Eric Dumazet [Sat, 29 May 2010 07:20:48 +0000 (00:20 -0700)]
net: fix sk_forward_alloc corruptions
As David found out, sock_queue_err_skb() should be called with socket
lock hold, or we risk sk_forward_alloc corruption, since we use non
atomic operations to update this field.
This patch adds bh_lock_sock()/bh_unlock_sock() pair to three spots.
(BH already disabled)
1) skb_tstamp_tx()
2) Before calling ip_icmp_error(), in __udp4_lib_err()
3) Before calling ipv6_icmp_error(), in __udp6_lib_err()
Reported-by: Anton Blanchard <anton@samba.org> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Carpenter [Wed, 26 May 2010 04:47:39 +0000 (04:47 +0000)]
be2net: remove superfluous externs
This fixes some sparse warnings:
drivers/net/benet/be_cmds.c:1503:12: warning: function
'be_cmd_enable_magic_wol' with external linkage has definition
drivers/net/benet/be_cmds.c:1668:12: warning: function
'be_cmd_get_seeprom_data' with external linkage has definition
Signed-off-by: Dan Carpenter <error27@gmail.com> Acked-by: Sarveshwar Bandi <sarveshwarb@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Carpenter [Wed, 26 May 2010 04:46:35 +0000 (04:46 +0000)]
be2net: add unlock on error path
The unlock accidentally got removed from the error path in dd131e76e5:
"be2net: Bug fix to avoid disabling bottom half during firmware upgrade."
Signed-off-by: Dan Carpenter <error27@gmail.com> Acked-by: Sarveshwar Bandi <sarveshwarb@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Julia Lawall [Wed, 26 May 2010 05:54:21 +0000 (05:54 +0000)]
net/rds: Add missing mutex_unlock
Add a mutex_unlock missing on the error path. In each case, whenever the
label out is reached from elsewhere in the function, mutex is not locked.
The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@@
expression E1;
@@
* mutex_lock(E1);
<+... when != E1
if (...) {
... when != E1
* return ...;
}
...+>
* mutex_unlock(E1);
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk> Reviewed-by: Zach Brown <zach.brown@oracle.com> Acked-by: Andy Grover <andy.grover@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Add a spin_unlock missing on the error path. The return value of write_reg
seems to be completely ignored, so it seems that the lock should be
released in every case.
The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@@
expression E1;
@@
* spin_lock(E1,...);
<+... when != E1
if (...) {
... when != E1
* return ...;
}
...+>
* spin_unlock(E1,...);
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk> Signed-off-by: David S. Miller <davem@davemloft.net>
Mark Ware [Sat, 29 May 2010 07:16:28 +0000 (00:16 -0700)]
fs_enet: Adjust BDs after tx error
This patch fixes an occasional transmit lockup in the mac-fcc which
occurs after a tx error. The test scenario had the local port set
to autoneg and the other end fixed at 100FD, resulting in a large
number of late collisions.
According to the MPC8280RM 30.10.1.3 (also 8272RM 29.10.1.3), after
a tx error occurs, TBPTR may sometimes point beyond BDs still marked
as ready. This patch walks back through the BDs and points TBPTR to
the earliest one marked as ready.
Tested on a custom board with a MPC8280.
Signed-off-by: Mark Ware <mware@elphinstone.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Brian Haley [Sat, 29 May 2010 06:02:35 +0000 (23:02 -0700)]
IPv6: fix Mobile IPv6 regression
Commit f4f914b5 (net: ipv6 bind to device issue) caused
a regression with Mobile IPv6 when it changed the meaning
of fl->oif to become a strict requirement of the route
lookup. Instead, only force strict mode when
sk->sk_bound_dev_if is set on the calling socket, getting
the intended behavior and fixing the regression.
