safe/jmp/linux-2.6
13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland... linus
Linus Torvalds [Thu, 8 Jul 2010 19:20:54 +0000 (12:20 -0700)]
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:
  IPoIB: Fix world-writable child interface control sysfs attributes
  IB/qib: Clean up properly if qib_init() fails
  IB/qib: Completion queue callback needs to be single threaded
  IB/qib: Update 7322 serdes tables
  IB/qib: Clear 6120 hardware error register
  IB/qib: Clear eager buffer memory for each new process
  IB/qib: Mask hardware error during link reset
  IB/qib: Don't mark VL15 bufs as WC to avoid a rare 7322 chip problem
  RDMA/cxgb4: Derive smac_idx from port viid
  RDMA/cxgb4: Avoid false GTS CIDX_INC overflows
  RDMA/cxgb4: Don't call abort_connection() for active connect failures
  RDMA/cxgb4: Use the DMA state API instead of the pci equivalents

13 years agoMerge branches 'cxgb4', 'ipoib' and 'qib' into for-next
Roland Dreier [Thu, 8 Jul 2010 16:10:24 +0000 (09:10 -0700)]
Merge branches 'cxgb4', 'ipoib' and 'qib' into for-next

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Thu, 8 Jul 2010 15:20:41 +0000 (08:20 -0700)]
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:
  ALSA: hda - Enable beep on Realtek codecs with PCI SSID override
  ALSA: usb-audio - Add volume resolution quirk for some Logitech webcams
  ALSA: hda - Add Macbook 5,2 quirk
  ALSA: hda - Fix uninitialized variable

13 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Thu, 8 Jul 2010 15:06:40 +0000 (08:06 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  writeback: simplify the write back thread queue
  writeback: split writeback_inodes_wb
  writeback: remove writeback_inodes_wbc
  fs-writeback: fix kernel-doc warnings
  splice: check f_mode for seekable file
  splice: direct_splice_actor() should not use pos in sd

13 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Thu, 8 Jul 2010 15:05:43 +0000 (08:05 -0700)]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
  powerpc: Fix userspace build of ptrace.h
  powerpc: Fix default_machine_crash_shutdown #ifdef botch
  powerpc: Fix logic error in fixup_irqs
  powerpc/iseries: Fix possible null pointer dereference in iSeries_pcibios_fixup_resources
  powerpc: Linux cannot run with 0 cores
  powerpc: Fix feature-fixup tests for gcc 4.5
  powerpc: Disable SPARSE_IRQ by default
  powerpc: Fix compile errors in prom_init_check for gcc 4.5
  powerpc: Fix module building for gcc 4.5 and 64 bit
  powerpc/perf_event: Fix for power_pmu_disable()

13 years agoMerge branch 's5p-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 8 Jul 2010 15:02:18 +0000 (08:02 -0700)]
Merge branch 's5p-fixes-for-linus' of git://git./linux/kernel/git/kgene/linux-samsung

* 's5p-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
  ARM: SAMSUNG: Fix on wrong function name for S5PV210 sdhci0
  ARM: S5P6442: Fix PLL setting announce message.
  ARM: SAMSUNG: Fix build without SDHCI controllers for S3C64XX
  ARM: S5PV210: Correct clock register properties
  ARM: S5P: Bug fix on external interrupt for S5P SoCs

13 years agoMerge branch 'fix/misc' into for-linus
Takashi Iwai [Thu, 8 Jul 2010 14:55:26 +0000 (16:55 +0200)]
Merge branch 'fix/misc' into for-linus

13 years agoMerge branch 'fix/hda' into for-linus
Takashi Iwai [Thu, 8 Jul 2010 14:55:02 +0000 (16:55 +0200)]
Merge branch 'fix/hda' into for-linus

13 years agopowerpc: Fix userspace build of ptrace.h
Sam Ravnborg [Sun, 9 May 2010 06:52:31 +0000 (08:52 +0200)]
powerpc: Fix userspace build of ptrace.h

Build of ptrace.h failed for assembly because it
pulls in stdint.h.
Use exportable types (__u32, __u64) to avoid the dependency
on stdint.h.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Andrey Volkov <avolkov@varma-el.com>
Cc: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Fix default_machine_crash_shutdown #ifdef botch
Paul E. McKenney [Tue, 15 Jun 2010 14:48:39 +0000 (14:48 +0000)]
powerpc: Fix default_machine_crash_shutdown #ifdef botch

crash_kexec_wait_realmode() is defined only if CONFIG_PPC_STD_MMU_64
and CONFIG_SMP, but is called if CONFIG_PPC_STD_MMU_64 even if !CONFIG_SMP.
Fix the conditional compilation around the invocation.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Fix logic error in fixup_irqs
Johannes Berg [Wed, 16 Jun 2010 00:09:35 +0000 (00:09 +0000)]
powerpc: Fix logic error in fixup_irqs

When SPARSE_IRQ is set, irq_to_desc() can
return NULL. While the code here has a
check for NULL, it's not really correct.
Fix it by separating the check for it.

This fixes CPU hot unplug for me.

