* Re: Off-by-one error in net/8021q/vlan.c
From: Michał Mirosław @ 2011-02-16 18:41 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Phil Karn, richard -rw- weinberger, kaber, netdev
In-Reply-To: <1297874372.30541.29.camel@edumazet-laptop>
2011/2/16 Eric Dumazet <eric.dumazet@gmail.com>:
> Le mercredi 16 février 2011 à 08:28 -0800, Phil Karn a écrit :
>> On 2/16/11 8:10 AM, richard -rw- weinberger wrote:
>> > On Wed, Feb 16, 2011 at 4:58 PM, Phil Karn <karn@ka9q.net> wrote:
>> >> On 2/16/11 4:51 AM, richard -rw- weinberger wrote:
>> >>> On Wed, Feb 16, 2011 at 11:58 AM, Phil Karn <karn@ka9q.net> wrote:
>> >>>> The range check on vlan_id in register_vlan_device is off by one, and it
>> >>>> prevents the creation of a vlan interface for vlan ID 4095. (OSX allows
>> >>>> this, I checked.)
>> >>>
>> >>> Then OSX should fix their code. 4095 is reserved.
>> >> If it's reserved, then it's up to the user to reserve it.
>> > No.
>> > See:
>> > http://standards.ieee.org/getieee802/download/802.1Q-2005.pdf
>> Well, then I guess we all know better than the user. That's the Windows
>> Way...no, wait, I thought this is Linux.
>>
>> The fact is that I did encounter a misconfigured switch using vlan 4095,
>> and because of this off-by-one error I was unable to talk to it and fix it.
>>
>> I was hoping I wouldn't have to patch every new kernel I install.
> You can use an OSX gateway ;)
>
> If we allow ID 4095, then some users will complain we violate rules.
>
> Really you cannot push this patch in official kernel only to ease your
> life ;)
The idea is that you don't have to use ID 4095 and if you don't -
nothing's broken by just allowing it. The same goes with ID 0 - it's
defined to be 802.1p packet, but people do use it as normal VLAN
(especially with hardware that can cope with only small number of
VLANs at once).
Allowing it but with a big fat warning in logs is even better: "You
want your network broken? Sure, can do, but you have been warned."
Best Regards,
Michał Mirosław
^ permalink raw reply
* [PATCH] bonding: added 802.3ad round-robin hashing policy and source mac selection mode
From: Oleg V. Ukhno @ 2011-02-16 19:13 UTC (permalink / raw)
To: netdev; +Cc: Jay Vosburgh, David S. Miller
Patch introduces two new (related) features to bonding module.
First feature is round-robin hashing policy, which is primarily
intended for use with 802.3ad mode, and puts every next IPv4 and
IPv6 packet into next availables slave without taling into account
which layer3 and above protocol is used.
Second feature makes possible choosing which MAC-address will be set
in the transmitted packet - when set to src-mac it will force setting
slave's interface real MAC address as source MAC address in every
packet, sent via this slave interface.
Main goal of this patch is to make possible single TCP stream
equally striped for both transmitted and received packets over all
available slaves.
This operating mode is not fully 802.3ad compliant, and will cause
some packet reordering in TCP stream, to some kernel tuning may be
required.
For correct working enabling round-robin hashing policy plus using
real slave's MAC addresses as source MAC addresses in transmitted
packets requires specific switch setting)hashing mode for port-channel
("etherchannel) should be set to src-mac or src-dst-mac to get
correct load-striping on the receiving host's etherchannel.
General requirements for using bonding in this operating mode are:
- even and preferrably equal number of slaves on sending and receiving
hosts;
- equal RTT between sending and receiving hosts on all slaves;
- switch capable of doing etherchannels and using src-mac or src-dst-mac
hashing policy for egress load striping
Signed-off-by: Oleg V. Ukhno <olegu@yandex-team.ru>
---
Documentation/networking/bonding.txt | 65 +++++++++++++++++++++++++++++++++++
drivers/net/bonding/bond_3ad.c | 2 -
drivers/net/bonding/bond_main.c | 60 +++++++++++++++++++++++++++++---
drivers/net/bonding/bond_sysfs.c | 50 ++++++++++++++++++++++++++
drivers/net/bonding/bonding.h | 7 +++
include/linux/if_bonding.h | 1
6 files changed, 178 insertions(+), 7 deletions(-)
diff -uprN -X linux-2.6/Documentation/dontdiff linux-2.6/Documentation/networking/bonding.txt linux-2.6.p/Documentation/networking/bonding.txt
--- linux-2.6/Documentation/networking/bonding.txt 2011-02-08 16:03:01.290281998 +0300
+++ linux-2.6.p/Documentation/networking/bonding.txt 2011-02-16 22:03:09.650281997 +0300
@@ -83,6 +83,7 @@ Table of Contents
12. Configuring Bonding for Maximum Throughput
12.1 Maximum Throughput in a Single Switch Topology
12.1.1 MT Bonding Mode Selection for Single Switch Topology
+12.1.1.1 Maximizing TCP Throughput for RX/TX for Single Switch Topology using layer2 mechanisms
12.1.2 MT Link Monitoring for Single Switch Topology
12.2 Maximum Throughput in a Multiple Switch Topology
12.2.1 MT Bonding Mode Selection for Multiple Switch Topology
@@ -761,6 +762,34 @@ xmit_hash_policy
conversations. Other implementations of 802.3ad may
or may not tolerate this noncompliance.
+ round-robin
+
+ This policy simply puts every next packet into next
+ slave interfaces, providing round-robin load striping
+ for transmitted data. This policy can be enabled with
+ any mode which supports choosing alternate hash policy,
+ but was initially done for 802.3ad mode.
+
+ Main goal for this policy is to stripe TX load without
+ taking into account which layer3 protocol is used, and
+ can be used for single TCP connection load striping. When
+ enabled, it will round-robin packets for IPv4 and IPv6
+ only.
+
+ There is also src_mac_select option, which can be used
+ to configure RX load-striping using switch hashing
+ algorhytms on the receiving side. See detailed description
+ below.
+
+ It is important to understand, that this hashing policy
+ will possibly cause TCP out-of-order packets when enabled
+ and must not be used when slaves have different bandwidth
+ and/or RTT in receiver's direction. This algorithm is not
+ fully 802.3ad compliant. Some implementations of 802.3ad
+ may or may not tolerate this noncompliance.
+
+ Hashing formula is transmitted packet number % slave count.
+
The default value is layer2. This option was added in bonding
version 2.6.3. In earlier versions of bonding, this parameter
does not exist, and the layer2 policy is the only policy. The
@@ -2190,6 +2219,42 @@ balance-alb: This mode is everything tha
device driver must support changing the hardware address while
the device is open.
+12.1.1.1 Maximizing TCP Throughput for RX/TX for Single Switch Topology
+ using layer2 mechanisms
+----------------------------------------------------------------------
+ Besides of methods of load striping and configuring HA, mentioned
+above, you can use round-robin hashing policy and src_mac_select "slave-src"
+setting to stripe TCP load near-equally over even number of slaves. Please
+note, that enabling round-robin policy for balance-xor mode should turn it
+into mode similar to balance-rr mode.
+ There is also specific switch configuration required to use all
+benefits of both round-robin hashing policy and src_mac_select "slave-src"
+setting.
+ When you enable round-robin xmit hashing policy plus set
+src_mac_select to slave-src mode, you will get every next packet
+transmitted over a new slave with every's packet source MAC address set
+to real MAC address of the according slave interface, not the aggregate
+interface.
+ Imagine, that you have two hosts(let's say A and B), each connected
+using 2 slave interfaces to switch with appropriate port-channels configured
+("etherchannels"). After you start transmitting TCP data from A to B, and
+round-robin hashing policy is enabled, you will see that TX load is equally
+striped over host A slaves, but all this traffic is received with only one
+machine's B slave.
+ Now, you set src_mac_select parameter to "slave-src" and
+configure switch for src-mac hashing for "outqoing" etherchannel load
+striping. Now every packet sent from host A has slave's MAC as source MAC
+address, and switch will send every packet from host A into receiving
+port-channel of host B taking into account source MAC address of packet being
+put into, so you will get near-equal RX load striping, which does not depend
+on layer3 and above protocols used for data transmission.
+ It is important to understand, that this load striping mode
+will only work correctly if number of slaves on each side is at least
+even, and preferrably equal and even.
+ This load striping mode also can cause TCP out-of-order packets,
+so you may need to tune your kernel for handling increased number of
+reordered packets.