Tested-by: Arnaud Ebalard <arno@natisbad.org> Signed-off-by: Brian Haley <brian.haley@hp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Prarit Bhargava [Thu, 27 May 2010 18:41:20 +0000 (14:41 -0400)]
libertas: fix uninitialized variable warning
Fixes:
drivers/net/wireless/libertas/rx.c: In function process_rxed_802_11_packet:
drivers/net/wireless/libertas/rx.c:354: error: radiotap_hdr.flags may be used uninitialized in this function
Signed-off-by: Prarit Bhargava <prarit@redhat.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath9k: Remove ATH9K_TX_SW_ABORTED and introduce a bool for this purpose
Wrong buffer is checked for bf_tx_aborted field in ath_tx_num_badfrms(),
this may result in a rate scaling with wrong feedback (number
of unacked frames in this case). It is the last one in the chain
of buffers for an aggregate frame that should be checked.
Also it misses the initialization of this field in the buffer,
this may lead to a situation where we stop the sw retransmission
of failed subframes associated to this buffer.
Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath_print in xmit.c should say "Reseting hardware"
instead of Resetting HAL!(since HAL is being fazed out).
dmesg shows:
[ 8660.899624] ath: Failed to stop TX DMA in 100 msec after killing last frame
[ 8660.899676] ath: Unable to stop TxDMA. Reset HAL!
Signed-off-by: Justin P. Mattock <justinmattock@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Scott Feldman [Fri, 28 May 2010 10:42:43 +0000 (03:42 -0700)]
netlink: bug fix: wrong size was calculated for vfinfo list blob
The wrong size was being calculated for vfinfo. In one case, it was over-
calculating using nlmsg_total_size on attrs, in another case, it was
under-calculating by assuming ifla_vf_* structs are packed together, but
each struct is it's own attr w/ hdr (and padding).
Signed-off-by: Scott Feldman <scofeldm@cisco.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Scott Feldman [Fri, 28 May 2010 10:42:18 +0000 (03:42 -0700)]
netlink: bug fix: don't overrun skbs on vf_port dump
Noticed by Patrick McHardy: was continuing to fill skb after a
nla_put_failure, ignoring the size calculated by upper layer. Now,
return -EMSGSIZE on any overruns, but also allow netdev to
fail ndo_get_vf_port with error other than -EMSGSIZE, thus unwinding
nest.
Signed-off-by: Scott Feldman <scofeldm@cisco.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Fri, 28 May 2010 10:41:17 +0000 (03:41 -0700)]
xt_tee: use skb_dst_drop()
After commit 7fee226a (net: add a noref bit on skb dst), its wrong to
use : dst_release(skb_dst(skb)), since we could decrement a refcount
while skb dst was not refcounted.
We should use skb_dst_drop(skb) instead.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Bryan Wu [Fri, 28 May 2010 10:40:39 +0000 (03:40 -0700)]
netdev/fec: fix ifconfig eth0 down hang issue
BugLink: http://bugs.launchpad.net/bugs/559065
In fec open/close function, we need to use phy_connect and phy_disconnect
operation before we start/stop phy. Otherwise it will cause system hang.
Only call fec_enet_mii_probe() in open function, because the first open
action will cause NULL pointer error.
Signed-off-by: Bryan Wu <bryan.wu@canonical.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Thu, 27 May 2010 23:31:41 +0000 (16:31 -0700)]
cnic: Fix context memory init. on 5709.
We need to zero context memory on 5709 in the function cnic_init_context().
Without this, iscsid restart on 5709 will not work because of stale data.
TX context blocks should not be initialized by cnic_init_context() because
of the special remapping on 5709.
Update version to 2.1.2.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Thu, 27 May 2010 23:14:30 +0000 (16:14 -0700)]
ipv6: Add GSO support on forwarding path
Currently we disallow GSO packets on the IPv6 forward path.
This patch fixes this.
Note that I discovered that our existing GSO MTU checks (e.g.,
IPv4 forwarding) are buggy in that they skip the check altogether,
when they really should be checking gso_size + header instead.
I have also been lazy here in that I haven't bothered to segment
the GSO packet by hand before generating an ICMP message. Someone
should add that to be 100% correct.
Reported-by: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Thu, 27 May 2010 23:09:39 +0000 (16:09 -0700)]
net: fix __neigh_event_send()
commit 7fee226ad23 (net: add a noref bit on skb dst) missed one spot
where an skb is enqueued, with a possibly not refcounted dst entry.
__neigh_event_send() inserts skb into arp_queue, so we must make sure
dst entry is refcounted, or dst entry can be freed by garbage collector
after caller exits from rcu protected section.