Reported-by: Alastair Bridgewater <alastair.bridgewater@gmail.com>
Cc: stable@kernel.org [2.6.32+]
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/iseries: Fix possible null pointer dereference in iSeries_pcibios_fixup_resources
Denis Kirjanov [Wed, 16 Jun 2010 05:16:59 +0000 (05:16 +0000)]
powerpc/iseries: Fix possible null pointer dereference in iSeries_pcibios_fixup_resources

I don't know if this is a right fix for the problem
since of_get_property can return NULL.
Since iseries_device_information is used only for informational purpose,
we can skip this function without valid HvSubBusNumber number.

Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Linux cannot run with 0 cores
Anton Blanchard [Thu, 17 Jun 2010 14:33:06 +0000 (14:33 +0000)]
powerpc: Linux cannot run with 0 cores

If we configure with CONFIG_SMP=n or set NR_CPUS less than the number of
SMT threads we will set the max cores property to 0 in the
ibm,client-architecture-support structure. On new versions of firmware that
understand this property it obliges and terminates our partition.

Use DIV_ROUND_UP so we handle not only the CONFIG_SMP=n case but also the
case where NR_CPUS isn't a multiple of the number of SMT threads.

Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Fix feature-fixup tests for gcc 4.5
Stephen Rothwell [Mon, 28 Jun 2010 21:08:29 +0000 (21:08 +0000)]
powerpc: Fix feature-fixup tests for gcc 4.5

The feature-fixup test declare some extern void variables and then take
their addresses.  Fix this by declaring them as extern u8 instead.

Fixes these warnings (treated as errors):

  CC      arch/powerpc/lib/feature-fixups.o
cc1: warnings being treated as errors
arch/powerpc/lib/feature-fixups.c: In function 'test_cpu_macros':
arch/powerpc/lib/feature-fixups.c:293:23: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:294:9: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:297:2: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:297:2: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c: In function 'test_fw_macros':
arch/powerpc/lib/feature-fixups.c:306:23: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:307:9: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:310:2: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:310:2: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c: In function 'test_lwsync_macros':
arch/powerpc/lib/feature-fixups.c:321:23: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:322:9: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:326:3: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:326:3: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:329:3: error: taking address of expression of type 'void'
arch/powerpc/lib/feature-fixups.c:329:3: error: taking address of expression of type 'void'

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Disable SPARSE_IRQ by default
Yang Li [Mon, 28 Jun 2010 23:32:57 +0000 (23:32 +0000)]
powerpc: Disable SPARSE_IRQ by default

The SPARSE_IRQ considerably adds overhead to critical path of IRQ
handling.  However it doesn't benefit much in space for most systems with
limited IRQ_NR.  Should be disabled unless really necessary.

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Fix compile errors in prom_init_check for gcc 4.5
Stephen Rothwell [Tue, 29 Jun 2010 20:04:22 +0000 (20:04 +0000)]
powerpc: Fix compile errors in prom_init_check for gcc 4.5

Just whitelist these extra compiler generated symbols.
Fixes these errors:

Error: External symbol '_restgpr0_14' referenced from prom_init.c
Error: External symbol '_restgpr0_20' referenced from prom_init.c
Error: External symbol '_restgpr0_22' referenced from prom_init.c
Error: External symbol '_restgpr0_24' referenced from prom_init.c
Error: External symbol '_restgpr0_25' referenced from prom_init.c
Error: External symbol '_restgpr0_26' referenced from prom_init.c
Error: External symbol '_restgpr0_27' referenced from prom_init.c
Error: External symbol '_restgpr0_28' referenced from prom_init.c
Error: External symbol '_restgpr0_29' referenced from prom_init.c
Error: External symbol '_restgpr0_31' referenced from prom_init.c
Error: External symbol '_savegpr0_14' referenced from prom_init.c
Error: External symbol '_savegpr0_20' referenced from prom_init.c
Error: External symbol '_savegpr0_22' referenced from prom_init.c
Error: External symbol '_savegpr0_24' referenced from prom_init.c
Error: External symbol '_savegpr0_25' referenced from prom_init.c
Error: External symbol '_savegpr0_26' referenced from prom_init.c
Error: External symbol '_savegpr0_27' referenced from prom_init.c
Error: External symbol '_savegpr0_28' referenced from prom_init.c
Error: External symbol '_savegpr0_29' referenced from prom_init.c
Error: External symbol '_savegpr0_31' referenced from prom_init.c

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc: Fix module building for gcc 4.5 and 64 bit
Stephen Rothwell [Tue, 29 Jun 2010 20:08:42 +0000 (20:08 +0000)]
powerpc: Fix module building for gcc 4.5 and 64 bit

Gcc 4.5 is now generating out of line register save and restore
in the function prefix and postfix when we use -Os.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agopowerpc/perf_event: Fix for power_pmu_disable()
Matt Evans [Mon, 5 Jul 2010 17:36:32 +0000 (17:36 +0000)]
powerpc/perf_event: Fix for power_pmu_disable()

When power_pmu_disable() removes the given event from a particular index into
cpuhw->event[], it shuffles down higher event[] entries.  But, this array is
paired with cpuhw->events[] and cpuhw->flags[] so should shuffle them
similarly.