+
12.1.2 MT Link Monitoring for Single Switch Topology
----------------------------------------------------
diff -uprN -X linux-2.6/Documentation/dontdiff linux-2.6/drivers/net/bonding/bond_3ad.c linux-2.6.p/drivers/net/bonding/bond_3ad.c
--- linux-2.6/drivers/net/bonding/bond_3ad.c 2011-02-16 00:59:18.710282002 +0300
+++ linux-2.6.p/drivers/net/bonding/bond_3ad.c 2011-02-16 01:30:47.770281998 +0300
@@ -2419,7 +2419,7 @@ int bond_3ad_xmit_xor(struct sk_buff *sk
goto out;
}
- slave_agg_no = bond->xmit_hash_policy(skb, slaves_in_agg);
+ slave_agg_no = bond->xmit_hash_policy(skb, slaves_in_agg, bond->rr_tx_counter++);
bond_for_each_slave(bond, slave, i) {
struct aggregator *agg = SLAVE_AD_INFO(slave).port.aggregator;
diff -uprN -X linux-2.6/Documentation/dontdiff linux-2.6/drivers/net/bonding/bonding.h linux-2.6.p/drivers/net/bonding/bonding.h
--- linux-2.6/drivers/net/bonding/bonding.h 2011-02-16 00:59:18.720282002 +0300
+++ linux-2.6.p/drivers/net/bonding/bonding.h 2011-02-16 01:33:11.610282004 +0300
@@ -162,6 +162,7 @@ struct bond_params {
int tx_queues;
int all_slaves_active;
int resend_igmp;
+ int src_mac_select;
};
struct bond_parm_tbl {
@@ -235,7 +236,7 @@ struct bonding {
#endif /* CONFIG_PROC_FS */
struct list_head bond_list;
struct netdev_hw_addr_list mc_list;
- int (*xmit_hash_policy)(struct sk_buff *, int);
+ int (*xmit_hash_policy)(struct sk_buff *, int, int);
__be32 master_ip;
u16 flags;
u16 rr_tx_counter;
@@ -308,6 +309,9 @@ static inline bool bond_is_lb(const stru
#define BOND_ARP_VALIDATE_ALL (BOND_ARP_VALIDATE_ACTIVE | \
BOND_ARP_VALIDATE_BACKUP)
+#define BOND_MAC_SRC_DEFAULT 0
+#define BOND_MAC_SRC_SLAVE 1
+
static inline int slave_do_arp_validate(struct bonding *bond,
struct slave *slave)
{
@@ -402,6 +406,7 @@ extern const struct bond_parm_tbl arp_va
extern const struct bond_parm_tbl fail_over_mac_tbl[];
extern const struct bond_parm_tbl pri_reselect_tbl[];
extern struct bond_parm_tbl ad_select_tbl[];
+extern const struct bond_parm_tbl src_mac_select_tbl[];
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
void bond_send_unsolicited_na(struct bonding *bond);
diff -uprN -X linux-2.6/Documentation/dontdiff linux-2.6/drivers/net/bonding/bond_main.c linux-2.6.p/drivers/net/bonding/bond_main.c
--- linux-2.6/drivers/net/bonding/bond_main.c 2011-02-16 00:59:18.720282002 +0300
+++ linux-2.6.p/drivers/net/bonding/bond_main.c 2011-02-16 22:08:22.650281997 +0300
@@ -111,6 +111,7 @@ static char *fail_over_mac;
static int all_slaves_active = 0;
static struct bond_params bonding_defaults;
static int resend_igmp = BOND_DEFAULT_RESEND_IGMP;
+static char *src_mac_select;
module_param(max_bonds, int, 0);
MODULE_PARM_DESC(max_bonds, "Max number of bonded devices");
@@ -152,7 +153,7 @@ module_param(ad_select, charp, 0);
MODULE_PARM_DESC(ad_select, "803.ad aggregation selection logic: stable (0, default), bandwidth (1), count (2)");
module_param(xmit_hash_policy, charp, 0);
MODULE_PARM_DESC(xmit_hash_policy, "XOR hashing method: 0 for layer 2 (default)"
- ", 1 for layer 3+4");
+ ", 1 for layer 3+4, 3 for round-robin");
module_param(arp_interval, int, 0);
MODULE_PARM_DESC(arp_interval, "arp interval in milliseconds");
module_param_array(arp_ip_target, charp, NULL, 0);
@@ -167,6 +168,9 @@ MODULE_PARM_DESC(all_slaves_active, "Kee
"0 for never (default), 1 for always.");
module_param(resend_igmp, int, 0);
MODULE_PARM_DESC(resend_igmp, "Number of IGMP membership reports to send on link failure");
+module_param(src_mac_select, charp, 0);
+MODULE_PARM_DESC(src_mac_select, "Source MAC selection mode: 0 or default (default),"
+ "1 or slave-src to use slave's MAC as packet's src MAC");
/*----------------------------- Global variables ----------------------------*/
@@ -206,6 +210,7 @@ const struct bond_parm_tbl xmit_hashtype
{ "layer2", BOND_XMIT_POLICY_LAYER2},
{ "layer3+4", BOND_XMIT_POLICY_LAYER34},
{ "layer2+3", BOND_XMIT_POLICY_LAYER23},
+{ "round-robin", BOND_XMIT_POLICY_LAYERRR},
{ NULL, -1},
};
@@ -238,6 +243,12 @@ struct bond_parm_tbl ad_select_tbl[] = {
{ NULL, -1},
};
+const struct bond_parm_tbl src_mac_select_tbl[] = {
+{ "default", BOND_MAC_SRC_DEFAULT},
+{ "slave-src", BOND_MAC_SRC_SLAVE},
+{ NULL, -1},
+};
+
/*-------------------------- Forward declarations ---------------------------*/
static void bond_send_gratuitous_arp(struct bonding *bond);
@@ -422,6 +433,7 @@ struct vlan_entry *bond_next_vlan(struct
int bond_dev_queue_xmit(struct bonding *bond, struct sk_buff *skb,
struct net_device *slave_dev)
{
+ struct ethhdr *eth_data;
skb->dev = slave_dev;
skb->priority = 1;
#ifdef CONFIG_NET_POLL_CONTROLLER
@@ -433,6 +445,15 @@ int bond_dev_queue_xmit(struct bonding *
slave_dev->priv_flags &= ~IFF_IN_NETPOLL;
} else
#endif
+ if (bond->params.src_mac_select == BOND_MAC_SRC_SLAVE &&
+ (skb->protocol == htons(ETH_P_IP) ||
+ skb->protocol == htons(ETH_P_IPV6))) {
+ skb_reset_mac_header(skb);
+ eth_data = eth_hdr(skb);
+ memcpy(eth_data->h_source, slave_dev->perm_addr,
+ ETH_ALEN);
+ }
+
dev_queue_xmit(skb);
return 0;
@@ -3261,6 +3282,13 @@ static void bond_info_show_master(struct
bond->params.xmit_policy);
}
+ if (bond->params.src_mac_select == BOND_MAC_SRC_DEFAULT ||
+ bond->params.src_mac_select == BOND_MAC_SRC_DEFAULT) {
+ seq_printf(seq, "Source MAC select is: %s (%d)\n",
+ src_mac_select_tbl[bond->params.src_mac_select].modename,
+ bond->params.src_mac_select);
+ }
+
if (USES_PRIMARY(bond->params.mode)) {
seq_printf(seq, "Primary Slave: %s",
(bond->primary_slave) ?
@@ -3717,7 +3745,8 @@ void bond_unregister_arp(struct bonding
* Hash for the output device based upon layer 2 and layer 3 data. If
* the packet is not IP mimic bond_xmit_hash_policy_l2()
*/
-static int bond_xmit_hash_policy_l23(struct sk_buff *skb, int count)
+static int bond_xmit_hash_policy_l23(struct sk_buff *skb, int count,
+ int pktcount)
{
struct ethhdr *data = (struct ethhdr *)skb->data;
struct iphdr *iph = ip_hdr(skb);
@@ -3735,7 +3764,8 @@ static int bond_xmit_hash_policy_l23(str
* the packet is a frag or not TCP or UDP, just use layer 3 data. If it is
* altogether not IP, mimic bond_xmit_hash_policy_l2()
*/
-static int bond_xmit_hash_policy_l34(struct sk_buff *skb, int count)
+static int bond_xmit_hash_policy_l34(struct sk_buff *skb, int count,
+ int pktcount)
{
struct ethhdr *data = (struct ethhdr *)skb->data;
struct iphdr *iph = ip_hdr(skb);
@@ -3759,13 +3789,29 @@ static int bond_xmit_hash_policy_l34(str
/*
* Hash for the output device based upon layer 2 data
*/
-static int bond_xmit_hash_policy_l2(struct sk_buff *skb, int count)
+static int bond_xmit_hash_policy_l2(struct sk_buff *skb, int count,
+ int pktcount)
{
struct ethhdr *data = (struct ethhdr *)skb->data;
return (data->h_dest[5] ^ data->h_source[5]) % count;
}
+/*
+ * Round-robin over all active slaves(one packet per slave) for IP and IPv6,
+ * otherwise mimic bond_xmit_hash_policy_l2()
+ */
+static int bond_xmit_hash_policy_rr(struct sk_buff *skb, int count,
+ int pktcount)
+{
+ struct ethhdr *data = (struct ethhdr *)skb->data;
+ if (skb->protocol == htons(ETH_P_IP)
+ || skb->protocol == htons(ETH_P_IPV6)) {
+ return pktcount % count;
+ }
+ return (data->h_dest[5] ^ data->h_source[5]) % count;
+}
+
/*-------------------------- Device entry points ----------------------------*/
static int bond_open(struct net_device *bond_dev)
@@ -4395,7 +4441,8 @@ static int bond_xmit_xor(struct sk_buff
if (!BOND_IS_OK(bond))
goto out;
- slave_no = bond->xmit_hash_policy(skb, bond->slave_cnt);
+ slave_no = bond->xmit_hash_policy(skb, bond->slave_cnt,
+ bond->rr_tx_counter++);
bond_for_each_slave(bond, slave, i) {
slave_no--;
@@ -4492,6 +4539,9 @@ static void bond_set_xmit_hash_policy(st
case BOND_XMIT_POLICY_LAYER34:
bond->xmit_hash_policy = bond_xmit_hash_policy_l34;
break;
+ case BOND_XMIT_POLICY_LAYERRR:
+ bond->xmit_hash_policy = bond_xmit_hash_policy_rr;
+ break;
case BOND_XMIT_POLICY_LAYER2:
default:
bond->xmit_hash_policy = bond_xmit_hash_policy_l2;
diff -uprN -X linux-2.6/Documentation/dontdiff linux-2.6/drivers/net/bonding/bond_sysfs.c linux-2.6.p/drivers/net/bonding/bond_sysfs.c
--- linux-2.6/drivers/net/bonding/bond_sysfs.c 2011-02-08 16:03:02.950282003 +0300
+++ linux-2.6.p/drivers/net/bonding/bond_sysfs.c 2011-02-16 02:05:58.650281999 +0300
@@ -1643,6 +1643,55 @@ out:
static DEVICE_ATTR(resend_igmp, S_IRUGO | S_IWUSR,
bonding_show_resend_igmp, bonding_store_resend_igmp);
+/*
+ * Show and set the bonding src_mac_select param.