Reported-by: Ingo Molnar <mingo@elte.hu> Tested-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Takuya Yoshikawa [Thu, 27 May 2010 10:01:58 +0000 (19:01 +0900)]
vhost-net: fix to check the return value of copy_to/from_user() correctly
copy_to/from_user() returns the number of bytes that could not be copied.
So we need to check if it is not zero, and in that case, we should return
the error number -EFAULT rather than directly return the return value from
copy_to/from_user().
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Takuya Yoshikawa [Thu, 27 May 2010 09:58:03 +0000 (18:58 +0900)]
vhost: fix to check the return value of copy_to/from_user() correctly
copy_to/from_user() returns the number of bytes that could not be copied.
So we need to check if it is not zero, and in that case, we should return
the error number -EFAULT rather than directly return the return value from
copy_to/from_user().
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Eric Dumazet [Wed, 26 May 2010 19:20:18 +0000 (19:20 +0000)]
net: fix lock_sock_bh/unlock_sock_bh
This new sock lock primitive was introduced to speedup some user context
socket manipulation. But it is unsafe to protect two threads, one using
regular lock_sock/release_sock, one using lock_sock_bh/unlock_sock_bh
This patch changes lock_sock_bh to be careful against 'owned' state.
If owned is found to be set, we must take the slow path.
lock_sock_bh() now returns a boolean to say if the slow path was taken,
and this boolean is used at unlock_sock_bh time to call the appropriate
unlock function.
After this change, BH are either disabled or enabled during the
lock_sock_bh/unlock_sock_bh protected section. This might be misleading,
so we rename these functions to lock_sock_fast()/unlock_sock_fast().
Reported-by: Anton Blanchard <anton@samba.org> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Tested-by: Anton Blanchard <anton@samba.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Brian Hill [Thu, 27 May 2010 03:44:30 +0000 (20:44 -0700)]
net: ll_temac: fix checksum offload logic
The current checksum offload code does not work and this corrects
that functionality. It also updates the interrupt coallescing
initialization so than there are fewer interrupts and performance
is increased.
Signed-off-by: Brian Hill <brian.hill@xilinx.com> Signed-off-by: John Linn <john.linn@xilinx.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Commit "ar9170: wait for asynchronous firmware loading"
introduced a bug, which is triggered by fatal errors
while the driver is initializing the device.
BUG: unable to handle kernel paging request at 6b6b6bf7
IP: [<c117b567>] kobject_put+0x7/0x70
*pde = 00000000
Oops: 0000 [#1] PREEMPT
last sysfs file: /sys/devices/platform/hdaps/position
Modules linked in: ar9170usb [...]
Dan Carpenter [Wed, 26 May 2010 07:40:11 +0000 (00:40 -0700)]
sctp: dubious bitfields in sctp_transport
Sparse complains because these one-bit bitfields are signed.
include/net/sctp/structs.h:879:24: error: dubious one-bit signed bitfield
include/net/sctp/structs.h:889:31: error: dubious one-bit signed bitfield
include/net/sctp/structs.h:895:26: error: dubious one-bit signed bitfield
include/net/sctp/structs.h:898:31: error: dubious one-bit signed bitfield
include/net/sctp/structs.h:901:27: error: dubious one-bit signed bitfield
It doesn't cause a problem in the current code, but it would be better
to clean it up. This was introduced by c0058a35aacc7: "sctp: Save some
room in the sctp_transport by using bitfields".
Signed-off-by: Dan Carpenter <error27@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Wed, 26 May 2010 01:53:57 +0000 (18:53 -0700)]
cls_cgroup: Initialise classid when module is absent
When the cls_cgroup module is not loaded, task_cls_classid will
return an uninitialised classid instead of zero.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Tue, 25 May 2010 23:24:03 +0000 (16:24 -0700)]
drivers/net/usb/asix.c: Fix pointer cast.
Stephen Rothwell reports the following new warning:
drivers/net/usb/asix.c: In function 'asix_rx_fixup':
drivers/net/usb/asix.c:325: warning: cast from pointer to integer of different size
drivers/net/usb/asix.c:354: warning: cast from pointer to integer of different size
The code just cares about the low alignment bits, so use
an "unsigned long" cast instead of one to "u32".
