11 years agomyri10ge: print MAC and serial number on probe failure
Brice Goglin [Tue, 6 Jan 2009 02:16:14 +0000]
myri10ge: print MAC and serial number on probe failure

To help board identification and diagnosis, print the MAC
and serial number on probe failure if they are available.

Signed-off-by: Brice Goglin <brice@myri.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agopkt_sched: cls_u32: Fix locking in u32_change()
Jarek Poplawski [Tue, 6 Jan 2009 02:14:19 +0000]
pkt_sched: cls_u32: Fix locking in u32_change()

New nodes are inserted in u32_change() under rtnl_lock() with wmb(),
so without tcf_tree_lock() like in other classifiers (e.g. cls_fw).
This isn't enough without rmb() on the read side, but on the other
hand adding such barriers doesn't give any savings, so the lock is
added instead.

Reported-by: m0sia <m0sia@plotinka.ru>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoiucv: fix cpu hotplug
Heiko Carstens [Tue, 6 Jan 2009 02:09:02 +0000]
iucv: fix cpu hotplug

If the iucv module is compiled in/loaded but no user is registered cpu
hot remove doesn't work. Reason for that is that the iucv cpu hotplug
notifier on CPU_DOWN_PREPARE checks if the iucv_buffer_cpumask would
be empty after the corresponding bit would be cleared. However the bit
was never set since iucv wasn't enable. That causes all cpu hot unplug
operations to fail in this scenario.
To fix this use iucv_path_table as an indicator wether iucv is enabled
or not.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoaf_iucv: Free iucv path/socket in path_pending callback
Hendrik Brueckner [Tue, 6 Jan 2009 02:08:23 +0000]
af_iucv: Free iucv path/socket in path_pending callback

Free iucv path after iucv_path_sever() calls in iucv_callback_connreq()
(path_pending() iucv callback).
If iucv_path_accept() fails, free path and free/kill newly created socket.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoaf_iucv: avoid left over IUCV connections from failing connects
Ursula Braun [Tue, 6 Jan 2009 02:07:46 +0000]
af_iucv: avoid left over IUCV connections from failing connects

For certain types of AFIUCV socket connect failures IUCV connections
are left over. Add some cleanup-statements to avoid cluttered IUCV
connections.

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoaf_iucv: New error return codes for connect()
Hendrik Brueckner [Tue, 6 Jan 2009 02:07:07 +0000]
af_iucv: New error return codes for connect()

If the iucv_path_connect() call fails then return an error code that
corresponds to the iucv_path_connect() failure condition; instead of
returning -ECONNREFUSED for any failure.

This helps to improve error handling for user space applications
(e.g.  inform the user that the z/VM guest is not authorized to
connect to other guest virtual machines).

The error return codes are based on those described in connect(2).

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet/ehea: bitops work on unsigned longs
Stephen Rothwell [Tue, 6 Jan 2009 00:06:02 +0000]
net/ehea: bitops work on unsigned longs

The flags field of struct ehea_port is only used with test_bit(),
clear_bit() and set_bit() and these interfaces only work on
"unsigned long"s, so change the field to be an "unsigned long".  Also,
this field only has two bits defined for it (0 and 1) so will still be
fine if someone builds this driver for a 32 bit arch (at least as far as
this flags field is concerned).

Also note that ehea_driver_flags is only used in ehca_main.c, so make it
static in there.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoRevert "net: Fix for initial link state in 2.6.28"
David S. Miller [Tue, 6 Jan 2009 00:01:51 +0000]
Revert "net: Fix for initial link state in 2.6.28"

This reverts commit 22604c866889c4b2e12b73cbf1683bda1b72a313.

We can't fix this issue in this way, because we now can try
to take the dev_base_lock rwlock as a writer in software interrupt
context and that is not allowed without major surgery elsewhere.

This initial link state problem needs to be solved in some other
way.

Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agotcp: Kill extraneous SPLICE_F_NONBLOCK checks.
David S. Miller [Mon, 5 Jan 2009 08:59:00 +0000]
tcp: Kill extraneous SPLICE_F_NONBLOCK checks.

In splice TCP receive, the SPLICE_F_NONBLOCK flag is used
to compute the "timeo" value.  So checking it again inside
of the main receive loop to trigger -EAGAIN processing is
entirely unnecessary.

Noticed by Jarek P. and Lennert Buytenhek.

Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agotcp: don't mask EOF and socket errors on nonblocking splice receive
Lennert Buytenhek [Mon, 5 Jan 2009 08:00:12 +0000]
tcp: don't mask EOF and socket errors on nonblocking splice receive

Currently, setting SPLICE_F_NONBLOCK on splice from a TCP socket
results in masking of EOF (RDHUP) and error conditions on the socket
by an -EAGAIN return.  Move the NONBLOCK check in tcp_splice_read()
to be after the EOF and error checks to fix this.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agodccp: Integrate the TFRC library with DCCP
Gerrit Renker [Mon, 5 Jan 2009 05:45:33 +0000]
dccp: Integrate the TFRC library with DCCP

This patch integrates the TFRC library, which is a dependency of CCID-3 (and
CCID-4), with the new use of CCIDs in the DCCP module.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agodccp: Clean up ccid.c after integration of CCID plugins
Gerrit Renker [Mon, 5 Jan 2009 05:43:23 +0000]
dccp: Clean up ccid.c after integration of CCID plugins

This patch cleans up after integrating the CCID modules and, in addition,

 * moves the if/else cases from ccid_delete() into ccid_hc_{tx,rx}_delete();
 * removes the 'gfp' argument to ccid_new() - since it is always gfp_any().

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agodccp: Lockless integration of CCID congestion-control plugins
Gerrit Renker [Mon, 5 Jan 2009 05:42:53 +0000]
dccp: Lockless integration of CCID congestion-control plugins

Based on Arnaldo's earlier patch, this patch integrates the standardised
CCID congestion control plugins (CCID-2 and CCID-3) of DCCP with dccp.ko:

 * enables a faster connection path by eliminating the need to always go
   through the CCID registration lock;

 * updates the implementation to use only a single array whose size equals
   the number of configured CCIDs instead of the maximum (256);

 * since the CCIDs are now fixed array elements, synchronization is no
   longer needed, simplifying use and implementation.

CCID-2 is suggested as minimum for a basic DCCP implementation (RFC 4340, 10);
CCID-3 is a standards-track CCID supported by RFC 4342 and RFC 5348.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqeth: get rid of extra argument after printk to dev_* conversion
Heiko Carstens [Mon, 5 Jan 2009 01:36:32 +0000]
qeth: get rid of extra argument after printk to dev_* conversion

drivers/s390/net/qeth_l3_main.c: In function 'qeth_l3_setadapter_parms':
drivers/s390/net/qeth_l3_main.c:1049: warning: too many arguments for format

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqeth: No large send using EDDP for HiperSockets.
Klaus-Dieter Wacker [Mon, 5 Jan 2009 01:36:05 +0000]
qeth: No large send using EDDP for HiperSockets.

From: Klaus-Dieter Wacker <kdwacker@de.ibm.com>

The device driver qeth dos not support large send using EDDP for
HiperSockets.

Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqeth: do not spin for SETIP ip assist command
Frank Blaschka [Mon, 5 Jan 2009 01:35:44 +0000]
qeth: do not spin for SETIP ip assist command

The ip assist hw command for setting an IP address last unacceptable
long so we can not spin while we waiting for the irq. Since we can
ensure process context for all occurrences of this command we can use
wait.

Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqeth: avoid crash in case of layer mismatch for VSWITCH
Ursula Braun [Mon, 5 Jan 2009 01:35:18 +0000]
qeth: avoid crash in case of layer mismatch for VSWITCH

For z/VM GuestLAN or VSWITCH devices the transport layer is
configured in z/VM. The layer2 attribute of a participating Linux
device has to match the z/VM definition. In case of a mismatch
Linux currently crashes in qeth recovery due to a reference to the
not yet existing net_device.
Solution: add a check for existence of net_device and add a message
pointing to the mismatch of layer definitions in Linux and z/VM.

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqeth: exploit source MAC address for inbound layer3 packets
Ursula Braun [Mon, 5 Jan 2009 01:34:52 +0000]
qeth: exploit source MAC address for inbound layer3 packets

OSA-devices operating in layer3 mode offer adding of the source MAC
address to the QDIO header of inbound packets. The qeth driver can
exploit this functionality to replace FAKELL-entries in the ethernet
header of received packets.

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqeth: HiperSockets mcl string conversion (pre z9 mach)
Klaus-Dieter Wacker [Mon, 5 Jan 2009 01:34:10 +0000]
qeth: HiperSockets mcl string conversion (pre z9 mach)

The pre z9 machines provide an mcl string in EBCDIC format,
z9 or later provide string in ASCII format.

Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agocan: update can-bcm for hrtimer hardirq callbacks
Oliver Hartkopp [Mon, 5 Jan 2009 01:31:18 +0000]
can: update can-bcm for hrtimer hardirq callbacks

Since commit ca109491f612aab5c8152207631c0444f63da97f ("hrtimer:
removing all ur callback modes") the hrtimer callbacks are processed
only in hardirq context.

This patch moves some functionality into tasklets to run in softirq
context.

Additionally some duplicated code was removed in bcm_rx_thr_flush()
and an avoidable memcpy was removed from bcm_rx_handler().

Signed-off-by: Oliver Hartkopp <oliver@hartkopp.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoDCB: fix kfree(skb)
Roel Kluin [Mon, 5 Jan 2009 01:29:21 +0000]
DCB: fix kfree(skb)

Use kfree_skb instead of kfree for struct sk_buff pointers.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoipv6: IPV6_PKTINFO relied userspace providing correct length
Ilpo Järvinen [Mon, 5 Jan 2009 01:27:31 +0000]
ipv6: IPV6_PKTINFO relied userspace providing correct length

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Eric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: Fix for initial link state in 2.6.28
Michael Marineau [Mon, 5 Jan 2009 01:18:51 +0000]
net: Fix for initial link state in 2.6.28

From: Michael Marineau <mike@marineau.org>

Commit b47300168e770b60ab96c8924854c3b0eb4260eb "Do not fire linkwatch
events until the device is registered." was made as a workaround for
drivers that call netif_carrier_off before registering the device.
Unfortunately this causes these drivers to incorrectly report their
link status as IF_OPER_UNKNOWN which can falsely set the IFF_RUNNING
flag when the interface is first brought up. This issues was
previously pointed out[1] but was dismissed saying that IFF_RUNNING is
not related to the link status. From my digging IFF_RUNNING, as
reported to userspace, is based on the link state. It is set based on
__LINK_STATE_START and IF_OPER_UP or IF_OPER_UNKNOWN. See [2], [3],
and [4]. (Whether or not the kernel has IFF_RUNNING set in flags is
not reported to user space so it may well be independent of the link,
I don't know if and when it may get set.)

The end result depends slightly depending on the driver. The the two I
tested were e1000e and b44. With e1000e if the system is booted
without a network cable attached the interface will falsely report
RUNNING when it is brought up causing NetworkManager to attempt to
start it and eventually time out. With b44 when the system is booted
with a network cable attached and brought up with dhcpcd it will time
out the first time.

The attached patch that will still set the operstate variable
correctly to IF_OPER_UP/DOWN/etc when linkwatch_fire_event is called
but then return rather than skipping the linkwatch_fire_event call
entirely as the previous fix did. (sorry it isn't inline, I don't have
a patch friendly email client at the moment)

Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agotun: Eliminate sparse signedness warning
Gerrit Renker [Mon, 5 Jan 2009 01:14:46 +0000]
tun: Eliminate sparse signedness warning

register_pernet_gen_device() expects 'int*', found via sparse.

 CHECK   drivers/net/tun.c
 drivers/net/tun.c:1245:36: warning: incorrect type in argument 1 (different signedness)
 drivers/net/tun.c:1245:36:    expected int *id
 drivers/net/tun.c:1245:36:    got unsigned int static [toplevel] *<noident>

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoe100: cosmetic cleanup
Bruce Allan [Mon, 5 Jan 2009 01:12:04 +0000]
e100: cosmetic cleanup

Add missing space after if, switch, for and while keywords.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet/rfkill/rfkill.c: fix unused rfkill_led_trigger() warning
Simon Holm Thøgersen [Mon, 5 Jan 2009 01:11:24 +0000]
net/rfkill/rfkill.c: fix unused rfkill_led_trigger() warning

commit 4dec9b807be757780ca3611a959ac22c28d292a7 ("rfkill: strip pointless
notifier chain") removed the only user of rfkill_led_trigger() that was not
guarded by #ifdef CONFIG_RFKILL_LEDS. Therefore, move rfkill_led_trigger()
completely inside #ifdef CONFIG_RFKILL_LEDS and avoid the compile time
warning:

net/rfkill/rfkill.c:59: warning: 'rfkill_led_trigger' defined but not used

Signed-off-by: Simon Holm Thøgersen <odie@cs.aau.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqlge: bugfix: Fix register access error checking.
Ron Mercer [Mon, 5 Jan 2009 01:08:29 +0000]
qlge: bugfix: Fix register access error checking.

Some indexed registers do not have error bits. In these cases a
value of zero should be used for error checking.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqlge: bugfix: Fix ring length setting for rx ring, large/small
Ron Mercer [Mon, 5 Jan 2009 01:08:11 +0000]
qlge: bugfix: Fix ring length setting for rx ring, large/small

The length field for these rings is 16-bits.  If the length is
the max supported 65536 then the setting should be zero.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqlge: bugfix: Fix shadow register endian issue.
Ron Mercer [Mon, 5 Jan 2009 01:07:50 +0000]
qlge: bugfix: Fix shadow register endian issue.

Shadow registers are consistent memory locations to which the chip
echos ring indexes in little endian format.  These values need to
be endian swapped before referencing.

Note:
The register pointer declaration uses the volatile modifier which
causes warnings in checkpatch.
Per Documentation/volatile-considered-harmful.txt:
  - Pointers to data structures in coherent memory which might be modified
    by I/O devices can, sometimes, legitimately be volatile.  A ring buffer
    used by a network adapter, where that adapter changes pointers to
    indicate which descriptors have been processed, is an example of this
    type of situation.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqlge: bugfix: Add missing pci_unmap_page call in receive path.
Ron Mercer [Mon, 5 Jan 2009 01:07:09 +0000]
qlge: bugfix: Add missing pci_unmap_page call in receive path.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoqlge: bugfix: Add missing pci_mapping_err checking.
Ron Mercer [Mon, 5 Jan 2009 01:06:46 +0000]
qlge: bugfix: Add missing pci_mapping_err checking.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoenc28j60: fix RX buffer overflow
Baruch Siach [Mon, 5 Jan 2009 00:23:01 +0000]
enc28j60: fix RX buffer overflow

The enc28j60 driver doesn't check whether the length of the packet as reported
by the hardware fits into the preallocated buffer. When stressed, the hardware
may report insanely large packets even tough the "Receive OK" bit is set. Fix
this.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoisdn: capi: &&/|| typos
Roel Kluin [Mon, 5 Jan 2009 00:22:04 +0000]
isdn: capi: &&/|| typos

Correct two typos.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Acked-by: Karsten Keil <kkeil@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agogro: Add page frag support
Herbert Xu [Mon, 5 Jan 2009 00:13:40 +0000]
gro: Add page frag support

This patch allows GRO to merge page frags (skb_shinfo(skb)->frags)
in one skb, rather than using the less efficient frag_list.

It also adds a new interface, napi_gro_frags to allow drivers
to inject page frags directly into the stack without allocating
an skb.  This is intended to be the GRO equivalent for LRO's
lro_receive_frags interface.

The existing GSO interface can already handle page frags with
or without an appended frag_list so nothing needs to be changed
there.

The merging itself is rather simple.  We store any new frag entries
after the last existing entry, without checking whether the first
new entry can be merged with the last existing entry.  Making this
check would actually be easy but since no existing driver can
produce contiguous frags anyway it would just be mental masturbation.

If the total number of entries would exceed the capacity of a
single skb, we simply resort to using frag_list as we do now.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agogro: Use gso_size to store MSS
Herbert Xu [Mon, 5 Jan 2009 00:13:19 +0000]
gro: Use gso_size to store MSS

In order to allow GRO packets without frag_list at all, we need to
store the MSS in the packet itself.  The obvious place is gso_size.
The only thing to watch out for is if the packet ends up not being
GRO then we need to clear gso_size before pushing the packet into
the stack.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agostarfire: use request_firmware()
Jaswinder Singh Rajput [Mon, 5 Jan 2009 00:12:11 +0000]
starfire: use request_firmware()

Firmware blob is big endian

Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agofirmware: convert tg3 driver to request_firmware()
Jaswinder Singh Rajput [Mon, 5 Jan 2009 00:11:25 +0000]
firmware: convert tg3 driver to request_firmware()

Firmware blob looks like this...
        u8 firmware_major
        u8 firmware_minor
        u8 firmware_fix
        u8 pad
        __be32 start_address
        __be32 length (total, including BSS sections to be zeroed)
        data... (in __be32 words, which is native for the firmware)

Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agofirmware: convert acenic driver to request_firmware()
Jaswinder Singh [Mon, 5 Jan 2009 00:09:40 +0000]
firmware: convert acenic driver to request_firmware()

We store the firmware in its native big-endian form now, so the loop in
ace_copy() is modified to use be32_to_cpup() when writing it out.

We can forget the BSS,SBSS sections of the firmware, since we were
clearing all the device's RAM anyway. And the text,rodata,data sections
can all be loaded as a single chunk since they're contiguous (give or
take a few dozen bytes in between).

Signed-off-by: Jaswinder Singh <jaswinder@infradead.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Acked-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoipv6: Fix sporadic sendmsg -EINVAL when sending to multicast groups.
David S. Miller [Mon, 5 Jan 2009 00:04:39 +0000]
ipv6: Fix sporadic sendmsg -EINVAL when sending to multicast groups.

Thanks to excellent diagnosis by Eduard Guzovsky.

The core problem is that on a network with lots of active
multicast traffic, the neighbour cache can fill up.  If
we try to allocate a new route and thus neighbour cache
entry, the bog-standard GC attempt the neighbour layer does
in ineffective because route entries hold a reference
to the existing neighbour entries and GC can only liberate
entries with no references.

IPV4 already has a way to handle this, by doing a route cache
GC in such situations (when neigh attach returns -ENOBUFS).

So simply mimick this on the ipv6 side.

Tested-by: Eduard Guzovsky <eguzovsky@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: Fix percpu counters deadlock
Herbert Xu [Tue, 30 Dec 2008 07:04:08 +0000]
net: Fix percpu counters deadlock

When we converted the protocol atomic counters such as the orphan
count and the total socket count deadlocks were introduced due to
the mismatch in BH status of the spots that used the percpu counter
operations.

Based on the diagnosis and patch by Peter Zijlstra, this patch
fixes these issues by disabling BH where we may be in process
context.

Reported-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agocpumask: prepare for iterators to only go to nr_cpu_ids/nr_cpumask_bits: net
Rusty Russell [Mon, 29 Dec 2008 12:23:42 +0000]
cpumask: prepare for iterators to only go to nr_cpu_ids/nr_cpumask_bits: net

In future all cpumask ops will only be valid (in general) for bit
numbers < nr_cpu_ids.  So use that instead of NR_CPUS in iterators
and other comparisons.

This is always safe: no cpu number can be >= nr_cpu_ids, and
nr_cpu_ids is initialized to NR_CPUS at boot.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Mike Travis <travis@sgi.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agodrivers/net/usb: use USB API functions rather than constants
Julia Lawall [Mon, 29 Dec 2008 00:21:07 +0000]
drivers/net/usb: use USB API functions rather than constants

This set of patches introduces calls to the following set of functions:

usb_endpoint_dir_in(epd)
usb_endpoint_dir_out(epd)
usb_endpoint_is_bulk_in(epd)
usb_endpoint_is_bulk_out(epd)
usb_endpoint_is_int_in(epd)
usb_endpoint_is_int_out(epd)
usb_endpoint_num(epd)
usb_endpoint_type(epd)
usb_endpoint_xfer_bulk(epd)
usb_endpoint_xfer_control(epd)
usb_endpoint_xfer_int(epd)
usb_endpoint_xfer_isoc(epd)

In some cases, introducing one of these functions is not possible, and it
just replaces an explicit integer value by one of the following constants:

USB_ENDPOINT_XFER_BULK
USB_ENDPOINT_XFER_CONTROL
USB_ENDPOINT_XFER_INT
USB_ENDPOINT_XFER_ISOC

In drivers/net/wireless/zd1211rw/zd_usb.c the code:

(endpoint->bEndpointAddress & USB_TYPE_MASK) == USB_DIR_OUT

is suspicious.  If it is intended to use USB_ENDPOINT_DIR_MASK rather than
USB_TYPE_MASK, then the whole conditional test could be converted to a call
to usb_endpoint_is_bulk_in.

An extract of the semantic patch that makes these changes is as follows:
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@r1@ struct usb_endpoint_descriptor *epd; @@

- ((epd->bmAttributes & \(USB_ENDPOINT_XFERTYPE_MASK\|3\)) ==
- \(USB_ENDPOINT_XFER_CONTROL\|0\))
+ usb_endpoint_xfer_control(epd)

@r5@ struct usb_endpoint_descriptor *epd; @@

- ((epd->bEndpointAddress & \(USB_ENDPOINT_DIR_MASK\|0x80\)) ==
-  \(USB_DIR_IN\|0x80\))
+ usb_endpoint_dir_in(epd)

@inc@
@@

#include <linux/usb.h>

@depends on !inc && (r1||r5)@
@@

+ #include <linux/usb.h>
  #include <linux/usb/...>
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agocls_cgroup: clean up Kconfig
Li Zefan [Sun, 28 Dec 2008 17:57:25 +0000]
cls_cgroup: clean up Kconfig

cls_cgroup can't be compiled as a module, since it's not supported by
cgroup.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agocls_cgroup: clean up for cgroup part
Li Zefan [Tue, 30 Dec 2008 03:39:03 +0000]
cls_cgroup: clean up for cgroup part

- It's better to use container_of() instead of casting cgroup_subsys_state *
  to cgroup_cls_state *.
- Add helper function task_cls_state().
- Rename net_cls_state() to cgrp_cls_state().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agocls_cgroup: fix an oops when removing a cgroup
Li Zefan [Sun, 28 Dec 2008 17:57:19 +0000]
cls_cgroup: fix an oops when removing a cgroup

When removing a cgroup, an oops was triggered immediately. The cause
is wrong kfree() in cgrp_destroy().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoEtherExpress16: fix printing timed out status
Roel Kluin [Tue, 30 Dec 2008 02:42:33 +0000]
EtherExpress16: fix printing timed out status

in drivers/net/eexpress.c:558, function unstick_cu()

while (!SCB_complete(rsst=scb_status(dev))) {
...
if (...)
printk(KERN_WARNING "%s: Reset timed out status %04x, retrying...\n",
                                                       dev->name,rsst);
}

but this will become

while (!((rsst = scb_status(dev) & 0x8000) != 0) ...

because of the macro:

#define SCB_complete(s) ((s&0x8000)!=0)

so rsst can only become either 0x8000 or 0, but in the latter case the
loop ends, I think the wrong timed out status is printed. This also
cleans up similar macros.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agomlx4_en: Added "set_ringparam" Ethtool interface implementation
Yevgeny Petrilin [Tue, 30 Dec 2008 02:39:20 +0000]
mlx4_en: Added "set_ringparam" Ethtool interface implementation

Now using Ethtool to determine ring sizes, removed the module parameters
that controlled those values.
Modifying ring size requires restart of the interface.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agomlx4_en: Always allocate RX ring for each interrupt vector
Yevgeny Petrilin [Tue, 30 Dec 2008 02:38:54 +0000]
mlx4_en: Always allocate RX ring for each interrupt vector

Removed module parameter specifying number of RX rings

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agomlx4_en: Verify number of RX rings doesn't exceed MAX_RX_RINGS
Yevgeny Petrilin [Tue, 30 Dec 2008 02:38:21 +0000]
mlx4_en: Verify number of RX rings doesn't exceed MAX_RX_RINGS

Required in cases were dev->caps.num_comp_vectors > MAX_RX_RINGS.
For current values this would happen on machines that have more
then 16 cores.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoIPVS: Make "no destination available" message more consistent between schedulers
Simon Horman [Tue, 30 Dec 2008 02:37:36 +0000]
IPVS: Make "no destination available" message more consistent between schedulers

Acked-by: Graeme Fowler <graeme@graemef.net>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: KS8695: removed duplicated #include
Huang Weiyi [Tue, 30 Dec 2008 02:30:05 +0000]
net: KS8695: removed duplicated #include

Removed duplicated include in drivers/net/arm/ks8695net.c.

Signed-off-by: Huang Weiyi <weiyi.huang@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agotun: Fix SIOCSIFHWADDR error.
Kusanagi Kouichi [Tue, 30 Dec 2008 02:23:28 +0000]
tun: Fix SIOCSIFHWADDR error.

Set proper operations.

Signed-off-by: Kusanagi Kouichi <slash@ma.neweb.ne.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosmsc911x: compile fix re netif_rx signature changes
Alexey Dobriyan [Tue, 30 Dec 2008 02:22:43 +0000]
smsc911x: compile fix re netif_rx signature changes

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Steve Glendinning <steve.glendinning@smsc.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonetns: foreach_netdev_safe is insufficient in default_device_exit
Eric W. Biederman [Tue, 30 Dec 2008 02:21:48 +0000]
netns: foreach_netdev_safe is insufficient in default_device_exit

During network namespace teardown we either move or delete
all of the network devices associated with a network namespace.
In the case of veth devices deleting one will also delete it's
pair device.  If both devices are in the same network namespace
then for_each_netdev_safe is insufficient as next may point
to the second veth device we have deleted.

To avoid problems I do what we do in __rtnl_kill_links and
restart the scan of the device list, after we have deleted
a device.

Currently dev_change_netnamespace does not appear to suffer from
this problem, but wireless devices are also paired and likely
should be moved between network namespaces together.  So I have
errored on the side of caution and restart the scan of the network
devices in that case as well.

Signed-off-by: Eric W. Biederman <ebiederm@aristanetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: make xfrm_statistics_seq_show use generic snmp_fold_field
Rusty Russell [Tue, 30 Dec 2008 02:20:06 +0000]
net: make xfrm_statistics_seq_show use generic snmp_fold_field

No reason to roll our own here.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: Fix more NAPI interface netdev argument drop fallout.
Kamalesh Babulal [Tue, 30 Dec 2008 02:18:24 +0000]
net: Fix more NAPI interface netdev argument drop fallout.

I hit similar build failure due to the change in the netif_rx_reschedule()

drivers/net/ehea/ehea_main.c: In function 'ehea_poll':
drivers/net/ehea/ehea_main.c:844: warning: passing argument 1 of 'netif_rx_reschedule' from incompatible pointer type
drivers/net/ehea/ehea_main.c:844: error: too many arguments to function 'netif_rx_reschedule'
make[3]: *** [drivers/net/ehea/ehea_main.o] Error 1

greping through the sources for the changes missed out, we have

./drivers/net/arm/ixp4xx_eth.c:507: netif_rx_reschedule(dev, napi)) {
./drivers/net/arm/ep93xx_eth.c:310:             if (more && netif_rx_reschedule(dev, napi))
./drivers/net/wan/ixp4xx_hss.c:657: netif_rx_reschedule(dev, napi)) {

Signed-off-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: Fix unused variable warnings in pasemi_mac.c and spider_net.c
David S. Miller [Mon, 29 Dec 2008 04:15:38 +0000]
net: Fix unused variable warnings in pasemi_mac.c and spider_net.c

As reported by Stephen Rothwell:

--------------------
Today's linux-next build (powerpc ppc64_defconfig) produced these new
warnings:

drivers/net/pasemi_mac.c: In function 'pasemi_mac_rx_intr':
drivers/net/pasemi_mac.c:957: warning: unused variable 'dev'
drivers/net/pasemi_mac.c: In function 'pasemi_mac_poll':
drivers/net/pasemi_mac.c:1637: warning: unused variable 'dev'
drivers/net/spider_net.c: In function 'spider_net_poll':
drivers/net/spider_net.c:1280: warning: unused variable 'netdev'

Probably caused by commit 908a7a16b852ffd618a9127be8d62432182d81b4 ("net:
Remove unused netdev arg from some NAPI interfaces").
--------------------

Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoMerge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Mon, 29 Dec 2008 00:54:33 +0000]
Merge branch 'next' of git://git./linux/kernel/git/paulus/powerpc

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (144 commits)
  powerpc/44x: Support 16K/64K base page sizes on 44x
  powerpc: Force memory size to be a multiple of PAGE_SIZE
  powerpc/32: Wire up the trampoline code for kdump
  powerpc/32: Add the ability for a classic ppc kernel to be loaded at 32M
  powerpc/32: Allow __ioremap on RAM addresses for kdump kernel
  powerpc/32: Setup OF properties for kdump
  powerpc/32/kdump: Implement crash_setup_regs() using ppc_save_regs()
  powerpc: Prepare xmon_save_regs for use with kdump
  powerpc: Remove default kexec/crash_kernel ops assignments
  powerpc: Make default kexec/crash_kernel ops implicit
  powerpc: Setup OF properties for ppc32 kexec
  powerpc/pseries: Fix cpu hotplug
  powerpc: Fix KVM build on ppc440
  powerpc/cell: add QPACE as a separate Cell platform
  powerpc/cell: fix build breakage with CONFIG_SPUFS disabled
  powerpc/mpc5200: fix error paths in PSC UART probe function
  powerpc/mpc5200: add rts/cts handling in PSC UART driver
  powerpc/mpc5200: Make PSC UART driver update serial errors counters
  powerpc/mpc5200: Remove obsolete code from mpc5200 MDIO driver
  powerpc/mpc5200: Add MDMA/UDMA support to MPC5200 ATA driver
  ...

Fix trivial conflict in drivers/char/Makefile as per Paul's directions

11 years agonet: ehea NAPI interface cleanup fix
Stephen Rothwell [Sun, 28 Dec 2008 23:46:13 +0000]
net: ehea NAPI interface cleanup fix

Commit 908a7a16b852ffd618a9127be8d62432182d81b4 ("net: Remove unused
netdev arg from some NAPI interfaces") missed two spots.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

11 years agocifs: update for new IP4/6 address printing
Stephen Rothwell [Wed, 3 Dec 2008 02:49:23 +0000]
cifs: update for new IP4/6 address printing

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Sun, 28 Dec 2008 23:15:08 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/security-testing-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/security-testing-2.6:
  smackfs: check for allocation failures in smk_set_access()

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-next
Linus Torvalds [Sun, 28 Dec 2008 23:13:48 +0000]
Merge git://git./linux/kernel/git/sam/kbuild-next

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-next: (25 commits)
  allow stripping of generated symbols under CONFIG_KALLSYMS_ALL
  kbuild: strip generated symbols from *.ko
  kbuild: simplify use of genksyms
  kernel-doc: check for extra kernel-doc notations
  kbuild: add headerdep used to detect inclusion cycles in header files
  kbuild: fix string equality testing in tags.sh
  kbuild: fix make tags/cscope
  kbuild: fix make incompatibility
  kbuild: remove TAR_IGNORE
  setlocalversion: add git-svn support
  setlocalversion: print correct subversion revision
  scripts: improve the decodecode script
  scripts/package: allow custom options to rpm
  genksyms: allow to ignore symbol checksum changes
  genksyms: track symbol checksum changes
  tags and cscope support really belongs in a shell script
  kconfig: fix options to check-lxdialog.sh
  kbuild: gen_init_cpio expands shell variables in file names
  remove bashisms from scripts/extract-ikconfig
  kbuild: teach mkmakfile to be silent
  ...

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-nvram
Linus Torvalds [Sun, 28 Dec 2008 23:12:35 +0000]
Merge git://git./linux/kernel/git/wim/linux-2.6-nvram

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-nvram:
  [PATCH] nvram - convert PRINT_PROC to seq_file
  [PATCH] nvram - CodingStyle

11 years agoMerge branch 'next' into for-linus
James Morris [Sun, 28 Dec 2008 22:57:38 +0000]
Merge branch 'next' into for-linus

11 years agopowerpc/44x: Support 16K/64K base page sizes on 44x
Ilya Yanok [Thu, 11 Dec 2008 01:55:41 +0000]
powerpc/44x: Support 16K/64K base page sizes on 44x

This adds support for 16k and 64k page sizes on PowerPC 44x processors.

The PGDIR table is much smaller than a page when using 16k or 64k
pages (512 and 32 bytes respectively) so we allocate the PGDIR with
kzalloc() instead of __get_free_pages().

One PTE table covers rather a large memory area when using 16k or 64k
pages (32MB or 512MB respectively), so we can easily put FIXMAP and
PKMAP in the area covered by one PTE table.

Signed-off-by: Yuri Tikhonov <yur@emcraft.com>
Signed-off-by: Vladimir Panfilov <pvr@emcraft.com>
Signed-off-by: Ilya Yanok <yanok@emcraft.com>
Acked-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

11 years agopowerpc: Force memory size to be a multiple of PAGE_SIZE
Hollis Blanchard [Wed, 26 Nov 2008 16:19:26 +0000]
powerpc: Force memory size to be a multiple of PAGE_SIZE

Ensure that total memory size is page-aligned, because otherwise
mark_bootmem() gets upset.

This error case was triggered by using 64 KiB pages in the kernel
while arch/powerpc/boot/4xx.c arbitrarily reduced the amount of memory
by 4096 (to work around a chip bug that affects the last 256 bytes of
physical memory).

Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux...
Linus Torvalds [Sun, 28 Dec 2008 20:54:07 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/geert/linux-m68k

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k: use the new byteorder headers
  fbcon: Protect free_irq() by MACH_IS_ATARI check
  fbcon: remove broken mac vbl handler
  m68k: fix trigraph ignored warning in setox.S
  macfb annotations and compiler warning fix
  m68k: mac baboon interrupt enable/disable
  m68k: machw.h cleanup
  m68k: Mac via cleanup and commentry
  m68k: Reinstate mac rtc

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next-2.6
Linus Torvalds [Sun, 28 Dec 2008 20:49:40 +0000]
Merge git://git./linux/kernel/git/davem/net-next-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next-2.6: (1429 commits)
  net: Allow dependancies of FDDI & Tokenring to be modular.
  igb: Fix build warning when DCA is disabled.
  net: Fix warning fallout from recent NAPI interface changes.
  gro: Fix potential use after free
  sfc: If AN is enabled, always read speed/duplex from the AN advertising bits
  sfc: When disabling the NIC, close the device rather than unregistering it
  sfc: SFT9001: Add cable diagnostics
  sfc: Add support for multiple PHY self-tests
  sfc: Merge top-level functions for self-tests
  sfc: Clean up PHY mode management in loopback self-test
  sfc: Fix unreliable link detection in some loopback modes
  sfc: Generate unique names for per-NIC workqueues
  802.3ad: use standard ethhdr instead of ad_header
  802.3ad: generalize out mac address initializer
  802.3ad: initialize ports LACPDU from const initializer
  802.3ad: remove typedef around ad_system
  802.3ad: turn ports is_individual into a bool
  802.3ad: turn ports is_enabled into a bool
  802.3ad: make ntt bool
  ixgbe: Fix set_ringparam in ixgbe to use the same memory pools.
  ...

Fixed trivial IPv4/6 address printing conflicts in fs/cifs/connect.c due
to the conversion to %pI (in this networking merge) and the addition of
doing IPv6 addresses (from the earlier merge of CIFS).

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Sun, 28 Dec 2008 20:37:14 +0000]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6: (31 commits)
  [CIFS] Remove redundant test
  [CIFS] make sure that DFS pathnames are properly formed
  Remove an already-checked error condition in SendReceiveBlockingLock
  Streamline SendReceiveBlockingLock: Use "goto out:" in an error condition
  Streamline SendReceiveBlockingLock: Use "goto out:" in an error condition
  [CIFS] Streamline SendReceive[2] by using "goto out:" in an error condition
  Slightly streamline SendReceive[2]
  Check the return value of cifs_sign_smb[2]
  [CIFS] Cleanup: Move the check for too large R/W requests
  [CIFS] Slightly simplify wait_for_free_request(), remove an unnecessary "else" branch
  Simplify allocate_mid() slightly: Remove some unnecessary "else" branches
  [CIFS] In SendReceive, move consistency check out of the mutexed region
  cifs: store password in tcon
  cifs: have calc_lanman_hash take more granular args
  cifs: zero out session password before freeing it
  cifs: fix wait_for_response to time out sleeping processes correctly
  [CIFS] Can not mount with prefixpath if root directory of share is inaccessible
  [CIFS] various minor cleanups pointed out by checkpatch script
  [CIFS] fix typo
  [CIFS] remove sparse warning
  ...

Fix trivial conflict in fs/cifs/cifs_fs_sb.h due to comment changes for
the CIFS_MOUNT_xyz bit definitions between cifs updates and security
updates.

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Sun, 28 Dec 2008 20:33:59 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (26 commits)
  IB/mlx4: Set ownership bit correctly when copying CQEs during CQ resize
  RDMA/nes: Remove tx_free_list
  RDMA/cma: Add IPv6 support
  RDMA/addr: Add support for translating IPv6 addresses
  mlx4_core: Delete incorrect comment
  mlx4_core: Add support for multiple completion event vectors
  IB/iser: Avoid recv buffer exhaustion caused by unexpected PDUs
  IB/ehca: Remove redundant test of vpage
  IB/ehca: Replace modulus operations in flush error completion path
  IB/ipath: Add locking for interrupt use of ipath_pd contexts vs free
  IB/ipath: Fix spi_pioindex value
  IB/ipath: Only do 1X workaround on rev1 chips
  IB/ipath: Don't count IB symbol and link errors unless link is UP
  IB/ipath: Check return value of dma_map_single()
  IB/ipath: Fix PSN of send WQEs after an RDMA read resend
  RDMA/nes: Cleanup warnings
  RDMA/nes: Add loopback check to make_cm_node()
  RDMA/nes: Check cqp_avail_reqs is empty after locking the list
  RDMA/nes: Fix TCP compliance test failures
  RDMA/nes: Forward packets for a new connection with stale APBVT entry
  ...

11 years agoMerge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Sun, 28 Dec 2008 20:33:21 +0000]
Merge branch 'for-linus' of git://git390.osdl.marist.edu/linux-2.6

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6: (85 commits)
  [S390] provide documentation for hvc_iucv kernel parameter.
  [S390] convert ctcm printks to dev_xxx and pr_xxx macros.
  [S390] convert zfcp printks to pr_xxx macros.
  [S390] convert vmlogrdr printks to pr_xxx macros.
  [S390] convert zfcp dumper printks to pr_xxx macros.
  [S390] convert cpu related printks to pr_xxx macros.
  [S390] convert qeth printks to dev_xxx and pr_xxx macros.
  [S390] convert sclp printks to pr_xxx macros.
  [S390] convert iucv printks to dev_xxx and pr_xxx macros.
  [S390] convert ap_bus printks to pr_xxx macros.
  [S390] convert dcssblk and extmem printks messages to pr_xxx macros.
  [S390] convert monwriter printks to pr_xxx macros.
  [S390] convert s390 debug feature printks to pr_xxx macros.
  [S390] convert monreader printks to pr_xxx macros.
  [S390] convert appldata printks to pr_xxx macros.
  [S390] convert setup printks to pr_xxx macros.
  [S390] convert hypfs printks to pr_xxx macros.
  [S390] convert time printks to pr_xxx macros.
  [S390] convert cpacf printks to pr_xxx macros.
  [S390] convert cio printks to pr_xxx macros.
  ...

11 years agoMerge branch 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 28 Dec 2008 20:27:58 +0000]
Merge branch 'sched-core-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (31 commits)
  sched: fix warning in fs/proc/base.c
  schedstat: consolidate per-task cpu runtime stats
  sched: use RCU variant of list traversal in for_each_leaf_rt_rq()
  sched, cpuacct: export percpu cpuacct cgroup stats
  sched, cpuacct: refactoring cpuusage_read / cpuusage_write
  sched: optimize update_curr()
  sched: fix wakeup preemption clock
  sched: add missing arch_update_cpu_topology() call
  sched: let arch_update_cpu_topology indicate if topology changed
  sched: idle_balance() does not call load_balance_newidle()
  sched: fix sd_parent_degenerate on non-numa smp machine
  sched: add uid information to sched_debug for CONFIG_USER_SCHED
  sched: move double_unlock_balance() higher
  sched: update comment for move_task_off_dead_cpu
  sched: fix inconsistency when redistribute per-cpu tg->cfs_rq shares
  sched/rt: removed unneeded defintion
  sched: add hierarchical accounting to cpu accounting controller
  sched: include group statistics in /proc/sched_debug
  sched: rename SCHED_NO_NO_OMIT_FRAME_POINTER => SCHED_OMIT_FRAME_POINTER
  sched: clean up SCHED_CPUMASK_ALLOC
  ...

11 years agoMerge branch 'tracing-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 28 Dec 2008 20:21:10 +0000]
Merge branch 'tracing-core-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (241 commits)
  sched, trace: update trace_sched_wakeup()
  tracing/ftrace: don't trace on early stage of a secondary cpu boot, v3
  Revert "x86: disable X86_PTRACE_BTS"
  ring-buffer: prevent false positive warning
  ring-buffer: fix dangling commit race
  ftrace: enable format arguments checking
  x86, bts: memory accounting
  x86, bts: add fork and exit handling
  ftrace: introduce tracing_reset_online_cpus() helper
  tracing: fix warnings in kernel/trace/trace_sched_switch.c
  tracing: fix warning in kernel/trace/trace.c
  tracing/ring-buffer: remove unused ring_buffer size
  trace: fix task state printout
  ftrace: add not to regex on filtering functions
  trace: better use of stack_trace_enabled for boot up code
  trace: add a way to enable or disable the stack tracer
  x86: entry_64 - introduce FTRACE_ frame macro v2
  tracing/ftrace: add the printk-msg-only option
  tracing/ftrace: use preempt_enable_no_resched_notrace in ring_buffer_time_stamp()
  x86, bts: correctly report invalid bts records
  ...

Fixed up trivial conflict in scripts/recordmcount.pl due to SH bits
being already partly merged by the SH merge.

11 years agoMerge branch 'x86-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 28 Dec 2008 20:07:57 +0000]
Merge branch 'x86-core-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (246 commits)
  x86: traps.c replace #if CONFIG_X86_32 with #ifdef CONFIG_X86_32
  x86: PAT: fix address types in track_pfn_vma_new()
  x86: prioritize the FPU traps for the error code
  x86: PAT: pfnmap documentation update changes
  x86: PAT: move track untrack pfnmap stubs to asm-generic
  x86: PAT: remove follow_pfnmap_pte in favor of follow_phys
  x86: PAT: modify follow_phys to return phys_addr prot and return value
  x86: PAT: clarify is_linear_pfn_mapping() interface
  x86: ia32_signal: remove unnecessary declaration
  x86: common.c boot_cpu_stack and boot_exception_stacks should be static
  x86: fix intel x86_64 llc_shared_map/cpu_llc_id anomolies
  x86: fix warning in arch/x86/kernel/microcode_amd.c
  x86: ia32.h: remove unused struct sigfram32 and rt_sigframe32
  x86: asm-offset_64: use rt_sigframe_ia32
  x86: sigframe.h: include headers for dependency
  x86: traps.c declare functions before they get used
  x86: PAT: update documentation to cover pgprot and remap_pfn related changes - v3
  x86: PAT: add pgprot_writecombine() interface for drivers - v3
  x86: PAT: change pgprot_noncached to uc_minus instead of strong uc - v3
  x86: PAT: implement track/untrack of pfnmap regions for x86 - v3
  ...

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Sun, 28 Dec 2008 19:43:54 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/security-testing-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/security-testing-2.6: (105 commits)
  SELinux: don't check permissions for kernel mounts
  security: pass mount flags to security_sb_kern_mount()
  SELinux: correctly detect proc filesystems of the form "proc/foo"
  Audit: Log TIOCSTI
  user namespaces: document CFS behavior
  user namespaces: require cap_set{ug}id for CLONE_NEWUSER
  user namespaces: let user_ns be cloned with fairsched
  CRED: fix sparse warnings
  User namespaces: use the current_user_ns() macro
  User namespaces: set of cleanups (v2)
  nfsctl: add headers for credentials
  coda: fix creds reference
  capabilities: define get_vfs_caps_from_disk when file caps are not enabled
  CRED: Allow kernel services to override LSM settings for task actions
  CRED: Add a kernel_service object class to SELinux
  CRED: Differentiate objective and effective subjective credentials on a task
  CRED: Documentation
  CRED: Use creds in file structs
  CRED: Prettify commoncap.c
  CRED: Make execve() take advantage of copy-on-write credentials
  ...

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Sun, 28 Dec 2008 19:43:22 +0000]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (57 commits)
  crypto: aes - Precompute tables
  crypto: talitos - Ack done interrupt in isr instead of tasklet
  crypto: testmgr - Correct comment about deflate parameters
  crypto: salsa20 - Remove private wrappers around various operations
  crypto: des3_ede - permit weak keys unless REQ_WEAK_KEY set
  crypto: sha512 - Switch to shash
  crypto: sha512 - Move message schedule W[80] to static percpu area
  crypto: michael_mic - Switch to shash
  crypto: wp512 - Switch to shash
  crypto: tgr192 - Switch to shash
  crypto: sha256 - Switch to shash
  crypto: md5 - Switch to shash
  crypto: md4 - Switch to shash
  crypto: sha1 - Switch to shash
  crypto: rmd320 - Switch to shash
  crypto: rmd256 - Switch to shash
  crypto: rmd160 - Switch to shash
  crypto: rmd128 - Switch to shash
  crypto: null - Switch to shash
  crypto: hash - Make setkey optional
  ...

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Sun, 28 Dec 2008 19:41:32 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6: (367 commits)
  ALSA: ASoC: fix a typo in omp-pcm.c
  ASoC: Fix DSP formats in SSM2602 audio codec
  ASoC: Fix incorrect DSP format in OMAP McBSP DAI and affected drivers
  ALSA: hda: fix incorrect mixer index values for 92hd83xx
  ALSA: hda: dinput_mux check
  ALSA: hda - Add quirk for another HP dv7
  ALSA: ASoC - Add missing __devexit annotation to wm8350.c
  ALSA: ASoc: DaVinci: davinci-evm use dsp_b mode
  ALSA: ASoC: DaVinci: i2s, evm, pass same value to codec and cpu_dai
  ALSA: ASoC: tlv320aic3x add dsp_a
  ALSA: ASoC: DaVinci: document I2S limitations
  ALSA: ASoC: DaVinci: davinci-i2s clean up
  ALSA: ASoC: DaVinci: davinci-i2s clean up
  ALSA: ASoC: DaVinci: davinci-i2s add comments to explain polarity
  ALSA: ASoC: DaVinci: davinvi-evm, make requests explicit
  ALSA: ca0106 - disable 44.1kHz capture
  ALSA: ca0106 - Add missing card->private_data initialization
  ALSA: ca0106 - Check ac97 availability at PM
  ALSA: hda - Power up always when no jack detection is available
  ALSA: hda - Fix unused variable warnings in patch_sigmatel.c
  ...

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6
Linus Torvalds [Sun, 28 Dec 2008 19:39:19 +0000]
Merge git://git./linux/kernel/git/lethal/sh-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (132 commits)
  sh: oprofile: Fix up the module build.
  sh: add UIO support for JPU on SH7722.
  serial: sh-sci: Fix up port pinmux for SH7366.
  sh: mach-rsk: Use uImage generation by default for rsk7201/7203.
  sh: mach-sh03: Fix up pata_platform build breakage.
  sh: enable deferred io LCDC on Migo-R
  video: sh_mobile_lcdcfb deferred io support
  video: deferred io with physically contiguous memory
  video: deferred io cleanup
  video: fix deferred io fsync()
  sh: add LCDC interrupt configuration to AP325 and Migo-R
  sh_mobile_lcdc: use FB_SYS helpers instead of FB_CFB
  sh: split coherent pages
  sh: dma: Kill off ISA DMA wrapper.
  sh: Conditionalize the code dumper on CONFIG_DUMP_CODE.
  sh: Kill off the unused SH_ALPHANUMERIC debug option.
  sh: Enable skipping of bss on debug platforms for sh32 also.
  doc: Update sh cpufreq documentation.
  sh: mrshpc_setup_windows() needs to be inline.
  serial: sh-sci: sci_poll_get_char() is only used by CONFIG_CONSOLE_POLL.
  ...

11 years agom68k: use the new byteorder headers
Harvey Harrison [Tue, 18 Nov 2008 19:45:23 +0000]
m68k: use the new byteorder headers

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agofbcon: Protect free_irq() by MACH_IS_ATARI check
Geert Uytterhoeven [Tue, 18 Nov 2008 19:45:23 +0000]
fbcon: Protect free_irq() by MACH_IS_ATARI check

Add missing check for Atari in free_irq() call, which could cause problems on
multi-platform m68k kernels.

Reported-by: Brad Boyer <flar@allandria.com>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agofbcon: remove broken mac vbl handler
Finn Thain [Tue, 18 Nov 2008 19:45:23 +0000]
fbcon: remove broken mac vbl handler

Remove the Mac VBL interrupt code as it doesn't work properly and
doesn't bring any benefit when fixed. Also remove unused
DEFAULT_CURSOR_BLINK_RATE macro and irqres variable.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agom68k: fix trigraph ignored warning in setox.S
Finn Thain [Tue, 18 Nov 2008 19:45:22 +0000]
m68k: fix trigraph ignored warning in setox.S

Fix the warning: trigraph ??/ ignored, use -trigraphs to enable
caused by the recent removal of -traditional option.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agomacfb annotations and compiler warning fix
Finn Thain [Tue, 18 Nov 2008 19:45:22 +0000]
macfb annotations and compiler warning fix

Add some __iomem annotations. Remove some volatile qualifiers to fix
several compiler warnings: "passing arg 1 of `iounmap' discards qualifiers
from pointer target type".

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agom68k: mac baboon interrupt enable/disable
Finn Thain [Tue, 18 Nov 2008 19:45:21 +0000]
m68k: mac baboon interrupt enable/disable

No-one seems to know how to mask individual baboon interrupts, so we just
mask the umbrella IRQ. This will work as long as only the IDE driver uses
the baboon chip (it can't deadlock). Use mac_enable_irq/mac_disable_irq
rather than enable_irq/disable_irq because the latter routines count the
depth of nested calls which triggers a warning and call trace because
IRQ_NUBUS_C is enabled twice in a row (once when the baboon handler is
registered, and once when the IDE IRQ is registered).

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agom68k: machw.h cleanup
Finn Thain [Tue, 18 Nov 2008 19:45:20 +0000]
m68k: machw.h cleanup

Remove some more cruft from machw.h and drop the #include where it isn't
needed.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agom68k: Mac via cleanup and commentry
Finn Thain [Tue, 18 Nov 2008 19:45:20 +0000]
m68k: Mac via cleanup and commentry

No behavioural changes, just cleanups and better documentation.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agom68k: Reinstate mac rtc
Finn Thain [Tue, 18 Nov 2008 19:45:20 +0000]
m68k: Reinstate mac rtc

Reinstate the Mac hardware clock for CUDA ADB and Mac II ADB models.
It doesn't work properly on Mac IIsi ADB and PMU ADB yet, so leave them
out.

Signed-off-by: Finn Thain <fthain@telegraphics.com.au>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

11 years agonet: Allow dependancies of FDDI & Tokenring to be modular.
Dave Jones [Sun, 28 Dec 2008 04:43:48 +0000]
net: Allow dependancies of FDDI & Tokenring to be modular.

I noticed it isn't possible to build token ring & fddi drivers
without causing LLC, and a bunch of other things to be forced
built-in.  For distro kernels, this means carrying a chunk of
code in the vmlinuz, even if the user doesn't use those protocols.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agoigb: Fix build warning when DCA is disabled.
David S. Miller [Fri, 26 Dec 2008 23:13:55 +0000]
igb: Fix build warning when DCA is disabled.

Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agonet: Fix warning fallout from recent NAPI interface changes.
David S. Miller [Fri, 26 Dec 2008 23:10:00 +0000]
net: Fix warning fallout from recent NAPI interface changes.

When we removed the network device argument from several
NAPI interfaces in 908a7a16b852ffd618a9127be8d62432182d81b4
("net: Remove unused netdev arg from some NAPI interfaces.")
several drivers now started getting unused variable warnings.

This fixes those up.

Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agogro: Fix potential use after free
Herbert Xu [Fri, 26 Dec 2008 22:57:42 +0000]
gro: Fix potential use after free

The initial skb may have been freed after napi_gro_complete in
napi_gro_receive if it was merged into an existing packet.  Thus
we cannot check same_flow (which indicates whether it was merged)
after calling napi_gro_complete.

This patch fixes this by saving the same_flow status before the
call to napi_gro_complete.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: If AN is enabled, always read speed/duplex from the AN advertising bits
Ben Hutchings [Fri, 26 Dec 2008 21:49:25 +0000]
sfc: If AN is enabled, always read speed/duplex from the AN advertising bits

When AN is enabled and the link is down the speed/duplex control bits
will not be meaningful.  Use the advertising bits instead, and mask
them with the LPA bits if and only if AN is complete (as before).

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: When disabling the NIC, close the device rather than unregistering it
Ben Hutchings [Fri, 26 Dec 2008 21:48:51 +0000]
sfc: When disabling the NIC, close the device rather than unregistering it

This should reduce user confusion and may also aid recovery (ioctls
will still be available).

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: SFT9001: Add cable diagnostics
Ben Hutchings [Fri, 26 Dec 2008 21:48:00 +0000]
sfc: SFT9001: Add cable diagnostics

The SFT9001 firmware implements cable diagnostics; run those and
include their results in a self-test.  In case of a cable fault, do
not fail the self-test as a whole; only faults in the NIC should cause
that.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: Add support for multiple PHY self-tests
Ben Hutchings [Fri, 26 Dec 2008 21:47:25 +0000]
sfc: Add support for multiple PHY self-tests

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: Merge top-level functions for self-tests
Ben Hutchings [Fri, 26 Dec 2008 21:47:04 +0000]
sfc: Merge top-level functions for self-tests

Pass in ethtool test flags to determine which tests to run.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: Clean up PHY mode management in loopback self-test
Ben Hutchings [Fri, 26 Dec 2008 21:46:38 +0000]
sfc: Clean up PHY mode management in loopback self-test

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: Fix unreliable link detection in some loopback modes
Ben Hutchings [Fri, 26 Dec 2008 21:46:12 +0000]
sfc: Fix unreliable link detection in some loopback modes

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years agosfc: Generate unique names for per-NIC workqueues
Ben Hutchings [Fri, 26 Dec 2008 21:44:39 +0000]
sfc: Generate unique names for per-NIC workqueues

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 years ago802.3ad: use standard ethhdr instead of ad_header
Holger Eitzenberger [Fri, 26 Dec 2008 21:41:53 +0000]
802.3ad: use standard ethhdr instead of ad_header

802.3ad has its own ethhdr-like structure in the form of an ad_header,
which is at the start of both the LACPDU and marker PDU.  Both are
the same from the struct values, both are packed as well.

It's therefore perfectly fine to replace the ad_header by the ethhdr
and to remove its definition.

Signed-off-by: Holger Eitzenberger <holger@eitzenberger.org>
Signed-off-by: David S. Miller <davem@davemloft.net>