GitHub/exynos8895/android_kernel_samsung_universal8895.git
14 years agotun: add ioctl to modify vnet header size
Michael S. Tsirkin [Wed, 17 Mar 2010 15:45:01 +0000 (17:45 +0200)]
tun: add ioctl to modify vnet header size

virtio added mergeable buffers mode where 2 bytes of extra info is put
after vnet header but before actual data (tun does not need this data).
In hindsight, it would have been better to add the new info *before* the
packet: as it is, users need a lot of tricky code to skip the extra 2
bytes in the middle of the iovec, and in fact applications seem to get
it wrong, and only work with specific iovec layout.  The fact we might
need to split iovec also means we might in theory overflow iovec max
size.

This patch adds a simpler way for applications to handle this,
and future proofs the interface against further extensions,
by making the size of the virtio net header configurable
from userspace. As a result, tun driver will simply
skip the extra 2 bytes on both input and output.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Mon, 3 May 2010 04:43:40 +0000 (21:43 -0700)]
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

14 years agonet: fix compile error due to double return type in SOCK_DEBUG
Jan Engelhardt [Sun, 2 May 2010 20:42:39 +0000 (13:42 -0700)]
net: fix compile error due to double return type in SOCK_DEBUG

Fix this one:
include/net/sock.h: error: two or more data types in declaration specifiers

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: Inline skb_pull() in eth_type_trans().
David S. Miller [Sun, 2 May 2010 09:21:44 +0000 (02:21 -0700)]
net: Inline skb_pull() in eth_type_trans().

In commit 6be8ac2f ("[NET]: uninline skb_pull, de-bloats a lot")
we uninlined skb_pull.

But in some critical paths it makes sense to inline this thing
and it helps performance significantly.

Create an skb_pull_inline() so that we can do this in a way that
serves also as annotation.

Based upon a patch by Eric Dumazet.

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/usb: initiate sync sequence in sierra_net.c driver
Elina Pasheva [Wed, 28 Apr 2010 13:28:24 +0000 (13:28 +0000)]
net/usb: initiate sync sequence in sierra_net.c driver

The following patch adds the initiation of the sync sequence to
"sierra_net_bind()". If this step is omitted, the modem will never sync up
with the host and it will not be possible to establish a data connection.

Signed-off-by: Elina Pasheva <epasheva@sierrawireless.com>
Signed-off-by: Rory Filer <rfiler@sierrawireless.com>
Tested-by: Elina Pasheva <epasheva@sierrawireless.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: sock_def_readable() and friends RCU conversion
Eric Dumazet [Thu, 29 Apr 2010 11:01:49 +0000 (11:01 +0000)]
net: sock_def_readable() and friends RCU conversion

sk_callback_lock rwlock actually protects sk->sk_sleep pointer, so we
need two atomic operations (and associated dirtying) per incoming
packet.

RCU conversion is pretty much needed :

1) Add a new structure, called "struct socket_wq" to hold all fields
that will need rcu_read_lock() protection (currently: a
wait_queue_head_t and a struct fasync_struct pointer).

[Future patch will add a list anchor for wakeup coalescing]

2) Attach one of such structure to each "struct socket" created in
sock_alloc_inode().

3) Respect RCU grace period when freeing a "struct socket_wq"

4) Change sk_sleep pointer in "struct sock" by sk_wq, pointer to "struct
socket_wq"

5) Change sk_sleep() function to use new sk->sk_wq instead of
sk->sk_sleep

6) Change sk_has_sleeper() to wq_has_sleeper() that must be used inside
a rcu_read_lock() section.

7) Change all sk_has_sleeper() callers to :
  - Use rcu_read_lock() instead of read_lock(&sk->sk_callback_lock)
  - Use wq_has_sleeper() to eventually wakeup tasks.
  - Use rcu_read_unlock() instead of read_unlock(&sk->sk_callback_lock)

8) sock_wake_async() is modified to use rcu protection as well.

9) Exceptions :
  macvtap, drivers/net/tun.c, af_unix use integrated "struct socket_wq"
instead of dynamically allocated ones. They dont need rcu freeing.

Some cleanups or followups are probably needed, (possible
sk_callback_lock conversion to a spinlock for example...).

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/usb: remove default in Kconfig for sierra_net driver
Elina Pasheva [Sat, 1 May 2010 02:05:28 +0000 (19:05 -0700)]
net/usb: remove default in Kconfig for sierra_net driver

The following patch removes the default from the Kconfig entry for sierra_net
driver as recommended.

Signed-off-by: Elina Pasheva <epasheva@sierrawireless.com>
Signed-off-by: Rory Filer <rfiler@sierrawireless.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv6: cleanup: remove unneeded null check
Dan Carpenter [Fri, 30 Apr 2010 23:42:08 +0000 (16:42 -0700)]
ipv6: cleanup: remove unneeded null check

We dereference "sk" unconditionally elsewhere in the function.

This was left over from:  b30bd282 "ip6_xmit: remove unnecessary NULL
ptr check".  According to that commit message, "the sk argument to
ip6_xmit is never NULL nowadays since the skb->priority assigment
expects a valid socket."

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoxfrm: potential uninitialized variable num_xfrms
Changli Gao [Tue, 27 Apr 2010 21:20:22 +0000 (21:20 +0000)]
xfrm: potential uninitialized variable num_xfrms

potential uninitialized variable num_xfrms

fix compiler warning: 'num_xfrms' may be used uninitialized in this function.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
----
 net/xfrm/xfrm_policy.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: speedup sock_recv_ts_and_drops()
Eric Dumazet [Wed, 28 Apr 2010 19:14:43 +0000 (19:14 +0000)]
net: speedup sock_recv_ts_and_drops()

sock_recv_ts_and_drops() is fat and slow (~ 4% of cpu time on some
profiles)

We can test all socket flags at once to make fast path fast again.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocdc_ether: Identify MBM devices by GUID in MDLM descriptor
Jonas Sjöquist [Fri, 23 Apr 2010 01:07:45 +0000 (01:07 +0000)]
cdc_ether: Identify MBM devices by GUID in MDLM descriptor

This patch removes vid/pid for Ericsson MBM devices from the whitelist set of
devices. The MBM devices are instead identified by GUID.

In order for cdc_ether to handle these devices the GUID in the MDLM descriptor
is tested. All MBM devices currently handled by cdc_ether as well as future
CDC Ethernet MBM devices can be identified by the GUID.

This is the same solution used in Carl Nordbeck's mbm driver,
http://kerneltrap.org/mailarchive/linux-usb/2008/11/17/4141384/thread

I post this as RFC to get feedback on however cdc_ether is the correct place to
do the binding, or if it should be done in a separate driver, e.g. zaurus.

Signed-off-by: Jonas Sjöquist <jonas.sjoquist@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agor8169: Fix rtl8169_rx_interrupt()
Eric Dumazet [Fri, 30 Apr 2010 23:20:39 +0000 (16:20 -0700)]
r8169: Fix rtl8169_rx_interrupt()

In case a reset is performed, rtl8169_rx_interrupt() is called from
process context instead of softirq context. Special care must be taken
to call appropriate network core services (netif_rx() instead of
netif_receive_skb()). VLAN handling also corrected.

Reported-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Tested-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Diagnosed-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoforcedeth: Stay in NAPI as long as there's work
stephen hemminger [Wed, 28 Apr 2010 08:25:28 +0000 (08:25 +0000)]
forcedeth: Stay in NAPI as long as there's work

The following does the same thing without the extra overhead
of testing all the registers. It also handles the out of memory
case.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Tested-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Fri, 30 Apr 2010 19:54:15 +0000 (12:54 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

14 years agoe1000e: Fix oops caused by ASPM patch.
Anton Blanchard [Wed, 28 Apr 2010 21:46:06 +0000 (21:46 +0000)]
e1000e: Fix oops caused by ASPM patch.