Signed-off-by: David S. Miller <davem@davemloft.net>
Sarveshwar Bandi [Tue, 25 May 2010 23:16:32 +0000 (16:16 -0700)]
be2net: Bug fix to avoid disabling bottom half during firmware upgrade.
Certain firmware commands/operations to upgrade firmware could take several
seconds to complete. The code presently disables bottom half during these
operations which could lead to unpredictable behaviour in certain cases. This
patch now does all firmware upgrade operations asynchronously using a
completion variable.
Signed-off-by: Sarveshwar Bandi <sarveshwarb@serverengines.com> Signed-off-by: David S. Miller <davem@davemloft.net>
J. R. Okajima [Tue, 25 May 2010 23:10:14 +0000 (16:10 -0700)]
proc_dointvec: write a single value
The commit 00b7c3395aec3df43de5bd02a3c5a099ca51169f
"sysctl: refactor integer handling proc code"
modified the behaviour of writing to /proc.
Before the commit, write("1\n") to /proc/sys/kernel/printk succeeded. But
now it returns EINVAL.
This commit supports writing a single value to a multi-valued entry.
Signed-off-by: J. R. Okajima <hooanon05@yahoo.co.jp> Reviewed-and-tested-by: WANG Cong <amwang@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Felix Fietkau [Thu, 13 May 2010 18:37:25 +0000 (20:37 +0200)]
ath9k: remove VEOL support for ad-hoc
With VEOL, Beacon transmission in ad-hoc does not currently work.
I believe for larger ad-hoc networks, VEOL is too unreliable, as
it can get beacon transmissions stuck during synchronization.
Use SWBA based beacon trasmission similar to AP mode instead.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Acked-by: Benoit Papillault <benoit.papillault@free.fr> Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath9k: Fix rx of mcast/bcast frames in PS mode with auto sleep
The functionality to keep the device awake until it is done with
the rx of any mcast/bcast frames which are pending on AP should
also be added to the hardwares which support auto sleep feature.
This patch fixes frequent failures in ARP resolution when it is
initiated by the other end. Currently auto sleep is enabled only
for ar9003 in ath9k.
Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Randy Dunlap [Fri, 21 May 2010 18:28:41 +0000 (11:28 -0700)]
wireless: fix sta_info.h kernel-doc warnings
Fix sta_info.h kernel-doc warnings:
Warning(net/mac80211/sta_info.h:164): No description found for parameter 'tid_active_rx[STA_TID_NUM]'
Warning(net/mac80211/sta_info.h:164): Excess struct/union/enum/typedef member 'tid_state_rx' description in 'sta_ampdu_mlme'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> Acked-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Randy Dunlap [Fri, 21 May 2010 18:28:33 +0000 (11:28 -0700)]
wireless: fix mac80211.h kernel-doc warnings
Fix kernel-doc warnings in mac80211.h:
Warning(include/net/mac80211.h:838): No description found for parameter 'ap_addr'
Warning(include/net/mac80211.h:1726): No description found for parameter 'get_survey'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Dan Carpenter [Fri, 14 May 2010 14:52:37 +0000 (16:52 +0200)]
ath9k_htc: rare leak in ath9k_hif_usb_alloc_tx_urbs()
This is obviously a small picky thing. The original error handling code
doesn't free the most recent allocations which haven't been added to the
hif_dev->tx.tx_buf list yet.
Signed-off-by: Dan Carpenter <error27@gmail.com> Acked-by: Sujith <Sujith.Manoharan@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Dan Carpenter [Fri, 14 May 2010 14:50:56 +0000 (16:50 +0200)]
ath9k_htc: dereferencing before check in hif_usb_tx_cb()
After c11d8f89d3b7: "ath9k_htc: Simplify TX URB management" we no longer
assume that tx_buf is a non-null pointer.
Signed-off-by: Dan Carpenter <error27@gmail.com> Acked-by: Sujith <Sujith.Manoharan@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
The recent changes to skb handling introduced a bug in the rt2800usb
TX descriptor writing whereby the length of the USB packet wasn't
calculated correctly.
Found via code inspection, as the devices themselves didn't seem to mind.