If these arrays get out of sync, code such as power_check_constraints() will
fail.  This caused a bug where events were temporarily disabled and then failed
to be re-enabled; subsequent code tried to write_pmc() with its (disabled) idx
of 0, causing a message "oops trying to write PMC0".  This triggers this bug on
POWER7, running a miss-heavy test:

  perf record -e L1-dcache-load-misses -e L1-dcache-store-misses ./misstest

Signed-off-by: Matt Evans <matt@ozlabs.org>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Thu, 8 Jul 2010 02:56:00 +0000 (19:56 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (35 commits)
  NET: SB1250: Initialize .owner
  vxge: show startup message with KERN_INFO
  ll_temac: Fix missing iounmaps
  bridge: Clear IPCB before possible entry into IP stack
  bridge br_multicast: BUG: unable to handle kernel NULL pointer dereference
  net: Fix definition of netif_vdbg() when VERBOSE_DEBUG is defined
  net/ne: fix memory leak in ne_drv_probe()
  xfrm: fix xfrm by MARK logic
  virtio_net: fix oom handling on tx
  virtio_net: do not reschedule rx refill forever
  s2io: resolve statistics issues
  linux/net.h: fix kernel-doc warnings
  net: decreasing real_num_tx_queues needs to flush qdisc
  sched: qdisc_reset_all_tx is calling qdisc_reset without qdisc_lock
  qlge: fix a eeh handler to not add a pending timer
  qlge: Replacing add_timer() to mod_timer()
  usbnet: Set parent device early for netdev_printk()
  net: Revert "rndis_host: Poll status channel before control channel"
  netfilter: ip6t_REJECT: fix a dst leak in ipv6 REJECT
  drivers: bluetooth: bluecard_cs.c: Fixed include error, changed to linux/io.h
  ...

13 years agoNET: SB1250: Initialize .owner
Ralf Baechle [Tue, 6 Jul 2010 05:18:11 +0000 (05:18 +0000)]
NET: SB1250: Initialize .owner

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
 drivers/net/sb1250-mac.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agovxge: show startup message with KERN_INFO
Wu Fengguang [Tue, 6 Jul 2010 03:02:03 +0000 (03:02 +0000)]
vxge: show startup message with KERN_INFO

The original KERN_CRIT will mess up terminals.

CC: Sreenivasa Honnur <Sreenivasa.Honnur@neterion.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoll_temac: Fix missing iounmaps
Denis Kirjanov [Mon, 5 Jul 2010 21:44:20 +0000 (21:44 +0000)]
ll_temac: Fix missing iounmaps

Fix missing iounmaps.

Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobridge: Clear IPCB before possible entry into IP stack
Herbert Xu [Mon, 5 Jul 2010 21:29:28 +0000 (21:29 +0000)]
bridge: Clear IPCB before possible entry into IP stack

The bridge protocol lives dangerously by having incestuous relations
with the IP stack.  In this instance an abomination has been created
where a bogus IPCB area from a bridged packet leads to a crash in
the IP stack because it's interpreted as IP options.

This patch papers over the problem by clearing the IPCB area in that
particular spot.  To fix this properly we'd also need to parse any
IP options if present but I'm way too lazy for that.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Cheers,
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Wed, 7 Jul 2010 18:43:28 +0000 (11:43 -0700)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm: correctly update connector DPMS status in drm_fb_helper
  drm/radeon/kms: fix shared ddc handling
  drm/ttm: Allocate the page pool manager in the heap.

13 years agodrm: correctly update connector DPMS status in drm_fb_helper
Jesse Barnes [Fri, 2 Jul 2010 17:48:08 +0000 (10:48 -0700)]
drm: correctly update connector DPMS status in drm_fb_helper

We don't currently update the DPMS status of the connector (both in the
connector itself and the connector's DPMS property) in the fb helper
code.  This means that if the kernel FB core has blanked the screen,
sysfs will still show a DPMS status of "on".  It also means that when X
starts, it will try to light up the connectors, but the drm_crtc_helper
code will ignore the DPMS change since according to the connector, the
DPMS status is already on.

Fixes https://bugs.freedesktop.org/show_bug.cgi?id=28436 (the annoying
"my screen was blanked when I started X and now it won't light up" bug).

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/radeon/kms: fix shared ddc handling
Alex Deucher [Thu, 1 Jul 2010 14:34:56 +0000 (10:34 -0400)]
drm/radeon/kms: fix shared ddc handling

Connectors with a shared ddc line can be connected to different
encoders.

Reported by Pasi Kärkkäinen <pasik@iki.fi> on dri-devel

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agodrm/ttm: Allocate the page pool manager in the heap.
Francisco Jerez [Sun, 4 Jul 2010 02:03:07 +0000 (04:03 +0200)]
drm/ttm: Allocate the page pool manager in the heap.

Repeated ttm_page_alloc_init/fini fails noisily because the pool
manager kobj isn't zeroed out between uses (we could do just that but
statically allocated kobjects are generally considered a bad thing).
Move it to kzalloc'ed memory.

Note that this patch drops the refcounting behavior of the pool
allocator init/fini functions: it would have led to a race condition
in its current form, and anyway it was never exploited.

This fixes a regression with reloading kms modules at runtime, since
page allocator was introduced.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agoVFS: introduce s_dirty accessors
Artem Bityutskiy [Thu, 10 Jun 2010 10:56:33 +0000 (13:56 +0300)]
VFS: introduce s_dirty accessors

This patch introduces 3 VFS accessors: 'sb_mark_dirty()',
'sb_mark_clean()', and 'sb_is_dirty()'. They simply
set 'sb->s_dirt' or test 'sb->s_dirt'. The plan is to make
every FS use these accessors later instead of manipulating
the 'sb->s_dirt' flag directly.

