Ivo van Doorn [Sat, 5 Jul 2008 13:11:57 +0000 (15:11 +0200)]
rt2x00: Fix NULL pointer error in adhoc/master mode
As soon as an interface is enabled, and that interface is in adhoc or master mode,
the device will start raising beacondone interrupts. But before the first interrupt is
raised, mac80211 will probably not have send any beacons to the device yet, which
results in a NULL pointer error when the skb is being freed.
Note that the "raise beacondone interrupts without a beacon" is also a bug,
and will be addressed later. The more important bug however is preventing
the NULL pointer failt itself, since there might be other conditions that could trigger
it as well.
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This corrects an error in the computation of the open loss interval I_0:
* the interval length is (highest_seqno - start_seqno) + 1
* and not (highest_seqno - start_seqno).
This condition was not fully clear in RFC 3448, but reflects the current
revision state of rfc3448bis and is also consistent with RFC 4340, 6.1.1.
Further changes:
----------------
* variable renamed due to line length constraints;
* explicit typecast to `s64' to avoid implicit signed/unsigned casting.
This fixes a bug in the logic of the TFRC loss detection:
* new_loss_indicated() should not be called while a loss is pending;
* but the code allows this;
* thus, for two subsequent gaps in the sequence space, when loss_count
has not yet reached NDUPACK=3, the loss_count is falsely reduced to 1.
To avoid further and similar problems, all loss handling and loss detection is
now done inside tfrc_rx_hist_handle_loss(), using an appropriate routine to
track new losses.
Further changes:
----------------
* added a reminder that no RX history operations should be performed when
rx_handle_loss() has identified a (new) loss, since the function takes
care of packet reordering during loss detection;
* made tfrc_rx_hist_loss_pending() bool (thanks to an earlier suggestion
by Arnaldo);
* removed unused functions.
RFC 4340, 7.7 specifies up to 6 bytes for the NDP Count option, whereas the code
is currently limited to up to 3 bytes. This seems to be a relict of an earlier
draft version and is brought up to date by the patch.
The TFRC loss detection code used the wrong loss condition (RFC 4340, 7.7.1):
* the difference between sequence numbers s1 and s2 instead of
* the number of packets missing between s1 and s2 (one less than the distance).
Since this condition appears in many places of the code, it has been put into a
separate function, dccp_loss_free().
Further changes:
----------------
* tidied up incorrect typing (it was using `int' for u64/s64 types);
* optimised conditional statements for common case of non-reordered packets;
* rewrote comments/documentation to match the changes.
Mark McLoughlin [Tue, 8 Jul 2008 07:10:42 +0000 (17:10 +1000)]
virtio_net: Set VIRTIO_NET_F_GUEST_CSUM feature
We can handle receiving partial csums, so set the
appropriate feature bit.
Signed-off-by: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:14:44 +0000 (15:14 -0700)]
igb: Improve multiqueue AIM support
Improve multiqueue performance
Change itr_val to reflect ITR timer value instead of ints/sec
Cleaned up AIM algorithms in general
Based on work by Mitch Williams
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Acked-by: Mitch Williams <mitch.a.williams@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:14:04 +0000 (15:14 -0700)]
igb: unused variable warning in igb remove
Wrap hw variable declaration in DCA flags to prevent unused variable
warning during compilation.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:13:38 +0000 (15:13 -0700)]
igb: update suspend resume
Updates the suspend and resume to better handle the possibility of MSIX
vector changes.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:13:05 +0000 (15:13 -0700)]
net: add netif_napi_del function to allow for removal of napistructs
Adds netif_napi_del function which is used to remove the napi struct from
the netdev napi_list in cases where CONFIG_NETPOLL was enabled.
The motivation for adding this is to handle the case in which the number of
queues on a device changes due to a configuration change. Previously the
napi structs for each queue would be left in the list until the netdev was
freed.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:12:13 +0000 (15:12 -0700)]
igb: add support for in kernel LRO
This patch adds support for the use of the inet_lro module to provide
software LRO support.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:11:40 +0000 (15:11 -0700)]
igb: add page recycling support
This patch adds support for page recycling by splitting the page into two
usable portions and tracking the reference count.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:10:46 +0000 (15:10 -0700)]
igb: Add support for quad port WOL and feature flags
Change igb from using a series of boolean operators to using a single flags
value that contains a number of different bit flags for all the different
features of the adapter.
This patch also adds WOL support for quad port adapters.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
We can remove a clunky workaround for not having the hardware
strip the CRC. 82575 silicon as well as the older PCI Express
e1000e hardware all work OK in this respect.