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com> Acked-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rt2x00: Fix failed SLEEP->AWAKE and AWAKE->SLEEP transitions.
(Based on a patch created by Ondrej Zary)
In some circumstances the Ralink devices do not properly go to sleep
or wake up, with timeouts occurring.
Fix this by retrying telling the device that it has to wake up or
sleep.
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com> Acked-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Helmut Schaa [Wed, 19 May 2010 06:47:59 +0000 (08:47 +0200)]
rt2x00: don't use to_pci_dev in rt2x00pci_uninitialize
Don't use to_pci_dev in rt2x00pci_uninitialize to get the allocated irq
as it won't work for platform devices (SoC). Instead, use the irq field
that's already used everywhere else.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com> Acked-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Bruno Randolf [Wed, 19 May 2010 01:18:16 +0000 (10:18 +0900)]
ath5k: consistently use rx_bufsize for RX DMA
We should use the same buffer size we set up for DMA also in the hardware
descriptor. Previously we used common->rx_bufsize for setting up the DMA
mapping, but used skb_tailroom(skb) for the size we tell to the hardware in the
descriptor itself. The problem is that skb_tailroom(skb) can give us a larger
value than the size we set up for DMA before. This allows the hardware to write
into memory locations not set up for DMA. In practice this should rarely happen
because all packets should be smaller than the maximum 802.11 packet size.
On the tested platform rx_bufsize is 2528, and we allocated an skb of 2559
bytes length (including padding for cache alignment) but sbk_tailroom() was
2592. Just consistently use rx_bufsize for all RX DMA memory sizes.
Also use the return value of the descriptor setup function.
Cc: stable@kernel.org Signed-off-by: Bruno Randolf <br1@einfach.org> Reviewed-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Herbert Xu [Mon, 24 May 2010 07:14:10 +0000 (00:14 -0700)]
tun: Update classid on packet injection
This patch makes tun update its socket classid every time we
inject a packet into the network stack. This is so that any
updates made by the admin to the process writing packets to
tun is effected.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Mon, 24 May 2010 07:12:34 +0000 (00:12 -0700)]
cls_cgroup: Store classid in struct sock
Up until now cls_cgroup has relied on fetching the classid out of
the current executing thread. This runs into trouble when a packet
processing is delayed in which case it may execute out of another
thread's context.
Furthermore, even when a packet is not delayed we may fail to
classify it if soft IRQs have been disabled, because this scenario
is indistinguishable from one where a packet unrelated to the
current thread is processed by a real soft IRQ.
In fact, the current semantics is inherently broken, as a single
skb may be constructed out of the writes of two different tasks.
A different manifestation of this problem is when the TCP stack
transmits in response of an incoming ACK. This is currently
unclassified.
As we already have a concept of packet ownership for accounting
purposes in the skb->sk pointer, this is a natural place to store
the classid in a persistent manner.
This patch adds the cls_cgroup classid in struct sock, filling up
an existing hole on 64-bit :)
The value is set at socket creation time. So all sockets created
via socket(2) automatically gains the ID of the thread creating it.
Whenever another process touches the socket by either reading or
writing to it, we will change the socket classid to that of the
process if it has a valid (non-zero) classid.
For sockets created on inbound connections through accept(2), we
inherit the classid of the original listening socket through
sk_clone, possibly preceding the actual accept(2) call.
In order to minimise risks, I have not made this the authoritative
classid. For now it is only used as a backup when we execute
with soft IRQs disabled. Once we're completely happy with its
semantics we can use it as the sole classid.
Footnote: I have rearranged the error path on cls_group module
creation. If we didn't do this, then there is a window where
someone could create a tc rule using cls_group before the cgroup
subsystem has been registered.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Graf Yang [Sat, 22 May 2010 22:00:10 +0000 (22:00 +0000)]
net/irda: bfin_sir: IRDA is not affected by anomaly 05000230
Anomaly 05000230 (over sampling of the UART STOP bit) applies only when
the peripheral is operating in UART mode. So drop the anomaly handling
in the IRDA code.
Signed-off-by: Graf Yang <graf.yang@analog.com> Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Dummy implementations for the optional CAPI controller operations
load_firmware and reset_ctr can cause userspace callers to hang
indefinitely. It's better not to implement them at all.