Ultimately, this change is a preparation for the periodic
superblock synchronization optimization which is about
preventing the "sync_supers" kernel thread from waking up
even if there is nothing to synchronize.

This patch does not do any functional change, just adds
accessor functions.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 7 Jul 2010 00:16:09 +0000 (17:16 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  rbtree: Undo augmented trees performance damage and regression
  x86, Calgary: Limit the max PHB number to 256

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Wed, 7 Jul 2010 00:15:15 +0000 (17:15 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  ceph: fix crush device 'out' threshold to 1.0, not 0.1
  ceph: fix caps usage accounting for import (non-reserved) case
  ceph: only release clean, unused caps with mds requests
  ceph: fix crush CHOOSE_LEAF when type is already a leaf
  ceph: fix crush recursion
  ceph: fix caps debugfs entry
  ceph: delay umount until all mds requests drop inode+dentry refs
  ceph: handle splice_dentry/d_materialize_unique error in readdir_prepopulate
  ceph: fix crush map update decoding
  ceph: fix message memory leak, uninitialized variable
  ceph: fix map handler error path
  ceph: some endianity fixes

13 years agoMerge branch 'urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/brodo/pcmcia-2.6
Linus Torvalds [Wed, 7 Jul 2010 00:14:53 +0000 (17:14 -0700)]
Merge branch 'urgent' of git://git./linux/kernel/git/brodo/pcmcia-2.6

* 'urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/brodo/pcmcia-2.6:
  pcmcia: do not initialize the present flag too late.

13 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Wed, 7 Jul 2010 00:14:33 +0000 (17:14 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus: (22 commits)
  MIPS: Return after handling coprocessor 2 exception
  MIPS: BCM47xx: Add NVRAM support devices
  MIPS: Loongson: Define rtc device on MC146818-equipped systems
  MIPS: MT: Fix FPU affinity.
  MIPS: Oprofile: Fixup of loongson2_exit()
  MIPS: Alchemy: sleepcode without compile-time cputype dependencies
  MIPS: Tracing: Cleanup of address space checking
  MIPS: Tracing: Cleanup of function graph tracer
  MIPS: Tracing: Reduce the overhead of dynamic Function Tracer
  MIPS: Tracing: Cleanup of instructions used
  MIPS: Tracing: Fix 32-bit support with -mmcount-ra-address
  MIPS: Tracing: Fix argument passing of the 32bit support with gcc 4.5
  MIPS: Tracing: Cleanup comments
  MIPS: Tracing: Cleanup the arguments passing of prepare_ftrace_return
  MIPS: Tracing: Merge adjacent #ifdefs with same condition.
  MIPS: AR7, BCM63xx: fix gpio_to_irq() return value
  MIPS: Restore signalling NaN behaviour for abs.[sd]
  MIPS: Loongson: CS5536: Fix ISA support
  MIPS: Loongson: Add a missing break statement in CS5536 IDE code
  MIPS: Loongson: CS5536: Add missing RDMSRs for IDE and USB
  ...

13 years agokbuild: Fix path to scripts/setlocalversion
Michal Marek [Mon, 5 Jul 2010 21:43:04 +0000 (23:43 +0200)]
kbuild: Fix path to scripts/setlocalversion

Commit 0a564b2 broke LOCALVERSION for O=... builds. Ouch.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reported-and-tested-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-by: Peter Anvin <hpa@zytor.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoIPoIB: Fix world-writable child interface control sysfs attributes
Or Gerlitz [Sun, 6 Jun 2010 04:59:16 +0000 (04:59 +0000)]
IPoIB: Fix world-writable child interface control sysfs attributes

Sumeet Lahorani <sumeet.lahorani@oracle.com> reported that the IPoIB
child entries are world-writable; however we don't want ordinary users
to be able to create and destroy child interfaces, so fix them to be
writable only by root.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Clean up properly if qib_init() fails
Ralph Campbell [Thu, 1 Jul 2010 20:25:45 +0000 (20:25 +0000)]
IB/qib: Clean up properly if qib_init() fails

If qib_init() fails, the driver fails to free memory, unregister
device files, and unregister with the PCIe framework. The driver will
unload without error but a subsequent driver load will cause the
system to panic.  This was found by changing the 7220 code to load the
serdes microcode separately and not installing the microcode file.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Completion queue callback needs to be single threaded
Ralph Campbell [Thu, 17 Jun 2010 23:14:15 +0000 (23:14 +0000)]
IB/qib: Completion queue callback needs to be single threaded

Workqueues aren't exactly equivalent to tasklets since the callback
function may be called from multiple CPUs before the callback returns.
This causes completion notification callbacks to have MT bugs since
they weren't expecting this behavior. The fix is to use a single
threaded work queue.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Update 7322 serdes tables
Ralph Campbell [Thu, 17 Jun 2010 23:14:09 +0000 (23:14 +0000)]
IB/qib: Update 7322 serdes tables

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Clear 6120 hardware error register
Ralph Campbell [Thu, 17 Jun 2010 23:14:04 +0000 (23:14 +0000)]
IB/qib: Clear 6120 hardware error register

The hardware error register needs to be cleared or another interrupt
will be generated, thus causing an infinite loop.  This is a
regression introduced when removing debug output.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Clear eager buffer memory for each new process
Ralph Campbell [Thu, 17 Jun 2010 23:13:59 +0000 (23:13 +0000)]
IB/qib: Clear eager buffer memory for each new process

The eager buffers are not being cleared before being mmapped into a
new user address space.  This is a potential security risk and should
be fixed.  Note that the eager header queue is already being cleared.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Mask hardware error during link reset
Ralph Campbell [Thu, 17 Jun 2010 23:13:54 +0000 (23:13 +0000)]
IB/qib: Mask hardware error during link reset

The HCA checks for certain hardware errors which can be falsely
triggered when the IB link is reset. The fix is to mask them rather
than report them.

Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoIB/qib: Don't mark VL15 bufs as WC to avoid a rare 7322 chip problem
Dave Olson [Thu, 17 Jun 2010 23:13:44 +0000 (23:13 +0000)]
IB/qib: Don't mark VL15 bufs as WC to avoid a rare 7322 chip problem

Don't set write combining via PAT on the VL15 buffers to avoid a rare
problem with unaligned writes from interrupt-flushed store buffers.

Signed-off-by: Dave Olson <dave.olson@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoRDMA/cxgb4: Derive smac_idx from port viid
Steve Wise [Wed, 23 Jun 2010 15:46:44 +0000 (15:46 +0000)]
RDMA/cxgb4: Derive smac_idx from port viid

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoRDMA/cxgb4: Avoid false GTS CIDX_INC overflows
Steve Wise [Thu, 10 Jun 2010 19:03:06 +0000 (19:03 +0000)]
RDMA/cxgb4: Avoid false GTS CIDX_INC overflows

The T4 IQ hw design assumes CIDX_INC credits will be returned on a
regular basis and always before the CIDX counter crosses over the PIDX
counter.  For RDMA CQs, however, returning CIDX_INC credits is only
needed and desired when and if the CQ is armed for notification.  This
can lead to a GTS write returning credits that causes the HW to reject
the credit update because it causes CIDX to pass PIDX.  Once this
happens, the CIDX/PIDX counters get out of whack and an application
can miss a notification and get stuck blocked awaiting a notification.

To avoid this, we allocate the HW IQ 2x times the requested size.
This seems to avoid the false overflow failures.  If we see more
issues with this, then we'll have to add code in the poll path to
return credits periodically like when the amount reaches 1/2 the queue
depth).  I would like to avoid this as it adds a PCI write transaction
for applications that never arm the CQ (like most MPIs).

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoRDMA/cxgb4: Don't call abort_connection() for active connect failures
Steve Wise [Thu, 10 Jun 2010 19:02:55 +0000 (19:02 +0000)]
RDMA/cxgb4: Don't call abort_connection() for active connect failures

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agoRDMA/cxgb4: Use the DMA state API instead of the pci equivalents
FUJITA Tomonori [Thu, 3 Jun 2010 05:37:50 +0000 (05:37 +0000)]
RDMA/cxgb4: Use the DMA state API instead of the pci equivalents

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

No functional change.

For further information about the background:

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

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
13 years agowriteback: simplify the write back thread queue
Christoph Hellwig [Tue, 6 Jul 2010 06:59:53 +0000 (08:59 +0200)]
writeback: simplify the write back thread queue

First remove items from work_list as soon as we start working on them.  This
means we don't have to track any pending or visited state and can get
rid of all the RCU magic freeing the work items - we can simply free
them once the operation has finished.  Second use a real completion for
tracking synchronous requests - if the caller sets the completion pointer
we complete it, otherwise use it as a boolean indicator that we can free
the work item directly.  Third unify struct wb_writeback_args and struct
bdi_work into a single data structure, wb_writeback_work.  Previous we
set all parameters into a struct wb_writeback_args, copied it into
struct bdi_work, copied it again on the stack to use it there.  Instead
of just allocate one structure dynamically or on the stack and use it
all the way through the stack.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
13 years agowriteback: split writeback_inodes_wb
Christoph Hellwig [Thu, 10 Jun 2010 10:07:54 +0000 (12:07 +0200)]
writeback: split writeback_inodes_wb

The case where we have a superblock doesn't require a loop here as we scan
over all inodes in writeback_sb_inodes. Split it out into a separate helper
to make the code simpler.  This also allows to get rid of the sb member in
struct writeback_control, which was rather out of place there.

Also update the comments in writeback_sb_inodes that explain the handling
of inodes from wrong superblocks.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
13 years agowriteback: remove writeback_inodes_wbc
Christoph Hellwig [Thu, 10 Jun 2010 10:07:27 +0000 (12:07 +0200)]
writeback: remove writeback_inodes_wbc

This was just an odd wrapper around writeback_inodes_wb.  Removing this
also allows to get rid of the bdi member of struct writeback_control
which was rather out of place there.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
13 years agobridge br_multicast: BUG: unable to handle kernel NULL pointer dereference
Herbert Xu [Mon, 5 Jul 2010 14:50:08 +0000 (14:50 +0000)]
bridge br_multicast: BUG: unable to handle kernel NULL pointer dereference

On Tue, Jul 06, 2010 at 08:48:35AM +0800, Herbert Xu wrote:
>
> bridge: Restore NULL check in br_mdb_ip_get

Resend with proper attribution.

bridge: Restore NULL check in br_mdb_ip_get

Somewhere along the line the NULL check in br_mdb_ip_get went
AWOL, causing crashes when we receive an IGMP packet with no
multicast table allocated.

This patch restores it and ensures all br_mdb_*_get functions
use it.