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Jeb Cramer [Tue, 8 Jul 2008 22:07:55 +0000 (15:07 -0700)]
igb: add DCA support
Add DCA support in the similar method that it was added to the ixgbe
driver recently. DCA allows the network device to put data in the
CPU cache and notify the chipset of that event. This reduces cache
misses during receives.
Signed-off-by: Jeb Cramer <cramerj@intel.com> Signed-off-by: Mitch Williams <mitch.a.williams@intel.com> Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com> Signed-off-by: Shannon Nelson <shannon.nelson@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Alexander Duyck [Tue, 8 Jul 2008 22:07:24 +0000 (15:07 -0700)]
igb: update ethtool stats to support multiqueue
Addesses problems seen earlier with igb driver not correctly reporting rx
and tx stats.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
igb: Introduce multiple TX queues with infrastructure
This code adds multiple Tx queue infrastructure much like we
previously did in ixgbe. The MSI-X vector mapping is the bulk of
the change.
IAM can now be safely enabled and we've verified that it does
work correctly. We can also eliminate the tx ring lock.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com> Signed-off-by: Mitch Williams <mitch.a.williams@intel.com> Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
rx cleanup should look more like our other drivers that have evolved
to nicer performance levels over time. Changes consist of refilling
tx buffers to hardware more often, some minor assignment cleanups.
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
a) kernel developers suggest LLTX is broken and unsafe to use, remove it.
b) remember to pre-stop the queue if we won't have room
c) removing lltx means we can remove our tx lock
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
some random coverage testing found that when changing mtu
under heavy traffic load, NAPI would use the rx_buffer_len variable
after it had been changed by change_mtu.
Similar to e1000 bugs found long ago.
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
ixgb: move time stamp set before setting dma pointer
a user pointed out that setting variables out of order with respect
to the checks we make for tx timeout handling could result in a race
where ->dma was set but ->time_stamp was set to the old value.
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
in order to prevent the case where poll_disable is waiting
on our device to permanently, check the flag to make sure we're not
down or closing down before re-enabling interrupts.
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
ixgb: fix bug in descriptor ring due to prefetch corruption
there was one more bug hidden in the prefetch routines in ixgb hardware
that force us to remove it completely. Writebacks were being done on
descriptors with stale data due to internal hardware fifo corruption.
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
back when maybe stop tx was added to the ixgb driver some mistakes
were made and the driver
a) didn't remove the tx lock, which is now un-necessary
b) didn't change the restart code to be compliant with maybe_stop
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
- Enable msi-x link interrupts because the timer based scheduler was getting
cancelled causing the link state to be lost with repetitive card up/downs
when changing the mtu.
- Unmask mac_rmac_link interrupts only for Xframe I and prevent a spurious
link interrupt in Xframe II.
- Stop the tx queue and indicate link down when card is down
Updates of the RBRDU have two different meanings depending on their
context:
1. the receiving process has not started - the value which is written
into the RBRDU register is supposed to be the free rx descriptors
count (rounded to a multiple of 4)
2. the receiving process is running - the value increments the count
above (sic)
The update is currently issued deep inside the rx replenish chain (see
velocity_give_many_rx_descs).
Let's propagate enough information to the caller so that the rx
replenish functions do not depend on hardware any more.
It is needed to perform the Rx/Tx buffers housekeeping when MTU changes.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Executive summary: the bounce buffers are in my way
- they use something like a 64 * 1500 bytes area of PCI
consistent area
- they are not resized when the MTU changes
- they are used
- to hand-pad undersized packets. skb_pad anyone ?
- to linearize fragmented skbs whose fragment count
goes beyond the 7 fragments hardware limit in order
to claim scatter-gather support
Actually the SG code is commented out and I wonder if it
could not be implemented (ab-)using the large send feature
of the chipset since the latter should support some
multi-descriptor packet transmitting.
Max Krasnyansky [Thu, 10 Jul 2008 23:59:11 +0000 (16:59 -0700)]
tun: Persistent devices can get stuck in xoff state
The scenario goes like this. App stops reading from tun/tap.
TX queue gets full and driver does netif_stop_queue().
App closes fd and TX queue gets flushed as part of the cleanup.
Next time the app opens tun/tap and starts reading from it but
the xoff state is not cleared. We're stuck.
Normally xoff state is cleared when netdev is brought up. But
in the case of persistent devices this happens only during
initial setup.
The fix is trivial. If device is already up when an app opens
it we clear xoff state and that gets things moving again.
