013-02-27  0:08 ` [ 61/86] powerpc/kexec: Disable hard IRQ before kexec Greg Kroah-Hartman
2013-02-27  0:08 ` [ 62/86] [PARISC] Purge existing TLB entries in set_pte_at and ptep_set_wrprotect Greg Kroah-Hartman
2013-02-27  0:08 ` [ 63/86] pcmcia/vrc4171: Add missing spinlock init Greg Kroah-Hartman
2013-02-27  0:08 ` [ 64/86] drivers/video: fsl-diu-fb: fix pixel formats for 24 and 16 bpp Greg Kroah-Hartman
2013-02-27  0:08 ` [ 65/86] fbcon: dont lose the console font across generic->chip driver switch Greg Kroah-Hartman
2013-02-27  0:08 ` [ 66/86] fb: rework locking to fix lock ordering on takeover Greg Kroah-Hartman
2013-02-27  0:08 ` [ 67/86] fb: Yet another band-aid for fixing lockdep mess Greg Kroah-Hartman
2013-02-27  0:08 ` [ 68/86] mmc: sdhci-esdhc-imx: fix host version read Greg Kroah-Hartman
2013-02-27  0:08 ` [ 69/86] HID: wiimote: fix nunchuck button parser Greg Kroah-Hartman
2013-02-27  0:08 ` [ 70/86] bridge: set priority of STP packets Greg Kroah-Hartman
2013-02-27  0:08 ` [ 71/86] net: fix infinite loop in __skb_recv_datagram() Greg Kroah-Hartman
2013-02-27  0:08 ` [ 72/86] xen-netback: correctly return errors from netbk_count_requests() Greg Kroah-Hartman
2013-02-27  0:08 ` [ 73/86] xen-netback: cancel the credit timer when taking the vif down Greg Kroah-Hartman
2013-02-27  0:08 ` [ 74/86] net: fix a compile error when SOCK_REFCNT_DEBUG is enabled Greg Kroah-Hartman
2013-02-27  0:08 ` [ 75/86] ipv4: fix a bug in ping_err() Greg Kroah-Hartman
2013-02-27  0:08 ` Greg Kroah-Hartman [this message]
2013-02-27  0:08 ` [ 77/86] sock_diag: Fix out-of-bounds access to sock_diag_handlers[] Greg Kroah-Hartman
2013-02-27  0:08 ` [ 78/86] vlan: adjust vlan_set_encap_proto() for its callers Greg Kroah-Hartman
2013-02-27  0:08 ` [ 79/86] USB: ehci-omap: Dont free gpios that we didnt request Greg Kroah-Hartman
2013-02-27  7:52   ` Roger Quadros
2013-02-27 17:20     ` Luis Henriques
2013-02-28 10:16       ` Roger Quadros
2013-02-28 10:37         ` Luis Henriques
2013-02-27 17:37     ` Greg Kroah-Hartman
2013-02-27  0:08 ` [ 80/86] dca: check against empty dca_domains list before unregister provider Greg Kroah-Hartman
2013-02-27  0:08 ` [ 81/86] USB: option: add and update Alcatel modems Greg Kroah-Hartman
2013-02-27  0:08 ` [ 82/86] USB: option: add Yota / Megafon M100-1 4g modem Greg Kroah-Hartman
2013-02-27  0:08 ` [ 83/86] USB: option: add Huawei "ACM" devices using protocol = vendor Greg Kroah-Hartman
2013-02-27  0:08 ` [ 84/86] USB: ehci-omap: Fix autoloading of module Greg Kroah-Hartman
2013-02-27  0:08 ` [ 85/86] USB: storage: properly handle the endian issues of idProduct Greg Kroah-Hartman
2013-02-27  0:08 ` [ 86/86] USB: usb-storage: unusual_devs update for Super TOP SATA bridge Greg Kroah-Hartman
2013-02-27 16:50 ` [ 00/86] 3.4.34-stable review Shuah Khan
2013-02-28 14:54 ` Satoru Takeuchi
2013-02-28 14:59   ` Greg Kroah-Hartman

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=20130226235920.825016157@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=ncardwell@google.com \
    --cc=stable@vger.kernel.org \
    --cc=ycheng@google.com \
    /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