01 Jun, 2011
3 commits
-
If the peer restart the asoc, we should not only fail any unsent/unacked
data, but also stop the T3-rtx, SACK, T4-rto timers, and teardown ASCONF
queues.Signed-off-by: Wei Yongjun
Signed-off-by: David S. Miller -
This patch fixes a driver crash during packet reception due to not enough
bytes allocated in the skb. Since the loop reads out 4 bytes at a time, we
need to allow for up to 3 bytes of slack space.Signed-off-by: Dennis Aberilla
Signed-off-by: David S. Miller -
The current code takes an unaligned pointer and does htonl() on it to
make it big-endian, then does a memcpy(). The problem is that the
compiler decides that since the pointer is to a __be32, it is legal
to optimize the copy into a processor word store. However, on an
architecture that does not handled unaligned writes in kernel space,
this produces an unaligned exception fault.The solution is to track the pointer as a "char *" (which removes a bunch
of unpleasant casts in any case), and then just use put_unaligned_be32()
to write the value to memory.Signed-off-by: Chris Metcalf
Signed-off-by: David S. Miller
28 May, 2011
12 commits
-
Ingo Molnar noticed that we have this unnecessary ratelimit.h
dependency in linux/net.h, which hid compilation problems from
people doing builds only with CONFIG_NET enabled.Move this stuff out to a seperate net/net_ratelimit.h file and
include that in the only two places where this thing is needed.Signed-off-by: David S. Miller
Acked-by: Ingo Molnar -
Several networking headers were depending upon the implicit
linux/sysctl.h include they get when including linux/net.hAdd explicit includes.
Signed-off-by: David S. Miller
-
This got missed back in 2006 when Jes Sorensen deleted
net/ethernet/sysctl_net_ether.cSigned-off-by: David S. Miller
-
Several crashes in cleanup_once() were reported in recent kernels.
Commit d6cc1d642de9 (inetpeer: various changes) added a race in
unlink_from_unused().One way to avoid taking unused_peers.lock before doing the list_empty()
test is to catch 0->1 refcnt transitions, using full barrier atomic
operations variants (atomic_cmpxchg() and atomic_inc_return()) instead
of previous atomic_inc() and atomic_add_unless() variants.We then call unlink_from_unused() only for the owner of the 0->1
transition.Add a new atomic_add_unless_return() static helper
With help from Arun Sharma.
Refs: https://bugzilla.kernel.org/show_bug.cgi?id=32772
Reported-by: Arun Sharma
Reported-by: Maximilian Engelhardt
Reported-by: Yann Dupont
Reported-by: Denys Fedoryshchenko
Signed-off-by: Eric Dumazet
Signed-off-by: David S. Miller -
It's currently exposed only through /proc which, besides requiring
screen-scraping, doesn't allow userspace to distinguish between two
identical ATM adapters with different ATM indexes. The ATM device index
is required when using PPPoATM on a system with multiple ATM adapters.Signed-off-by: Dan Williams
Reviewed-by: Eric Dumazet
Tested-by: David Woodhouse
Cc: stable@kernel.org
Signed-off-by: David S. Miller -
The AR9287 calibration code was not being called because of an
incorrect MAC revision check.
This forced the AR9287 to use the AR9285 initial calibration code and
bypass the AR9287 code entirely.Signed-off-by: Adrian Chadd
Signed-off-by: John W. Linville -
Commit 79f460ca49d8d5700756ab7071c951311c7f29cc add a duplicate
linux/slab.h include to net/mac80211/scan.c - remove it.Signed-off-by: Jesper Juhl
Acked-by: Luciano Coelho
Signed-off-by: John W. Linville -
local->ps_data wasn't cleared on disassociation, which
(in some corner cases) caused reconnections to enter
psm before association completed.Signed-off-by: Eliad Peller
Signed-off-by: John W. Linville -
We make oldconfig every time when a new kernel arrives, but
if we don't have such a device(I guess this is the most common
case for a new device), the default value should be 'n' so
that the kernel size we build doesn't grow up too much quickly.