Signed-off-by: Tilman Schmidt <tilman@imap.cc> Acked-by: Karsten Keil <isdn@linux-pingi.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Tilman Schmidt [Sun, 23 May 2010 01:02:08 +0000 (01:02 +0000)]
isdn/capi: make reset_ctr op truly optional
The CAPI controller operation reset_ctr is marked as optional, and
not all drivers do implement it. Add a check to the kernel CAPI
whether it exists before trying to call it.
Signed-off-by: Tilman Schmidt <tilman@imap.cc> Acked-by: Karsten Keil <isdn@linux-pingi.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Sjur Braendeland [Fri, 21 May 2010 02:16:10 +0000 (02:16 +0000)]
caif: Bugfix - Poll can't return POLLHUP while connecting.
Discovered bug when testing async connect.
While connecting poll should not return POLLHUP,
but POLLOUT when connected.
Also fixed the sysfs flow-control-counters.
Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com> Signed-off-by: David S. Miller <davem@davemloft.net>
introduced a bug when there is a hash collision making impossible
to rename a device with eth%d. This bug is very hard to reproduce
and appears rarely.
The problem is coming from we don't pass a temporary buffer to
__dev_alloc_name but 'dev->name' which is modified by the function.
Changelog:
V2 : replaced strings comparison by pointers comparison
Signed-off-by: Daniel Lezcano <daniel.lezcano@free.fr> Reviewed-by: Octavian Purdila <opurdila@ixiacom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Author: Chris Wright <chrisw@sous-sol.org>
Date: Sun May 16 01:05:45 2010 -0700
Subject: rtnetlink: make SR-IOV VF interface symmetric
adds broken error handling to do_setlink() in net/core/rtnetlink.c. The
problem is the following chunk of code:
if (tb[IFLA_VFINFO_LIST]) {
struct nlattr *attr;
int rem;
nla_for_each_nested(attr, tb[IFLA_VFINFO_LIST], rem) {
if (nla_type(attr) != IFLA_VF_INFO)
----> goto errout;
err = do_setvfinfo(dev, attr);
if (err < 0)
goto errout;
modified = 1;
}
}
which can get to errout without setting err, resulting in the following error:
net/core/rtnetlink.c: In function 'do_setlink':
net/core/rtnetlink.c:904: warning: 'err' may be used uninitialized in this function
Change the code to return -EINVAL in this case. Note that this might not be
the appropriate error though.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Chris Wright <chrisw@sous-sol.org>
cc: David S. Miller <davem@davemloft.net> Acked-by: Chris Wright <chrisw@sous-sol.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg [Tue, 18 May 2010 12:36:34 +0000 (14:36 +0200)]
cfg80211: add missing braces
Specifying a valid channel type will get
goto out rather than continuing, due to
missing braces. This affects both remain
on channel and action frame TX commands.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Jussi Kivilinna [Tue, 18 May 2010 08:20:51 +0000 (11:20 +0300)]
rndis_wlan: replace wireless_send_event with cfg80211_disconnected
Remove (hopefully) last use of WEXT in rndis_wlan. Replace wireless_send_event
with missing cfg80211_disconnected in rndis_wlan_do_link_down_work.
Reported-by: "Rogério Brito" <rbrito@ime.usp.br> Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi> Signed-off-by: John W. Linville <linville@tuxdriver.com>
We tried to squeeze as much AR9003 support into this kernel
release cycle but there are a few features which are still
being tested and developed. Some of these features are critical
to the stable operation of AR9003 so for now disable AR9003 support
all together. This will get re-enabled once all necessary features
are in place but very likely will not happen for 2.6.35.
Reviewed-by: Don Breslin <don.breslin@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Juuso Oikarinen [Fri, 14 May 2010 07:46:24 +0000 (10:46 +0300)]
wl1271: Fix RX data path frame lengths
The current frame length used by the driver for RX frames is the SPI bus
transfer length. This length has padding bytes, which do not belong to the
WLAN frame.
As there is no other length information in the WLAN frame except the skb
length this problem caused for instance extra ESSID's to be listed at the
end of scan results (IE id 0) with zero length.
