These are full of unresolved problems, mainly that conversions don't
work 1-1 from hrtimers to tasklet_hrtimers because unlike hrtimers
tasklets can't be killed from softirq context.

And when a qdisc gets reset, that's exactly what we need to do here.

We'll work this out in the net-next-2.6 tree and if warranted we'll
backport that work to -stable.

This reverts the following 3 changesets:

a2cb6a4d
("pkt_sched: Fix bogon in tasklet_hrtimer changes.")

38acce2d
("pkt_sched: Convert CBQ to tasklet_hrtimer.")

ee5f9757
("pkt_sched: Convert qdisc_watchdog to tasklet_hrtimer")
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After commit 2b85a34e
(net: No more expensive sock_hold()/sock_put() on each tx)
sk_free() frees socks conditionally and depends
on sk_wmem_alloc being set e.g. in sock_init_data(). But in some
cases sk_free() is called earlier, usually after other alloc errors.

Fix is to move sk_wmem_alloc initialization from sock_init_data()
to sk_alloc() itself.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch solves the problem that the Oops(BUG_ON) occurs by rmmod.

	# rmmod gianfar_driver
	------------[ cut here ]------------
	Kernel BUG at c01fec48 [verbose debug info unavailable]
	Oops: Exception in kernel mode, sig: 5 [#1]
	MPC837x MDS
	Modules linked in: gianfar_driver(-) usb_storage scsi_wait_scan
	NIP: c01fec48 LR: c01febf4 CTR: c01feba8
	REGS: dec5bd60 TRAP: 0700   Tainted: G        W   (2.6.31-rc2)
	MSR: 00029032 <EE,ME,CE,IR,DR>  CR: 22000424  XER: 20000000
	TASK = dec4cac0[1135] 'rmmod' THREAD: dec5a000
	GPR00: 00000002 dec5be10 dec4cac0 dfba1820 c035d444 c035d478 ffffffff 00000000
	GPR08: 0000002b 00000001 dfba193c 00000001 22000424 10019b34 1ffcb000 00000000
	GPR16: 10012008 00000000 bf82ebe0 100017ec bf82ebec bf82ebe8 bf82ebd0 00000880
	GPR24: 00000000 bf82ebf0 c03532f0 c03532e4 c036b594 dfba183c dfba1800 dfba1820
	NIP [c01fec48] free_netdev+0xa0/0xb8
	LR [c01febf4] free_netdev+0x4c/0xb8
	Call Trace:
	[dec5be10] [c01febf4] free_netdev+0x4c/0xb8 (unreliable)
	[dec5be30] [e105f290] gfar_remove+0x50/0x68 [gianfar_driver]
	[dec5be40] [c01ec534] of_platform_device_remove+0x30/0x44
	[dec5be50] [c0181760] __device_release_driver+0x68/0xc8
	[dec5be60] [c0181868] driver_detach+0xa8/0xac
	[dec5be80] [c0180814] bus_remove_driver+0x9c/0xd8
	[dec5bea0] [c0181efc] driver_unregister+0x60/0x98
	[dec5beb0] [c01ec650] of_unregister_driver+0x14/0x24
	[dec5bec0] [e10631bc] gfar_exit+0x18/0x4bc [gianfar_driver]
	[dec5bed0] [c0047584] sys_delete_module+0x16c/0x228
	[dec5bf40] [c00116bc] ret_from_syscall+0x0/0x38
	--- Exception: c01 at 0xff3669c
	    LR = 0x10000f34
	Instruction dump:
	409e0024 a07e00c0 7c63f050 4be74429 80010024 bba10014 38210020 7c0803a6
	4e800020 68000003 3160ffff 7d2b0110 <0f090000> 38000004 387e01f0 901e01d4
	---[ end trace 8c595bcd37230a0f ]---
	 localhost kernel: ------------[ cut here ]------------

Signed-off-by: Toru UCHIYAMA uchiyama.toru@jp.fujitsu.com
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Bartlomiej Zolnierkiewicz reported an atomic order-6 allocation failure
for ipw2200 firmware loading in kernel 2.6.30. High order allocation is
likely to fail and should always be avoided.

The patch fixes this problem by replacing the original order-6
pci_alloc_consistent() with an array of order-1 pages from a pci pool.
This utilized the ipw2200 DMA command blocks (up to 64 slots). The
maximum firmware size support remains the same (64*8K).

This patch fixes bug http://bugzilla.kernel.org/show_bug.cgi?id=14016

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: NZhu Yi <yi.zhu@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

If we run out of memory, use keventd to fill the buffer.  There's a
report of this happening: "Page allocation failures in guest",
Message-ID: <20090713115158.0a4892b0@mjolnir.ossman.eu>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch is based on commit d2f3ad4c (pxaficp-ir: remove incorrect
net_device_ops). Do the same for sa1100_ir.
Untested.
Signed-off-by: NAlexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch is based on commit d2f3ad4c (pxaficp-ir: remove incorrect
net_device_ops). Do the same for au1k_ir.
Untested.
Signed-off-by: NAlexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Reported by Stephen Rothwell, luckily it's harmless:

net/sched/sch_api.c: In function 'qdisc_watchdog':
net/sched/sch_api.c:460: warning: initialization from incompatible pointer type
net/sched/sch_cbq.c: In function 'cbq_undelay':
net/sched/sch_cbq.c:595: warning: initialization from incompatible pointer type
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The NETPOLL requires that interrupts remain disabled in its callbacks.

Using *_irq_save()/irq_restore() to replace *_irq_disable()/irq_enable()
functions in NETPOLL's callbacks of smc91x, so that it doesn't enable
interrupts when already disabled, and kgdboe/netconsole would work
properly over smc91x.
Signed-off-by: NDongdong Deng <dongdong.deng@windriver.com>
Acked-by: NNicolas Pitre <nico@cam.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch fixes broken pxaficp-ir. The problem was in incorrect
net_device_ops being specified which prevented the driver from
operating. The symptoms were:
 - failing ifconfig for IrLAN, resulting in
	SIOCSIFFLAGS: Cannot assign requested address
 - irattach working for IrCOMM, but the port stayed disabled

Moreover this patch corrects missing sysfs device link.
Signed-off-by: NMarek Vasut <marek.vasut@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sllc_arphrd member of sockaddr_llc might not be changed. Zero sllc
before copying to the above layer's structure.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The NETPOLL API requires that interrupts remain disabled in
netpoll_send_skb(). The use of "A functions set" in the NETPOLL API
callbacks causes the interrupts to get enabled and can lead to kernel
instability.

The solution is to use "B functions set" to prevent the irqs from
getting enabled while in netpoll_send_skb().

A functions set:
local_irq_disable()/local_irq_enable()
spin_lock_irq()/spin_unlock_irq()
spin_trylock_irq()/spin_unlock_irq()

B functions set:
local_irq_save()/local_irq_restore()
spin_lock_irqsave()/spin_unlock_irqrestore()
spin_trylock_irqsave()/spin_unlock_irqrestore()
Signed-off-by: NDongdong Deng <dongdong.deng@windriver.com>
Acked-by: NMatt Mackall <mpm@selenic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

WARN_ONCE for ndo_start_xmit() enable interrupts in netpoll_send_skb(),
because the NETPOLL API requires that interrupts remain disabled in
netpoll_send_skb().
Signed-off-by: NDongdong Deng <dongdong.deng@windriver.com>
Acked-by: NMatt Mackall <mpm@selenic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NAndreas Mohr <andi@lisas.de>
Acked-by: NRichard Röjfors <richard.rojfors.ext@mocean-labs.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Success was indicated on a memory allocation failure, thereby causing
a crash due to a later NULL deref.
(Affects v2.6.30-rc1 up to here.)
Signed-off-by: NJan Engelhardt <jengelh@medozas.de>
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ipv6: Fix commit 63d9950b (ipv6: Make v4-mapped bindings consistent with IPv4)

Commit 63d9950b
  (ipv6: Make v4-mapped bindings consistent with IPv4)
changes behavior of inet6_bind() for v4-mapped addresses so it should
behave the same way as inet_bind().

During this change setting of err to -EADDRNOTAVAIL got lost:

af_inet.c:469 inet_bind()
	err = -EADDRNOTAVAIL;
	if (!sysctl_ip_nonlocal_bind &&
	    !(inet->freebind || inet->transparent) &&
	    addr->sin_addr.s_addr != htonl(INADDR_ANY) &&
	    chk_addr_ret != RTN_LOCAL &&
	    chk_addr_ret != RTN_MULTICAST &&
	    chk_addr_ret != RTN_BROADCAST)
		goto out;


af_inet6.c:463 inet6_bind()
	if (addr_type == IPV6_ADDR_MAPPED) {
		int chk_addr_ret;

		/* Binding to v4-mapped address on a v6-only socket                         
		 * makes no sense                                                           
		 */
		if (np->ipv6only) {
			err = -EINVAL;
			goto out; 
		}

		/* Reproduce AF_INET checks to make the bindings consitant */               
		v4addr = addr->sin6_addr.s6_addr32[3];                                      
		chk_addr_ret = inet_addr_type(net, v4addr);                                 
		if (!sysctl_ip_nonlocal_bind &&                                             
		    !(inet->freebind || inet->transparent) &&                               
		    v4addr != htonl(INADDR_ANY) &&
		    chk_addr_ret != RTN_LOCAL &&                                            
		    chk_addr_ret != RTN_MULTICAST &&                                        
		    chk_addr_ret != RTN_BROADCAST)
			goto out;
	} else {


Signed-off-by Bruno Prémont <bonbons@linux-vserver.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

E100 places it's RX packet descriptors inside skb->data and uses them
with bidirectional streaming DMA mapping. Data in descriptors is
accessed simultaneously by the chip (writing status and size when
a packet is received) and CPU (reading to check if the packet was
received). This isn't a valid usage of PCI DMA API, which requires use
of the coherent (consistent) memory for such purpose. Unfortunately e100
chips working in "simplified" RX mode have to store received data
directly after the descriptor. Fixing the driver to conform to the API
would require using unsupported "flexible" RX mode or receiving data
into a coherent memory and using CPU to copy it to network buffers.

This patch, while not yet making the driver conform to the PCI DMA API,
allows it to work correctly on X86 with swiotlb (while not breaking
other architectures).
Signed-off-by: NKrzysztof Hałasa <khc@pm.waw.pl>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This code expects to run in softirq context, and bare hrtimers
run in hw IRQ context.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NThomas Gleixner <tglx@linutronix.de>

None of this stuff should execute in hw IRQ context, therefore
use a tasklet_hrtimer so that it runs in softirq context.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NThomas Gleixner <tglx@linutronix.de>

RTL8187B always needs MSR_LINK_ENEDCA flag to be set even when it is in
no link mode, otherwise it'll not be able to associate when this flag is
not set after the change "mac80211: fix managed mode BSSID handling".

By accident, setting BSSID of AP before association makes 8187B to
successfuly associate even when ENEDCA flag isn't set, which was the
case before the mac80211 change. But now the BSSID of AP we are trying
to associate is only available after association is successful, and
any attempt to associate without the needed flag doesn't work.
Signed-off-by: NHerton Ronaldo Krzesinski <herton@mandriva.com.br>
Tested-by: NLarry Finger <Larry.Finger@lwfinger.net>
Acked-by: NHin-Tak Leung <htl10@users.sourceforge.net>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

When ibm_newemac netdev instance is shutdown with "ifconfig down",
the netdev interface does not go properly down. netif_carrier_ok()
keeps returning TRUE even after "ifconfig down".

The problem can be seen when ibm_newemac instances are slaves of
a bonding interface. The bonding interface code uses netif_carrier_ok()
to determine the link status of its slaves. When ibm_newemac slave is
shutdown with "ifconfig down", the bonding interface won't detect any
link status change because netif_carrier_ok() keeps returning TRUE.
Signed-off-by: NPetri Gynther <pgynther@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix build errors due to missing Kconfig select of CRC32:

ks8851.c:(.text+0x7d2ee): undefined reference to `crc32_le'
ks8851.c:(.text+0x7d2f5): undefined reference to `bitrev32'
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Due to I modified the corresponding platform device name, 
so I make the patch to rename MAC platform driver
for w90p910 platform.
Signed-off-by: NWan ZongShun <mcuos.com@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

yellowfin_init_ring() needs to clean up if dev_alloc_skb() fails and
should pass an error status up to the caller. This also prevents an
buffer underrun if failure occurred in the first iteration.
yellowfin_open() which calls yellowfin_init_ring() should free its
requested irq upon failure.
Signed-off-by: NRoel Kluin <roel.kluin@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In 5e140dfc "net: reorder struct Qdisc
for better SMP performance" the definition of struct gnet_stats_basic
changed incompatibly, as copies of this struct are shipped to
userland via netlink.

Restoring old behavior is not welcome, for performance reason.

Fix is to use a private structure for kernel, and
teach gnet_stats_copy_basic() to convert from kernel to user land,
using legacy structure (struct gnet_stats_basic)

Based on a report and initial patch from Michael Spang.
Reported-by: NMichael Spang <mspang@csclub.uwaterloo.ca>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The static variable used by nr_call_to_digi might result in corruption if
multiple threads are trying to usee a node or neighbour via ioctl.  Fixed
by having the caller pass a structure in.  This is safe because nr_add_node
rsp. nr_add_neigh will allocate a permanent structure, if needed.
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If key is 4 that is an array out of bounds.
Reported-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The key todo lock can be taken from different locks
that require it to be _bh to avoid lock inversion
due to (soft)irqs.

This should fix the two problems reported by Bob and
Gabor:
http://mid.gmane.org/20090619113049.GB18956@hash.localnet
http://mid.gmane.org/4A3FA376.8020307@openwrt.orgSigned-off-by: NJohannes Berg <johannes@sipsolutions.net>
Cc: Bob Copeland <me@bobcopeland.com>
Cc: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

commit 111b9dc5 ("e1000e: add aer support") introduces pcie aer
support for e1000e, but it is not reasonable to disable it in
e1000_remove but enable it in e1000_resume.  This patch enables aer
support in e1000_probe.
Signed-off-by: NXiaotian Feng <dfeng@redhat.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

With manageability (Intel AMT) enabled via BIOS, PHY wakeup does not get
configured on newer parts which use PHY wakeup vs. MAC wakeup which causes
WoL to not work.  The driver should configure PHY wakeup whether or not
manageability is enabled.
Signed-off-by: NBruce Allan <bruce.w.allan@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The slow path ulp_init and ulp_exit calls to the bnx2i driver
are sleepable calls and therefore should not be protected using
rcu_read_lock.  Fix it by using mutex and refcount during these
calls.  cnic_unregister_driver() will now wait for the refcount
to go to zero before completing the call.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The slow path ulp_start and ulp_stop calls to the bnx2i driver
are sleepable calls and therefore should not be protected using
rcu_read_lock.  Fix it by using mutex and setting a bit during
these calls.  cnic_unregister_device() will now wait for the bit
to clear before completing the call.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The slow path calls to the cnic driver are sleepable calls so we
cannot use rcu_read_lock().  Use mutex for these slow path calls
instead.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Register and unregister with bnx2 during NETDEV_UP and NETDEV_DOWN
events.  This simplifies the sequence of events and allows locking
fixes in the next patch.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the cnic driver tries to grab a symbol from bnx2 when bnx2 is
running init code, symbol_get() will succeed but symbol_put_addr()
will hit BUG() a moment later.  module_text_address() fails because
bnx2 is still in init code.

This is fixed by using symbol_put() instead which does the exact
opposite of symbol_get().
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The GRE header length should be subtracted when the tunnel MTU is
calculated.  This just corrects for the associativity change
introduced by commit 42aa9162
("gre: Move MTU setting out of ipgre_tunnel_bind_dev").
Signed-off-by: NTom Goff <thomas.goff@boeing.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>