For anyone who has the device, it is OK for them to turn it on
by themselves.Cc: "John W. Linville"
Cc: Johannes Berg
Signed-off-by: Tao Ma
Signed-off-by: John W. Linville -
rx_status.band is used uninitialized, what disallow to work on 5GHz .
Cc: stable@kernel.org # 2.6.39+
Signed-off-by: Stanislaw Gruszka
Signed-off-by: John W. Linville
27 May, 2011
20 commits
-
When ip_vs was adapted to netns the ftp application was not adapted
in a correct way.
However this is a fix to avoid kernel errors. In the long term another solution
might be chosen. I.e the ports that the ftp appl, uses should be per netns.Signed-off-by: Hans Schillstrom
Acked-by: Julian Anastasov
Signed-off-by: Pablo Neira Ayuso -
Stop tx queues before updating rate control to ensure
proper rate selection. Otherwise packets can be transmitted
in 40 Mhz whereas hw is configured in HT20.Signed-off-by: Rajkumar Manoharan
Signed-off-by: John W. Linville -
Whenever there is a channel width change from 40 Mhz to 20 Mhz,
the hardware is reconfigured to ht20. Meantime before doing
the rate control updation, the packets are being transmitted are
selected rate with IEEE80211_TX_RC_40_MHZ_WIDTH.While transmitting ht40 rate packets in ht20 mode is causing
baseband panic with AR9003 based chips.==== BB update: BB status=0x02001109 ====
ath: ** BB state: wd=1 det=1 rdar=0 rOFDM=1 rCCK=1 tOFDM=0 tCCK=0 agc=2
src=0 **
ath: ** BB WD cntl: cntl1=0xffff0085 cntl2=0x00000004 **
ath: ** BB mode: BB_gen_controls=0x000033c0 **
ath: ** BB busy times: rx_clear=99%, rx_frame=0%, tx_frame=0% **
ath: ==== BB update: done ====Cc: stable@kernel.org
Signed-off-by: Rajkumar Manoharan
Signed-off-by: John W. Linville -
While receiving unsupported rate frame rx state machine
gets into a state 0xb and if phy_restart happens in that
state, BB would go hang. If RXSM is in 0xb state after
first bb panic, ensure to disable the phy_restart.Signed-off-by: Rajkumar Manoharan
Signed-off-by: John W. Linville -
Resetting hardware helps to recover from baseband
hang/panic for AR9003 based chips.Cc: stable@kernel.org
Signed-off-by: Rajkumar Manoharan
Signed-off-by: John W. Linville -
Reported-by: Larry Finger
Signed-off-by: Rafał Miłecki
Signed-off-by: John W. Linville -
Although a previous fix handles the kernel panics that result from
failure to allocate a new RX buffer, memory fragmentation can be
reduced if the amsdu_8k capability is disabled as new buffers need only
be of O(0), not O(2).Signed-off-by: Larry Finger
Signed-off-by: John W. Linville -
To handle amsdu_8k capability, the PCI routine of this driver must
allocate receive buffers of order 2. Under heavy load, this causes
fragmentation of memory. The present code releases the current buffer
before checking to see if a new one is availble. Recovery from
allocation failures is not possible, which results in kernel panics.The fix is to reorder the code to check that a new buffer can be
allocated before the old one is released. If not possible, the
received frame is dropped and the old one is reused. Without this
change, it is impossible to transfer a 2 GB file without a kernel panic.Signed-off-by: Larry Finger
Cc: Stable [2.6.{37,38,39}]
Signed-off-by: John W. Linville -
In both trigger_scan and sched_scan operations, we were checking for
the SSID length before assigning the value correctly. Since the
memory was just kzalloc'ed, the check was always failing and SSID with
over 32 characters were allowed to go through.This was causing a buffer overflow when copying the actual SSID to the
proper place.This bug has been there since 2.6.29-rc4.
Cc: stable@kernel.org
Signed-off-by: Luciano Coelho
Signed-off-by: John W. Linville -
While decoding received event packet from firmware, 4 bytes
of interface header are already removed unconditionally.
So for handling event only 4 more bytes needs to be pulled.
This is achieved by changing event header length to 4.Almost all the events, except BA stream related and AMSDU
aggregation control events, do not have the payload in their
event skb. Such events handling depends only on the event ID.