Commit 6f461f6c7c961f0b1b73c0f27becf472a0ac606b
("e1000e: enable/disable ASPM L0s and L1 and ERT according to hardware errata")
oopses on one of my ppc64 boxes with a NULL pointer (0x4a):

Unable to handle kernel paging request for data at address 0x0000004a
Faulting instruction address: 0xc0000000004d2f1c
cpu 0xe: Vector: 300 (Data Access) at [c000000bec1833a0]
    pc: c0000000004d2f1c: .e1000e_disable_aspm+0xe0/0x150
    lr: c0000000004d2f0c: .e1000e_disable_aspm+0xd0/0x150
   dar: 4a

[c000000bec1836d0c00000000069b9d8 .e1000_probe+0x84/0xe8c
[c000000bec1837b0c000000000386d90 .local_pci_probe+0x4c/0x68
[c000000bec183840c0000000003872ac .pci_device_probe+0xfc/0x148
[c000000bec183900c000000000409e8c .driver_probe_device+0xe4/0x1d0
[c000000bec1839a0c00000000040a024 .__driver_attach+0xac/0xf4
[c000000bec183a40c000000000409124 .bus_for_each_dev+0x9c/0x10c
[c000000bec183b00c000000000409c1c .driver_attach+0x40/0x60
[c000000bec183b90c0000000004085dc .bus_add_driver+0x150/0x328
[c000000bec183c40c00000000040a58c .driver_register+0x100/0x1c4
[c000000bec183cf0c00000000038764c .__pci_register_driver+0x78/0x128

Seems like pdev->bus->self == NULL. I haven't touched pci in a long time
so I'm trying to remember what this means (no pcie bridge perhaps?)

The patch below fixes the oops for me.

Signed-off-by: Anton Blanchard <anton@samba.org>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: ip_queue_rcv_skb() helper
Eric Dumazet [Wed, 28 Apr 2010 22:31:51 +0000 (15:31 -0700)]
net: ip_queue_rcv_skb() helper

When queueing a skb to socket, we can immediately release its dst if
target socket do not use IP_CMSG_PKTINFO.

tcp_data_queue() can drop dst too.

This to benefit from a hot cache line and avoid the receiver, possibly
on another cpu, to dirty this cache line himself.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: speedup udp receive path
Eric Dumazet [Wed, 28 Apr 2010 21:35:48 +0000 (14:35 -0700)]
net: speedup udp receive path

Since commit 95766fff ([UDP]: Add memory accounting.),
each received packet needs one extra sock_lock()/sock_release() pair.

This added latency because of possible backlog handling. Then later,
ticket spinlocks added yet another latency source in case of DDOS.

This patch introduces lock_sock_bh() and unlock_sock_bh()
synchronization primitives, avoiding one atomic operation and backlog
processing.

skb_free_datagram_locked() uses them instead of full blown
lock_sock()/release_sock(). skb is orphaned inside locked section for
proper socket memory reclaim, and finally freed outside of it.

UDP receive path now take the socket spinlock only once.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/sb1250: register mdio bus in probe
Sebastian Siewior [Wed, 28 Apr 2010 09:57:01 +0000 (09:57 +0000)]
net/sb1250: register mdio bus in probe

"ifconfig eth0 up && ifconfig eth0 down" triggers:
| kobject (a8000000cfa5a480): tried to init an initialized object, something is seriously wrong.
| Call Trace:
| [<ffffffff8010aabc>] dump_stack+0x8/0x34
| [<ffffffff80293128>] kobject_init+0xe8/0xf0
| [<ffffffff802d922c>] device_initialize+0x2c/0x98
| [<ffffffff802d9cfc>] device_register+0x14/0x28
| [<ffffffff80312cd4>] mdiobus_register+0xdc/0x1e0
| [<ffffffff80314cf0>] sbmac_open+0x58/0x220
| [<ffffffff803519bc>] __dev_open+0x11c/0x180
| [<ffffffff8034d578>] __dev_change_flags+0x120/0x180
| [<ffffffff80351848>] dev_change_flags+0x20/0x78
| [<ffffffff803a753c>] devinet_ioctl+0x7cc/0x820
| [<ffffffff80339ac8>] sock_do_ioctl+0x38/0x90
| [<ffffffff8033a258>] compat_sock_ioctl_trans+0x408/0x1030
| [<ffffffff8033af30>] compat_sock_ioctl+0xb0/0xd0
| [<ffffffff80208b08>] compat_sys_ioctl+0xa0/0x18b8
| [<ffffffff80102f94>] handle_sys+0x114/0x130
|
| sb1250-mac-mdio: probed

mdiobus_register() calls device_register() which initializes the kobj of
the device. mdiobus_unregister() calls only device_del() so we have one
reference left. That one is leaving with mdiobus_free() which is only
called on remove.
Since I don't see any reason why mdiobus_register()/mdiobus_unregister()
should happen in ->open()/->close() I move them to probe & exit.

Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoigb: Clean up left over prototype of igb_get_hw_dev_name()
Emil Tantilov [Wed, 28 Apr 2010 21:24:51 +0000 (14:24 -0700)]
igb: Clean up left over prototype of igb_get_hw_dev_name()

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agowireless: Fix merge.
Hauke Mehrtens [Wed, 28 Apr 2010 21:23:15 +0000 (14:23 -0700)]
wireless: Fix merge.

in your merge in 5c01d5669356e13f0fb468944c1dd4c6a7e978ad you added "int
i;" into wl1271_main.c which is unused in that function.

This patch fixes the merge problem:

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosctp: Fix skb_over_panic resulting from multiple invalid parameter errors (CVE-2010...
Neil Horman [Wed, 28 Apr 2010 10:30:59 +0000 (10:30 +0000)]
sctp: Fix skb_over_panic resulting from multiple invalid parameter errors (CVE-2010-1173) (v4)

Ok, version 4

Change Notes:
1) Minor cleanups, from Vlads notes

Summary:

Hey-
Recently, it was reported to me that the kernel could oops in the
following way:

<5> kernel BUG at net/core/skbuff.c:91!
<5> invalid operand: 0000 [#1]
<5> Modules linked in: sctp netconsole nls_utf8 autofs4 sunrpc iptable_filter
ip_tables cpufreq_powersave parport_pc lp parport vmblock(U) vsock(U) vmci(U)
vmxnet(U) vmmemctl(U) vmhgfs(U) acpiphp dm_mirror dm_mod button battery ac md5
ipv6 uhci_hcd ehci_hcd snd_ens1371 snd_rawmidi snd_seq_device snd_pcm_oss
snd_mixer_oss snd_pcm snd_timer snd_page_alloc snd_ac97_codec snd soundcore
pcnet32 mii floppy ext3 jbd ata_piix libata mptscsih mptsas mptspi mptscsi
mptbase sd_mod scsi_mod
<5> CPU:    0
<5> EIP:    0060:[<c02bff27>]    Not tainted VLI
<5> EFLAGS: 00010216   (2.6.9-89.0.25.EL)
<5> EIP is at skb_over_panic+0x1f/0x2d
<5> eax: 0000002c   ebx: c033f461   ecx: c0357d96   edx: c040fd44
<5> esi: c033f461   edi: df653280   ebp: 00000000   esp: c040fd40
<5> ds: 007b   es: 007b   ss: 0068
<5> Process swapper (pid: 0, threadinfo=c040f000 task=c0370be0)
<5> Stack: c0357d96 e0c29478 00000084 00000004 c033f461 df653280 d7883180
e0c2947d
<5>        00000000 00000080 df653490 00000004 de4f1ac0 de4f1ac0 00000004
df653490
<5>        00000001 e0c2877a 08000800 de4f1ac0 df653490 00000000 e0c29d2e
00000004
<5> Call Trace:
<5>  [<e0c29478>] sctp_addto_chunk+0xb0/0x128 [sctp]
<5>  [<e0c2947d>] sctp_addto_chunk+0xb5/0x128 [sctp]
<5>  [<e0c2877a>] sctp_init_cause+0x3f/0x47 [sctp]
<5>  [<e0c29d2e>] sctp_process_unk_param+0xac/0xb8 [sctp]
<5>  [<e0c29e90>] sctp_verify_init+0xcc/0x134 [sctp]
<5>  [<e0c20322>] sctp_sf_do_5_1B_init+0x83/0x28e [sctp]
<5>  [<e0c25333>] sctp_do_sm+0x41/0x77 [sctp]
<5>  [<c01555a4>] cache_grow+0x140/0x233
<5>  [<e0c26ba1>] sctp_endpoint_bh_rcv+0xc5/0x108 [sctp]
<5>  [<e0c2b863>] sctp_inq_push+0xe/0x10 [sctp]
<5>  [<e0c34600>] sctp_rcv+0x454/0x509 [sctp]
<5>  [<e084e017>] ipt_hook+0x17/0x1c [iptable_filter]
<5>  [<c02d005e>] nf_iterate+0x40/0x81
<5>  [<c02e0bb9>] ip_local_deliver_finish+0x0/0x151
<5>  [<c02e0c7f>] ip_local_deliver_finish+0xc6/0x151
<5>  [<c02d0362>] nf_hook_slow+0x83/0xb5
<5>  [<c02e0bb2>] ip_local_deliver+0x1a2/0x1a9
<5>  [<c02e0bb9>] ip_local_deliver_finish+0x0/0x151
<5>  [<c02e103e>] ip_rcv+0x334/0x3b4
<5>  [<c02c66fd>] netif_receive_skb+0x320/0x35b
<5>  [<e0a0928b>] init_stall_timer+0x67/0x6a [uhci_hcd]
<5>  [<c02c67a4>] process_backlog+0x6c/0xd9
<5>  [<c02c690f>] net_rx_action+0xfe/0x1f8
<5>  [<c012a7b1>] __do_softirq+0x35/0x79
<5>  [<c0107efb>] handle_IRQ_event+0x0/0x4f
<5>  [<c01094de>] do_softirq+0x46/0x4d

Its an skb_over_panic BUG halt that results from processing an init chunk in
which too many of its variable length parameters are in some way malformed.

The problem is in sctp_process_unk_param:
if (NULL == *errp)
*errp = sctp_make_op_error_space(asoc, chunk,
 ntohs(chunk->chunk_hdr->length));

if (*errp) {
sctp_init_cause(*errp, SCTP_ERROR_UNKNOWN_PARAM,
 WORD_ROUND(ntohs(param.p->length)));
sctp_addto_chunk(*errp,
WORD_ROUND(ntohs(param.p->length)),
  param.v);

When we allocate an error chunk, we assume that the worst case scenario requires
that we have chunk_hdr->length data allocated, which would be correct nominally,
given that we call sctp_addto_chunk for the violating parameter.  Unfortunately,
we also, in sctp_init_cause insert a sctp_errhdr_t structure into the error
chunk, so the worst case situation in which all parameters are in violation
requires chunk_hdr->length+(sizeof(sctp_errhdr_t)*param_count) bytes of data.

The result of this error is that a deliberately malformed packet sent to a
listening host can cause a remote DOS, described in CVE-2010-1173:
http://cve.mitre.org/cgi-bin/cvename.cgi?name=2010-1173

I've tested the below fix and confirmed that it fixes the issue.  We move to a
strategy whereby we allocate a fixed size error chunk and ignore errors we don't
have space to report.  Tested by me successfully

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoBugfix: Link selection was swapped in switch.
Sjur Braendeland [Wed, 28 Apr 2010 08:54:40 +0000 (08:54 +0000)]
Bugfix: Link selection was swapped in switch.

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocaif: Bugfixes in CAIF netdevice for close and flow control
Sjur Braendeland [Wed, 28 Apr 2010 08:54:39 +0000 (08:54 +0000)]
caif: Bugfixes in CAIF netdevice for close and flow control

Changes:
o Bugfix: Flow control was causing the device to be destroyed.
o Bugfix: Handle CAIF channel connect failures.
o If the underlying link layer is gone the net-device is no longer removed,
  but closed.

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocaif: Rewritten socket implementation
Sjur Braendeland [Wed, 28 Apr 2010 08:54:38 +0000 (08:54 +0000)]
caif: Rewritten socket implementation

Changes:
 This is a complete re-write of the socket layer. Making the socket
 implementation more aligned with the other socket layers and using more
 of the support functions available in sock.c. Lots of code is copied
 from af_unix (and some from af_irda).
 Non-blocking mode should be working as well.

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocaif: Disconnect without waiting for response
Sjur Braendeland [Wed, 28 Apr 2010 08:54:37 +0000 (08:54 +0000)]
caif: Disconnect without waiting for response

Changes:
o Function cfcnfg_disconn_adapt_layer is changed to do asynchronous
  disconnect, not waiting for any response from the modem. Due to this
  the function cfcnfg_linkdestroy_rsp does nothing anymore.
o Because disconnect may take down a connection before a connect response
  is received the function cfcnfg_linkup_rsp is checking if the client is
  still waiting for the response, if not a disconnect request is sent to
  the modem.
o cfctrl is no longer keeping track of pending disconnect requests.
o Added function cfctrl_cancel_req, which is used for deleting a pending
  connect request if disconnect is done before connect response is received.
o Removed unused function cfctrl_insert_req2
o Added better handling of connect reject from modem.

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocaif: Add reference counting to service layer
Sjur Braendeland [Wed, 28 Apr 2010 08:54:36 +0000 (08:54 +0000)]
caif: Add reference counting to service layer

Changes:
o Added functions cfsrvl_get and cfsrvl_put.
o Added support release_client to use by socket and net device.
o Increase reference counting for in-flight packets from cfmuxl

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocaif: Rename functions in cfcnfg and caif_dev
Sjur Braendeland [Wed, 28 Apr 2010 08:54:35 +0000 (08:54 +0000)]
caif: Rename functions in cfcnfg and caif_dev

Changes:
 o Renamed cfcnfg_del_adapt_layer to cfcnfg_disconn_adapt_layer
 o Fixed typo cfcfg to cfcnfg
 o Renamed linkid to channel_id
 o Updated documentation in caif_dev.h
 o Minor formatting changes

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocaif: Ldisc add permission check and mem-alloc error check
Sjur Braendeland [Wed, 28 Apr 2010 08:54:34 +0000 (08:54 +0000)]
caif: Ldisc add permission check and mem-alloc error check

Changes:
   o Added permission checks for installing. CAP_SYS_ADMIN and
     CAP_SYS_TTY_CONFIG can install the ldisc.
   o Check if allocation of skb was successful.

Signed-off-by: Sjur Braendeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Create multiple TX queues
Ben Hutchings [Wed, 28 Apr 2010 09:30:43 +0000 (09:30 +0000)]
sfc: Create multiple TX queues

Create a core TX queue and 2 hardware TX queues for each channel.
If separate_tx_channels is set, create equal numbers of RX and TX
channels instead.