Fix the frame length by removing padding.
Signed-off-by: Juuso Oikarinen <juuso.oikarinen@nokia.com> Reviewed-by: Luciano Coelho <luciano.coelho@nokia.com> Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Reinette Chatre [Thu, 13 May 2010 21:49:44 +0000 (14:49 -0700)]
iwlwifi: fix internal scan race
It is possible for internal scan to race against itself if the device is
not returning the scan results from first requests. What happens in this
case is the cleanup done during the abort of the first internal scan also
cleans up part of the new scan, causing it to access memory it shouldn't.
Here are details:
* First internal scan is triggered and scan command sent to device.
* After seven seconds there is no scan results so the watchdog timer
triggers a scan abort.
* The scan abort succeeds and a SCAN_COMPLETE_NOTIFICATION is received for
failed scan.
* During processing of SCAN_COMPLETE_NOTIFICATION we clear STATUS_SCANNING
and queue the "scan_completed" work.
** At this time, since the problem that caused the internal scan in first
place is still present, a new internal scan is triggered.
The behavior at this point is a bit different between 2.6.34 and 2.6.35
since 2.6.35 has a lot of this synchronized. The rest of the race
description will thus be generalized.
** As part of preparing for the scan "is_internal_short_scan" is set to
true.
* At this point the completion work for fist scan is run. As part of this
there is some locking missing around the "is_internal_short_scan"
variable and it is set to "false".
** Now the second scan runs and it considers itself a real (not internal0
scan and thus causes problems with wrong memory being accessed.
The fix is twofold.
* Since "is_internal_short_scan" should be protected by mutex, fix this in
scan completion work so that changes to it can be serialized.
* Do not queue a new internal scan if one is in progress.
This fixes https://bugzilla.kernel.org/show_bug.cgi?id=15824
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Sujith [Thu, 13 May 2010 05:04:08 +0000 (10:34 +0530)]
cfg80211: Fix signal_type comparison
signal_type is enum cfg80211_signal_type.
This fixes the gcc warning:
scan.c: In function `cfg80211_inform_bss':
scan.c:518:6: warning: comparison between `enum cfg80211_signal_type' and `enum nl80211_bss'
scan.c: In function `cfg80211_inform_bss_frame':
scan.c:574:6: warning: comparison between `enum cfg80211_signal_type' and `enum nl80211_bss'
Signed-off-by: Sujith <Sujith.Manoharan@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Tom Herbert [Thu, 20 May 2010 18:37:59 +0000 (18:37 +0000)]
net: fix problem in dequeuing from input_pkt_queue
Fix some issues introduced in batch skb dequeuing for input_pkt_queue.
The primary issue it that the queue head must be incremented only
after a packet has been processed, that is only after
__netif_receive_skb has been called. This is needed for the mechanism
to prevent OOO packet in RFS. Also when flushing the input_pkt_queue
and process_queue, the process queue should be done first to prevent
OOO packets.
Because the input_pkt_queue has been effectively split into two queues,
the calculation of the tail ptr is no longer correct. The correct value
would be head+input_pkt_queue->len+process_queue->len. To avoid
this calculation we added an explict input_queue_tail in softnet_data.
The tail value is simply incremented when queuing to input_pkt_queue.
Signed-off-by: Tom Herbert <therbert@google.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Oliver Hartkopp [Wed, 19 May 2010 06:46:38 +0000 (06:46 +0000)]
can: SJA1000 add missing spin_lock_init()
As remarked by Sam Ravnborg the spin_lock variable, that has been introduced
in commit 57c8a456640fa3ca777652f11f2db4179a3e66b6 ("can: Fix SJA1000 command
register writes on SMP systems") has not been initialized properly.
This patch adds the initialization to allow spinlock debugging.
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> CC: Sam Ravnborg <sam@ravnborg.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Denis Kirjanov [Thu, 20 May 2010 04:00:59 +0000 (04:00 +0000)]
sh_eth: Fix memleak in sh_mdio_release
Allocated memory for IRQs should be freed when releasing the mii_bus
Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org> Acked-by: Nobuhiro Iwamatsu <iwamatsu@nigauri.org> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / mirror_ubuntu-disco-kernel.git / log