+ */
+
+static ssize_t bonding_show_src_mac_select(struct device *d,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct bonding *bond = to_bond(d);
+
+ return sprintf(buf, "%s %d\n",
+ src_mac_select_tbl[bond->params.src_mac_select].modename,
+ bond->params.src_mac_select);
+}
+
+static ssize_t bonding_store_src_mac_select(struct device *d,
+ struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ int new_value, ret = count;
+ struct bonding *bond = to_bond(d);
+
+ if (bond->dev->flags & IFF_UP) {
+ pr_err("%s: Interface is up. Unable to update src mac select policy.\n",
+ bond->dev->name);
+ ret = -EPERM;
+ goto out;
+ }
+
+ new_value = bond_parse_parm(buf, src_mac_select_tbl);
+ if (new_value < 0) {
+ pr_err("%s: Ignoring invalid src mac select policy value %.*s.\n",
+ bond->dev->name,
+ (int)strlen(buf) - 1, buf);
+ ret = -EINVAL;
+ goto out;
+ } else {
+ bond->params.src_mac_select = new_value;
+ pr_info("%s: setting src mac select policy to %s (%d).\n",
+ bond->dev->name,
+ src_mac_select_tbl[new_value].modename, new_value);
+ }
+out:
+ return ret;
+}
+
+static DEVICE_ATTR(src_mac_select, S_IRUGO | S_IWUSR,
+ bonding_show_src_mac_select, bonding_store_src_mac_select);
+
static struct attribute *per_bond_attrs[] = {
&dev_attr_slaves.attr,
&dev_attr_mode.attr,
@@ -1671,6 +1720,7 @@ static struct attribute *per_bond_attrs[
&dev_attr_queue_id.attr,
&dev_attr_all_slaves_active.attr,
&dev_attr_resend_igmp.attr,
+ &dev_attr_src_mac_select.attr,
NULL,
};
diff -uprN -X linux-2.6/Documentation/dontdiff linux-2.6/include/linux/if_bonding.h linux-2.6.p/include/linux/if_bonding.h
--- linux-2.6/include/linux/if_bonding.h 2011-02-16 00:59:18.720282002 +0300
+++ linux-2.6.p/include/linux/if_bonding.h 2011-02-16 01:23:38.660282000 +0300
@@ -91,6 +91,7 @@
#define BOND_XMIT_POLICY_LAYER2 0 /* layer 2 (MAC only), default */
#define BOND_XMIT_POLICY_LAYER34 1 /* layer 3+4 (IP ^ (TCP || UDP)) */
#define BOND_XMIT_POLICY_LAYER23 2 /* layer 2+3 (IP ^ MAC) */
+#define BOND_XMIT_POLICY_LAYERRR 3 /* round-robin mode */
typedef struct ifbond {
__s32 bond_mode;
^ permalink raw reply
* Re: [PATCH -next] PCI: fix tlan build when CONFIG_PCI is not enabled
From: Jesse Barnes @ 2011-02-16 19:46 UTC (permalink / raw)
To: Randy Dunlap
Cc: Stephen Rothwell, netdev, linux-pci, linux-next, LKML, davem,
Sakari Ailus
In-Reply-To: <20110214122750.e1e03bc8.randy.dunlap@oracle.com>
On Mon, 14 Feb 2011 12:27:50 -0800
Randy Dunlap <randy.dunlap@oracle.com> wrote:
> From: Randy Dunlap <randy.dunlap@oracle.com>
>
> When CONFIG_PCI is not enabled, tlan.c has a build error:
> drivers/net/tlan.c:503: error: implicit declaration of function 'pci_wake_from_d3'
>
> so add an inline function stub for this function to pci.h when
> PCI is not enabled, similar to other stubbed PCI functions.
>
> Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
> Cc: Sakari Ailus <sakari.ailus@iki.fi>
> ---
> include/linux/pci.h | 5 +++++
> 1 file changed, 5 insertions(+)
>
> --- linux-next-20110214.orig/include/linux/pci.h
> +++ linux-next-20110214/include/linux/pci.h
> @@ -1191,6 +1191,11 @@ static inline int pci_set_power_state(st
> return 0;
> }
>
> +static inline int pci_wake_from_d3(struct pci_dev *dev, bool enable)
> +{
> + return 0;
> +}
> +
> static inline pci_power_t pci_choose_state(struct pci_dev *dev,
> pm_message_t state)
> {
>
Applied to linux-next, thanks guys.
--
Jesse Barnes, Intel Open Source Technology Center
^ permalink raw reply
* AVB QoS support (IEEE802.1 Qav and Qat)
From: Eliot Blennerhassett @ 2011-02-16 19:51 UTC (permalink / raw)
To: netdev
In-Reply-To: <4D488D9B.5080409@audioscience.com>
Greetings
Can linux network stacks can support the various protocols required by AVB?
It is not clear to me if it already does, and in any case how the setup
would be accessed?
In particular these two standards and the interaction between them
802.1Qat - Stream Reservation Protocol
802.1Qav - Forwarding and Queuing Enhancements for Time-Sensitive
Streams (approved on Dec 10th, 2009)
Qav requires traffic shaping to smooth the traffic flow in guaranteed
stream class. This is afaik based on 1 packet per stream per 125us
Note that in future (maybe already) there will be hardware assist for
this feature in NIC chips.
--
Eliot Blennerhassett
AudioScience Inc.
[1] http://www.ieee802.org/1/pages/avbridges.html
[2]
http://en.wikipedia.org/wiki/Audio_Video_Bridging#Traffic_shaping_for_AV_streams
^ permalink raw reply
* [BUG] behaviour mismatch between ipv4 and ipv6 in UDP rx path
From: Chris Friesen @ 2011-02-16 20:18 UTC (permalink / raw)
To: netdev
Hi,
I sent this out a week ago but didn't see a reply, so I'm sending it out
again.
One of our guys is seeing occasional dropped ipv4 packets coming in on
an ipv6 udp socket obtained via socket(AF_INET6, SOCK_DGRAM, IPPROTO_UDP).
Here's what he says:
"The problem happens when release_sock() goes down an interesting code
path. If (sk->sk_backlog.tail) is non-NULL then release_sock() invokes
__release_sock() which loops over all queue packets and invokes the
socket's backlog receive function for each previously queued packet.
Now for the interesting part. The UDPv6 backlog receive function (in
net/ipv6/udp.c, udpv6_queue_rcv_skb()) invokes xfrm6_policy_check() to
confirm that the packet is allowed, but the problem is that it calls
this function regardless of whether the packet is IPv4 or IPv6. The
xfrm6_policy_check() function then assumes that it is an IPv6 packet and
tries to match a policy based on its packet header... but that clearly
won't work because the addresses that it finds when it decodes the skb
are completely bogus."
Looking at the ipv4 code, git commit 9382177 split __udp_queue_rcv_skb()
out of udp_queue_rcv_skb(). It was done for locking purposes, but it
also means that backlog_rcv is bound to __udp_queue_rcv_skb(), which
doesn't call xfrm4_policy_check().
Should a new function __udpv6_queue_rcv_skb() be split out from
udpv6_queue_rcv_skb() and bound to backlog_rcv to resolve the xfrm
issue? What about the locking that was the reason for the split in the
ipv4 case--is there a similar problem with ipv6?
Thanks,
Chris
--
Chris Friesen
Software Developer
GENBAND
chris.friesen@genband.com
www.genband.com
^ permalink raw reply
* [PATCH v2] bnx2x: Support for managing RX indirection table
From: Tom Herbert @ 2011-02-16 20:27 UTC (permalink / raw)
To: davem, eilong, netdev
Support fetching and retrieving RX indirection table via ethtool.