Rewrite the channel and queue iteration macros accordingly.
Eliminate efx_channel::used_flags as redundant.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Test only the first pair of TX queues
Ben Hutchings [Wed, 28 Apr 2010 09:30:30 +0000 (09:30 +0000)]
sfc: Test only the first pair of TX queues

This makes no immediate difference, but we definitely do not want
to test all TX queues once we allocate a pair of TX queues to each
channel.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Add Siena PHY BIST and cable diagnostic support
Steve Hodgson [Wed, 28 Apr 2010 09:30:22 +0000 (09:30 +0000)]
sfc: Add Siena PHY BIST and cable diagnostic support

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Clean up efx_nic::irq_zero_count
Ben Hutchings [Wed, 28 Apr 2010 09:30:00 +0000 (09:30 +0000)]
sfc: Clean up efx_nic::irq_zero_count

There is no need for this to be unsigned long; make it unsigned int.
It does need a line in kernel-doc, so add that.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Add necessary parentheses to macro definitions in net_driver.h
Ben Hutchings [Wed, 28 Apr 2010 09:29:50 +0000 (09:29 +0000)]
sfc: Add necessary parentheses to macro definitions in net_driver.h

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Break NAPI processing after one ring-full of TX completions
Ben Hutchings [Wed, 28 Apr 2010 09:29:42 +0000 (09:29 +0000)]
sfc: Break NAPI processing after one ring-full of TX completions

Currently TX completions do not count towards the NAPI budget.  This
means a continuous stream of TX completions can cause the polling
function to loop indefinitely with scheduling disabled.  To avoid
this, follow the common practice of reporting the budget spent after
processing one ring-full of TX completions.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Set PERIODIC_NOEVENT flag for MC_CMD_MAC_STATS
Steve Hodgson [Wed, 28 Apr 2010 09:29:32 +0000 (09:29 +0000)]
sfc: Set PERIODIC_NOEVENT flag for MC_CMD_MAC_STATS

When set, an event is not sent whenever periodic MAC statistics are
raised.  This avoids unnecessary wake-ups.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Update MCDI protocol definitions
Ben Hutchings [Wed, 28 Apr 2010 09:29:14 +0000 (09:29 +0000)]
sfc: Update MCDI protocol definitions

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Enable IPv6 RSS using random key for Toeplitz hash
Ben Hutchings [Wed, 28 Apr 2010 09:29:02 +0000 (09:29 +0000)]
sfc: Enable IPv6 RSS using random key for Toeplitz hash

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Read MEM_STAT for SRM_PERR as well as MEM_PERR errors
Steve Hodgson [Wed, 28 Apr 2010 09:28:52 +0000 (09:28 +0000)]
sfc: Read MEM_STAT for SRM_PERR as well as MEM_PERR errors

Parity errors in different blocks of SRAM may set one of two different
interrupt flags.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Log specific message for failure of NVRAM self-test
Ben Hutchings [Wed, 28 Apr 2010 09:28:36 +0000 (09:28 +0000)]
sfc: Log specific message for failure of NVRAM self-test

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Extend the legacy interrupt workarounds
Steve Hodgson [Wed, 28 Apr 2010 09:28:27 +0000 (09:28 +0000)]
sfc: Extend the legacy interrupt workarounds

Siena has two problems with legacy interrupts:
  1. There is no synchronisation between the ISR read completion,
     and the interrupt deassert message.
  2. A downstream read at the "wrong" moment can return 0, and
     suppress generating the next interrupt.

Falcon should suffer from both of these, and it appears it does.
Enable EFX_WORKAROUND_15783 on Falcon as well.

Also, when we see queues == 0, ensure we always schedule or rearm
every event queue.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Reconfigure the XAUI serdes after an EM reset
Steve Hodgson [Wed, 28 Apr 2010 09:28:10 +0000 (09:28 +0000)]
sfc: Reconfigure the XAUI serdes after an EM reset

Fix a regression introduced in d3245b28ef2a45ec4e115062a38100bd06229289
"sfc: Refactor link configuration".

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Stop masking out XGMII faults over reconfigures
Steve Hodgson [Wed, 28 Apr 2010 09:27:54 +0000 (09:27 +0000)]
sfc: Stop masking out XGMII faults over reconfigures

The aim of this code was to avoid a spurious XGMII fault over a MAC
reconfigure. It's less relevant now that the PHY reconfigure isn't
called from the MAC reconfigure.

After applying this patch, our link stress test passed 48 hours of
testing without ever resetting the PHY.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Handle serious errors in exactly one interrupt handler
Steve Hodgson [Wed, 28 Apr 2010 09:27:36 +0000 (09:27 +0000)]
sfc: Handle serious errors in exactly one interrupt handler

'Fatal' errors set an interrupt flag associated with a specific event
queue; only read the syndrome vector if we see that queue's flag set
(legacy interrupts) or in the interrupt handler for that queue (MSI).

Do not ignore an interrupt if the fatal error flag is set but specific
error flags are all zero.  Even if we don't schedule a reset, we must
respect the queue mask and rearm the appropriate event queues.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Consistently report short MCDI responses as EIO
Ben Hutchings [Wed, 28 Apr 2010 09:27:14 +0000 (09:27 +0000)]
sfc: Consistently report short MCDI responses as EIO

In some cases failing functions were returning 0 which is obviously wrong.
In other cases they were returning inappropriate error codes.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Ignore parity errors in the other port's SRAM
Steve Hodgson [Wed, 28 Apr 2010 09:25:22 +0000 (09:25 +0000)]
sfc: Ignore parity errors in the other port's SRAM

Siena has a separate SRAM bank for each port.  On single-port boards
these can be merged together, so each port has an interrupt flag for
parity errors in the other port's SRAM.  Currently we do not enable
such merging and should mask this interrupt source.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Change falcon_probe_board() to fail for unsupported boards
Ben Hutchings [Wed, 28 Apr 2010 09:01:50 +0000 (09:01 +0000)]
sfc: Change falcon_probe_board() to fail for unsupported boards

The driver needs specific PHY and board support code for each SFC4000
board; there is no point trying to continue if it is missing.
Currently unsupported boards can trigger an 'oops'.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Cc: stable@kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Always close net device at the end of a disabling reset
Ben Hutchings [Wed, 28 Apr 2010 09:01:33 +0000 (09:01 +0000)]
sfc: Always close net device at the end of a disabling reset

This fixes a regression introduced by commit
eb9f6744cbfa97674c13263802259b5aa0034594 "sfc: Implement ethtool
reset operation".

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Cc: stable@kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Wait at most 10ms for the MC to finish reading out MAC statistics
Ben Hutchings [Wed, 28 Apr 2010 09:00:35 +0000 (09:00 +0000)]
sfc: Wait at most 10ms for the MC to finish reading out MAC statistics

The original code would wait indefinitely if MAC stats DMA failed.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Cc: stable@kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosctp: Fix oops when sending queued ASCONF chunks
Vlad Yasevich [Wed, 28 Apr 2010 08:47:22 +0000 (08:47 +0000)]
sctp: Fix oops when sending queued ASCONF chunks

When we finish processing ASCONF_ACK chunk, we try to send
the next queued ASCONF.  This action runs the sctp state
machine recursively and it's not prepared to do so.

kernel BUG at kernel/timer.c:790!
invalid opcode: 0000 [#1] SMP
last sysfs file: /sys/module/ipv6/initstate
Modules linked in: sha256_generic sctp libcrc32c ipv6 dm_multipath
uinput 8139too i2c_piix4 8139cp mii i2c_core pcspkr virtio_net joydev
floppy virtio_blk virtio_pci [last unloaded: scsi_wait_scan]