Reported-by: Frank Arnold <frank.arnold@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Thanks,
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: Fix definition of netif_vdbg() when VERBOSE_DEBUG is defined
Ben Hutchings [Fri, 2 Jul 2010 07:08:44 +0000 (07:08 +0000)]
net: Fix definition of netif_vdbg() when VERBOSE_DEBUG is defined

netif_vdbg() was originally defined as entirely equivalent to
netdev_vdbg(), but I assume that it was intended to take the same
parameters as netif_dbg() etc.  (Currently it is only used by the
sfc driver, in which I worked on that assumption.)

In commit a4ed89c I changed the definition used when VERBOSE_DEBUG is
not defined, but I failed to notice that the definition used when
VERBOSE_DEBUG is defined was also not as I expected.  Change that to
match netif_dbg() as well.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet/ne: fix memory leak in ne_drv_probe()
Kulikov Vasiliy [Sat, 3 Jul 2010 05:20:42 +0000 (05:20 +0000)]
net/ne: fix memory leak in ne_drv_probe()

net_device allocated with alloc_eip_netdev() must be freed.

Signed-off-by: Kulikov Vasiliy <segooon@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoceph: fix crush device 'out' threshold to 1.0, not 0.1
Sage Weil [Mon, 5 Jul 2010 16:44:17 +0000 (09:44 -0700)]
ceph: fix crush device 'out' threshold to 1.0, not 0.1

Fix a typo that made any OSD weighted between 0.1 and 1.0 effectively
weighted as 1.0 (fully in).

Signed-off-by: Sage Weil <sage@newdream.net>
13 years agoMIPS: Return after handling coprocessor 2 exception
Jesper Nilsson [Thu, 17 Jun 2010 13:25:54 +0000 (15:25 +0200)]
MIPS: Return after handling coprocessor 2 exception

Breaking here dropped us to the default code which always sends a SIGILL
to the current process, no matter what the CU2 notifier says.

[Ralf: Currently this only hurts on Cavium and possibly some out of tree
platforms.]

Signed-off-by: Jesper Nilsson <jesper@jni.nu>
To: linux-mips@linux-mips.org
To: linux-kernel@vger.kernel.org
Patchwork: http://patchwork.linux-mips.org/patch/1391/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: BCM47xx: Add NVRAM support devices
Waldemar Brodkorb [Tue, 8 Jun 2010 17:06:01 +0000 (19:06 +0200)]
MIPS: BCM47xx: Add NVRAM support devices

When trying to netboot a Linksys WRT54GS WLAN router, the bootup fails,
because of following error message:

...
[    0.424000] b44: b44.c:v2.0
[    0.424000] b44: Invalid MAC address found in EEPROM
[    0.432000] b44 ssb0:1: Problem fetching invariants of chip,aborting
[    0.436000] b44: probe of ssb0:1 failed with error -22
...

The router uses a CFE bootloader, but most of the needed environment
variables for network card initialization, are not available from CFE
via printenv and even though not via cfe_getenv().
The required environment variables are saved in a special partition
in flash memory. The attached patch implement nvram_getenv and enables
bootup via NFS root on my router.

Most of the patch is extracted from the OpenWrt subversion repository and
stripped down and cleaned up to just fix this issue.

[Ralf: sorted out header file inclusions.  Lots of unneded headers and such
that should have been included.]

Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Reviewed-by: Phil Sutter <phil@nwl.cc>
To: linux-mips@linux-mips.org
Cc: Hauke Mehrtens <hauke@hauke-m.de>
Patchwork: http://patchwork.linux-mips.org/patch/1359/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Loongson: Define rtc device on MC146818-equipped systems
Arnaud Patard [Tue, 1 Jun 2010 22:39:54 +0000 (00:39 +0200)]
MIPS: Loongson: Define rtc device on MC146818-equipped systems

This patch declare the rtc device present on systems with clock compatible
with the mc146818 and handled by rtc-cmos.  Introduce a new Kconfig entry
because there are some systems without rtc_cmos compatible clock.

Signed-off-by: Arnaud Patard <apatard@mandriva.com>
To: linux-mips@linux-mips.org
Cc: aba@not.so.argh.org
Patchwork: http://patchwork.linux-mips.org/patch/1320/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: MT: Fix FPU affinity.
Ralf Baechle [Sat, 29 May 2010 02:19:57 +0000 (03:19 +0100)]
MIPS: MT: Fix FPU affinity.

The fragile MT sys_sched_setaffinity wrapper needs its regular dose of
fixes.

Nose-poked-at-pile-o-crap-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Oprofile: Fixup of loongson2_exit()
Wu Zhangjin [Thu, 6 May 2010 17:03:49 +0000 (01:03 +0800)]
MIPS: Oprofile: Fixup of loongson2_exit()

When exiting from loongson2_exit(), we need to reset the counter
register too, this patch adds a function reset_counters() to do it, by
the way, this function will be shared by Perf.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1199/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Alchemy: sleepcode without compile-time cputype dependencies
Manuel Lauss [Mon, 24 May 2010 17:42:52 +0000 (19:42 +0200)]
MIPS: Alchemy: sleepcode without compile-time cputype dependencies

Split the low-level sleepcode into per-cpu functions instead of
relying on compile-time-defined cpu type.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
To: Linux-MIPS <linux-mips@linux-mips.org>
Patchwork: http://patchwork.linux-mips.org/patch/1281/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Cleanup of address space checking
Wu Zhangjin [Fri, 14 May 2010 11:08:34 +0000 (19:08 +0800)]
MIPS: Tracing: Cleanup of address space checking

This patch adds an inline function in_module() to check which space the
instruction pointer in, kernel space or module space.