Signed-off-by: Tom Herbert <therbert@google.com>
---
drivers/net/bnx2x/bnx2x.h | 2 +
drivers/net/bnx2x/bnx2x_ethtool.c | 56 +++++++++++++++++++++++++++++++++++++
drivers/net/bnx2x/bnx2x_main.c | 22 +++++++++++---
3 files changed, 75 insertions(+), 5 deletions(-)
diff --git a/drivers/net/bnx2x/bnx2x.h b/drivers/net/bnx2x/bnx2x.h
index 236d79a..c0dd30d 100644
--- a/drivers/net/bnx2x/bnx2x.h
+++ b/drivers/net/bnx2x/bnx2x.h
@@ -1076,6 +1076,7 @@ struct bnx2x {
int num_queues;
int disable_tpa;
int int_mode;
+ u32 *rx_indir_table;
struct tstorm_eth_mac_filter_config mac_filters;
#define BNX2X_ACCEPT_NONE 0x0000
@@ -1799,5 +1800,6 @@ static inline u32 reg_poll(struct bnx2x *bp, u32 reg, u32 expected, int ms,
BNX2X_EXTERN int load_count[2][3]; /* per path: 0-common, 1-port0, 2-port1 */
extern void bnx2x_set_ethtool_ops(struct net_device *netdev);
+void bnx2x_push_indir_table(struct bnx2x *bp);
#endif /* bnx2x.h */
diff --git a/drivers/net/bnx2x/bnx2x_ethtool.c b/drivers/net/bnx2x/bnx2x_ethtool.c
index 816fef6..8d19d12 100644
--- a/drivers/net/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/bnx2x/bnx2x_ethtool.c
@@ -2134,6 +2134,59 @@ static int bnx2x_phys_id(struct net_device *dev, u32 data)
return 0;
}
+static int bnx2x_get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
+ void *rules __always_unused)
+{
+ struct bnx2x *bp = netdev_priv(dev);
+
+ switch (info->cmd) {
+ case ETHTOOL_GRXRINGS:
+ info->data = BNX2X_NUM_ETH_QUEUES(bp);
+ return 0;
+
+ default:
+ return -EOPNOTSUPP;
+ }
+}
+
+static int bnx2x_get_rxfh_indir(struct net_device *dev,
+ struct ethtool_rxfh_indir *indir)
+{
+ struct bnx2x *bp = netdev_priv(dev);
+ size_t copy_size =
+ min_t(size_t, indir->size, TSTORM_INDIRECTION_TABLE_SIZE);
+
+ if (bp->multi_mode == ETH_RSS_MODE_DISABLED)
+ return -EOPNOTSUPP;
+
+ indir->size = TSTORM_INDIRECTION_TABLE_SIZE;
+ memcpy(indir->ring_index, bp->rx_indir_table,
+ copy_size * sizeof(bp->rx_indir_table[0]));
+ return 0;
+}
+
+static int bnx2x_set_rxfh_indir(struct net_device *dev,
+ const struct ethtool_rxfh_indir *indir)
+{
+ struct bnx2x *bp = netdev_priv(dev);
+ size_t i;
+
+ if (bp->multi_mode == ETH_RSS_MODE_DISABLED)
+ return -EOPNOTSUPP;
+
+ /* Validate size and indices */
+ if (indir->size != TSTORM_INDIRECTION_TABLE_SIZE)
+ return -EINVAL;
+ for (i = 0; i < TSTORM_INDIRECTION_TABLE_SIZE; i++)
+ if (indir->ring_index[i] >= BNX2X_NUM_ETH_QUEUES(bp))
+ return -EINVAL;
+
+ memcpy(bp->rx_indir_table, indir->ring_index,
+ indir->size * sizeof(bp->rx_indir_table[0]));
+ bnx2x_push_indir_table(bp);
+ return 0;
+}
+
static const struct ethtool_ops bnx2x_ethtool_ops = {
.get_settings = bnx2x_get_settings,
.set_settings = bnx2x_set_settings,
@@ -2170,6 +2223,9 @@ static const struct ethtool_ops bnx2x_ethtool_ops = {
.get_strings = bnx2x_get_strings,
.phys_id = bnx2x_phys_id,
.get_ethtool_stats = bnx2x_get_ethtool_stats,
+ .get_rxnfc = bnx2x_get_rxnfc,
+ .get_rxfh_indir = bnx2x_get_rxfh_indir,
+ .set_rxfh_indir = bnx2x_set_rxfh_indir,
};
void bnx2x_set_ethtool_ops(struct net_device *netdev)
diff --git a/drivers/net/bnx2x/bnx2x_main.c b/drivers/net/bnx2x/bnx2x_main.c
index c238c4d..6c7745e 100644
--- a/drivers/net/bnx2x/bnx2x_main.c
+++ b/drivers/net/bnx2x/bnx2x_main.c
@@ -4254,7 +4254,7 @@ static void bnx2x_init_eq_ring(struct bnx2x *bp)
min_t(int, MAX_SP_DESC_CNT - MAX_SPQ_PENDING, NUM_EQ_DESC) - 1);
}
-static void bnx2x_init_ind_table(struct bnx2x *bp)
+void bnx2x_push_indir_table(struct bnx2x *bp)
{
int func = BP_FUNC(bp);
int i;
@@ -4262,13 +4262,20 @@ static void bnx2x_init_ind_table(struct bnx2x *bp)
if (bp->multi_mode == ETH_RSS_MODE_DISABLED)
return;
- DP(NETIF_MSG_IFUP,
- "Initializing indirection table multi_mode %d\n", bp->multi_mode);
for (i = 0; i < TSTORM_INDIRECTION_TABLE_SIZE; i++)
REG_WR8(bp, BAR_TSTRORM_INTMEM +
TSTORM_INDIRECTION_TABLE_OFFSET(func) + i,
- bp->fp->cl_id + (i % (bp->num_queues -
- NONE_ETH_CONTEXT_USE)));
+ bp->fp->cl_id + bp->rx_indir_table[i]);
+}
+
+static void bnx2x_init_ind_table(struct bnx2x *bp)
+{
+ int i;
+
+ for (i = 0; i < TSTORM_INDIRECTION_TABLE_SIZE; i++)
+ bp->rx_indir_table[i] = i % BNX2X_NUM_ETH_QUEUES(bp);
+
+ bnx2x_push_indir_table(bp);
}
void bnx2x_set_storm_rx_mode(struct bnx2x *bp)
@@ -6016,6 +6023,8 @@ void bnx2x_free_mem(struct bnx2x *bp)
BNX2X_PCI_FREE(bp->eq_ring, bp->eq_mapping,
BCM_PAGE_SIZE * NUM_EQ_PAGES);
+ BNX2X_FREE(bp->rx_indir_table);
+
#undef BNX2X_PCI_FREE
#undef BNX2X_KFREE
}
@@ -6146,6 +6155,9 @@ int bnx2x_alloc_mem(struct bnx2x *bp)
/* EQ */
BNX2X_PCI_ALLOC(bp->eq_ring, &bp->eq_mapping,
BCM_PAGE_SIZE * NUM_EQ_PAGES);
+
+ BNX2X_ALLOC(bp->rx_indir_table, sizeof(bp->rx_indir_table[0]) *
+ TSTORM_INDIRECTION_TABLE_SIZE);
return 0;
alloc_mem_err:
--
1.7.3.1
^ permalink raw reply related
* Re: [BUG] behaviour mismatch between ipv4 and ipv6 in UDP rx path
From: Eric Dumazet @ 2011-02-16 20:59 UTC (permalink / raw)
To: Chris Friesen; +Cc: netdev, Herbert Xu
In-Reply-To: <4D5C3128.4080101@genband.com>
Le mercredi 16 février 2011 à 14:18 -0600, Chris Friesen a écrit :
> Hi,
>
> I sent this out a week ago but didn't see a reply, so I'm sending it out
> again.
>
> One of our guys is seeing occasional dropped ipv4 packets coming in on
> an ipv6 udp socket obtained via socket(AF_INET6, SOCK_DGRAM, IPPROTO_UDP).
>
> Here's what he says:
>
>
> "The problem happens when release_sock() goes down an interesting code
> path. If (sk->sk_backlog.tail) is non-NULL then release_sock() invokes
> __release_sock() which loops over all queue packets and invokes the
> socket's backlog receive function for each previously queued packet.
>
> Now for the interesting part. The UDPv6 backlog receive function (in
> net/ipv6/udp.c, udpv6_queue_rcv_skb()) invokes xfrm6_policy_check() to
> confirm that the packet is allowed, but the problem is that it calls
> this function regardless of whether the packet is IPv4 or IPv6. The
> xfrm6_policy_check() function then assumes that it is an IPv6 packet and
> tries to match a policy based on its packet header... but that clearly
> won't work because the addresses that it finds when it decodes the skb
> are completely bogus."
>
>
> Looking at the ipv4 code, git commit 9382177 split __udp_queue_rcv_skb()
> out of udp_queue_rcv_skb(). It was done for locking purposes, but it
> also means that backlog_rcv is bound to __udp_queue_rcv_skb(), which
> doesn't call xfrm4_policy_check().
>
>
> Should a new function __udpv6_queue_rcv_skb() be split out from
> udpv6_queue_rcv_skb() and bound to backlog_rcv to resolve the xfrm
> issue? What about the locking that was the reason for the split in the
> ipv4 case--is there a similar problem with ipv6?
>
Yes, please submit a patch ?
Ideally, __udp_queue_rcv_skb() should be the common .backlog
In practice, because of sock_rps_save_rxhash() and MIB counters, I
suspect a __udp6_queue_rcv_skb() is OK.