Pid: 0, comm: swapper Not tainted 2.6.34-rc4 #15 /Bochs
EIP: 0060:[<c044a2ef>] EFLAGS: 00010286 CPU: 0
EIP is at add_timer+0xd/0x1b
EAX: cecbab14 EBX: 000000f0 ECX: c0957b1c EDX: 03595cf4
ESI: cecba800 EDI: cf276f00 EBP: c0957aa0 ESP: c0957aa0
 DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
Process swapper (pid: 0, ti=c0956000 task=c0988ba0 task.ti=c0956000)
Stack:
 c0957ae0 d1851214 c0ab62e4 c0ab5f26 0500ffff 00000004 00000005 00000004
<0> 00000000 d18694fd 00000004 1666b892 cecba800 cecba800 c0957b14
00000004
<0> c0957b94 d1851b11 ceda8b00 cecba800 cf276f00 00000001 c0957b14
000000d0
Call Trace:
 [<d1851214>] ? sctp_side_effects+0x607/0xdfc [sctp]
 [<d1851b11>] ? sctp_do_sm+0x108/0x159 [sctp]
 [<d1863386>] ? sctp_pname+0x0/0x1d [sctp]
 [<d1861a56>] ? sctp_primitive_ASCONF+0x36/0x3b [sctp]
 [<d185657c>] ? sctp_process_asconf_ack+0x2a4/0x2d3 [sctp]
 [<d184e35c>] ? sctp_sf_do_asconf_ack+0x1dd/0x2b4 [sctp]
 [<d1851ac1>] ? sctp_do_sm+0xb8/0x159 [sctp]
 [<d1863334>] ? sctp_cname+0x0/0x52 [sctp]
 [<d1854377>] ? sctp_assoc_bh_rcv+0xac/0xe1 [sctp]
 [<d1858f0f>] ? sctp_inq_push+0x2d/0x30 [sctp]
 [<d186329d>] ? sctp_rcv+0x797/0x82e [sctp]

Tested-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: Yuansong Qiao <ysqiao@research.ait.ie>
Signed-off-by: Shuaijun Zhang <szhang@research.ait.ie>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosctp: fix to calc the INIT/INIT-ACK chunk length correctly is set
Wei Yongjun [Wed, 28 Apr 2010 08:47:21 +0000 (08:47 +0000)]
sctp: fix to calc the INIT/INIT-ACK chunk length correctly is set

When calculating the INIT/INIT-ACK chunk length, we should not
only account the length of parameters, but also the parameters
zero padding length, such as AUTH HMACS parameter and CHUNKS
parameter. Without the parameters zero padding length we may get
following oops.

skb_over_panic: text:ce2068d2 len:130 put:6 head:cac3fe00 data:cac3fe00 tail:0xcac3fe82 end:0xcac3fe80 dev:<NULL>
------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:127!
invalid opcode: 0000 [#2] SMP
last sysfs file: /sys/module/aes_generic/initstate
Modules linked in: authenc ......

Pid: 4102, comm: sctp_darn Tainted: G      D    2.6.34-rc2 #6
EIP: 0060:[<c0607630>] EFLAGS: 00010282 CPU: 0
EIP is at skb_over_panic+0x37/0x3e
EAX: 00000078 EBX: c07c024b ECX: c07c02b9 EDX: cb607b78
ESI: 00000000 EDI: cac3fe7a EBP: 00000002 ESP: cb607b74
 DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
Process sctp_darn (pid: 4102, ti=cb607000 task=cabdc990 task.ti=cb607000)
Stack:
 c07c02b9 ce2068d2 00000082 00000006 cac3fe00 cac3fe00 cac3fe82 cac3fe80
<0> c07c024b cac3fe7c cac3fe7a c0608dec ca986e80 ce2068d2 00000006 0000007a
<0> cb8120ca ca986e80 cb812000 00000003 cb8120c4 ce208a25 cb8120ca cadd9400
Call Trace:
 [<ce2068d2>] ? sctp_addto_chunk+0x45/0x85 [sctp]
 [<c0608dec>] ? skb_put+0x2e/0x32
 [<ce2068d2>] ? sctp_addto_chunk+0x45/0x85 [sctp]
 [<ce208a25>] ? sctp_make_init+0x279/0x28c [sctp]
 [<c0686a92>] ? apic_timer_interrupt+0x2a/0x30
 [<ce1fdc0b>] ? sctp_sf_do_prm_asoc+0x2b/0x7b [sctp]
 [<ce202823>] ? sctp_do_sm+0xa0/0x14a [sctp]
 [<ce2133b9>] ? sctp_pname+0x0/0x14 [sctp]
 [<ce211d72>] ? sctp_primitive_ASSOCIATE+0x2b/0x31 [sctp]
 [<ce20f3cf>] ? sctp_sendmsg+0x7a0/0x9eb [sctp]
 [<c064eb1e>] ? inet_sendmsg+0x3b/0x43
 [<c04244b7>] ? task_tick_fair+0x2d/0xd9
 [<c06031e1>] ? sock_sendmsg+0xa7/0xc1
 [<c0416afe>] ? smp_apic_timer_interrupt+0x6b/0x75
 [<c0425123>] ? dequeue_task_fair+0x34/0x19b
 [<c0446abb>] ? sched_clock_local+0x17/0x11e
 [<c052ea87>] ? _copy_from_user+0x2b/0x10c
 [<c060ab3a>] ? verify_iovec+0x3c/0x6a
 [<c06035ca>] ? sys_sendmsg+0x186/0x1e2
 [<c042176b>] ? __wake_up_common+0x34/0x5b
 [<c04240c2>] ? __wake_up+0x2c/0x3b
 [<c057e35c>] ? tty_wakeup+0x43/0x47
 [<c04430f2>] ? remove_wait_queue+0x16/0x24
 [<c0580c94>] ? n_tty_read+0x5b8/0x65e
 [<c042be02>] ? default_wake_function+0x0/0x8
 [<c0604e0e>] ? sys_socketcall+0x17f/0x1cd
 [<c040264c>] ? sysenter_do_call+0x12/0x22
Code: 0f 45 de 53 ff b0 98 00 00 00 ff b0 94 ......
EIP: [<c0607630>] skb_over_panic+0x37/0x3e SS:ESP 0068:cb607b74

To reproduce:

# modprobe sctp
# echo 1 > /proc/sys/net/sctp/addip_enable
# echo 1 > /proc/sys/net/sctp/auth_enable
# sctp_test -H 3ffe:501:ffff:100:20c:29ff:fe4d:f37e -P 800 -l
# sctp_darn -H 3ffe:501:ffff:100:20c:29ff:fe4d:f37e -P 900 -h 192.168.0.21 -p 800 -I -s -t
sctp_darn ready to send...
3ffe:501:ffff:100:20c:29ff:fe4d:f37e:900-192.168.0.21:800 Interactive mode> bindx-add=192.168.0.21
3ffe:501:ffff:100:20c:29ff:fe4d:f37e:900-192.168.0.21:800 Interactive mode> bindx-add=192.168.1.21
3ffe:501:ffff:100:20c:29ff:fe4d:f37e:900-192.168.0.21:800 Interactive mode> snd=10

------------------------------------------------------------------
eth0 has addresses: 3ffe:501:ffff:100:20c:29ff:fe4d:f37e and 192.168.0.21
eth1 has addresses: 192.168.1.21
------------------------------------------------------------------

Reported-by: George Cheimonidis <gchimon@gmail.com>
Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosctp: per_cpu variables should be in bh_disabled section
Vlad Yasevich [Wed, 28 Apr 2010 08:47:20 +0000 (08:47 +0000)]
sctp: per_cpu variables should be in bh_disabled section

Since the change of the atomics to percpu variables, we now
have to disable BH in process context when touching percpu variables.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosctp: fix potential reference of a freed pointer
Vlad Yasevich [Wed, 28 Apr 2010 08:47:19 +0000 (08:47 +0000)]
sctp: fix potential reference of a freed pointer

When sctp attempts to update an assocition, it removes any
addresses that were not in the updated INITs.  However, the loop
may attempt to refrence a transport with address after removing it.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosctp: avoid irq lock inversion while call sk->sk_data_ready()
Wei Yongjun [Wed, 28 Apr 2010 08:47:18 +0000 (08:47 +0000)]
sctp: avoid irq lock inversion while call sk->sk_data_ready()

sk->sk_data_ready() of sctp socket can be called from both BH and non-BH
contexts, but the default sk->sk_data_ready(), sock_def_readable(), can
not be used in this case. Therefore, we have to make a new function
sctp_data_ready() to grab sk->sk_data_ready() with BH disabling.