Signed-off-by: Max Krasnyansky <maxk@qualcomm.com> Tested-by: Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
xfrm: Add a XFRM_STATE_AF_UNSPEC flag to xfrm_usersa_info
Add a XFRM_STATE_AF_UNSPEC flag to handle the AF_UNSPEC behavior for
the selector family. Userspace applications can set this flag to leave
the selector family of the xfrm_state unspecified. This can be used
to to handle inter family tunnels if the selector is not set from
userspace.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Ben Hutchings [Thu, 10 Jul 2008 23:52:52 +0000 (16:52 -0700)]
ipv4: fib_trie: Fix lookup error return
In commit a07f5f508a4d9728c8e57d7f66294bf5b254ff7f "[IPV4] fib_trie: style
cleanup", the changes to check_leaf() and fn_trie_lookup() were wrong - where
fn_trie_lookup() would previously return a negative error value from
check_leaf(), it now returns 0.
Now fn_trie_lookup() doesn't appear to care about plen, so we can revert
check_leaf() to returning the error value.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com> Tested-by: William Boughton <bill@boughton.de> Acked-by: Stephen Heminger <shemminger@vyatta.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Take out the confusing language in tcp_frto, and organize the
undocumented values.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com> Acked-by: Randy Dunlap <rdunlap@xenotime.net> Signed-off-by: David S. Miller <davem@davemloft.net>
David Howells [Wed, 9 Jul 2008 22:06:45 +0000 (15:06 -0700)]
netfilter: nf_nat_snmp_basic: fix a range check in NAT for SNMP
Fix a range check in netfilter IP NAT for SNMP to always use a big enough size
variable that the compiler won't moan about comparing it to ULONG_MAX/8 on a
64-bit platform.
Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Patrick McHardy [Wed, 9 Jul 2008 22:06:12 +0000 (15:06 -0700)]
netfilter: nf_conntrack_tcp: fix endless loop
When a conntrack entry is destroyed in process context and destruction
is interrupted by packet processing and the packet is an attempt to
reopen a closed connection, TCP conntrack tries to kill the old entry
itself and returns NF_REPEAT to pass the packet through the hook
again. This may lead to an endless loop: TCP conntrack repeatedly
finds the old entry, but can not kill it itself since destruction
is already in progress, but destruction in process context can not
complete since TCP conntrack is keeping the CPU busy.
Drop the packet in TCP conntrack if we can't kill the connection
ourselves to avoid this.
Reported by: hemao77@gmail.com [ Kernel bugzilla #11058 ] Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
If a mesh or ad-hoc interface is brought up and later it is replaced
by managed interface, the managed interface will keep transmitting
the beacons that were configured for the former interface. This patch
fixes that behaviour.
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Ivo van Doorn [Tue, 8 Jul 2008 11:45:20 +0000 (13:45 +0200)]
rt2x00: Disable synchronization during initialization
As soon as init_registers() was called, the rt2400/rt2500
would start raising beacondone interrupts. Since this is highly
premature since no beacons were provided yet, we should
initialize the synchronization register to 0.
This will make all drivers initialize it to 0 regardless
if they are raising beacondone interrupts or not, since it only
makes sense to have it completely disabled.
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This removes the fast_start parameter from the rc_pid parameters
information and instead uses the parameter macro when initializing
the rc_pid state. Since the parameter is only used on initialization,
there is no point of making exporting it via debugfs. This also fixes
uninitialized memory references to the fast_start and norm_offset
parameters detected by the kmemcheck utility. Thanks to Vegard Nossum
for reporting the bug.
Signed-off-by: Mattias Nissler <mattias.nissler@gmx.de> Signed-off-by: John W. Linville <linville@tuxdriver.com>
David S. Miller [Wed, 9 Jul 2008 05:49:00 +0000 (22:49 -0700)]
netdev: Kill qdisc_ingress, use netdev->rx_queue.qdisc instead.
Now that our qdisc management is bi-directional, per-queue, and fully
orthogonal, there is no reason to have a special ingress qdisc pointer
in struct net_device.
Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Wed, 9 Jul 2008 00:33:13 +0000 (17:33 -0700)]
netdev: The ingress_lock member is no longer needed.
Every qdisc is assosciated with a queue, and in the case of ingress
qdiscs that will now be netdev->rx_queue so using that queue's lock is
the thing to do.
Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Wed, 9 Jul 2008 00:18:23 +0000 (17:18 -0700)]
netdev: Move queue_lock into struct netdev_queue.
The lock is now an attribute of the device queue.
One thing to notice is that "suspicious" places
emerge which will need specific training about
multiple queue handling. They are so marked with
explicit "netdev->rx_queue" and "netdev->tx_queue"
references.
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com> Acked-by: Randy Dunlap <randy.dunlap@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-2.6 / log