^ permalink raw reply
* Re: [PATCH] bonding/vlan: Avoid mangled NAs on slaves without VLAN tag insertion
From: Greg KH @ 2011-02-16 21:58 UTC (permalink / raw)
To: David Miller; +Cc: bhutchings, netdev, fubar, stable, bonding-devel
In-Reply-To: <20110207.131754.59684095.davem@davemloft.net>
On Mon, Feb 07, 2011 at 01:17:54PM -0800, David Miller wrote:
> From: Ben Hutchings <bhutchings@solarflare.com>
> Date: Mon, 07 Feb 2011 19:20:55 +0000
>
> > This is related to commit f88a4a9b65a6f3422b81be995535d0e69df11bb8
> > upstream, but the bug cannot be properly fixed without the other
> > changes to VLAN tagging in 2.6.37.
> >
> > bond_na_send() attempts to insert a VLAN tag in between building and
> > sending packets of the respective formats. If the slave does not
> > implement hardware VLAN tag insertion then vlan_put_tag() will mangle
> > the network-layer header because the Ethernet header is not present at
> > this point (unlike in bond_arp_send()).
> >
> > Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
>
> Acked-by: David S. Miller <davem@davemloft.net>
Great, thanks for the patch, now queued up for the next .32-stable
release.
greg k-h
^ permalink raw reply
* macb:
From: Marc Kleine-Budde @ 2011-02-16 22:31 UTC (permalink / raw)
To: Netdev; +Cc: Nicolas Ferre
[-- Attachment #1: Type: text/plain, Size: 650 bytes --]
Hello,
I added type checking[1] to platform_set_drvdata and got the following
warning:
drivers/net/macb.c: In function 'macb_mii_init':
drivers/net/macb.c:263: warning: passing argument 1 of 'platform_set_drvdata' from incompatible pointer type
I'm new to the mii_bus stuff and don't see how to fix it.
regards, Marc
[1] https://lkml.org/lkml/2011/2/16/380
--
Pengutronix e.K. | Marc Kleine-Budde |
Industrial Linux Solutions | Phone: +49-231-2826-924 |
Vertretung West/Dortmund | Fax: +49-5121-206917-5555 |
Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 262 bytes --]
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: Greg KH @ 2011-02-16 22:39 UTC (permalink / raw)
To: Matt Carlson; +Cc: davem, netdev, stable
In-Reply-To: <1297810270-4690-1-git-send-email-mcarlson@broadcom.com>
On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
> If management firmware is present and the device is down, the firmware
> will assume control of the phy. If a phy access were allowed from the
> host, it will collide with firmware phy accesses, resulting in
> unpredictable behavior. This patch fixes the problem by disallowing phy
> accesses during the problematic condition.
>
> Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
There is no such upstream git commit id in Linus's tree. What am I
doing wrong here?
confused,
greg k-h
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: Matt Carlson @ 2011-02-16 23:06 UTC (permalink / raw)
To: Greg KH
Cc: Matthew Carlson, davem@davemloft.net, netdev@vger.kernel.org,
stable@kernel.org
In-Reply-To: <20110216223935.GF22056@kroah.com>
On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
> > If management firmware is present and the device is down, the firmware
> > will assume control of the phy. If a phy access were allowed from the
> > host, it will collide with firmware phy accesses, resulting in
> > unpredictable behavior. This patch fixes the problem by disallowing phy
> > accesses during the problematic condition.
> >
> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
>
> There is no such upstream git commit id in Linus's tree. What am I
> doing wrong here?
The commit is in Dave Miller's net-next-2.6 tree.
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: David Miller @ 2011-02-16 23:11 UTC (permalink / raw)
To: mcarlson; +Cc: greg, netdev, stable
In-Reply-To: <20110216230613.GA11053@mcarlson.broadcom.com>
From: "Matt Carlson" <mcarlson@broadcom.com>
Date: Wed, 16 Feb 2011 15:06:13 -0800
> On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
>> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
>> > If management firmware is present and the device is down, the firmware
>> > will assume control of the phy. If a phy access were allowed from the
>> > host, it will collide with firmware phy accesses, resulting in
>> > unpredictable behavior. This patch fixes the problem by disallowing phy
>> > accesses during the problematic condition.
>> >
>> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
>>
>> There is no such upstream git commit id in Linus's tree. What am I
>> doing wrong here?
>
> The commit is in Dave Miller's net-next-2.6 tree.
>
If it wasn't appropriate for net-2.6, it absolutely it not appropriate
for -stable.
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: Matt Carlson @ 2011-02-16 23:52 UTC (permalink / raw)
To: David Miller
Cc: Matthew Carlson, greg@kroah.com, netdev@vger.kernel.org,
stable@kernel.org
In-Reply-To: <20110216.151103.189711682.davem@davemloft.net>
On Wed, Feb 16, 2011 at 03:11:03PM -0800, David Miller wrote:
> From: "Matt Carlson" <mcarlson@broadcom.com>
> Date: Wed, 16 Feb 2011 15:06:13 -0800
>
> > On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
> >> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
> >> > If management firmware is present and the device is down, the firmware
> >> > will assume control of the phy. If a phy access were allowed from the
> >> > host, it will collide with firmware phy accesses, resulting in
> >> > unpredictable behavior. This patch fixes the problem by disallowing phy
> >> > accesses during the problematic condition.
> >> >
> >> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
> >>
> >> There is no such upstream git commit id in Linus's tree. What am I
> >> doing wrong here?
> >
> > The commit is in Dave Miller's net-next-2.6 tree.
> >
>
> If it wasn't appropriate for net-2.6, it absolutely it not appropriate
> for -stable.
net-2.6 was the target tree for the patch. The stable_kernel_rules.txt
seemed to suggest that I could just CC stable@kernel.org with the
commit ID, and Greg would pull it in as the process dictates. If that
isn't correct, what is the preferred way to expedite the integration of
a patch?
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: Greg KH @ 2011-02-17 0:00 UTC (permalink / raw)
To: Matt Carlson; +Cc: David Miller, stable@kernel.org, netdev@vger.kernel.org
In-Reply-To: <20110216235248.GA11108@mcarlson.broadcom.com>
On Wed, Feb 16, 2011 at 03:52:48PM -0800, Matt Carlson wrote:
> On Wed, Feb 16, 2011 at 03:11:03PM -0800, David Miller wrote:
> > From: "Matt Carlson" <mcarlson@broadcom.com>
> > Date: Wed, 16 Feb 2011 15:06:13 -0800
> >
> > > On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
> > >> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
> > >> > If management firmware is present and the device is down, the firmware
> > >> > will assume control of the phy. If a phy access were allowed from the
> > >> > host, it will collide with firmware phy accesses, resulting in
> > >> > unpredictable behavior. This patch fixes the problem by disallowing phy
> > >> > accesses during the problematic condition.
> > >> >
> > >> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
> > >>
> > >> There is no such upstream git commit id in Linus's tree. What am I
> > >> doing wrong here?
> > >
> > > The commit is in Dave Miller's net-next-2.6 tree.
> > >
> >
> > If it wasn't appropriate for net-2.6, it absolutely it not appropriate
> > for -stable.
>
> net-2.6 was the target tree for the patch. The stable_kernel_rules.txt
> seemed to suggest that I could just CC stable@kernel.org with the
> commit ID, and Greg would pull it in as the process dictates. If that
> isn't correct, what is the preferred way to expedite the integration of
> a patch?
Keep reading that file, it says to put the Cc: in the signed-off-by area
of the original patch.
Also, that file says the patch has to be in Linus's tree, otherwise
sending me a git commit id of some other tree isn't going to help at
all.
thanks,
greg k-h
^ permalink raw reply
* state of rtcache removal...
From: David Miller @ 2011-02-17 0:08 UTC (permalink / raw)
To: netdev
So I've been testing out the routing cache removal patch to see
what the impact is on performance.
I'm using a UDP flood to a single IP address over a dummy interface
with hard coded ARP entries, so that pretty much just the main IP
output and routing paths are being exercised.
The UDP flood tool I cooked up based upon a description sent to me by
Eric Dumazet of a similar utility he uses for testing. I've included
the code to this tool at the end of this email, as well as the dummy
interface setup script. Basically, you go:
bash# ./udpflood_setup.sh
bash# time ./udpflood -l 10000 10.2.2.11
The IP output path is about twice as slow with the routing cache
removed entirely. Here are the numbers I have:
net-next-2.6, rt_cache on:
davem@maramba:~$ time udpflood -l 10000000 10.2.2.11
real 1m47.012s
user 0m8.670s
sys 1m38.370s
net-next-2.6, rt_cache turned off via sysctl:
davem@maramba:~$ time udpflood -l 10000000 10.2.2.11
real 3m12.662s
user 0m9.490s
sys 3m3.220s
net-next-2.6 + "BONUS" rt_cache deletion patch:
maramba:/home/davem# time ./bin/udpflood -l 10000000 10.2.2.11
real 3m9.921s
user 0m9.520s
sys 3m0.440s
I then worked on some simplifications of the code in net/ipv4/route.c
that remains after the cache removal. I'll post those patches after
I've chewed on them some more, but they knock a couple seconds back off
of the benchmark:
The profile output is what you'd expect, with fib_table_lookup() topping
the charts taking ~%10 of the time.