Note:  This will not work when the kernel space and module space are the
same. If they are the same, we need to modify scripts/recordmcount.pl,
ftrace_make_nop/call() and the other related parts to ensure the
enabling/disabling of the calling site to _mcount is right for both
kernel and module.

[Ralf: It also is still incorrect for some 64-bit kernels.]

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1232/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Cleanup of function graph tracer
Wu Zhangjin [Fri, 14 May 2010 11:08:33 +0000 (19:08 +0800)]
MIPS: Tracing: Cleanup of function graph tracer

Cleans up comments and ftrace_get_parent_addr() of function graph tracer.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1231/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Reduce the overhead of dynamic Function Tracer
Wu Zhangjin [Fri, 14 May 2010 11:08:32 +0000 (19:08 +0800)]
MIPS: Tracing: Reduce the overhead of dynamic Function Tracer

With the help of uasm this patch encodes the instructions of the dynamic
function tracer in ftrace_dyn_arch_init() when initializing it.

As a result we can remove the dynamic encoding of instructions in
ftrace_make_nop()/call(), ftrace_enable_ftrace_graph_caller() and remove
the macro jump_insn_encode() and at last this reduce the overhead of
dynamic Function Tracer.  This also is cleaner.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1230/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Cleanup of instructions used
Wu Zhangjin [Fri, 14 May 2010 11:08:31 +0000 (19:08 +0800)]
MIPS: Tracing: Cleanup of instructions used

This patch adds some cleanups of the instructions:
  o use macros instead of magic numbers
  o use macros instead of variables to reduce some overhead
  o add new macro for the jal instruction

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1229/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Fix 32-bit support with -mmcount-ra-address
Wu Zhangjin [Fri, 14 May 2010 11:08:30 +0000 (19:08 +0800)]
MIPS: Tracing: Fix 32-bit support with -mmcount-ra-address

For 32-bit kernel the -mmcount-ra-address option of gcc 4.5 emits one
extra instruction before calling to _mcount so we need to use a different
"b 1f" for it.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1228/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Fix argument passing of the 32bit support with gcc 4.5
Wu Zhangjin [Fri, 14 May 2010 11:08:29 +0000 (19:08 +0800)]
MIPS: Tracing: Fix argument passing of the 32bit support with gcc 4.5

As documented in the GCC 4.5 docs [1] -mmcount-ra-address uses register
$12 to pass the stack offset of the return address to the _mcount function.
On 64-bit kernels $12 is t0 but in 32-bit kernels it is t4 so we need to
use $12 instead of t0 here to be correct for both kernel types.

[1] GCC documentation: MIPS Options
http://gcc.gnu.org/onlinedocs/gcc/MIPS-Options.html

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1227/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Cleanup comments
Wu Zhangjin [Fri, 14 May 2010 11:08:28 +0000 (19:08 +0800)]
MIPS: Tracing: Cleanup comments

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1225/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Cleanup the arguments passing of prepare_ftrace_return
Wu Zhangjin [Fri, 14 May 2010 11:08:27 +0000 (19:08 +0800)]
MIPS: Tracing: Cleanup the arguments passing of prepare_ftrace_return

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1226/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Tracing: Merge adjacent #ifdefs with same condition.
Wu Zhangjin [Fri, 14 May 2010 11:08:26 +0000 (19:08 +0800)]
MIPS: Tracing: Merge adjacent #ifdefs with same condition.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips <linux-mips@linux-mips.org>
Cc: David Daney <david.s.daney@gmail.com>
Patchwork: http://patchwork.linux-mips.org/patch/1224/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: AR7, BCM63xx: fix gpio_to_irq() return value
Yoichi Yuasa [Mon, 24 May 2010 08:36:24 +0000 (17:36 +0900)]
MIPS: AR7, BCM63xx: fix gpio_to_irq() return value

The return value of gpio_to_irq() is not a pointer but an integer.

Signed-off-by: Yoichi Yuasa <yuasa@linux-mips.org>
Cc: linux-mips <linux-mips@linux-mips.org>
Patchwork: http://patchwork.linux-mips.org/patch/1280/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Restore signalling NaN behaviour for abs.[sd]
Chris Dearman [Wed, 12 May 2010 01:30:34 +0000 (18:30 -0700)]
MIPS: Restore signalling NaN behaviour for abs.[sd]

Atsushi Nemoto <anemo@mba.ocn.ne.jp> spotted that this had been incorrectly
removed in a previous patch

Signed-off-by: Chris Dearman <chris@mips.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1213/
Tested-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Loongson: CS5536: Fix ISA support
Wu Zhangjin [Wed, 19 May 2010 01:14:18 +0000 (09:14 +0800)]
MIPS: Loongson: CS5536: Fix ISA support

The function _wrmsr() called by divil_lbar_disable()/enable() should be
called with the offset as the argument.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: Zhang Le <r0bertz@gentoo.org>
Patchwork: http://patchwork.linux-mips.org/patch/1252/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Loongson: Add a missing break statement in CS5536 IDE code
Wu Zhangjin [Wed, 19 May 2010 01:13:38 +0000 (09:13 +0800)]
MIPS: Loongson: Add a missing break statement in CS5536 IDE code

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: Zhang Le <r0bertz@gentoo.org>
Cc: Hu Hongbing <huhb@lemote.com>
Patchwork: http://patchwork.linux-mips.org/patch/1251/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: Loongson: CS5536: Add missing RDMSRs for IDE and USB
Wu Zhangjin [Wed, 19 May 2010 01:12:17 +0000 (09:12 +0800)]
MIPS: Loongson: CS5536: Add missing RDMSRs for IDE and USB

Add several missing RDMSRs for IDE and USB are missing to avoid the
agressive modification of the high 32 bits of the MSR.