=========================================================
[ INFO: possible irq lock inversion dependency detected ]
2.6.33-rc6 #129
---------------------------------------------------------
sctp_darn/1517 just changed the state of lock:
 (clock-AF_INET){++.?..}, at: [<c06aab60>] sock_def_readable+0x20/0x80
but this lock took another, SOFTIRQ-unsafe lock in the past:
 (slock-AF_INET){+.-...}

and interrupts could create inverse lock ordering between them.

other info that might help us debug this:
1 lock held by sctp_darn/1517:
 #0:  (sk_lock-AF_INET){+.+.+.}, at: [<cdfe363d>] sctp_sendmsg+0x23d/0xc00 [sctp]

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoRevert "tcp: bind() fix when many ports are bound"
David S. Miller [Wed, 28 Apr 2010 18:25:59 +0000 (11:25 -0700)]
Revert "tcp: bind() fix when many ports are bound"

This reverts two commits:

fda48a0d7a8412cedacda46a9c0bf8ef9cd13559
tcp: bind() fix when many ports are bound

and a follow-on fix for it:

6443bb1fc2050ca2b6585a3fa77f7833b55329ed
ipv6: Fix inet6_csk_bind_conflict()

It causes problems with binding listening sockets when time-wait
sockets from a previous instance still are alive.

It's too late to keep fiddling with this so late in the -rc
series, and we'll deal with it in net-next-2.6 instead.

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosky2: use the DMA state API instead of the pci equivalents
FUJITA Tomonori [Tue, 27 Apr 2010 14:57:05 +0000 (14:57 +0000)]
sky2: use the DMA state API instead of the pci equivalents

This replace the PCI DMA state API (include/linux/pci-dma.h) with the
DMA equivalents since the PCI DMA state API will be obsolete.

No functional change.

For further information about the background:

http://marc.info/?l=linux-netdev&m=127037540020276&w=2

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2x: Remove two prefetch()
Eric Dumazet [Tue, 27 Apr 2010 12:18:13 +0000 (12:18 +0000)]
bnx2x: Remove two prefetch()

1) Even on 64bit arches, sizeof(struct sk_buff) < 256
2) No need to prefetch same pointer twice.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Acked-by: Eliezer Tamir <eliezer@tamir.org.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: disable MSI-X by default on certain Cisco adapters
Nicholas Nunley [Wed, 28 Apr 2010 02:47:49 +0000 (19:47 -0700)]
ixgbe: disable MSI-X by default on certain Cisco adapters

Due to an errata in 82598 parts MSI-X needs to be disabled
in certain ixgbe devices designed to transfer peer-to-peer
traffic on the PCIe bus. This patch sets the default
interrupt type to MSI rather than MSI-X for specific Cisco
ixgbe adapters.

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Acked-by: John Ronciak <john.ronciak@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobridge: multicast_flood cleanup
stephen hemminger [Tue, 27 Apr 2010 15:01:07 +0000 (15:01 +0000)]
bridge: multicast_flood cleanup

Move some declarations around to make it clearer which variables
are being used inside loop.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobridge: multicast port group RCU fix
stephen hemminger [Tue, 27 Apr 2010 15:01:06 +0000 (15:01 +0000)]
bridge: multicast port group RCU fix

The recently introduced bridge mulitcast port group list was only
partially using RCU correctly. It was missing rcu_dereference()
and missing the necessary barrier on deletion.

The code should have used one of the standard list methods (list or hlist)
instead of open coding a RCU based link list.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobridge: multicast flood
stephen hemminger [Tue, 27 Apr 2010 15:01:05 +0000 (15:01 +0000)]
bridge: multicast flood

Fix unsafe usage of RCU. Would never work on Alpha SMP because
of lack of rcu_dereference()

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobridge: simplify multicast_add_router
stephen hemminger [Tue, 27 Apr 2010 15:01:04 +0000 (15:01 +0000)]
bridge: simplify multicast_add_router

By coding slightly differently, there are only two cases
to deal with: add at head and add after previous entry.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/usb: add sierra_net.c driver
Elina Pasheva [Wed, 28 Apr 2010 01:06:41 +0000 (18:06 -0700)]
net/usb: add sierra_net.c driver

Re-submitted based on comments from netdev community.
Summary of the changes:
1. Improved error handling.
2. Added the missing timeout arguments to usb_control_msg().

The following is a new Linux driver which exposes certain models of Sierra
Wireless modems to the operating system as Network Interface Cards (NICs).

This driver requires a version of the sierra.c driver which supports
blacklisting to work properly. The blacklist in sierra.c rejects the interfaces
claimed by sierra_net.c. Likewise, the sierra_net.c driver only accepts
(i.e. whitelists) the interface(s) used for USB-to-WWAN traffic.
The version of sierra.c which supports blacklisting is
available from the sierra wireless knowledge base page for older kernels. It is
also available in Linux kernel starting from version 2.6.31.

This driver works with all Sierra Wireless devices configured with PID=68A3
like USB305, USB306 provided the corresponding firmware version is I2.0
(for USB305) or M3.0 (for USB306) and later.
This driver will not work with earlier firmware versions than the ones shown
above. In this case the driver will issue an error message indicating
incompatibility and will not serve the device's USB-to-WWAN interface.

Sierra_net.c sits atop a pre-existing Linux driver called usbnet.c.
A series of hook functions are provided in sierra_net.c which are called by
usbnet.c in response to a particular condition such as receipt or transmission
of a data packet. As such, usbnet.c does most of the work of making
a modem appear to the system as a network device and for properly exchanging
traffic between the USB subsystem and the Network card interface.
Sierra_net.c is concerned with managing the data exchanged between the
USB-to-WWAN interface and the upper layers of the operating system.

Signed-off-by: Elina Pasheva <epasheva@sierrawireless.com>
Signed-off-by: Rory Filer <rfiler@sierrawireless.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: add registers etc. printout code just before resetting adapters
Taku Izumi [Tue, 27 Apr 2010 14:39:53 +0000 (14:39 +0000)]
ixgbe: add registers etc. printout code just before resetting adapters

This patch adds registers (,tx/rx rings' status and so on) printout
code just before resetting adapters. This will be helpful for detecting
the root cause of adapters reset.

Signed-off-by: Taku Izumi <izumi.taku@jp.fujitsu.com>
Signed-off-by: Koki Sanagi <sanagi.koki@jp.fujitsu.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoigb: add registers etc. printout code just before resetting adapters
Taku Izumi [Tue, 27 Apr 2010 14:39:30 +0000 (14:39 +0000)]
igb: add registers etc. printout code just before resetting adapters

This patch adds registers (,tx/rx rings' status and so on) printout
code just before resetting adapters. This will be helpful for detecting
the root cause of adapters reset.