What might not be initially apparent is that each output route lookup
results in two calls to fib_table_lookup() and thus two trie lookups.
Why? Because we have two routing tables (3 with IP_MULTIPLE_TABLES
enabled) that get searched, first the LOCAL then the MAIN table (then
with mutliple-tables enabled, the DEFAULT). And most external
outgoing routes sit in the MAIN table.
We do this so we can store all the interface address network,
broadcast, loopback network, et al. routes in the LOCAL table, then all
globally visible routes in the MAIN table.
Anyways, the long and short of this is that route lookups take two
trie lookups instead of just one. On input there are even more, for
source address validation done by fib_validate_source(). That can be
up to 4 more fib_table_lookup() invocations.
Add in another level of complexity if you have a series of FIB rules
installed.
So, to me, this means that spending time micro-optiming fib_trie is
not going to help much. Getting rid of that multiplier somehow, on
the other hand, might.
I plan to play with some ideas, such as sticking fib_alias entries into
the flow cache and consulting/populating the flow cache on fib_lookup()
calls.
-------------------- udpflood.c --------------------
/* An adaptation of Eric Dumazet's udpflood tool. */
#include <stdio.h>
#include <stddef.h>
#include <malloc.h>
#include <string.h>
#include <errno.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#define _GNU_SOURCE
#include <getopt.h>
static int usage(void)
{
printf("usage: udpflood [ -l count ] [ -m message_size ] IP_ADDRESS\n");
return -1;
}
static int send_packets(in_addr_t addr, int port, int count, int msg_sz)
{
char *msg = malloc(msg_sz);
struct sockaddr_in saddr;
int fd, i, err;
if (!msg)
return -ENOMEM;
memset(msg, 0, msg_sz);
memset(&saddr, 0, sizeof(saddr));
saddr.sin_family = AF_INET;
saddr.sin_port = port;
saddr.sin_addr.s_addr = addr;
fd = socket(PF_INET, SOCK_DGRAM, IPPROTO_IP);
if (fd < 0) {
perror("socket");
err = fd;
goto out_nofd;
}
err = connect(fd, (struct sockaddr *) &saddr, sizeof(saddr));
if (err < 0) {
perror("connect");
close(fd);
goto out;
}
for (i = 0; i < count; i++) {
err = sendto(fd, msg, msg_sz, 0,
(struct sockaddr *) &saddr, sizeof(saddr));
if (err < 0) {
perror("sendto");
goto out;
}
}
err = 0;
out:
close(fd);
out_nofd:
free(msg);
return err;
}
int main(int argc, char **argv, char **envp)
{
int port, msg_sz, count, ret;
in_addr_t addr;
port = 6000;
msg_sz = 32;
count = 10000000;
while ((ret = getopt(argc, argv, "l:s:p:")) >= 0) {
switch (ret) {
case 'l':
sscanf(optarg, "%d", &count);
break;
case 's':
sscanf(optarg, "%d", &msg_sz);
break;
case 'p':
sscanf(optarg, "%d", &port);
break;
case '?':
return usage();
}
}
if (!argv[optind])
return usage();
addr = inet_addr(argv[optind]);
if (addr == INADDR_NONE)
return usage();
return send_packets(addr, port, count, msg_sz);
}
-------------------- udpflood_setup.sh --------------------
#!/bin/sh
modprobe dummy
ifconfig dummy0 10.2.2.254 netmask 255.255.255.0 up
for f in $(seq 11 26)
do
arp -H ether -i dummy0 -s 10.2.2.$f 00:00:0c:07:ac:$f
done
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: David Miller @ 2011-02-17 0:10 UTC (permalink / raw)
To: mcarlson; +Cc: greg, netdev, stable
In-Reply-To: <20110216235248.GA11108@mcarlson.broadcom.com>
From: "Matt Carlson" <mcarlson@broadcom.com>
Date: Wed, 16 Feb 2011 15:52:48 -0800
> On Wed, Feb 16, 2011 at 03:11:03PM -0800, David Miller wrote:
>> From: "Matt Carlson" <mcarlson@broadcom.com>
>> Date: Wed, 16 Feb 2011 15:06:13 -0800
>>
>> > On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
>> >> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
>> >> > If management firmware is present and the device is down, the firmware
>> >> > will assume control of the phy. If a phy access were allowed from the
>> >> > host, it will collide with firmware phy accesses, resulting in
>> >> > unpredictable behavior. This patch fixes the problem by disallowing phy
>> >> > accesses during the problematic condition.
>> >> >
>> >> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
>> >>
>> >> There is no such upstream git commit id in Linus's tree. What am I
>> >> doing wrong here?
>> >
>> > The commit is in Dave Miller's net-next-2.6 tree.
>> >
>>
>> If it wasn't appropriate for net-2.6, it absolutely it not appropriate
>> for -stable.
>
> net-2.6 was the target tree for the patch. The stable_kernel_rules.txt
> seemed to suggest that I could just CC stable@kernel.org with the
> commit ID, and Greg would pull it in as the process dictates. If that
> isn't correct, what is the preferred way to expedite the integration of
> a patch?
You are posting a commit ID for the net-next-2.6 tree, that's what triggered
my response.
Unless it also went into the net-2.6 tree (in which case you should
give Greg the net-2.6 commit ID, which is also what the commit ID must
be in Linus's tree right now), the change is not appropriate for
-stable submission.
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: Matt Carlson @ 2011-02-17 0:11 UTC (permalink / raw)
To: Greg KH
Cc: Matthew Carlson, David Miller, stable@kernel.org,
netdev@vger.kernel.org
In-Reply-To: <20110217000035.GB6296@kroah.com>
On Wed, Feb 16, 2011 at 04:00:35PM -0800, Greg KH wrote:
> On Wed, Feb 16, 2011 at 03:52:48PM -0800, Matt Carlson wrote:
> > On Wed, Feb 16, 2011 at 03:11:03PM -0800, David Miller wrote:
> > > From: "Matt Carlson" <mcarlson@broadcom.com>
> > > Date: Wed, 16 Feb 2011 15:06:13 -0800
> > >
> > > > On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
> > > >> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
> > > >> > If management firmware is present and the device is down, the firmware
> > > >> > will assume control of the phy. If a phy access were allowed from the
> > > >> > host, it will collide with firmware phy accesses, resulting in
> > > >> > unpredictable behavior. This patch fixes the problem by disallowing phy
> > > >> > accesses during the problematic condition.
> > > >> >
> > > >> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
> > > >>
> > > >> There is no such upstream git commit id in Linus's tree. What am I
> > > >> doing wrong here?
> > > >
> > > > The commit is in Dave Miller's net-next-2.6 tree.
> > > >
> > >
> > > If it wasn't appropriate for net-2.6, it absolutely it not appropriate
> > > for -stable.
> >
> > net-2.6 was the target tree for the patch. The stable_kernel_rules.txt
> > seemed to suggest that I could just CC stable@kernel.org with the
> > commit ID, and Greg would pull it in as the process dictates. If that
> > isn't correct, what is the preferred way to expedite the integration of
> > a patch?
>
> Keep reading that file, it says to put the Cc: in the signed-off-by area
> of the original patch.
Ah. Yes. I see that now.
> Also, that file says the patch has to be in Linus's tree, otherwise
> sending me a git commit id of some other tree isn't going to help at
> all.
I see. Thanks for the tips.
^ permalink raw reply
* Re: [RFC, PATCH 1/4] net: sh_eth: modify the definitions of register
From: Nobuhiro Iwamatsu @ 2011-02-17 0:21 UTC (permalink / raw)
To: Yoshihiro Shimoda; +Cc: netdev, SH-Linux
In-Reply-To: <4D5A67CC.80107@renesas.com>
2011/2/15 Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>:
> The previous code cannot handle the ETHER and GETHER both as same time
> because the definitions of register was hardcoded.
>
> Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
> ---
> arch/sh/include/asm/sh_eth.h | 7 +
> drivers/net/sh_eth.c | 322 +++++++++++-----------
> drivers/net/sh_eth.h | 623 ++++++++++++++++++++++++------------------
> 3 files changed, 537 insertions(+), 415 deletions(-)
>
<snip>
>
> +static const u16 *sh_eth_get_register_offset(int register_type)
> +{
> + const u16 *reg_offset = NULL;
> +
> + switch (register_type) {
> + case SH_ETH_REG_GIGABIT:
> + reg_offset = sh_eth_offset_gigabit;
> + break;
> + case SH_ETH_REG_FAST_SH4:
> + reg_offset = sh_eth_offset_fast_sh4;
> + break;
> + case SH_ETH_REG_FAST_SH3_SH2:
> + reg_offset = sh_eth_offset_fast_sh3_sh2;
> + break;
> + case SH_ETH_REG_DEFAULT:
> +#if defined(CONFIG_CPU_SUBTYPE_SH7763)
> + reg_offset = sh_eth_offset_gigabit;
> +#elif defined(CONFIG_CPU_SH4)
> + reg_offset = sh_eth_offset_fast_sh4;
> +#else
> + reg_offset = sh_eth_offset_fast_sh3_sh2;
> +#endif
> + break;
> + default:
> + printk(KERN_ERR "Unknown register type (%d)\n", register_type);
> + break;
> + }
> +
> + return reg_offset;
> +}
> +
Is the handling of SH_ETH_REG_DEFAULT necessary?