This event ID is the first four bytes of the event skb, which
is copied to a separate variable before pulling the skb header.
Hence event handling worked only for those events that didn't
have payload in event skb.This patch fixes the broken event path of the events with
payload in their event skb without harming existing working
event path for the events without payload.Signed-off-by: Yogesh Ashok Powar
Signed-off-by: Kiran Divekar
Signed-off-by: Bing Zhao
Signed-off-by: John W. Linville -
As reported by Ingo Molnar, we still have configuration combinations
where use of the WARN_RATELIMIT interfaces break the build because
dependencies don't get met.Instead of going down the long road of trying to make it so that
ratelimit.h can get included by kernel.h or asm-generic/bug.h,
just move the interface into ratelimit.h and make users have
to include that.Reported-by: Ingo Molnar
Signed-off-by: David S. Miller
Acked-by: Randy Dunlap -
Weiping Pan noticed that the module option description for
xmit_hash_policy was incorrect and was nice enough to post a patch to
fix it. The text was correct, but created a line over 80 characters and
I would rather not add those. I realized I could take a few minutes and
clean up all the descriptions and things would look much better. This
is the result.Based on patch from Weiping Pan .
Signed-off-by: Andy Gospodarek
CC: Weiping Pan
Reviewed-by: Weiping Pan
Signed-off-by: David S. Miller -
The below patch removes vlan_buggyright and vlan_copyright from vlan_proto_init,
so that it prints out just the fullname of vlan and the version number.before:
[ 30.438203] 802.1Q VLAN Support v1.8 Ben Greear
[ 30.441542] All bugs added by David S. Millerafter:
[ 31.513910] 802.1Q VLAN Support v1.8
Signed-off-by: Justin P. Mattock
CC: Joe Perches
CC: David S. Miller
CC: Ben Greear
Signed-off-by: David S. Miller -
Use platform device rather than net device in dev_err calls before net
device has been registered to avoid messages such as(null): DaVinci EMAC: Failed to get EMAC clock
Also replace remaining printks in probe with dev_{err,warn}.
Signed-off-by: Johan Hovold
Signed-off-by: David S. Miller -
As these pointers have been printed without using %p they were missed in the
big network kptr_restrict conversion patch %p -> %pK from Dan Rosenberg.Signed-off-by: Oliver Hartkopp
Signed-off-by: David S. Miller -
Current code squashes flags to bool - this makes set_flags fail whenever
some ETH_FLAG_* equivalent features are set. Fix this.Signed-off-by: Michał Mirosław
Signed-off-by: David S. Miller -
Kill set but not used 'entry_offset'.
Add a default case to the switch statement so the compiler
can see that we always initialize off and size_kern before
using them.Signed-off-by: David S. Miller
Signed-off-by: Pablo Neira Ayuso -
ip_set_flush returned -EPROTO instead of -IPSET_ERR_PROTOCOL, fixed
Signed-off-by: Jozsef Kadlecsik
Signed-off-by: Pablo Neira Ayuso -
Variable 'ret' is set in type_pf_tdel() but not used, remove.
Signed-off-by: Jozsef Kadlecsik
Signed-off-by: Pablo Neira Ayuso -
Signed-off-by: Jozsef Kadlecsik
Signed-off-by: Pablo Neira Ayuso
26 May, 2011
5 commits
-
Improves the documentation about how IGMP resend parameter
works, fix two missing checks and coding style issues.Signed-off-by: Flavio Leitner
Acked-by: Rick Jones
Signed-off-by: David S. Miller -
This soft lockup was recently reported:
[root@dell-per715-01 ~]# echo +bond5 > /sys/class/net/bonding_masters
[root@dell-per715-01 ~]# echo +eth1 > /sys/class/net/bond5/bonding/slaves
bonding: bond5: doing slave updates when interface is down.