Signed-off-by: Taku Izumi <izumi.taku@jp.fujitsu.com>
Signed-off-by: Koki Sanagi <sanagi.koki@jp.fujitsu.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: add registers etc. printout code just before resetting adapters
Taku Izumi [Tue, 27 Apr 2010 14:39:08 +0000 (14:39 +0000)]
e1000e: add registers etc. printout code just before resetting adapters

This patch adds registers (,tx/rx rings' status and so on) printout
code just before resetting adapters. This will be helpful for detecting
the root cause of adapters reset.

Signed-off-by: Taku Izumi <izumi.taku@jp.fujitsu.com>
Signed-off-by: Koki Sanagi <sanagi.koki@jp.fujitsu.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocdc_ether: fix autosuspend for mbm devices
Torgny Johansson [Wed, 28 Apr 2010 00:07:40 +0000 (17:07 -0700)]
cdc_ether: fix autosuspend for mbm devices

Autosuspend works until you bring the wwan interface up, then the
device does not enter autosuspend anymore.

The following patch fixes the problem by setting the .manage_power
field in the mbm_info struct to the same as in the cdc_info struct
(cdc_manager_power).

Signed-off-by: Torgny Johansson <torgny.johansson@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000: Use netdev_<level>, pr_<level> and dev_<level>
Emil Tantilov [Tue, 27 Apr 2010 14:02:58 +0000 (14:02 +0000)]
e1000: Use netdev_<level>, pr_<level> and dev_<level>

This patch is an alternative to similar patch provided by Joe Perches.

Substitute DPRINTK macro for e_<level> that uses netdev_<level> and dev_<level>
similar to e1000e.
- Convert printk to pr_<level> where applicable.
- Use common #define pr_fmt for the driver.
- Use dev_<level> for displaying text in parts of the driver where the interface
  name is not assigned (like e1000_param.c).
- Better align test with the new macros.

CC: Joe Perches <joe@perches.com>
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobluetooth: handle l2cap_create_connless_pdu() errors
Dan Carpenter [Wed, 21 Apr 2010 23:52:01 +0000 (23:52 +0000)]
bluetooth: handle l2cap_create_connless_pdu() errors

l2cap_create_connless_pdu() can sometimes return ERR_PTR(-ENOMEM) or
ERR_PTR(-EFAULT).

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoRevert "bridge: Use hlist_for_each_entry_rcu() in br_multicast_add_router()"
David S. Miller [Tue, 27 Apr 2010 23:49:58 +0000 (16:49 -0700)]
Revert "bridge: Use hlist_for_each_entry_rcu() in br_multicast_add_router()"

This reverts commit ff65e8275f6c96a5eda57493bd84c4555decf7b3.

As explained by Stephen Hemminger, the traversal doesn't require
RCU handling as we hold a lock.

The list addition et al. calls, on the other hand, do.

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agogianfar: Wait for both RX and TX to stop
Andy Fleming [Tue, 27 Apr 2010 23:43:31 +0000 (16:43 -0700)]
gianfar: Wait for both RX and TX to stop

When gracefully stopping the controller, the driver was continuing if
*either* RX or TX had stopped.  We need to wait for both, or the
controller could get into an invalid state.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbevf: use DMA API instead of PCI DMA functions
Nick Nunley [Tue, 27 Apr 2010 13:10:50 +0000 (13:10 +0000)]
ixgbevf: use DMA API instead of PCI DMA functions

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: use DMA API instead of PCI DMA functions
Nick Nunley [Tue, 27 Apr 2010 13:10:27 +0000 (13:10 +0000)]
ixgbe: use DMA API instead of PCI DMA functions

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgb: use DMA API instead of PCI DMA functions
Nick Nunley [Tue, 27 Apr 2010 13:10:03 +0000 (13:10 +0000)]
ixgb: use DMA API instead of PCI DMA functions

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoigbvf: use DMA API instead of PCI DMA functions
Nick Nunley [Tue, 27 Apr 2010 13:09:44 +0000 (13:09 +0000)]
igbvf: use DMA API instead of PCI DMA functions

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoigb: convert igb from using PCI DMA functions to using DMA API functions
Alexander Duyck [Tue, 27 Apr 2010 13:09:25 +0000 (13:09 +0000)]
igb: convert igb from using PCI DMA functions to using DMA API functions

This patch makes it so that igb now uses the DMA API functions instead of
the PCI API functions.  To do this the pci_dev pointer that was in the
rings has been replaced with a device pointer, and as a result all
references to [tr]x_ring->pdev have been replaced with [tr]x_ring->dev.

This patch is based of of work originally done by Nicholas Nunley.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: use DMA API instead of PCI DMA functions
Nick Nunley [Tue, 27 Apr 2010 13:09:05 +0000 (13:09 +0000)]
e1000e: use DMA API instead of PCI DMA functions

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000: use DMA API instead of PCI DMA functions
Nick Nunley [Tue, 27 Apr 2010 13:08:45 +0000 (13:08 +0000)]
e1000: use DMA API instead of PCI DMA functions

Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobridge: Use hlist_for_each_entry_rcu() in br_multicast_add_router()
David S. Miller [Tue, 27 Apr 2010 23:26:49 +0000 (16:26 -0700)]
bridge: Use hlist_for_each_entry_rcu() in br_multicast_add_router()

Noticed by Michał Mirosław.

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocxgb4: set skb->rxhash
Dimitris Michailidis [Tue, 27 Apr 2010 23:22:42 +0000 (16:22 -0700)]
cxgb4: set skb->rxhash

Implement the ->set_flags ethtool method to control NETIF_F_RXHASH and
set skb->rxhash to the HW calculated hash accordingly.

Follow Eric Dumazet's suggestion and use the hash value raw.

Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/sb1250: setup the pdevice within the soc code
Sebastian Andrzej Siewior [Tue, 27 Apr 2010 22:54:50 +0000 (15:54 -0700)]
net/sb1250: setup the pdevice within the soc code

doing it within the driver does not look good.
And surely isn't how platform devices were meat to be used.

Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/sb1250: remove CONFIG_SIBYTE_STANDALONE
Sebastian Andrzej Siewior [Tue, 27 Apr 2010 22:53:50 +0000 (15:53 -0700)]
net/sb1250: remove CONFIG_SIBYTE_STANDALONE

CONFIG_SIBYTE_STANDALONE is gone since v2.6.31-rc1 ("MIPS: Sibyte:
Remove standalone kernel support")
This is a missing piece.

Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: disallow to use net_assign_generic externally
Jiri Pirko [Fri, 23 Apr 2010 01:40:47 +0000 (01:40 +0000)]
net: disallow to use net_assign_generic externally

Now there's no need to use this fuction directly because it's handled by
register_pernet_device. So to make this simple and easy to understand,
make this static to do not tempt potentional users.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocxgb4: increase serial number length
Dimitris Michailidis [Tue, 27 Apr 2010 12:24:16 +0000 (12:24 +0000)]
cxgb4: increase serial number length

Some boards have longer serial numbers in their VPD, up to 24 bytes.

Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocxgb4: parse the VPD instead of relying on a static VPD layout
Dimitris Michailidis [Tue, 27 Apr 2010 12:24:15 +0000 (12:24 +0000)]
cxgb4: parse the VPD instead of relying on a static VPD layout

Some boards' VPDs contain additional keywords or have longer serial numbers,
meaning the keyword locations are variable.  Ditch the static layout and
use the pci_vpd_* family of functions to parse the VPD instead.

Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: sk_add_backlog() take rmem_alloc into account
Eric Dumazet [Tue, 27 Apr 2010 22:13:20 +0000 (15:13 -0700)]
net: sk_add_backlog() take rmem_alloc into account

Current socket backlog limit is not enough to really stop DDOS attacks,
because user thread spend many time to process a full backlog each
round, and user might crazy spin on socket lock.

We should add backlog size and receive_queue size (aka rmem_alloc) to
pace writers, and let user run without being slow down too much.

Introduce a sk_rcvqueues_full() helper, to avoid taking socket lock in
stress situations.

Under huge stress from a multiqueue/RPS enabled NIC, a single flow udp
receiver can now process ~200.000 pps (instead of ~100 pps before the
patch) on a 8 core machine.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: batch skb dequeueing from softnet input_pkt_queue
Changli Gao [Tue, 27 Apr 2010 22:07:33 +0000 (15:07 -0700)]
net: batch skb dequeueing from softnet input_pkt_queue

batch skb dequeueing from softnet input_pkt_queue to reduce potential lock
contention when RPS is enabled.

Note: in the worst case, the number of packets in a softnet_data may
be double of netdev_max_backlog.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: Make RFS socket operations not be inet specific.
David S. Miller [Tue, 27 Apr 2010 22:05:31 +0000 (15:05 -0700)]
net: Make RFS socket operations not be inet specific.

Idea from Eric Dumazet.

As for placement inside of struct sock, I tried to choose a place
that otherwise has a 32-bit hole on 64-bit systems.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
14 years agoipheth: potential null dereferences on error path
Dan Carpenter [Mon, 26 Apr 2010 23:20:12 +0000 (23:20 +0000)]
ipheth: potential null dereferences on error path

The calls to usb_free_buffer() dereference rx_urb and tx_urb in the
parameter list but those could be NULL.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: L. Alberto Giménez <agimenez@sysvalve.es>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosmc91c92_cs: spin_unlock_irqrestore before calling smc_interrupt()
Ken Kawasaki [Sat, 24 Apr 2010 10:37:09 +0000 (10:37 +0000)]
smc91c92_cs: spin_unlock_irqrestore before calling smc_interrupt()

smc91c92_cs:
  * spin_unlock_irqrestore before calling smc_interrupt() in media_check()
     to avoid lockup.
  * use spin_lock_irqsave for ethtool function.

Signed-off-by: Ken Kawasaki <ken_kawasaki@spring.nifty.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/usb/net/kaweth.c: add device "Allied Telesyn AT-USB10 USB Ethernet Adapter"
Andreas Hartmann [Tue, 27 Apr 2010 21:39:33 +0000 (14:39 -0700)]
drivers/usb/net/kaweth.c: add device "Allied Telesyn AT-USB10 USB Ethernet Adapter"

akpm: reluctantly typed in from
https://bugzilla.kernel.org/show_bug.cgi?id=15599

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Update version to 2.0.9.
Michael Chan [Tue, 27 Apr 2010 11:28:11 +0000 (11:28 +0000)]
bnx2: Update version to 2.0.9.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Prevent "scheduling while atomic" warning with cnic, bonding and vlan.
Michael Chan [Tue, 27 Apr 2010 11:28:10 +0000 (11:28 +0000)]
bnx2: Prevent "scheduling while atomic" warning with cnic, bonding and vlan.

The bonding driver calls ndo_vlan_rx_register() while holding bond->lock.
The bnx2 driver calls bnx2_netif_stop() to stop the rx handling while
changing the vlgrp.  The call also stops the cnic driver which sleeps
while the bond->lock is held and cause the warning.

This code path only needs to stop the NAPI rx handling while we are
changing the vlgrp.  Since no reset is going to occur, there is no need
to stop cnic in this case.  By adding a parameter to bnx2_netif_stop()
to skip stopping cnic, we can avoid the warning.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobnx2: Fix lost MSI-X problem on 5709 NICs.
Michael Chan [Tue, 27 Apr 2010 11:28:09 +0000 (11:28 +0000)]
bnx2: Fix lost MSI-X problem on 5709 NICs.

It has been reported that under certain heavy traffic conditions in MSI-X
mode, the driver can lose an MSI-X vector causing all packets in the
associated rx/tx ring pair to be dropped.  The problem is caused by
the chip dropping the write to unmask the MSI-X vector by the kernel
(when migrating the IRQ for example).

This can be prevented by increasing the GRC timeout value for these
register read and write operations.

Thanks to Dell for helping us debug this problem.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: Properly display 1 gig downshift warning for backplane
Anjali Singhai [Tue, 27 Apr 2010 11:31:25 +0000 (11:31 +0000)]
ixgbe: Properly display 1 gig downshift warning for backplane

Description: When using Intel smartspeed, the patch displays a
warning when the link down shifts to 1 Gig.

Signed-off-by: Anjali Singhai <anjali.singhai@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: cleanup ethtool autoneg input
Don Skidmore [Tue, 27 Apr 2010 11:31:06 +0000 (11:31 +0000)]
ixgbe: cleanup ethtool autoneg input

The way we were setting autoneg via ethtool was inconstant with that
of our other drivers.  It will change the following:

If autoneg is off:
>ethtool -a eth0
Pause parameters for eth0:

Autonegotiate:  off
RX:             off
TX:             off

Before:
>ethtool -A eth0 autoneg on
>ethtool -a eth0
Pause parameters for eth0:

Autonegotiate:  off
RX:             off
TX:             off

Now:
>ethtool -A eth0 autoneg on
>ethtool -a eth0
Pause parameters for eth0:

Autonegotiate:  on
RX:             on
TX:             on

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbevf: Fix link speed display
Greg Rose [Tue, 27 Apr 2010 11:31:45 +0000 (11:31 +0000)]
ixgbevf: Fix link speed display

The ixgbevf driver would always report 10Gig speeds even when the link
speed is downshifted to 1Gig.  This patch fixes that problem.

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: reimplement softnet_data.output_queue as a FIFO queue
Changli Gao [Mon, 26 Apr 2010 23:06:24 +0000 (23:06 +0000)]
net: reimplement softnet_data.output_queue as a FIFO queue

reimplement softnet_data.output_queue as a FIFO queue to keep the
fairness among the qdiscs rescheduled.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
----
 include/linux/netdevice.h |    1 +
 net/core/dev.c            |   22 ++++++++++++----------
 2 files changed, 13 insertions(+), 10 deletions(-)
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/kaber/ipmr-2.6
David S. Miller [Tue, 27 Apr 2010 19:57:39 +0000 (12:57 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/kaber/ipmr-2.6

14 years agoixgb: Use pr_<level> and netdev_<level>
Joe Perches [Tue, 27 Apr 2010 00:50:58 +0000 (00:50 +0000)]
ixgb: Use pr_<level> and netdev_<level>

Convert DEBUGOUTx to pr_debug
Convert DEBUGFUNC to more commonly used ENTER
Convert mac address output to %pM
Use #define pr_fmt
Convert a few printks to pr_<level>
Improve ixgb_mc_addr_list_update: use a temporary for current mc address
Use etherdevice.h functions for mac address testing

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: ixgbe_down needs to stop dev_watchdog
John Fastabend [Tue, 27 Apr 2010 02:13:39 +0000 (02:13 +0000)]
ixgbe: ixgbe_down needs to stop dev_watchdog

There is a small race between when the tx queues are stopped
and when netif_carrier_off() is called in ixgbe_down.  If the
dev_watchdog() timer fires during this time it is possible for
a false tx timeout to occur.

This patch moves the netif_carrier_off() so that it is called before
the tx queues are stopped preventing the dev_watchdog timer from
detecting false tx timeouts.  The race is seen occosionally when
FCoE or DCB settings are being configured or changed.

Testing note, running ifconfig up/down will not reproduce this
issue because dev_open/dev_close call dev_deactivate() and then
dev_activate().

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>