>
> +static inline void sh_eth_write(struct net_device *ndev, unsigned long data,
> + int enum_index)
> +{
> + struct sh_eth_private *mdp = netdev_priv(ndev);
> +
> + writel(data, ndev->base_addr + mdp->reg_offset[enum_index]);
> +}
> +
> +static inline unsigned long sh_eth_read(struct net_device *ndev,
> + int enum_index)
> +{
> + struct sh_eth_private *mdp = netdev_priv(ndev);
> +
> + return readl(ndev->base_addr + mdp->reg_offset[enum_index]);
> +}
> +
> +static inline void sh_eth_tsu_write(struct sh_eth_private *mdp,
> + unsigned long data, int enum_index)
> +{
> + writel(data, mdp->tsu_addr + mdp->reg_offset[enum_index]);
> +}
> +
> +static inline unsigned long sh_eth_tsu_read(struct sh_eth_private *mdp,
> + int enum_index)
> +{
> + return readl(mdp->tsu_addr + mdp->reg_offset[enum_index]);
> +}
> +
> #endif /* #ifndef __SH_ETH_H__ */
I do not think that a new function is necessary.
I think it use MACRO.
For example,
#define sh_eth_write(data,reg) \
{\
writel(data, ndev->base_addr + mdp->reg_offset[reg]); \
}
Best regards,
Nobuhiro
--
Nobuhiro Iwamatsu
iwamatsu at {nigauri.org / debian.org}
GPG ID: 40AD1FA6
^ permalink raw reply
* Re: [RFC, PATCH 2/4] net: sh_eth: remove the SH_TSU_ADDR
From: Nobuhiro Iwamatsu @ 2011-02-17 0:24 UTC (permalink / raw)
To: Yoshihiro Shimoda; +Cc: netdev, SH-Linux
In-Reply-To: <4D5A67CF.3040406@renesas.com>
2011/2/15 Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>:
> The defination is hardcoded in this driver for some CPUs. This patch
> modifies to get resource of TSU address from platform_device.
>
> Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
> ---
> drivers/net/sh_eth.c | 16 ++++++++++++----
> drivers/net/sh_eth.h | 15 ---------------
> 2 files changed, 12 insertions(+), 19 deletions(-)
>
> diff --git a/drivers/net/sh_eth.c b/drivers/net/sh_eth.c
> index 3b6d545..0593f29 100644
> --- a/drivers/net/sh_eth.c
> +++ b/drivers/net/sh_eth.c
> @@ -1446,7 +1446,7 @@ static const struct net_device_ops sh_eth_netdev_ops = {
> static int sh_eth_drv_probe(struct platform_device *pdev)
> {
> int ret, devno = 0;
> - struct resource *res;
> + struct resource *res, *res_tsu;
> struct net_device *ndev = NULL;
> struct sh_eth_private *mdp;
> struct sh_eth_plat_data *pd;
> @@ -1520,9 +1520,13 @@ static int sh_eth_drv_probe(struct platform_device *pdev)
> mdp->cd->chip_reset(ndev);
>
> #if defined(SH_ETH_HAS_TSU)
> - /* TSU init (Init only)*/
> - mdp->tsu_addr = SH_TSU_ADDR;
> - sh_eth_tsu_init(mdp);
> + res_tsu = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> + if (res_tsu) {
> + mdp->tsu_addr = ioremap(res_tsu->start,
> + resource_size(res_tsu));
> + /* TSU init (Init only)*/
> + sh_eth_tsu_init(mdp);
> + }
> #endif
> }
>
> @@ -1549,6 +1553,8 @@ out_unregister:
>
> out_release:
> /* net_dev free */
> + if (mdp->tsu_addr)
> + iounmap(mdp->tsu_addr);
> if (ndev)
> free_netdev(ndev);
>
> @@ -1559,7 +1565,9 @@ out:
> static int sh_eth_drv_remove(struct platform_device *pdev)
> {
> struct net_device *ndev = platform_get_drvdata(pdev);
> + struct sh_eth_private *mdp = netdev_priv(ndev);
>
> + iounmap(mdp->tsu_addr);
> sh_mdio_release(ndev);
> unregister_netdev(ndev);
> pm_runtime_disable(&pdev->dev);
You forget to fix for ARSTR.
Best regards,
Nobuhiro
--
Nobuhiro Iwamatsu
iwamatsu at {nigauri.org / debian.org}
GPG ID: 40AD1FA6
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: Matt Carlson @ 2011-02-17 0:39 UTC (permalink / raw)
To: David Miller
Cc: Matthew Carlson, greg@kroah.com, netdev@vger.kernel.org,
stable@kernel.org
In-Reply-To: <20110216.161025.59672084.davem@davemloft.net>
On Wed, Feb 16, 2011 at 04:10:25PM -0800, David Miller wrote:
> From: "Matt Carlson" <mcarlson@broadcom.com>
> Date: Wed, 16 Feb 2011 15:52:48 -0800
>
> > On Wed, Feb 16, 2011 at 03:11:03PM -0800, David Miller wrote:
> >> From: "Matt Carlson" <mcarlson@broadcom.com>
> >> Date: Wed, 16 Feb 2011 15:06:13 -0800
> >>
> >> > On Wed, Feb 16, 2011 at 02:39:35PM -0800, Greg KH wrote:
> >> >> On Tue, Feb 15, 2011 at 02:51:10PM -0800, Matt Carlson wrote:
> >> >> > If management firmware is present and the device is down, the firmware
> >> >> > will assume control of the phy. If a phy access were allowed from the
> >> >> > host, it will collide with firmware phy accesses, resulting in
> >> >> > unpredictable behavior. This patch fixes the problem by disallowing phy
> >> >> > accesses during the problematic condition.
> >> >> >
> >> >> > Upstream commit ID f746a3136a61ae535c5d0b49a9418fa21edc61b5
> >> >>
> >> >> There is no such upstream git commit id in Linus's tree. What am I
> >> >> doing wrong here?
> >> >
> >> > The commit is in Dave Miller's net-next-2.6 tree.
> >> >
> >>
> >> If it wasn't appropriate for net-2.6, it absolutely it not appropriate
> >> for -stable.
> >
> > net-2.6 was the target tree for the patch. The stable_kernel_rules.txt
> > seemed to suggest that I could just CC stable@kernel.org with the
> > commit ID, and Greg would pull it in as the process dictates. If that
> > isn't correct, what is the preferred way to expedite the integration of
> > a patch?
>
> You are posting a commit ID for the net-next-2.6 tree, that's what triggered
> my response.
>
> Unless it also went into the net-2.6 tree (in which case you should
> give Greg the net-2.6 commit ID, which is also what the commit ID must
> be in Linus's tree right now), the change is not appropriate for
> -stable submission.
So the proper thing to do here is recall the patch, submit a new patch
to net-2.6 with a CC: stabel@kernel.org in the signed-off-by section.
Would I do the exact same thing if I were posting to net-next-2.6?
(i.e. the CC line tells you I want this patch to go to net-next-2.6,
net-2.6, then Linus's tree, then stable?) Or would you rather I posted
a completely different patchset against net-2.6?
^ permalink raw reply
* Re: [stable] [PATCH net-2.6/stable] tg3: Restrict phy ioctl access
From: David Miller @ 2011-02-17 0:56 UTC (permalink / raw)
To: mcarlson; +Cc: greg, netdev, stable
In-Reply-To: <20110217003947.GA11185@mcarlson.broadcom.com>
From: "Matt Carlson" <mcarlson@broadcom.com>
Date: Wed, 16 Feb 2011 16:39:47 -0800
> So the proper thing to do here is recall the patch, submit a new patch
> to net-2.6 with a CC: stabel@kernel.org in the signed-off-by section.
No.
I have not applied your patch yet, but when I do and I also get my
tree pulled next time into Linus's tree, you can ask stable to apply
it.
Because only at that point will it exist as a commit in Linus's tree.
Before that happens you cannot ask Greg to apply it to his tree.
^ permalink raw reply
* Re: [RFC, PATCH 1/4] net: sh_eth: modify the definitions of register
From: Nobuhiro Iwamatsu @ 2011-02-17 0:56 UTC (permalink / raw)
To: Yoshihiro Shimoda; +Cc: netdev, SH-Linux
In-Reply-To: <AANLkTimvaLOuBXcqJ=co1LakTxJqGEUQC3Cif+wBoJhm@mail.gmail.com>
2011/2/17 Nobuhiro Iwamatsu <iwamatsu@nigauri.org>:
> 2011/2/15 Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>:
>> The previous code cannot handle the ETHER and GETHER both as same time
>> because the definitions of register was hardcoded.
