- 22 May, 2008 12 commits
-
-
Jay Vosburgh authored
Add a "follow" selection for fail_over_mac. This option causes the MAC address to move from slave to slave as the active slave changes. This is in addition to the existing fail_over_mac option that causes the bond's MAC address to change during failover. This new option is useful for devices that cannot tolerate multiple ports using the same MAC address simultaneously, either because it confuses them or incurs a performance penalty (as is the case with some LPAR-aware multiport devices). Because the MAC of the bond itself does not change, the "follow" option is slightly more reliable during failover and doesn't change the MAC of the bond during operation. This patch requires a previous ARP monitor change to properly handle RTNL during failovers. Signed-off-by:
Jay Vosburgh <fubar@us.ibm.com> Signed-off-by:
Jeff Garzik <jgarzik@redhat.com>
-
Jay Vosburgh authored
Refactor ARP monitor for active-backup mode. The motivation for this is to take care of locking issues in a clear manner (particularly to correctly handle RTNL vs. the bonding locks). Currently, the a-b ARP monitor does not hold RTNL at all, but future changes will require RTNL during ARP monitor failovers. Rather than using conditional locking, this patch instead breaks up the ARP monitor into three discrete steps: inspection, commit changes, and probe. The inspection phase marks slaves that require link state changes. The commit phase is only called if inspection detects that changes are needed, and is called with RTNL. Lastly, the probe phase issues the ARP probes that the inspection phase uses to determine link state. Signed-off-by:
-
Moni Shoua authored
With IPoIB, reception of gratuitous ARP by neighboring hosts is essential for a successful change of slaves in case of failure. Otherwise, they won't learn about the HW address change and need to wait a long time until the neighboring system gives up and sends an ARP request to learn the new HW address. This patch decreases the chance for a lost of a gratuitous ARP packet by sending it more than once. The number retries is configurable and can be set with a module param. Signed-off-by:
Moni Shoua <monis@voltaire.com> Acked-by:
-
Pavel Emelyanov authored
Some places iterate over the checked list right after the check itself, so even if the list is empty, the list_for_each_xxx iterator will make everything right by himself. Signed-off-by:
Pavel Emelyanov <xemul@openvz.org> Acked-by:
-
Pavel Emelyanov authored
Many places either do not modify the list under the list_for_each_xxx, or break out of the loop as soon as the first element is removed. Thus, this _safe iteration just occupies some unneeded .text space and requires an additional variable. Signed-off-by:
-
Pavel Emelyanov authored
While we're fixing the bond_create, I hope it's OK to polish it a bit after the fixes. The third argument is NULL at the first caller and is ignored by the second one, so remove it. Signed-off-by:
-
Jay Vosburgh authored
Remove bond_has_ip and all references to it. With this change, the ARP monitor will always send ARP probes if the master is up and has at least one slave. If the bond has an IP address, it is used in the ARP probe; if not, the probes are sent with all zeros in the sender's IP address (which is consistent with an RFC 2131 4.4.1 duplicate address probe). This is useful for cases when bonding itself is hidden underneath a layer of virtual devices, e.g., with Xen. Change suggested by Tsutomu Fujii <t-fujii@nb.jp.nec.com>, who included a one-line patch that only affected active-backup mode. Signed-off-by:
-
Jay Vosburgh authored
Convert bonding to use msecs_to_jiffies instead of doing the math. For the ARP monitor, there was an underflow problem that could result in an infinite loop. The miimon already had that worked around, but this is cleaner. Originally by Nicolas de Pesloüan <nicolas.2p.debian@free.fr> Jay Vosburgh corrected a math error in the original; Nicolas' original commit message is: When setting arp_interval parameter to a very low value, delta_in_ticks for next arp might become 0, causing an infinite loop. See http://bugzilla.kernel.org/show_bug.cgi?id=10680 Same problem for miimon parameter already fixed, but fix might be enhanced, by using msecs_to_jiffies() function. Signed-off-by:
Nicolas de Pesloüan <nicolas.2p.debian@free.fr> Signed-off-by:
-
Al Viro authored
Signed-off-by:
Al Viro <viro@zeniv.linux.org.uk> Signed-off-by:
-
Divy Le Ray authored
Add LRO support. Signed-off-by:
Divy Le Ray <divy@chelsio.com> Signed-off-by:
-
Divy Le Ray authored
Add page support to Jumbo frame Rx queues. Signed-off-by:
-
Divy Le Ray authored
Take potential dma mapping errors in account. Signed-off-by:
-
- 21 May, 2008 15 commits
-
-
Allan Stephens authored
This patch ensures that TIPC's topology service and configuration service are shut down before switching into "network mode". This ensures that TIPC does not mistakenly try to send unnecessary "publication withdraw" messages to other nodes before it is fully initialized for sending off-node messages. Note that the node's current network address is now updated only after the two services are shut down; this ensures that any existing connections to the topology server are terminated correctly using the old address. Signed-off-by:
Allan Stephens <allan.stephens@windriver.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Allan Stephens authored
This patch ensures that the "previous node" field in any existing TIPC port message header templates is updated properly when a TIPC network address is assigned to the node. (Previously, only the "originating node" field was updated.) Signed-off-by:
-
Allan Stephens authored
This patch optimizes TIPC neighbor discovery code to avoid testing for a null node pointer when the pointer is already known to be non-null. Signed-off-by:
-
Allan Stephens authored
This patch ensures that the simultaneous discovery of the same neighboring node by multiple interfaces does not cause TIPC to add the node into its internal data structures more than once. Signed-off-by:
-
Allan Stephens authored
This patch prevents a TIPC configuration command requiring network administrator privileges from triggering an skbuff underrun if it is issued by a process lacking those privileges. The revised error handling code avoids the use of a potentially uninitialized global variable by transforming the unauthorized command into a new command, then following the standard command processing path to generate the required error message. Signed-off-by:
-
Pavel Emelyanov authored
Similar to ipmr. [ Fix build failures -DaveM ] Signed-off-by:
-
Pavel Emelyanov authored
These devices use the private area of appropriate size for statistics. Turning them to use on-device ones make them "privless" and thus - really small wrt kmalloc cache, they are allocated from. Besides, code looks nicer, because of absence of multi-braced type casts and dereferences. [ Fix build failures -DaveM ] Signed-off-by:
-
Pavel Emelyanov authored
This tunnel uses its own private structure and requires separate patch to switch from private stats to on-device ones. Signed-off-by:
-
Pavel Emelyanov authored
All users already use on-device statistics, so this field can be safely removed. Signed-off-by:
-
Pavel Emelyanov authored
The ipmr uses ipip tunnels for its purposes and updates the tunnels' stats, but the ipip driver is already switched to use on-device ones. Actually, this is a part of the patch #4 from this set. Signed-off-by:
-
Pavel Emelyanov authored
Signed-off-by:
-
Pavel Emelyanov authored
Signed-off-by:
-
Pavel Emelyanov authored
Just switch from tunnel->stat to tunnel->dev->stats. The ip_tunnel->stat member itself will be removed after I fix its other users (very soon). Signed-off-by:
-
Pavel Emelyanov authored
Even though bridges require 6 fields from struct net_device_stats, the on-device stats are always there, so we may just use them. The br_dev_get_stats is no longer required after this. Signed-off-by:
-
Pavel Emelyanov authored
And so does the pointer is returns, but sysfs and netlinks still check for both cases. Signed-off-by:
-
- 19 May, 2008 11 commits
-
-
Pavel Emelyanov authored
There are some sysctls left to be switched to read-only, but they are all in ipv6, so complete with them. Signed-off-by:
-
Pavel Emelyanov authored
Parts of fragments-related sysctls are read-only, but this is done by cloning all the tables and dropping write-bits from mode. Do the same but with read-only root. Signed-off-by:
-
Pavel Emelyanov authored
The error code is ignored now, but ipv6 is a module and one can be loaded under memory pressure, so the error may occur (in theory). Besides, I'm going to handle error returned from registering a read-only part of the table, so ignoring this one, while handing the other one would look strange. (However, this possibility of error is rather small, so I'm not sure whether this is a candidate for current net tree). Signed-off-by:
-
Pavel Emelyanov authored
The fragments sysctls also contains some, that are to be visible, but read-only in net namespaces. The naming in net/core/sysctl_net_core.c is - tables, that are to be registered in namespaces have a "ns" word in their names. So rename ones in ipv4/ip_fragment.c and ipv6/reassembly.c to fit this. Signed-off-by:
-
Pavel Emelyanov authored
Most of the net/core/xxx sysctls are read-only now, but this goal is achieved with excessive memory consumption in each namespace - the whole table is cloned and most of the entries in it are ~= 0222. Split it into two parts and register (the largest) one at the read-only root. Signed-off-by:
-
Pavel Emelyanov authored
This one stores all ctl-heads in one list and restricts the permissions not give write access to non-init net namespaces. Signed-off-by:
-
Allan Stephens authored
This patch contains a set of cosmetic changes to TIPC's network topology service subsystem, including: - updates to comments (including copyright dates) - re-ordering structure fields to group them more logically - removal of optional debugging code that is no longer required - minor changes to whitespace to conform to Linux coding conventions Signed-off-by:
-
Allan Stephens authored
This patch modifies TIPC's network topology service so that it only requires a single reference table entry per subscriber connection, rather than two. This is achieved by letting the reference to the server port communicating with the subscriber act as the reference to the subscriber object itself. (Since the subscriber cannot exist without its port, and vice versa, this dual role for the reference is perfectly natural.) This consolidation reduces the size of the reference table by 50% in the default configuration. Signed-off-by:
-
Allan Stephens authored
This patch fixes TIPC's topology server so that it does byte swapping correctly when endianness conversion is required. (Note: This bug only impacted an application if it issues a subscription request to a topology server on another node, rather than the server on it's own node; since the topology server is normally not accessible by off-node applications, most TIPC applications were not impacted by the bug.) Signed-off-by:
-
Allan Stephens authored
This patch enables TIPC's topology server code to do customized endianness conversions on a per-subscription basis. (This capability is needed to support the upcoming consolidation of subscriber and subscription object references.) Signed-off-by:
-
Allan Stephens authored
This patch enables TIPC's topology server code to do customized overlap detection handling on a per-subscription basis. (This capability is needed to support the upcoming introduction of multi-cluster TIPC networks.) Signed-off-by:
-
- 17 May, 2008 2 commits
-
-
Michael Chan authored
Signed-off-by:
Michael Chan <mchan@broadcom.com> Signed-off-by:
-
Benjamin Li authored
Instead of assigning values for the struct cpu_reg's at runtime, we already know these values at compile time. Therefore, we can use designated initializers, to initialize these structures and not have to incur this assignment cost at run-time. Signed-off-by:
Benjamin Li <benli@broadcom.com> Signed-off-by:
-