bonding bond5: master_dev is not up in bond_enslave
[root@dell-per715-01 ~]# echo -eth1 > /sys/class/net/bond5/bonding/slaves
bonding: bond5: doing slave updates when interface is down.BUG: soft lockup - CPU#12 stuck for 60s! [bash:6444]
CPU 12:
Modules linked in: bonding autofs4 hidp rfcomm l2cap bluetooth lockd sunrpc
be2d
Pid: 6444, comm: bash Not tainted 2.6.18-262.el5 #1
RIP: 0010:[] []
.text.lock.spinlock+0x26/00
RSP: 0018:ffff810113167da8 EFLAGS: 00000286
RAX: ffff810113167fd8 RBX: ffff810123a47800 RCX: 0000000000ff1025
RDX: 0000000000000000 RSI: ffff810123a47800 RDI: ffff81021b57f6f8
RBP: ffff81021b57f500 R08: 0000000000000000 R09: 000000000000000c
R10: 00000000ffffffff R11: ffff81011d41c000 R12: ffff81021b57f000
R13: 0000000000000000 R14: 0000000000000282 R15: 0000000000000282
FS: 00002b3b41ef3f50(0000) GS:ffff810123b27940(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00002b3b456dd000 CR3: 000000031fc60000 CR4: 00000000000006e0Call Trace:
[] _spin_lock_bh+0x9/0x14
[] :bonding:tlb_clear_slave+0x22/0xa1
[] :bonding:bond_alb_deinit_slave+0xba/0xf0
[] :bonding:bond_release+0x1b4/0x450
[] __down_write_nested+0x12/0x92
[] :bonding:bonding_store_slaves+0x25c/0x2f7
[] sysfs_write_file+0xb9/0xe8
[] vfs_write+0xce/0x174
[] sys_write+0x45/0x6e
[] tracesys+0xd5/0xe0It occurs because we are able to change the slave configuarion of a bond while
the bond interface is down. The bonding driver initializes some data structures
only after its ndo_open routine is called. Among them is the initalization of
the alb tx and rx hash locks. So if we add or remove a slave without first
opening the bond master device, we run the risk of trying to lock/unlock a
spinlock that has garbage for data in it, which results in our above softlock.Note that sometimes this works, because in many cases an unlocked spinlock has
the raw_lock parameter initialized to zero (meaning that the kzalloc of the
net_device private data is equivalent to calling spin_lock_init), but thats not
true in all cases, and we aren't guaranteed that condition, so we need to pass
the relevant spinlocks through the spin_lock_init function.Fix it by moving the spin_lock_init calls for the tx and rx hashtable locks to
the ndo_init path, so they are ready for use by the bond_store_slaves path.Change notes:
v2) Based on conversation with Jay and Nicolas it seems that the ability to
enslave devices while the bond master is down should be safe to do. As such
this is an outlier bug, and so instead we'll just initalize the errant spinlocks
in the init path rather than the open path, solving the problem. We'll also
remove the warnings about the bond being down during enslave operations, since
it should be safev3) Fix spelling error
Signed-off-by: Neil Horman
Reported-by: jtluka@redhat.com
CC: Jay Vosburgh
CC: Andy Gospodarek
CC: nicolas.2p.debian@gmail.com
CC: "David S. Miller"
Signed-off-by: Jay Vosburgh
Signed-off-by: David S. Miller -
Commit e67f88dd12f6 (dont hold rtnl mutex during netlink dump callbacks)
missed fact that rtnl_fill_ifinfo() must be called with rtnl held.Because of possible deadlocks between two mutexes (cb_mutex and rtnl),
its not easy to solve this problem, so revert this part of the patch.It also forgot one rcu_read_unlock() in FIB dump_rules()
Add one ASSERT_RTNL() in rtnl_fill_ifinfo() to remind us the rule.
Signed-off-by: Eric Dumazet
CC: Patrick McHardy
CC: Stephen Hemminger
Signed-off-by: David S. Miller -
This patch adds the PCI ID of Fujitsu 1000base-SX NIC to tg3 driver.
Tested to detect the card, MAC and serdes, not tested with link at the
moment since I have no fiber switch here. I did not add new constants to
the pci_ids.h header file since these constants are used only here.Signed-off-by: Meelis Roos
Signed-off-by: David S. Miller -
Signed-off-by: Dmitry Kravkov
Signed-off-by: Eilon Greenstein
Signed-off-by: David S. Miller