>>
>> Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
>> ---
>> arch/sh/include/asm/sh_eth.h | 7 +
>> drivers/net/sh_eth.c | 322 +++++++++++-----------
>> drivers/net/sh_eth.h | 623 ++++++++++++++++++++++++------------------
>> 3 files changed, 537 insertions(+), 415 deletions(-)
>>
>
> <snip>
>
>>
>> +static const u16 *sh_eth_get_register_offset(int register_type)
>> +{
>> + const u16 *reg_offset = NULL;
>> +
>> + switch (register_type) {
>> + case SH_ETH_REG_GIGABIT:
>> + reg_offset = sh_eth_offset_gigabit;
>> + break;
>> + case SH_ETH_REG_FAST_SH4:
>> + reg_offset = sh_eth_offset_fast_sh4;
>> + break;
>> + case SH_ETH_REG_FAST_SH3_SH2:
>> + reg_offset = sh_eth_offset_fast_sh3_sh2;
>> + break;
>> + case SH_ETH_REG_DEFAULT:
>> +#if defined(CONFIG_CPU_SUBTYPE_SH7763)
>> + reg_offset = sh_eth_offset_gigabit;
>> +#elif defined(CONFIG_CPU_SH4)
>> + reg_offset = sh_eth_offset_fast_sh4;
>> +#else
>> + reg_offset = sh_eth_offset_fast_sh3_sh2;
>> +#endif
>> + break;
>> + default:
>> + printk(KERN_ERR "Unknown register type (%d)\n", register_type);
>> + break;
>> + }
>> +
>> + return reg_offset;
>> +}
>> +
>
> Is the handling of SH_ETH_REG_DEFAULT necessary?
>
>>
>> +static inline void sh_eth_write(struct net_device *ndev, unsigned long data,
>> + int enum_index)
>> +{
>> + struct sh_eth_private *mdp = netdev_priv(ndev);
>> +
>> + writel(data, ndev->base_addr + mdp->reg_offset[enum_index]);
>> +}
>> +
>> +static inline unsigned long sh_eth_read(struct net_device *ndev,
>> + int enum_index)
>> +{
>> + struct sh_eth_private *mdp = netdev_priv(ndev);
>> +
>> + return readl(ndev->base_addr + mdp->reg_offset[enum_index]);
>> +}
>> +
>> +static inline void sh_eth_tsu_write(struct sh_eth_private *mdp,
>> + unsigned long data, int enum_index)
>> +{
>> + writel(data, mdp->tsu_addr + mdp->reg_offset[enum_index]);
>> +}
>> +
>> +static inline unsigned long sh_eth_tsu_read(struct sh_eth_private *mdp,
>> + int enum_index)
>> +{
>> + return readl(mdp->tsu_addr + mdp->reg_offset[enum_index]);
>> +}
>> +
>> #endif /* #ifndef __SH_ETH_H__ */
>
> I do not think that a new function is necessary.
> I think it use MACRO.
>
> For example,
> #define sh_eth_write(data,reg) \
> {\
> writel(data, ndev->base_addr + mdp->reg_offset[reg]); \
> }
>
Oh sorry, this is bad coding style.
Please ignore this comment.
Nobuhiro
--
Nobuhiro Iwamatsu
iwamatsu at {nigauri.org / debian.org}
GPG ID: 40AD1FA6
^ permalink raw reply
* ixgbe: 82599 and Westmere with HT
From: Andrew Dickinson @ 2011-02-17 1:21 UTC (permalink / raw)
To: netdev
Hi,
I've got a dual Westmere board (X5675) with an X520 card with dual
10G. I see 24-cores exposed to me and the ixgbe driver exposes 24 tx
and 24 rx interrupts per NIC. I then pin the interrupts to cores for
each NIC (each interrupt gets its own core, standard stuff).
Anyway.... I'm only seeing RX interrupts on 16 of the 24 cores (random
src/dest pairs across a /16 each, so I should be getting good flow
hashing). Did I miss some magic somewhere?
I'm running 2.6.32.4, perhaps this has been fixed upstream. If not,
any thoughts on how to make this work?
-A
^ permalink raw reply
* Fwd: IGMP and rwlock: Dead ocurred again on TILEPro
From: Cypher Wu @ 2011-02-17 2:17 UTC (permalink / raw)
To: linux-kernel, Chris Metcalf, Américo Wang, Eric Dumazet,
netdev
---------- Forwarded message ----------
From: Cypher Wu <cypher.w@gmail.com>
Date: Wed, Feb 16, 2011 at 5:58 PM
Subject: GMP and rwlock: Dead ocurred again on TILEPro
To: linux-kernel@vger.kernel.org
The rwlock and spinlock of TILEPro platform use TNS instruction to
test the value of lock, but if interrupt is not masked, read_lock()
have another chance to deadlock while read_lock() called in bh of
interrupt.
The original code:
void __raw_read_lock_slow(raw_
rwlock_t *rwlock, u32 val)
{
u32 iterations = 0;
do {
if (!(val & 1))
rwlock->lock = val;
delay_backoff(iterations++);
val = __insn_tns((int *)&rwlock->lock);
} while ((val << RD_COUNT_WIDTH) != 0);
rwlock->lock = val + (1 << RD_COUNT_SHIFT);
}
I've modified it to get some information:
void __raw_read_lock_slow(raw_rwlock_t *rwlock, u32 val)
{
u32 iterations = 0;
do {
if (!(val & 1))
{
rwlock->lock = val;
iterations = 0;
}
delay_backoff(iterations++);
if (iterations > 0x1000000)
{
dump_stack();
iterations = 0;
}
val = __insn_tns((int *)&rwlock->lock);
} while ((val << RD_COUNT_WIDTH) != 0);
rwlock->lock = val + (1 << RD_COUNT_SHIFT);
}
And this is the stack info:
Starting stack dump of tid 837, pid 837 (ff0) on cpu 55 at cycle 10180633928773
frame 0: 0xfd3bfbe0 dump_stack+0x0/0x20 (sp 0xe4b5f9d8)
frame 1: 0xfd3c0b50 __raw_read_lock_slow.cold+0x50/0x90 (sp 0xe4b5f9d8)
frame 2: 0xfd184a58 igmpv3_send_cr+0x60/0x440 (sp 0xe4b5f9f0)
frame 3: 0xfd3bd928 igmp_ifc_timer_expire+0x30/0x90 (sp 0xe4b5fa20)
frame 4: 0xfd047698 run_timer_softirq+0x258/0x3c8 (sp 0xe4b5fa30)
frame 5: 0xfd0563f8 __do_softirq+0x138/0x220 (sp 0xe4b5fa70)
frame 6: 0xfd097d48 do_softirq+0x88/0x110 (sp 0xe4b5fa98)
frame 7: 0xfd1871f8 irq_exit+0xf8/0x120 (sp 0xe4b5faa8)
frame 8: 0xfd1afda0 do_timer_interrupt+0xa0/0xf8 (sp 0xe4b5fab0)
frame 9: 0xfd187b98 handle_interrupt+0x2d8/0x2e0 (sp 0xe4b5fac0)
<interrupt 25 while in kernel mode>
frame 10: 0xfd0241c8 _read_lock+0x8/0x40 (sp 0xe4b5fc38)
frame 11: 0xfd1bb008 ip_mc_del_src+0xc8/0x378 (sp 0xe4b5fc40)
frame 12: 0xfd2681e8 ip_mc_leave_group+0xf8/0x1e0 (sp 0xe4b5fc70)
frame 13: 0xfd0a4d70 do_ip_setsockopt+0xe48/0x1560 (sp 0xe4b5fc90)
frame 14: 0xfd2b4168 sys_setsockopt+0x150/0x170 (sp 0xe4b5fe98)
frame 15: 0xfd14e550 handle_syscall+0x2d0/0x320 (sp 0xe4b5fec0)
<syscall while in user mode>
frame 16: 0x3342a0 (sp 0xbfddfb00)
frame 17: 0x16130 (sp 0xbfddfb08)
frame 18: 0x16640 (sp 0xbfddfb38)
frame 19: 0x16ee8 (sp 0xbfddfc58)
frame 20: 0x345a08 (sp 0xbfddfc90)
frame 21: 0x10218 (sp 0xbfddfe48)
Stack dump complete
I don't know the clear definition of rwlock & spinlock in Linux, but
the implementation of other platforms
like x86, PowerPC, ARM don't have that issue. The use of TNS cause a
race condition between system
call and interrupt.
Through the call tree of packet sending, there are also some other
rwlock will be tried, say
read_lock(&fib_hash_lock) in fn_hash_lookup() which is called in
ip_route_output_slow(). I've seen deadlock
on fib_hash_lock, but haven't reproduced with that debug information yet.
Maybe IGMP is not the only one, TCP timer will retransmit data and
will also call read_lock(&fib_hash_lock).
--
Cyberman Wu
--
Cyberman Wu
^ permalink raw reply
* Re: state of rtcache removal...
From: Tom Herbert @ 2011-02-17 2:59 UTC (permalink / raw)
To: David Miller; +Cc: netdev
In-Reply-To: <20110216.160838.39164069.davem@davemloft.net>
On Wed, Feb 16, 2011 at 4:08 PM, David Miller <davem@davemloft.net> wrote:
>
> So I've been testing out the routing cache removal patch to see
> what the impact is on performance.
>
Interesting results.
I assume that this test is purposely using sento on a connected socket
to force sendmsg to go through the route lookup :-), so this is
showing what the benefits of rtcache are is when cache hit rate is
100%. For comparison, it might interesting to see what the
performance is when rate is < 100%. For instance, we often see hit
rates < 20% on front end servers. This could be done flooding to
random addresses in 10/8 or even 0/0... I'm hoping that without the
rtcache performance actually improves in that case!
Tom
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox