GitHub/moto-9609/android_kernel_motorola_exynos9610.git
11 years agoveth: extend device features
Eric Dumazet [Sat, 29 Dec 2012 16:26:10 +0000 (16:26 +0000)]
veth: extend device features

veth is lacking most modern facilities, like SG, checksums, TSO.

It makes sense to extend dev->features to get them, or GRO aggregation
is defeated by a forced segmentation.

Reported-by: Andrew Vagin <avagin@parallels.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoveth: reduce stat overhead
Eric Dumazet [Sat, 29 Dec 2012 16:02:43 +0000 (16:02 +0000)]
veth: reduce stat overhead

veth stats are a bit bloated. There is no need to account transmit
and receive stats, since they are absolutely symmetric.

Also use a per device atomic64_t for the dropped counter, as it
should never be used in fast path.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoteam: implement carrier change
Flavio Leitner [Sat, 29 Dec 2012 15:31:01 +0000 (15:31 +0000)]
team: implement carrier change

The user space teamd daemon may need to control the
master's carrier state depending on the selected mode.

Signed-off-by: Flavio Leitner <fbl@redhat.com>
Acked-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobridge: respect RFC2863 operational state
stephen hemminger [Fri, 28 Dec 2012 18:15:22 +0000 (18:15 +0000)]
bridge: respect RFC2863 operational state

The bridge link detection should follow the operational state
of the lower device, rather than the carrier bit. This allows devices
like tunnels that are controlled by userspace control plane to work
with bridge STP link management.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Reviewed-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: filter: return -EINVAL if BPF_S_ANC* operation is not supported
Daniel Borkmann [Fri, 28 Dec 2012 10:50:17 +0000 (10:50 +0000)]
net: filter: return -EINVAL if BPF_S_ANC* operation is not supported

Currently, we return -EINVAL for malformed or wrong BPF filters.
However, this is not done for BPF_S_ANC* operations, which makes it
more difficult to detect if it's actually supported or not by the
BPF machine. Therefore, we should also return -EINVAL if K is within
the SKF_AD_OFF universe and the ancillary operation did not match.

Why exactly is it needed? If tools such as libpcap/tcpdump want to
make use of new ancillary operations (like filtering VLAN in kernel
space), there is currently no sane way to test if this feature /
BPF_S_ANC* op is present or not, since no error is returned. This
patch will make life easier for that and allow for a proper usage
for user space applications.

There was concern, if this patch will break userland. Short answer: Yes
and no. Long answer: It will "break" only for code that calls ...

  { BPF_LD | BPF_(W|H|B) | BPF_ABS, 0, 0, <K> },

... where <K> is in [0xfffff000, 0xffffffff] _and_ <K> is *not* an
ancillary. And here comes the BUT: assuming some *old* code will have
such an instruction where <K> is between [0xfffff000, 0xffffffff] and
it doesn't know ancillary operations, then this will give a
non-expected / unwanted behavior as well (since we do not return the
BPF machine with 0 after a failed load_pointer(), which was the case
before introducing ancillary operations, but load sth. into the
accumulator instead, and continue with the next instruction, for
instance). Thus, user space code would already have been broken by
introducing ancillary operations into the BPF machine per se. Code
that does such a direct load, e.g. "load word at packet offset
0xffffffff into accumulator" ("ld [0xffffffff]") is quite broken,
isn't it? The whole assumption of ancillary operations is that no-one
intentionally calls things like "ld [0xffffffff]" and expect this
word to be loaded from such a packet offset. Hence, we can also safely
make use of this feature testing patch and facilitate application
development. Therefore, at least from this patch onwards, we have
*for sure* a check whether current or in future implemented BPF_S_ANC*
ops are supported in the kernel. Patch was tested on x86_64.

(Thanks to Eric for the previous review.)

Cc: Eric Dumazet <eric.dumazet@gmail.com>
Reported-by: Ani Sinha <ani@aristanetworks.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoskbuff: make __kmalloc_reserve static
stephen hemminger [Fri, 28 Dec 2012 18:24:28 +0000 (18:24 +0000)]
skbuff: make __kmalloc_reserve static

Sparse detected case where this local function should be static.
It may even allow some compiler optimizations.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: make proc_tcp_fastopen_key static
stephen hemminger [Fri, 28 Dec 2012 18:20:24 +0000 (18:20 +0000)]
tcp: make proc_tcp_fastopen_key static

Detected by sparse.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosctp: make sctp_addr_wq_timeout_handler static
stephen hemminger [Fri, 28 Dec 2012 18:18:55 +0000 (18:18 +0000)]
sctp: make sctp_addr_wq_timeout_handler static

Fix sparse warning about local function that should be static.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: use per task frag allocator in skb_append_datato_frags
Eric Dumazet [Fri, 28 Dec 2012 06:06:37 +0000 (06:06 +0000)]
net: use per task frag allocator in skb_append_datato_frags

Use the new per task frag allocator in skb_append_datato_frags(),
to reduce number of frags and page allocator overhead.

Tested:
 ifconfig lo mtu 16436
 perf record netperf -t UDP_STREAM ; perf report

before :
 Throughput: 32928 Mbit/s
    51.79%  netperf  [kernel.kallsyms]  [k] copy_user_generic_string
     5.98%  netperf  [kernel.kallsyms]  [k] __alloc_pages_nodemask
     5.58%  netperf  [kernel.kallsyms]  [k] get_page_from_freelist
     5.01%  netperf  [kernel.kallsyms]  [k] __rmqueue
     3.74%  netperf  [kernel.kallsyms]  [k] skb_append_datato_frags
     1.87%  netperf  [kernel.kallsyms]  [k] prep_new_page
     1.42%  netperf  [kernel.kallsyms]  [k] next_zones_zonelist
     1.28%  netperf  [kernel.kallsyms]  [k] __inc_zone_state
     1.26%  netperf  [kernel.kallsyms]  [k] alloc_pages_current
     0.78%  netperf  [kernel.kallsyms]  [k] sock_alloc_send_pskb
     0.74%  netperf  [kernel.kallsyms]  [k] udp_sendmsg
     0.72%  netperf  [kernel.kallsyms]  [k] zone_watermark_ok
     0.68%  netperf  [kernel.kallsyms]  [k] __cpuset_node_allowed_softwall
     0.67%  netperf  [kernel.kallsyms]  [k] fib_table_lookup
     0.60%  netperf  [kernel.kallsyms]  [k] memcpy_fromiovecend
     0.55%  netperf  [kernel.kallsyms]  [k] __udp4_lib_lookup

 after:
  Throughput: 47185 Mbit/s
61.74% netperf  [kernel.kallsyms] [k] copy_user_generic_string
 2.07% netperf  [kernel.kallsyms] [k] prep_new_page
 1.98% netperf  [kernel.kallsyms] [k] skb_append_datato_frags
 1.02% netperf  [kernel.kallsyms] [k] sock_alloc_send_pskb
 0.97% netperf  [kernel.kallsyms] [k] enqueue_task_fair
 0.97% netperf  [kernel.kallsyms] [k] udp_sendmsg
 0.91% netperf  [kernel.kallsyms] [k] __ip_route_output_key
 0.88% netperf  [kernel.kallsyms] [k] __netif_receive_skb
 0.87% netperf  [kernel.kallsyms] [k] fib_table_lookup
 0.85% netperf  [kernel.kallsyms] [k] resched_task
 0.78% netperf  [kernel.kallsyms] [k] __udp4_lib_lookup
 0.77% netperf  [kernel.kallsyms] [k] _raw_spin_lock_irqsave

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agodummy: implement carrier change
Jiri Pirko [Thu, 27 Dec 2012 23:49:40 +0000 (23:49 +0000)]
dummy: implement carrier change

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agortnl: expose carrier value with possibility to set it
Jiri Pirko [Thu, 27 Dec 2012 23:49:39 +0000 (23:49 +0000)]
rtnl: expose carrier value with possibility to set it

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: allow to change carrier via sysfs
Jiri Pirko [Thu, 27 Dec 2012 23:49:38 +0000 (23:49 +0000)]
net: allow to change carrier via sysfs

Make carrier writable

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: add change_carrier netdev op
Jiri Pirko [Thu, 27 Dec 2012 23:49:37 +0000 (23:49 +0000)]
net: add change_carrier netdev op

This allows a driver to register change_carrier callback which will be
called whenever user will like to change carrier state. This is useful
for devices like dummy, gre, team and so on.

Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Flavio Leitner <fbl@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: use ARRAY_SIZE where possible
Sasha Levin [Thu, 20 Dec 2012 09:11:24 +0000 (09:11 +0000)]
bnx2x: use ARRAY_SIZE where possible

Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck...
Linus Torvalds [Thu, 27 Dec 2012 18:46:47 +0000 (10:46 -0800)]
Merge tag 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:

 - Report i2c errors to userspace in lm73 driver

 - Fix problem with DIV_ROUND_CLOSEST and unsigned divisors in emc6w201
   driver

* tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (emc6w201) Fix DIV_ROUND_CLOSEST problem with unsigned divisors
  hwmon: (lm73} Detect and report i2c bus errors

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm...
Linus Torvalds [Thu, 27 Dec 2012 18:42:46 +0000 (10:42 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ebiederm/user-namespace

Pull namespace fixes from Eric Biederman:
 "This tree includes two bug fixes for problems Oleg spotted on his
  review of the recent pid namespace work.  A small fix to not enable
  bottom halves with irqs disabled, and a trivial build fix for f2fs
  with user namespaces enabled."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace:
  f2fs: Don't assign e_id in f2fs_acl_from_disk
  proc: Allow proc_free_inum to be called from any context
  pidns: Stop pid allocation when init dies
  pidns: Outlaw thread creation after unshare(CLONE_NEWPID)

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Thu, 27 Dec 2012 18:40:30 +0000 (10:40 -0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

1) GRE tunnel drivers don't set the transport header properly, they also
   blindly deref the inner protocol ipv4 and needs some checks.  Fixes
   from Isaku Yamahata.

2) Fix sleeps while atomic in netdevice rename code, from Eric Dumazet.

3) Fix double-spinlock in solos-pci driver, from Dan Carpenter.

4) More ARP bug fixes.  Fix lockdep splat in arp_solicit() and then the
   bug accidentally added by that fix.  From Eric Dumazet and Cong Wang.

5) Remove some __dev* annotations that slipped back in, as well as all
   HOTPLUG references.  From Greg KH

6) RDS protocol uses wrong interfaces to access scatter-gather elements,
   causing a regression.  From Mike Marciniszyn.

7) Fix build error in cpts driver, from Richard Cochran.

8) Fix arithmetic in packet scheduler, from Stefan Hasko.

9) Similarly, fix association during calculation of random backoff in
   batman-adv.  From Akinobu Mita.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (21 commits)
  ipv6/ip6_gre: set transport header correctly
  ipv4/ip_gre: set transport header correctly to gre header
  IB/rds: suppress incompatible protocol when version is known
  IB/rds: Correct ib_api use with gs_dma_address/sg_dma_len
  net/vxlan: Use the underlying device index when joining/leaving multicast groups
  tcp: should drop incoming frames without ACK flag set
  netprio_cgroup: define sk_cgrp_prioidx only if NETPRIO_CGROUP is enabled
  cpts: fix a run time warn_on.
  cpts: fix build error by removing useless code.
  batman-adv: fix random jitter calculation
  arp: fix a regression in arp_solicit()
  net: sched: integer overflow fix
  CONFIG_HOTPLUG removal from networking core
  Drivers: network: more __dev* removal
  bridge: call br_netpoll_disable in br_add_if
  ipv4: arp: fix a lockdep splat in arp_solicit()
  tuntap: dont use a private kmem_cache
  net: devnet_rename_seq should be a seqcount
  ip_gre: fix possible use after free
  ip_gre: make ipgre_tunnel_xmit() not parse network header as IP unconditionally
  ...

12 years agoipv6/ip6_gre: set transport header correctly
Isaku Yamahata [Mon, 24 Dec 2012 16:51:04 +0000 (16:51 +0000)]
ipv6/ip6_gre: set transport header correctly

ip6gre_xmit2() incorrectly sets transport header to inner payload
instead of GRE header. It seems copy-and-pasted from ipip.c.
Set transport header to gre header.
(In ipip case the transport header is the inner ip header, so that's
correct.)

Found by inspection. In practice the incorrect transport header
doesn't matter because the skb usually is sent to another net_device
or socket, so the transport header isn't referenced.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv4/ip_gre: set transport header correctly to gre header
Isaku Yamahata [Mon, 24 Dec 2012 16:51:03 +0000 (16:51 +0000)]
ipv4/ip_gre: set transport header correctly to gre header

ipgre_tunnel_xmit() incorrectly sets transport header to inner payload
instead of GRE header. It seems copy-and-pasted from ipip.c.
So set transport header to gre header.
(In ipip case the transport header is the inner ip header, so that's
correct.)

Found by inspection. In practice the incorrect transport header
doesn't matter because the skb usually is sent to another net_device
or socket, so the transport header isn't referenced.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoIB/rds: suppress incompatible protocol when version is known
Marciniszyn, Mike [Fri, 21 Dec 2012 08:01:54 +0000 (08:01 +0000)]
IB/rds: suppress incompatible protocol when version is known

Add an else to only print the incompatible protocol message
when version hasn't been established.

Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoIB/rds: Correct ib_api use with gs_dma_address/sg_dma_len
Marciniszyn, Mike [Fri, 21 Dec 2012 08:01:49 +0000 (08:01 +0000)]
IB/rds: Correct ib_api use with gs_dma_address/sg_dma_len

0b088e00 ("RDS: Use page_remainder_alloc() for recv bufs")
added uses of sg_dma_len() and sg_dma_address(). This makes
RDS DOA with the qib driver.

IB ulps should use ib_sg_dma_len() and ib_sg_dma_address
respectively since some HCAs overload ib_sg_dma* operations.

Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/vxlan: Use the underlying device index when joining/leaving multicast groups
Yan Burman [Thu, 20 Dec 2012 03:36:08 +0000 (03:36 +0000)]
net/vxlan: Use the underlying device index when joining/leaving multicast groups

The socket calls from vxlan to join/leave multicast group aren't
using the index of the underlying device, as a result the stack uses
the first interface that is up. This results in vxlan being non functional
over a device which isn't the 1st to be up.
Fix this by providing the iflink field to the vxlan instance
to the multicast calls.

Signed-off-by: Yan Burman <yanb@mellanox.com>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: should drop incoming frames without ACK flag set
Eric Dumazet [Wed, 26 Dec 2012 12:44:34 +0000 (12:44 +0000)]
tcp: should drop incoming frames without ACK flag set

In commit 96e0bf4b5193d (tcp: Discard segments that ack data not yet
sent) John Dykstra enforced a check against ack sequences.

In commit 354e4aa391ed5 (tcp: RFC 5961 5.2 Blind Data Injection Attack
Mitigation) I added more safety tests.

But we missed fact that these tests are not performed if ACK bit is
not set.

RFC 793 3.9 mandates TCP should drop a frame without ACK flag set.

" fifth check the ACK field,
      if the ACK bit is off drop the segment and return"

Not doing so permits an attacker to only guess an acceptable sequence
number, evading stronger checks.

Many thanks to Zhiyun Qian for bringing this issue to our attention.

See :
http://web.eecs.umich.edu/~zhiyunq/pub/ccs12_TCP_sequence_number_inference.pdf

Reported-by: Zhiyun Qian <zhiyunq@umich.edu>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Nandita Dukkipati <nanditad@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: John Dykstra <john.dykstra1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomm: Fix PageHead when !CONFIG_PAGEFLAGS_EXTENDED
Christoffer Dall [Fri, 21 Dec 2012 18:03:50 +0000 (13:03 -0500)]
mm: Fix PageHead when !CONFIG_PAGEFLAGS_EXTENDED

Unfortunately with !CONFIG_PAGEFLAGS_EXTENDED, (!PageHead) is false, and
(PageHead) is true, for tail pages.  If this is indeed the intended
behavior, which I doubt because it breaks cache cleaning on some ARM
systems, then the nomenclature is highly problematic.

This patch makes sure PageHead is only true for head pages and PageTail
is only true for tail pages, and neither is true for non-compound pages.

[ This buglet seems ancient - seems to have been introduced back in Apr
  2008 in commit 6a1e7f777f61: "pageflags: convert to the use of new
  macros".  And the reason nobody noticed is because the PageHead()
  tests are almost all about just sanity-checking, and only used on
  pages that are actual page heads.  The fact that the old code returned
  true for tail pages too was thus not really noticeable.   - Linus ]

Signed-off-by: Christoffer Dall <cdall@cs.columbia.edu>
Acked-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Will Deacon <Will.Deacon@arm.com>
Cc: Steve Capper <Steve.Capper@arm.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: stable@kernel.org # 2.6.26+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agonetprio_cgroup: define sk_cgrp_prioidx only if NETPRIO_CGROUP is enabled
Li Zefan [Tue, 25 Dec 2012 20:48:24 +0000 (20:48 +0000)]
netprio_cgroup: define sk_cgrp_prioidx only if NETPRIO_CGROUP is enabled

sock->sk_cgrp_prioidx won't be used at all if CONFIG_NETPRIO_CGROUP=n.

Signed-off-by: Li Zefan <lizefan@huawei.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agocpts: fix a run time warn_on.
Richard Cochran [Sun, 23 Dec 2012 21:19:10 +0000 (21:19 +0000)]
cpts: fix a run time warn_on.

This patch fixes a warning in clk_enable by calling clk_prepare_enable
instead.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agocpts: fix build error by removing useless code.
Richard Cochran [Sun, 23 Dec 2012 21:19:09 +0000 (21:19 +0000)]
cpts: fix build error by removing useless code.

The cpts driver tries to obtain the input clock frequency by calling the
clock's internal 'recalc' method. Since <plat/clock.h> has been removed,
this code can no longer compile.

However, the driver never makes use of the frequency value, so this patch
fixes the issue by removing the offending code altogether.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobatman-adv: fix random jitter calculation
Akinobu Mita [Wed, 26 Dec 2012 02:32:10 +0000 (02:32 +0000)]
batman-adv: fix random jitter calculation

batadv_iv_ogm_emit_send_time() attempts to calculates a random integer
in the range of 'orig_interval +- BATADV_JITTER' by the below lines.

        msecs = atomic_read(&bat_priv->orig_interval) - BATADV_JITTER;
        msecs += (random32() % 2 * BATADV_JITTER);

But it actually gets 'orig_interval' or 'orig_interval - BATADV_JITTER'
because '%' and '*' have same precedence and associativity is
left-to-right.

This adds the parentheses at the appropriate position so that it matches
original intension.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Cc: Marek Lindner <lindner_marek@yahoo.de>
Cc: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Cc: Antonio Quartulli <ordex@autistici.org>
Cc: b.a.t.m.a.n@lists.open-mesh.org
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agof2fs: Don't assign e_id in f2fs_acl_from_disk
Eric W. Biederman [Sat, 22 Dec 2012 09:52:39 +0000 (01:52 -0800)]
f2fs: Don't assign e_id in f2fs_acl_from_disk

With user namespaces enabled building f2fs fails with:

 CC      fs/f2fs/acl.o
fs/f2fs/acl.c: In function ‘f2fs_acl_from_disk’:
fs/f2fs/acl.c:85:21: error: ‘struct posix_acl_entry’ has no member named ‘e_id’
make[2]: *** [fs/f2fs/acl.o] Error 1
make[2]: Target `__build' not remade because of errors.

e_id is a backwards compatibility field only used for file systems
that haven't been converted to use kuids and kgids.  When the posix
acl tag field is neither ACL_USER nor ACL_GROUP assigning e_id is
unnecessary.  Remove the assignment so f2fs will build with user
namespaces enabled.

Cc: Namjae Jeon <namjae.jeon@samsung.com>
Cc: Amit Sahrawat <a.sahrawat@samsung.com>
Acked-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
12 years agoproc: Allow proc_free_inum to be called from any context
Eric W. Biederman [Sat, 22 Dec 2012 04:38:00 +0000 (20:38 -0800)]
proc: Allow proc_free_inum to be called from any context

While testing the pid namespace code I hit this nasty warning.

[  176.262617] ------------[ cut here ]------------
[  176.263388] WARNING: at /home/eric/projects/linux/linux-userns-devel/kernel/softirq.c:160 local_bh_enable_ip+0x7a/0xa0()
[  176.265145] Hardware name: Bochs
[  176.265677] Modules linked in:
[  176.266341] Pid: 742, comm: bash Not tainted 3.7.0userns+ #18
[  176.266564] Call Trace:
[  176.266564]  [<ffffffff810a539f>] warn_slowpath_common+0x7f/0xc0
[  176.266564]  [<ffffffff810a53fa>] warn_slowpath_null+0x1a/0x20
[  176.266564]  [<ffffffff810ad9ea>] local_bh_enable_ip+0x7a/0xa0
[  176.266564]  [<ffffffff819308c9>] _raw_spin_unlock_bh+0x19/0x20
[  176.266564]  [<ffffffff8123dbda>] proc_free_inum+0x3a/0x50
[  176.266564]  [<ffffffff8111d0dc>] free_pid_ns+0x1c/0x80
[  176.266564]  [<ffffffff8111d195>] put_pid_ns+0x35/0x50
[  176.266564]  [<ffffffff810c608a>] put_pid+0x4a/0x60
[  176.266564]  [<ffffffff8146b177>] tty_ioctl+0x717/0xc10
[  176.266564]  [<ffffffff810aa4d5>] ? wait_consider_task+0x855/0xb90
[  176.266564]  [<ffffffff81086bf9>] ? default_spin_lock_flags+0x9/0x10
[  176.266564]  [<ffffffff810cab0a>] ? remove_wait_queue+0x5a/0x70
[  176.266564]  [<ffffffff811e37e8>] do_vfs_ioctl+0x98/0x550
[  176.266564]  [<ffffffff810b8a0f>] ? recalc_sigpending+0x1f/0x60
[  176.266564]  [<ffffffff810b9127>] ? __set_task_blocked+0x37/0x80
[  176.266564]  [<ffffffff810ab95b>] ? sys_wait4+0xab/0xf0
[  176.266564]  [<ffffffff811e3d31>] sys_ioctl+0x91/0xb0
[  176.266564]  [<ffffffff810a95f0>] ? task_stopped_code+0x50/0x50
[  176.266564]  [<ffffffff81939199>] system_call_fastpath+0x16/0x1b
[  176.266564] ---[ end trace 387af88219ad6143 ]---

It turns out that spin_unlock_bh(proc_inum_lock) is not safe when
put_pid is called with another spinlock held and irqs disabled.

For now take the easy path and use spin_lock_irqsave(proc_inum_lock)
in proc_free_inum and spin_loc_irq in proc_alloc_inum(proc_inum_lock).

Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
12 years agopidns: Stop pid allocation when init dies
Eric W. Biederman [Sat, 22 Dec 2012 04:27:12 +0000 (20:27 -0800)]
pidns: Stop pid allocation when init dies

Oleg pointed out that in a pid namespace the sequence.
- pid 1 becomes a zombie
- setns(thepidns), fork,...
- reaping pid 1.
- The injected processes exiting.

Can lead to processes attempting access their child reaper and
instead following a stale pointer.

That waitpid for init can return before all of the processes in
the pid namespace have exited is also unfortunate.

Avoid these problems by disabling the allocation of new pids in a pid
namespace when init dies, instead of when the last process in a pid
namespace is reaped.

Pointed-out-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
12 years agopidns: Outlaw thread creation after unshare(CLONE_NEWPID)
Eric W. Biederman [Fri, 21 Dec 2012 03:26:06 +0000 (19:26 -0800)]
pidns: Outlaw thread creation after unshare(CLONE_NEWPID)

The sequence:
unshare(CLONE_NEWPID)
clone(CLONE_THREAD|CLONE_SIGHAND|CLONE_VM)

Creates a new process in the new pid namespace without setting
pid_ns->child_reaper.  After forking this results in a NULL
pointer dereference.

Avoid this and other nonsense scenarios that can show up after
creating a new pid namespace with unshare by adding a new
check in copy_prodcess.

Pointed-out-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
12 years agoarp: fix a regression in arp_solicit()
Cong Wang [Sun, 23 Dec 2012 15:23:16 +0000 (15:23 +0000)]
arp: fix a regression in arp_solicit()

Sedat reported the following commit caused a regression:

commit 9650388b5c56578fdccc79c57a8c82fb92b8e7f1
Author: Eric Dumazet <edumazet@google.com>
Date:   Fri Dec 21 07:32:10 2012 +0000

    ipv4: arp: fix a lockdep splat in arp_solicit

This is due to the 6th parameter of arp_send() needs to be NULL
for the broadcast case, the above commit changed it to an all-zero
array by mistake.

Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Cc: Sedat Dilek <sedat.dilek@gmail.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux
Linus Torvalds [Sun, 23 Dec 2012 17:48:33 +0000 (09:48 -0800)]
Merge branch 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux

Pull i2c __dev* attribute removal from Wolfram Sang:
 "The squashed patches from Bill to get rid of the __dev* annotations in
  the i2c subsystem.  I couldn't include it in my previous pull request
  due to some dependency with the mfd subsystem.  I had this patch in
  linux-next for two days before rc1 and nothing popped up."

* 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux:
  i2c: remove __dev* attributes from subsystem

12 years agomm: modify pgdat_balanced() so that it also handles order-0
Zlatko Calusic [Sun, 23 Dec 2012 14:12:54 +0000 (15:12 +0100)]
mm: modify pgdat_balanced() so that it also handles order-0

Teach pgdat_balanced() about order-0 allocations so that we can simplify
code in a few places in vmstat.c.

Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Zlatko Calusic <zlatko.calusic@iskon.hr>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoPartly revert "[media] uvcvideo: Set error_idx properly for extended controls API...
Rafael J. Wysocki [Sun, 23 Dec 2012 13:39:32 +0000 (14:39 +0100)]
Partly revert "[media] uvcvideo: Set error_idx properly for extended controls API failures"

Commit f0ed2ce840b3 ("[media] uvcvideo: Set error_idx properly for
extended controls API failures") causes user space to behave incorrectly
on one of my test machines (there is no sound under KDE 4.9.4 using
pulseaudio and there is a knotify4 process occupying one of the CPU
cores 100% of the time).  Reverting that commit entirely fixes the
problem for me.

However, commit f0ed2ce840b3 appears to do more than it follows from its
changelog, because the changelog only says about the changes related to
ctrls->error_idx, while the commit additionally changes error codes
returned by various functions in uvc_ctrl.c and uvc_v4l2.c.  It turns
out that the changes of the returned error codes confuse the user spce,
so it is sufficient to revert the part of commit f0ed2ce840b3 not
mentioned in its changelog to fix the problem.

[ 'ENOENT' is not a valid error return from an ioctl to begin with, and
  I don't understand how anybody ever even thought it would be.  - Linus ]

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>
Cc: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoi2c: remove __dev* attributes from subsystem
Bill Pemberton [Tue, 27 Nov 2012 20:59:38 +0000 (15:59 -0500)]
i2c: remove __dev* attributes from subsystem

CONFIG_HOTPLUG is going away as an option.  As result the __dev*
markings will be going away.

Remove use of __devinit, __devexit_p, __devinitdata, __devinitconst,
and __devexit.

Signed-off-by: Bill Pemberton <wfp5p@virginia.edu>
Acked-by: Peter Korsgaard <peter.korsgaard@barco.com> (for ocores and mux-gpio)
Acked-by: Havard Skinnemoen <hskinnemoen@gmail.com> (for i2c-gpio)
Acked-by: Guan Xuetao <gxt@mprc.pku.edu.cn> (for puf3)
Acked-by: Barry Song <baohua.song@csr.com> (for sirf)
Reviewed-by: Jean Delvare <khali@linux-fr.org>
[wsa: Fixed "foo* bar" flaws while we are here]
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
12 years agohwmon: (emc6w201) Fix DIV_ROUND_CLOSEST problem with unsigned divisors
Guenter Roeck [Wed, 19 Dec 2012 02:16:08 +0000 (18:16 -0800)]
hwmon: (emc6w201) Fix DIV_ROUND_CLOSEST problem with unsigned divisors

Result of DIV_ROUND_CLOSEST is undefined for negative dividends if the divisor
variable type is unsigned. Fix by declaring divisor as signed variable.

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Acked-by: Jean Delvare <khali@linux-fr.org>
12 years agonet: sched: integer overflow fix
Stefan Hasko [Fri, 21 Dec 2012 15:04:59 +0000 (15:04 +0000)]
net: sched: integer overflow fix

Fixed integer overflow in function htb_dequeue

Signed-off-by: Stefan Hasko <hasko.stevo@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoCONFIG_HOTPLUG removal from networking core
Greg KH [Fri, 21 Dec 2012 13:44:29 +0000 (13:44 +0000)]
CONFIG_HOTPLUG removal from networking core

CONFIG_HOTPLUG is always enabled now, so remove the unused code that was
trying to be compiled out when this option was disabled, in the
networking core.

Cc: Bill Pemberton <wfp5p@virginia.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoDrivers: network: more __dev* removal
Greg KH [Fri, 21 Dec 2012 13:42:15 +0000 (13:42 +0000)]
Drivers: network: more __dev* removal

Remove some __dev* markings that snuck in the 3.8-rc1 merge window in
the drivers/net/* directory.

Cc: Bill Pemberton <wfp5p@virginia.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agohwmon: (lm73} Detect and report i2c bus errors
Chris Verges [Fri, 21 Dec 2012 09:58:34 +0000 (01:58 -0800)]
hwmon: (lm73} Detect and report i2c bus errors

If an LM73 device does not exist on an I2C bus, attempts to communicate
with the device result in an error code returned from the i2c read/write
functions.  The current lm73 driver casts that return value from a s32
type to a s16 type, then converts it to a temperature in celsius.
Because negative temperatures are valid, it is difficult to distinguish
between an error code printed to the response buffer and a negative
temperature recorded by the sensor.

The solution is to evaluate the return value from the i2c functions
before performing any temperature calculations.  If the i2c function did
not succeed, the error code should be passed back through the virtual
file system layer instead of being printed into the response buffer.

Before:

   $ cat /sys/class/hwmon/hwmon0/device/temp1_input
   -46

After:

   $ cat /sys/class/hwmon/hwmon0/device/temp1_input
   cat: read error: No such device or address

Signed-off-by: Chris Verges <kg4ysn@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
12 years agoLinux 3.8-rc1
Linus Torvalds [Sat, 22 Dec 2012 01:19:00 +0000 (17:19 -0800)]
Linux 3.8-rc1

12 years agoMerge git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Sat, 22 Dec 2012 01:10:29 +0000 (17:10 -0800)]
Merge git://www.linux-watchdog.org/linux-watchdog

Pull watchdog updates from Wim Van Sebroeck:
 "This includes some fixes and code improvements (like
  clk_prepare_enable and clk_disable_unprepare), conversion from the
  omap_wdt and twl4030_wdt drivers to the watchdog framework, addition
  of the SB8x0 chipset support and the DA9055 Watchdog driver and some
  OF support for the davinci_wdt driver."

* git://www.linux-watchdog.org/linux-watchdog: (22 commits)
  watchdog: mei: avoid oops in watchdog unregister code path
  watchdog: Orion: Fix possible null-deference in orion_wdt_probe
  watchdog: sp5100_tco: Add SB8x0 chipset support
  watchdog: davinci_wdt: add OF support
  watchdog: da9052: Fix invalid free of devm_ allocated data
  watchdog: twl4030_wdt: Change TWL4030_MODULE_PM_RECEIVER to TWL_MODULE_PM_RECEIVER
  watchdog: remove depends on CONFIG_EXPERIMENTAL
  watchdog: Convert dev_printk(KERN_<LEVEL> to dev_<level>(
  watchdog: DA9055 Watchdog driver
  watchdog: omap_wdt: eliminate goto
  watchdog: omap_wdt: delete redundant platform_set_drvdata() calls
  watchdog: omap_wdt: convert to devm_ functions
  watchdog: omap_wdt: convert to new watchdog core
  watchdog: WatchDog Timer Driver Core: fix comment
  watchdog: s3c2410_wdt: use clk_prepare_enable and clk_disable_unprepare
  watchdog: imx2_wdt: Select the driver via ARCH_MXC
  watchdog: cpu5wdt.c: add missing del_timer call
  watchdog: hpwdt.c: Increase version string
  watchdog: Convert twl4030_wdt to watchdog core
  davinci_wdt: preparation for switch to common clock framework
  ...

12 years agoMerge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sat, 22 Dec 2012 01:09:07 +0000 (17:09 -0800)]
Merge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6

Pull CIFS fixes from Steve French:
 "Misc small cifs fixes"

* 'for-next' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: eliminate cifsERROR variable
  cifs: don't compare uniqueids in cifs_prime_dcache unless server inode numbers are in use
  cifs: fix double-free of "string" in cifs_parse_mount_options

12 years agoMerge tag 'dm-3.8-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm
Linus Torvalds [Sat, 22 Dec 2012 01:08:06 +0000 (17:08 -0800)]
Merge tag 'dm-3.8-fixes' of git://git./linux/kernel/git/agk/linux-dm

Pull dm update from Alasdair G Kergon:
 "Miscellaneous device-mapper fixes, cleanups and performance
  improvements.

  Of particular note:
   - Disable broken WRITE SAME support in all targets except linear and
     striped.  Use it when kcopyd is zeroing blocks.
   - Remove several mempools from targets by moving the data into the
     bio's new front_pad area(which dm calls 'per_bio_data').
   - Fix a race in thin provisioning if discards are misused.
   - Prevent userspace from interfering with the ioctl parameters and
     use kmalloc for the data buffer if it's small instead of vmalloc.
   - Throttle some annoying error messages when I/O fails."

* tag 'dm-3.8-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm: (36 commits)
  dm stripe: add WRITE SAME support
  dm: remove map_info
  dm snapshot: do not use map_context
  dm thin: dont use map_context
  dm raid1: dont use map_context
  dm flakey: dont use map_context
  dm raid1: rename read_record to bio_record
  dm: move target request nr to dm_target_io
  dm snapshot: use per_bio_data
  dm verity: use per_bio_data
  dm raid1: use per_bio_data
  dm: introduce per_bio_data
  dm kcopyd: add WRITE SAME support to dm_kcopyd_zero
  dm linear: add WRITE SAME support
  dm: add WRITE SAME support
  dm: prepare to support WRITE SAME
  dm ioctl: use kmalloc if possible
  dm ioctl: remove PF_MEMALLOC
  dm persistent data: improve improve space map block alloc failure message
  dm thin: use DMERR_LIMIT for errors
  ...

12 years agoRevert "nfsd: warn on odd reply state in nfsd_vfs_read"
J. Bruce Fields [Sat, 22 Dec 2012 00:48:59 +0000 (19:48 -0500)]
Revert "nfsd: warn on odd reply state in nfsd_vfs_read"

This reverts commit 79f77bf9a4e3dd5ead006b8f17e7c4ff07d8374e.

This is obviously wrong, and I have no idea how I missed seeing the
warning in testing: I must just not have looked at the right logs.  The
caller bumps rq_resused/rq_next_page, so it will always be hit on a
large enough read.

Reported-by: Dave Jones <davej@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Sat, 22 Dec 2012 00:40:26 +0000 (16:40 -0800)]
Merge tag 'rdma-for-linus' of git://git./linux/kernel/git/roland/infiniband

Pull more infiniband changes from Roland Dreier:
 "Second batch of InfiniBand/RDMA changes for 3.8:
   - cxgb4 changes to fix lookup engine hash collisions
   - mlx4 changes to make flow steering usable
   - fix to IPoIB to avoid pinning dst reference for too long"

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  RDMA/cxgb4: Fix bug for active and passive LE hash collision path
  RDMA/cxgb4: Fix LE hash collision bug for passive open connection
  RDMA/cxgb4: Fix LE hash collision bug for active open connection
  mlx4_core: Allow choosing flow steering mode
  mlx4_core: Adjustments to Flow Steering activation logic for SR-IOV
  mlx4_core: Fix error flow in the flow steering wrapper
  mlx4_core: Add QPN enforcement for flow steering rules set by VFs
  cxgb4: Add LE hash collision bug fix path in LLD driver
  cxgb4: Add T4 filter support
  IPoIB: Call skb_dst_drop() once skb is enqueued for sending

12 years agoMerge tag 'asm-generic' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm...
Linus Torvalds [Sat, 22 Dec 2012 00:39:08 +0000 (16:39 -0800)]
Merge tag 'asm-generic' of git://git./linux/kernel/git/arnd/asm-generic

Pull asm-generic cleanup from Arnd Bergmann:
 "These are a few cleanups for asm-generic:

   - a set of patches from Lars-Peter Clausen to generalize asm/mmu.h
     and use it in the architectures that don't need any special
     handling.
   - A patch from Will Deacon to remove the {read,write}s{b,w,l} as
     discussed during the arm64 review
   - A patch from James Hogan that helps with the meta architecture
     series."

* tag 'asm-generic' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  xtensa: Use generic asm/mmu.h for nommu
  h8300: Use generic asm/mmu.h
  c6x: Use generic asm/mmu.h
  asm-generic/mmu.h: Add support for FDPIC
  asm-generic/mmu.h: Remove unused vmlist field from mm_context_t
  asm-generic: io: remove {read,write} string functions
  asm-generic/io.h: remove asm/cacheflush.h include

12 years agoARM: dts: fix duplicated build target and alphabetical sort out for exynos
Kukjin Kim [Fri, 21 Dec 2012 18:02:13 +0000 (10:02 -0800)]
ARM: dts: fix duplicated build target and alphabetical sort out for exynos

Commit db5b0ae00712 ("Merge tag 'dt' of git://git.kernel.org/.../arm-soc")
causes a duplicated build target.  This patch fixes it and sorts out the
build target alphabetically so that we can recognize something wrong
easily.

Cc: Olof Johansson <olof@lixom.net>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agobridge: call br_netpoll_disable in br_add_if
Gao feng [Wed, 19 Dec 2012 23:41:43 +0000 (23:41 +0000)]
bridge: call br_netpoll_disable in br_add_if

When netdev_set_master faild in br_add_if, we should
call br_netpoll_disable to do some cleanup jobs,such
as free the memory of struct netpoll which allocated
in br_netpoll_enable.

Signed-off-by: Gao feng <gaofeng@cn.fujitsu.com>
Acked-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv4: arp: fix a lockdep splat in arp_solicit()
Eric Dumazet [Fri, 21 Dec 2012 07:32:10 +0000 (07:32 +0000)]
ipv4: arp: fix a lockdep splat in arp_solicit()

Yan Burman reported following lockdep warning :

=============================================
[ INFO: possible recursive locking detected ]
3.7.0+ #24 Not tainted
---------------------------------------------
swapper/1/0 is trying to acquire lock:
  (&n->lock){++--..}, at: [<ffffffff8139f56e>] __neigh_event_send
+0x2e/0x2f0

but task is already holding lock:
  (&n->lock){++--..}, at: [<ffffffff813f63f4>] arp_solicit+0x1d4/0x280

other info that might help us debug this:
  Possible unsafe locking scenario:

        CPU0
        ----
   lock(&n->lock);
   lock(&n->lock);

  *** DEADLOCK ***

  May be due to missing lock nesting notation

4 locks held by swapper/1/0:
  #0:  (((&n->timer))){+.-...}, at: [<ffffffff8104b350>]
call_timer_fn+0x0/0x1c0
  #1:  (&n->lock){++--..}, at: [<ffffffff813f63f4>] arp_solicit
+0x1d4/0x280
  #2:  (rcu_read_lock_bh){.+....}, at: [<ffffffff81395400>]
dev_queue_xmit+0x0/0x5d0
  #3:  (rcu_read_lock_bh){.+....}, at: [<ffffffff813cb41e>]
ip_finish_output+0x13e/0x640

stack backtrace:
Pid: 0, comm: swapper/1 Not tainted 3.7.0+ #24
Call Trace:
  <IRQ>  [<ffffffff8108c7ac>] validate_chain+0xdcc/0x11f0
  [<ffffffff8108d570>] ? __lock_acquire+0x440/0xc30
  [<ffffffff81120565>] ? kmem_cache_free+0xe5/0x1c0
  [<ffffffff8108d570>] __lock_acquire+0x440/0xc30
  [<ffffffff813c3570>] ? inet_getpeer+0x40/0x600
  [<ffffffff8108d570>] ? __lock_acquire+0x440/0xc30
  [<ffffffff8139f56e>] ? __neigh_event_send+0x2e/0x2f0
  [<ffffffff8108ddf5>] lock_acquire+0x95/0x140
  [<ffffffff8139f56e>] ? __neigh_event_send+0x2e/0x2f0
  [<ffffffff8108d570>] ? __lock_acquire+0x440/0xc30
  [<ffffffff81448d4b>] _raw_write_lock_bh+0x3b/0x50
  [<ffffffff8139f56e>] ? __neigh_event_send+0x2e/0x2f0
  [<ffffffff8139f56e>] __neigh_event_send+0x2e/0x2f0
  [<ffffffff8139f99b>] neigh_resolve_output+0x16b/0x270
  [<ffffffff813cb62d>] ip_finish_output+0x34d/0x640
  [<ffffffff813cb41e>] ? ip_finish_output+0x13e/0x640
  [<ffffffffa046f146>] ? vxlan_xmit+0x556/0xbec [vxlan]
  [<ffffffff813cb9a0>] ip_output+0x80/0xf0
  [<ffffffff813ca368>] ip_local_out+0x28/0x80
  [<ffffffffa046f25a>] vxlan_xmit+0x66a/0xbec [vxlan]
  [<ffffffffa046f146>] ? vxlan_xmit+0x556/0xbec [vxlan]
  [<ffffffff81394a50>] ? skb_gso_segment+0x2b0/0x2b0
  [<ffffffff81449355>] ? _raw_spin_unlock_irqrestore+0x65/0x80
  [<ffffffff81394c57>] ? dev_queue_xmit_nit+0x207/0x270
  [<ffffffff813950c8>] dev_hard_start_xmit+0x298/0x5d0
  [<ffffffff813956f3>] dev_queue_xmit+0x2f3/0x5d0
  [<ffffffff81395400>] ? dev_hard_start_xmit+0x5d0/0x5d0
  [<ffffffff813f5788>] arp_xmit+0x58/0x60
  [<ffffffff813f59db>] arp_send+0x3b/0x40
  [<ffffffff813f6424>] arp_solicit+0x204/0x280
  [<ffffffff813a1a70>] ? neigh_add+0x310/0x310
  [<ffffffff8139f515>] neigh_probe+0x45/0x70
  [<ffffffff813a1c10>] neigh_timer_handler+0x1a0/0x2a0
  [<ffffffff8104b3cf>] call_timer_fn+0x7f/0x1c0
  [<ffffffff8104b350>] ? detach_if_pending+0x120/0x120
  [<ffffffff8104b748>] run_timer_softirq+0x238/0x2b0
  [<ffffffff813a1a70>] ? neigh_add+0x310/0x310
  [<ffffffff81043e51>] __do_softirq+0x101/0x280
  [<ffffffff814518cc>] call_softirq+0x1c/0x30
  [<ffffffff81003b65>] do_softirq+0x85/0xc0
  [<ffffffff81043a7e>] irq_exit+0x9e/0xc0
  [<ffffffff810264f8>] smp_apic_timer_interrupt+0x68/0xa0
  [<ffffffff8145122f>] apic_timer_interrupt+0x6f/0x80
  <EOI>  [<ffffffff8100a054>] ? mwait_idle+0xa4/0x1c0
  [<ffffffff8100a04b>] ? mwait_idle+0x9b/0x1c0
  [<ffffffff8100a6a9>] cpu_idle+0x89/0xe0
  [<ffffffff81441127>] start_secondary+0x1b2/0x1b6

Bug is from arp_solicit(), releasing the neigh lock after arp_send()
In case of vxlan, we eventually need to write lock a neigh lock later.

Its a false positive, but we can get rid of it without lockdep
annotations.

We can instead use neigh_ha_snapshot() helper.

Reported-by: Yan Burman <yanb@mellanox.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotuntap: dont use a private kmem_cache
Eric Dumazet [Fri, 21 Dec 2012 07:17:21 +0000 (07:17 +0000)]
tuntap: dont use a private kmem_cache

Commit 96442e42429 (tuntap: choose the txq based on rxq)
added a per tun_struct kmem_cache.

As soon as several tun_struct are used, we get an error
because two caches cannot have same name.

Use the default kmalloc()/kfree_rcu(), as it reduce code
size and doesn't have performance impact here.

Reported-by: Paul Moore <pmoore@redhat.com>
Tested-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: devnet_rename_seq should be a seqcount
Eric Dumazet [Thu, 20 Dec 2012 17:25:08 +0000 (17:25 +0000)]
net: devnet_rename_seq should be a seqcount

Using a seqlock for devnet_rename_seq is not a good idea,
as device_rename() can sleep.

As we hold RTNL, we dont need a protection for writers,
and only need a seqcount so that readers can catch a change done
by a writer.

Bug added in commit c91f6df2db4972d3 (sockopt: Change getsockopt() of
SO_BINDTODEVICE to return an interface name)

Reported-by: Dave Jones <davej@redhat.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoip_gre: fix possible use after free
Eric Dumazet [Thu, 20 Dec 2012 16:00:27 +0000 (16:00 +0000)]
ip_gre: fix possible use after free

Once skb_realloc_headroom() is called, tiph might point to freed memory.

Cache tiph->ttl value before the reallocation, to avoid unexpected
behavior.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Isaku Yamahata <yamahata@valinux.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoip_gre: make ipgre_tunnel_xmit() not parse network header as IP unconditionally
Isaku Yamahata [Thu, 20 Dec 2012 15:12:52 +0000 (15:12 +0000)]
ip_gre: make ipgre_tunnel_xmit() not parse network header as IP unconditionally

ipgre_tunnel_xmit() parses network header as IP unconditionally.
But transmitting packets are not always IP packet. For example such packet
can be sent by packet socket with sockaddr_ll.sll_protocol set.
So make the function check if skb->protocol is IP.

Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosolos-pci: double lock in geos_gpio_store()
Dan Carpenter [Wed, 19 Dec 2012 21:48:45 +0000 (21:48 +0000)]
solos-pci: double lock in geos_gpio_store()

There is a typo here so we do a double lock instead of an unlock.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agodm stripe: add WRITE SAME support
Mike Snitzer [Fri, 21 Dec 2012 20:23:41 +0000 (20:23 +0000)]
dm stripe: add WRITE SAME support

Rename stripe_map_discard to stripe_map_range and reuse it for WRITE
SAME bio processing.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm: remove map_info
Mikulas Patocka [Fri, 21 Dec 2012 20:23:41 +0000 (20:23 +0000)]
dm: remove map_info

This patch removes map_info from bio-based device mapper targets.
map_info is still used for request-based targets.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm snapshot: do not use map_context
Mikulas Patocka [Fri, 21 Dec 2012 20:23:41 +0000 (20:23 +0000)]
dm snapshot: do not use map_context

Eliminate struct map_info from dm-snap.

map_info->ptr was used in dm-snap to indicate if the bio was tracked.
If map_info->ptr was non-NULL, the bio was linked in tracked_chunk_hash.

This patch removes the use of map_info->ptr. We determine if the bio was
tracked based on hlist_unhashed(&c->node). If hlist_unhashed is true,
the bio is not tracked, if it is false, the bio is tracked.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: dont use map_context
Mikulas Patocka [Fri, 21 Dec 2012 20:23:40 +0000 (20:23 +0000)]
dm thin: dont use map_context

This patch removes endio_hook_pool from dm-thin and uses per-bio data instead.

This patch removes any use of map_info in preparation for the next patch
that removes map_info from bio-based device mapper.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm raid1: dont use map_context
Mikulas Patocka [Fri, 21 Dec 2012 20:23:40 +0000 (20:23 +0000)]
dm raid1: dont use map_context

Don't use map_info any more in dm-raid1.

map_info was used for writes to hold the region number. For this purpose
we add a new field dm_bio_details to dm_raid1_bio_record.

map_info was used for reads to hold a pointer to dm_raid1_bio_record (if
the pointer was non-NULL, bio details were saved; if the pointer was
NULL, bio details were not saved). We use
dm_raid1_bio_record.details->bi_bdev for this purpose. If bi_bdev is
NULL, details were not saved, if bi_bdev is non-NULL, details were
saved.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm flakey: dont use map_context
Mikulas Patocka [Fri, 21 Dec 2012 20:23:39 +0000 (20:23 +0000)]
dm flakey: dont use map_context

Replace map_info with a per-bio structure "struct per_bio_data" in dm-flakey.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm raid1: rename read_record to bio_record
Mikulas Patocka [Fri, 21 Dec 2012 20:23:39 +0000 (20:23 +0000)]
dm raid1: rename read_record to bio_record

Rename struct read_record to bio_record in dm-raid1.

In the following patch, the structure will be used for both read and
write bios, so rename it.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm: move target request nr to dm_target_io
Mikulas Patocka [Fri, 21 Dec 2012 20:23:39 +0000 (20:23 +0000)]
dm: move target request nr to dm_target_io

This patch moves target_request_nr from map_info to dm_target_io and
makes it accessible with dm_bio_get_target_request_nr.

This patch is a preparation for the next patch that removes map_info.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm snapshot: use per_bio_data
Mikulas Patocka [Fri, 21 Dec 2012 20:23:38 +0000 (20:23 +0000)]
dm snapshot: use per_bio_data

Replace tracked_chunk_pool with per_bio_data in dm-snap.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm verity: use per_bio_data
Mikulas Patocka [Fri, 21 Dec 2012 20:23:38 +0000 (20:23 +0000)]
dm verity: use per_bio_data

Replace io_mempool with per_bio_data in dm-verity.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm raid1: use per_bio_data
Mikulas Patocka [Fri, 21 Dec 2012 20:23:38 +0000 (20:23 +0000)]
dm raid1: use per_bio_data

Replace read_record_pool with per_bio_data in dm-raid1.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm: introduce per_bio_data
Mikulas Patocka [Fri, 21 Dec 2012 20:23:38 +0000 (20:23 +0000)]
dm: introduce per_bio_data

Introduce a field per_bio_data_size in struct dm_target.

Targets can set this field in the constructor. If a target sets this
field to a non-zero value, "per_bio_data_size" bytes of auxiliary data
are allocated for each bio submitted to the target. These data can be
used for any purpose by the target and help us improve performance by
removing some per-target mempools.

Per-bio data is accessed with dm_per_bio_data. The
argument data_size must be the same as the value per_bio_data_size in
dm_target.

If the target has a pointer to per_bio_data, it can get a pointer to
the bio with dm_bio_from_per_bio_data() function (data_size must be the
same as the value passed to dm_per_bio_data).

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm kcopyd: add WRITE SAME support to dm_kcopyd_zero
Mike Snitzer [Fri, 21 Dec 2012 20:23:37 +0000 (20:23 +0000)]
dm kcopyd: add WRITE SAME support to dm_kcopyd_zero

Add WRITE SAME support to dm-io and make it accessible to
dm_kcopyd_zero().  dm_kcopyd_zero() provides an asynchronous interface
whereas the blkdev_issue_write_same() interface is synchronous.

WRITE SAME is a SCSI command that can be leveraged for more efficient
zeroing of a specified logical extent of a device which supports it.
Only a single zeroed logical block is transfered to the target for each
WRITE SAME and the target then writes that same block across the
specified extent.

The dm thin target uses this.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm linear: add WRITE SAME support
Mike Snitzer [Fri, 21 Dec 2012 20:23:37 +0000 (20:23 +0000)]
dm linear: add WRITE SAME support

The linear target can already support WRITE SAME requests so signal
this by setting num_write_same_requests to 1.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm: add WRITE SAME support
Mike Snitzer [Fri, 21 Dec 2012 20:23:37 +0000 (20:23 +0000)]
dm: add WRITE SAME support

WRITE SAME bios have a payload that contain a single page.  When
cloning WRITE SAME bios DM has no need to modify the bi_io_vec
attributes (and doing so would be detrimental).  DM need only alter the
start and end of the WRITE SAME bio accordingly.

Rather than duplicate __clone_and_map_discard, factor out a common
function that is also used by __clone_and_map_write_same.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm: prepare to support WRITE SAME
Mike Snitzer [Fri, 21 Dec 2012 20:23:36 +0000 (20:23 +0000)]
dm: prepare to support WRITE SAME

Allow targets to opt in to WRITE SAME support by setting
'num_write_same_requests' in the dm_target structure.

A dm device will only advertise WRITE SAME support if all its
targets and all its underlying devices support it.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm ioctl: use kmalloc if possible
Mikulas Patocka [Fri, 21 Dec 2012 20:23:36 +0000 (20:23 +0000)]
dm ioctl: use kmalloc if possible

If the parameter buffer is small enough, try to allocate it with kmalloc()
rather than vmalloc().

vmalloc is noticeably slower than kmalloc because it has to manipulate
page tables.

In my tests, on PA-RISC this patch speeds up activation 13 times.
On Opteron this patch speeds up activation by 5%.

This patch introduces a new function free_params() to free the
parameters and this uses new flags that record whether or not vmalloc()
was used and whether or not the input buffer must be wiped after use.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm ioctl: remove PF_MEMALLOC
Mikulas Patocka [Fri, 21 Dec 2012 20:23:36 +0000 (20:23 +0000)]
dm ioctl: remove PF_MEMALLOC

When allocating memory for the userspace ioctl data, set some
appropriate GPF flags directly instead of using PF_MEMALLOC.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm persistent data: improve improve space map block alloc failure message
Joe Thornber [Fri, 21 Dec 2012 20:23:36 +0000 (20:23 +0000)]
dm persistent data: improve improve space map block alloc failure message

Improve space map error message when unable to allocate a new
metadata block.

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: use DMERR_LIMIT for errors
Mike Snitzer [Fri, 21 Dec 2012 20:23:34 +0000 (20:23 +0000)]
dm thin: use DMERR_LIMIT for errors

Throttle all errors logged from the IO path by dm thin.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm persistent data: use DMERR_LIMIT for errors
Mike Snitzer [Fri, 21 Dec 2012 20:23:34 +0000 (20:23 +0000)]
dm persistent data: use DMERR_LIMIT for errors

Nearly all of persistent-data is in the IO path so throttle error
messages with DMERR_LIMIT to limit the amount logged when
something has gone wrong.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm block manager: reinstate message when validator fails
Mike Snitzer [Fri, 21 Dec 2012 20:23:34 +0000 (20:23 +0000)]
dm block manager: reinstate message when validator fails

Reinstate a useful error message when the block manager buffer validator fails.
This was mistakenly eliminated when the block manager was converted to use
dm-bufio.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm raid: round region_size to power of two
Jonathan Brassow [Fri, 21 Dec 2012 20:23:33 +0000 (20:23 +0000)]
dm raid: round region_size to power of two

If the user does not supply a bitmap region_size to the dm raid target,
a reasonable size is computed automatically.  If this is not a power of 2,
the md code will report an error later.

This patch catches the problem early and rounds the region_size to the
next power of two.

Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: cleanup dead code
Joe Thornber [Fri, 21 Dec 2012 20:23:33 +0000 (20:23 +0000)]
dm thin: cleanup dead code

Remove unused @data_block parameter from cell_defer.
Change thin_bio_map to use many returns rather than setting a variable.

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: rename cell_defer_except to cell_defer_no_holder
Joe Thornber [Fri, 21 Dec 2012 20:23:33 +0000 (20:23 +0000)]
dm thin: rename cell_defer_except to cell_defer_no_holder

Rename cell_defer_except() to cell_defer_no_holder() which describes
its function more clearly.

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm snapshot: optimize track_chunk
Mikulas Patocka [Fri, 21 Dec 2012 20:23:33 +0000 (20:23 +0000)]
dm snapshot: optimize track_chunk

track_chunk is always called with interrupts enabled. Consequently, we
do not need to save and restore interrupt state in "flags" variable.
This patch changes spin_lock_irqsave to spin_lock_irq and
spin_unlock_irqrestore to spin_unlock_irq.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm raid: use DM_ENDIO_INCOMPLETE
Mikulas Patocka [Fri, 21 Dec 2012 20:23:32 +0000 (20:23 +0000)]
dm raid: use DM_ENDIO_INCOMPLETE

Use a defined macro DM_ENDIO_INCOMPLETE instead of a numeric constant.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm raid1: remove impossible mempool_alloc error test
Mikulas Patocka [Fri, 21 Dec 2012 20:23:32 +0000 (20:23 +0000)]
dm raid1: remove impossible mempool_alloc error test

mempool_alloc can't fail if __GFP_WAIT is specified, so the condition
that tests if read_record is non-NULL is always true.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: emit ignore_discard in status when discards disabled
Mike Snitzer [Fri, 21 Dec 2012 20:23:32 +0000 (20:23 +0000)]
dm thin: emit ignore_discard in status when discards disabled

If "ignore_discard" is specified when creating the thin pool device then
discard support is disabled for that device.  The pool device's status
should reflect this fact rather than stating "no_discard_passdown"
(which implies discards are enabled but passdown is disabled).

Reported-by: Zdenek Kabelac <zkabelac@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm persistent data: fix nested btree deletion
Joe Thornber [Fri, 21 Dec 2012 20:23:32 +0000 (20:23 +0000)]
dm persistent data: fix nested btree deletion

When deleting nested btrees, the code forgets to delete the innermost
btree.  The thin-metadata code serendipitously compensates for this by
claiming there is one extra layer in the tree.

This patch corrects both problems.

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: wake worker when discard is prepared
Joe Thornber [Fri, 21 Dec 2012 20:23:31 +0000 (20:23 +0000)]
dm thin: wake worker when discard is prepared

When discards are prepared it is best to directly wake the worker that
will process them.  The worker will be woken anyway, via periodic
commit, but there is no reason to not wake_worker here.

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm thin: fix race between simultaneous io and discards to same block
Joe Thornber [Fri, 21 Dec 2012 20:23:31 +0000 (20:23 +0000)]
dm thin: fix race between simultaneous io and discards to same block

There is a race when discard bios and non-discard bios are issued
simultaneously to the same block.

Discard support is expensive for all thin devices precisely because you
have to be careful to quiesce the area you're discarding.  DM thin must
handle this conflicting IO pattern (simultaneous non-discard vs discard)
even though a sane application shouldn't be issuing such IO.

The race manifests as follows:

1. A non-discard bio is mapped in thin_bio_map.
   This doesn't lock out parallel activity to the same block.

2. A discard bio is issued to the same block as the non-discard bio.

3. The discard bio is locked in a dm_bio_prison_cell in process_discard
   to lock out parallel activity against the same block.

4. The non-discard bio's mapping continues and its all_io_entry is
   incremented so the bio is accounted for in the thin pool's all_io_ds
   which is a dm_deferred_set used to track time locality of non-discard IO.

5. The non-discard bio is finally locked in a dm_bio_prison_cell in
   process_bio.

The race can result in deadlock, leaving the block layer hanging waiting
for completion of a discard bio that never completes, e.g.:

INFO: task ruby:15354 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
ruby            D ffffffff8160f0e0     0 15354  15314 0x00000000
 ffff8802fb08bc58 0000000000000082 ffff8802fb08bfd8 0000000000012900
 ffff8802fb08a010 0000000000012900 0000000000012900 0000000000012900
 ffff8802fb08bfd8 0000000000012900 ffff8803324b9480 ffff88032c6f14c0
Call Trace:
 [<ffffffff814e5a19>] schedule+0x29/0x70
 [<ffffffff814e3d85>] schedule_timeout+0x195/0x220
 [<ffffffffa06b9bc1>] ? _dm_request+0x111/0x160 [dm_mod]
 [<ffffffff814e589e>] wait_for_common+0x11e/0x190
 [<ffffffff8107a170>] ? try_to_wake_up+0x2b0/0x2b0
 [<ffffffff814e59ed>] wait_for_completion+0x1d/0x20
 [<ffffffff81233289>] blkdev_issue_discard+0x219/0x260
 [<ffffffff81233e79>] blkdev_ioctl+0x6e9/0x7b0
 [<ffffffff8119a65c>] block_ioctl+0x3c/0x40
 [<ffffffff8117539c>] do_vfs_ioctl+0x8c/0x340
 [<ffffffff8119a547>] ? block_llseek+0x67/0xb0
 [<ffffffff811756f1>] sys_ioctl+0xa1/0xb0
 [<ffffffff810561f6>] ? sys_rt_sigprocmask+0x86/0xd0
 [<ffffffff814ef099>] system_call_fastpath+0x16/0x1b

The thinp-test-suite's test_discard_random_sectors reliably hits this
deadlock on fast SSD storage.

The fix for this race is that the all_io_entry for a bio must be
incremented whilst the dm_bio_prison_cell is held for the bio's
associated virtual and physical blocks.  That cell locking wasn't
occurring early enough in thin_bio_map.  This patch fixes this.

Care is taken to always call the new function inc_all_io_entry() with
the relevant cells locked, but they are generally unlocked before
calling issue() to try to avoid holding the cells locked across
generic_submit_request.

Also, now that thin_bio_map may lock bios in a cell, process_bio() is no
longer the only thread that will do so.  Because of this we must be sure
to use cell_defer_except() to release all non-holder entries, that
were added by the other thread, because they must be deferred.

This patch depends on "dm thin: replace dm_cell_release_singleton with
cell_defer_except".

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: stable@vger.kernel.org
12 years agodm thin: replace dm_cell_release_singleton with cell_defer_except
Joe Thornber [Fri, 21 Dec 2012 20:23:31 +0000 (20:23 +0000)]
dm thin: replace dm_cell_release_singleton with cell_defer_except

Change existing users of the function dm_cell_release_singleton to share
cell_defer_except instead, and then remove the now-unused function.

Everywhere that calls dm_cell_release_singleton, the bio in question
is the holder of the cell.

If there are no non-holder entries in the cell then cell_defer_except
behaves exactly like dm_cell_release_singleton.  Conversely, if there
*are* non-holder entries then dm_cell_release_singleton must not be used
because those entries would need to be deferred.

Consequently, it is safe to replace use of dm_cell_release_singleton
with cell_defer_except.

This patch is a pre-requisite for "dm thin: fix race between
simultaneous io and discards to same block".

Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm: disable WRITE SAME
Mike Snitzer [Fri, 21 Dec 2012 20:23:30 +0000 (20:23 +0000)]
dm: disable WRITE SAME

WRITE SAME bios are not yet handled correctly by device-mapper so
disable their use on device-mapper devices by setting
max_write_same_sectors to zero.

As an example, a ciphertext device is incompatible because the data
gets changed according to the location at which it written and so the
dm crypt target cannot support it.

Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
Cc: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agodm ioctl: prevent unsafe change to dm_ioctl data_size
Alasdair G Kergon [Fri, 21 Dec 2012 20:23:30 +0000 (20:23 +0000)]
dm ioctl: prevent unsafe change to dm_ioctl data_size

Abort dm ioctl processing if userspace changes the data_size parameter
after we validated it but before we finished copying the data buffer
from userspace.

The dm ioctl parameters are processed in the following sequence:
 1. ctl_ioctl() calls copy_params();
 2. copy_params() makes a first copy of the fixed-sized portion of the
    userspace parameters into the local variable "tmp";
 3. copy_params() then validates tmp.data_size and allocates a new
    structure big enough to hold the complete data and copies the whole
    userspace buffer there;
 4. ctl_ioctl() reads userspace data the second time and copies the whole
    buffer into the pointer "param";
 5. ctl_ioctl() reads param->data_size without any validation and stores it
    in the variable "input_param_size";
 6. "input_param_size" is further used as the authoritative size of the
    kernel buffer.

The problem is that userspace code could change the contents of user
memory between steps 2 and 4.  In particular, the data_size parameter
can be changed to an invalid value after the kernel has validated it.
This lets userspace force the kernel to access invalid kernel memory.

The fix is to ensure that the size has not changed at step 4.

This patch shouldn't have a security impact because CAP_SYS_ADMIN is
required to run this code, but it should be fixed anyway.

Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: stable@kernel.org
12 years agodm persistent data: rename node to btree_node
Mikulas Patocka [Fri, 21 Dec 2012 20:23:30 +0000 (20:23 +0000)]
dm persistent data: rename node to btree_node

This patch fixes a compilation failure on sparc32 by renaming struct node.

struct node is already defined in include/linux/node.h. On sparc32, it
happens to be included through other dependencies and persistent-data
doesn't compile because of conflicting declarations.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
12 years agoNFS: Kill fscache warnings when mounting without -ofsc
Trond Myklebust [Fri, 21 Dec 2012 16:02:32 +0000 (11:02 -0500)]
NFS: Kill fscache warnings when mounting without -ofsc

The fscache code will currently bleat a "non-unique superblock keys"
warning even if the user is mounting without the 'fsc' option.

There should be no reason to even initialise the superblock cache cookie
unless we're planning on using fscache for something, so ensure that we
check for the NFS_OPTION_FSCACHE flag before calling into the fscache
code.

Reported-by: Paweł Sikora <pawel.sikora@agmk.net>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: David Howells <dhowells@redhat.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoNFS: Provide stub nfs_fscache_wait_on_invalidate() for when CONFIG_NFS_FSCACHE=n
David Howells [Fri, 21 Dec 2012 12:15:05 +0000 (12:15 +0000)]
NFS: Provide stub nfs_fscache_wait_on_invalidate() for when CONFIG_NFS_FSCACHE=n

Provide a stub nfs_fscache_wait_on_invalidate() function for when
CONFIG_NFS_FSCACHE=n lest the following error appear:

  fs/nfs/inode.c: In function 'nfs_invalidate_mapping':
  fs/nfs/inode.c:887:2: error: implicit declaration of function 'nfs_fscache_wait_on_invalidate' [-Werror=implicit-function-declaration]
  cc1: some warnings being treated as errors

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Reported-by: Vineet Gupta <Vineet.Gupta1@synopsys.com>
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge tag 'vfio-for-v3.8-v2' of git://github.com/awilliam/linux-vfio
Linus Torvalds [Fri, 21 Dec 2012 05:30:12 +0000 (21:30 -0800)]
Merge tag 'vfio-for-v3.8-v2' of git://github.com/awilliam/linux-vfio

Pull vfio update from Alex Williamson.

* tag 'vfio-for-v3.8-v2' of git://github.com/awilliam/linux-vfio:
  vfio-pci: Enable device before attempting reset
  VFIO: fix out of order labels for error recovery in vfio_pci_init()
  VFIO: use ACCESS_ONCE() to guard access to dev->driver
  VFIO: unregister IOMMU notifier on error recovery path
  vfio-pci: Re-order device reset
  vfio: simplify kmalloc+copy_from_user to memdup_user

12 years agoMerge branch 'for-next' of git://git.infradead.org/users/eparis/notify
Linus Torvalds [Fri, 21 Dec 2012 04:11:52 +0000 (20:11 -0800)]
Merge branch 'for-next' of git://git.infradead.org/users/eparis/notify

Pull filesystem notification updates from Eric Paris:
 "This pull mostly is about locking changes in the fsnotify system.  By
  switching the group lock from a spin_lock() to a mutex() we can now
  hold the lock across things like iput().  This fixes a problem
  involving unmounting a fs and having inodes be busy, first pointed out
  by FAT, but reproducible with tmpfs.

  This also restores signal driven I/O for inotify, which has been
  broken since about 2.6.32."

Ugh.  I *hate* the timing of this.  It was rebased after the merge
window opened, and then left to sit with the pull request coming the day
before the merge window closes.  That's just crap.  But apparently the
patches themselves have been around for over a year, just gathering
dust, so now it's suddenly critical.

Fixed up semantic conflict in fs/notify/fdinfo.c as per Stephen
Rothwell's fixes from -next.

* 'for-next' of git://git.infradead.org/users/eparis/notify:
  inotify: automatically restart syscalls
  inotify: dont skip removal of watch descriptor if creation of ignored event failed
  fanotify: dont merge permission events
  fsnotify: make fasync generic for both inotify and fanotify
  fsnotify: change locking order
  fsnotify: dont put marks on temporary list when clearing marks by group
  fsnotify: introduce locked versions of fsnotify_add_mark() and fsnotify_remove_mark()
  fsnotify: pass group to fsnotify_destroy_mark()
  fsnotify: use a mutex instead of a spinlock to protect a groups mark list
  fanotify: add an extra flag to mark_remove_from_mask that indicates wheather a mark should be destroyed
  fsnotify: take groups mark_lock before mark lock
  fsnotify: use reference counting for groups
  fsnotify: introduce fsnotify_get_group()
  inotify, fanotify: replace fsnotify_put_group() with fsnotify_destroy_group()

12 years agoMerge branch 'akpm' (Andrew's patch-bomb)
Linus Torvalds [Fri, 21 Dec 2012 04:00:43 +0000 (20:00 -0800)]
Merge branch 'akpm' (Andrew's patch-bomb)

Merge the rest of Andrew's patches for -rc1:
 "A bunch of fixes and misc missed-out-on things.

  That'll do for -rc1.  I still have a batch of IPC patches which still
  have a possible bug report which I'm chasing down."

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (25 commits)
  keys: use keyring_alloc() to create module signing keyring
  keys: fix unreachable code
  sendfile: allows bypassing of notifier events
  SGI-XP: handle non-fatal traps
  fat: fix incorrect function comment
  Documentation: ABI: remove testing/sysfs-devices-node
  proc: fix inconsistent lock state
  linux/kernel.h: fix DIV_ROUND_CLOSEST with unsigned divisors
  memcg: don't register hotcpu notifier from ->css_alloc()
  checkpatch: warn on uapi #includes that #include <uapi/...
  revert "rtc: recycle id when unloading a rtc driver"
  mm: clean up transparent hugepage sysfs error messages
  hfsplus: add error message for the case of failure of sync fs in delayed_sync_fs() method
  hfsplus: rework processing of hfs_btree_write() returned error
  hfsplus: rework processing errors in hfsplus_free_extents()
  hfsplus: avoid crash on failed block map free
  kcmp: include linux/ptrace.h
  drivers/rtc/rtc-imxdi.c: must include <linux/spinlock.h>
  mm: cma: WARN if freed memory is still in use
  exec: do not leave bprm->interp on stack
  ...

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Fri, 21 Dec 2012 02:14:31 +0000 (18:14 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull VFS update from Al Viro:
 "fscache fixes, ESTALE patchset, vmtruncate removal series, assorted
  misc stuff."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (79 commits)
  vfs: make lremovexattr retry once on ESTALE error
  vfs: make removexattr retry once on ESTALE
  vfs: make llistxattr retry once on ESTALE error
  vfs: make listxattr retry once on ESTALE error
  vfs: make lgetxattr retry once on ESTALE
  vfs: make getxattr retry once on an ESTALE error
  vfs: allow lsetxattr() to retry once on ESTALE errors
  vfs: allow setxattr to retry once on ESTALE errors
  vfs: allow utimensat() calls to retry once on an ESTALE error
  vfs: fix user_statfs to retry once on ESTALE errors
  vfs: make fchownat retry once on ESTALE errors
  vfs: make fchmodat retry once on ESTALE errors
  vfs: have chroot retry once on ESTALE error
  vfs: have chdir retry lookup and call once on ESTALE error
  vfs: have faccessat retry once on an ESTALE error
  vfs: have do_sys_truncate retry once on an ESTALE error
  vfs: fix renameat to retry on ESTALE errors
  vfs: make do_unlinkat retry once on ESTALE errors
  vfs: make do_rmdir retry once on ESTALE errors
  vfs: add a flags argument to user_path_parent
  ...

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/signal
Linus Torvalds [Fri, 21 Dec 2012 02:05:28 +0000 (18:05 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/signal

Pull signal handling cleanups from Al Viro:
 "sigaltstack infrastructure + conversion for x86, alpha and um,
  COMPAT_SYSCALL_DEFINE infrastructure.

  Note that there are several conflicts between "unify
  SS_ONSTACK/SS_DISABLE definitions" and UAPI patches in mainline;
  resolution is trivial - just remove definitions of SS_ONSTACK and
  SS_DISABLED from arch/*/uapi/asm/signal.h; they are all identical and
  include/uapi/linux/signal.h contains the unified variant."

Fixed up conflicts as per Al.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/signal:
  alpha: switch to generic sigaltstack
  new helpers: __save_altstack/__compat_save_altstack, switch x86 and um to those
  generic compat_sys_sigaltstack()
  introduce generic sys_sigaltstack(), switch x86 and um to it
  new helper: compat_user_stack_pointer()
  new helper: restore_altstack()
  unify SS_ONSTACK/SS_DISABLE definitions
  new helper: current_user_stack_pointer()
  missing user_stack_pointer() instances
  Bury the conditionals from kernel_thread/kernel_execve series
  COMPAT_SYSCALL_DEFINE: infrastructure