Without this patch some usb devices may fail after printing "reset ehci
host ....." when reading the partition information.

Signed-off-by: Hu Hongbing <huhb@lemote.com>
Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: Zhang Le <r0bertz@gentoo.org>
Cc: Hu Hongbing <huhb@lemote.com>
Patchwork: http://patchwork.linux-mips.org/patch/1250/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: AR7: Fix typo in ar7.h
Florian Fainelli [Sun, 16 May 2010 13:25:30 +0000 (15:25 +0200)]
MIPS: AR7: Fix typo in ar7.h

This fixes a typo on the AR7_RESET_PERIPHERAL define.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1247/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoMIPS: AR7: Use correct UART port type
Florian Fainelli [Sun, 16 May 2010 13:25:17 +0000 (15:25 +0200)]
MIPS: AR7: Use correct UART port type

PORT_AR7 has the correct TRIG flag (UART_FCR_R_TRIG_00) as well as
UART_CAP_AFE being set. This fixes UART on TNETD7300 revision 0x02,
which would otherwise mangle some characters, no side effects on
other revisions.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1246/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
13 years agoALSA: hda - Enable beep on Realtek codecs with PCI SSID override
Takashi Iwai [Mon, 5 Jul 2010 15:28:17 +0000 (17:28 +0200)]
ALSA: hda - Enable beep on Realtek codecs with PCI SSID override

When the PCI SSID gives an overriding SKU assno, PC-beep bit isn't
detected (since it's located over 16bit), resulting in no PC beep.
Also, many devices seem ignoring the requirement by Realtek's spec
for SSID numbers, and it also confuses the PC beep detection.

This patch assumes the PC beep is available on every machine with
PCI SSID override.  It's a regression fix from 2.6.34.

Reference: Kernel bug 16251
http://bugzilla.kernel.org/show_bug.cgi?id=16251

Signed-off-by: Takashi Iwai <tiwai@suse.de>
13 years agorbtree: Undo augmented trees performance damage and regression
Peter Zijlstra [Sat, 29 May 2010 13:31:43 +0000 (15:31 +0200)]
rbtree: Undo augmented trees performance damage and regression

Reimplement augmented RB-trees without sprinkling extra branches
all over the RB-tree code (which lives in the scheduler hot path).

This approach is 'borrowed' from Fabio's BFQ implementation and
relies on traversing the rebalance path after the RB-tree-op to
correct the heap property for insertion/removal and make up for
the damage done by the tree rotations.

For insertion the rebalance path is trivially that from the new
node upwards to the root, for removal it is that from the deepest
node in the path from the to be removed node that will still
be around after the removal.

[ This patch also fixes a video driver regression reported by
  Ali Gholami Rudi - the memtype->subtree_max_end was updated
  incorrectly. ]

Acked-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Venkatesh Pallipadi <venki@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Tested-by: Ali Gholami Rudi <ali@rudi.ir>
Cc: Fabio Checconi <fabio@gandalf.sssup.it>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
LKML-Reference: <1275414172.27810.27961.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
13 years agopcmcia: do not initialize the present flag too late.
Dominik Brodowski [Sat, 19 Jun 2010 12:33:56 +0000 (14:33 +0200)]
pcmcia: do not initialize the present flag too late.

The "present" flag was initialized too late -- possibly, a card
was already registered at this time, so re-setting the flag to 0
caused pcmcia_dev_present() to fail.

Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
13 years agoARM: SAMSUNG: Fix on wrong function name for S5PV210 sdhci0
Hyuk Lee [Thu, 10 Jun 2010 03:40:39 +0000 (12:40 +0900)]
ARM: SAMSUNG: Fix on wrong function name for S5PV210 sdhci0

This patch fixes on wrong function name in include/plat/sdhci.h for Samsung.
The 's5pc100_default_sdhci0()' function should be chnaged to
's5pv210_default_sdhci0()'. Because 's5pv210_default_sdhci0()' must be pair.

Signed-off-by: Hyuk Lee <hyuk1.lee@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
13 years agoARM: S5P6442: Fix PLL setting announce message.
Thomas Abraham [Mon, 21 Jun 2010 08:51:54 +0000 (14:21 +0530)]
ARM: S5P6442: Fix PLL setting announce message.

The S5P6442 PLL setting announce message incorrectly displays S5P6440
as the SoC. Change it to S5P6442.

Signed-off-by: Thomas Abraham <thomas.ab@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
13 years agoARM: SAMSUNG: Fix build without SDHCI controllers for S3C64XX
Marek Szyprowski [Tue, 22 Jun 2010 06:45:26 +0000 (15:45 +0900)]
ARM: SAMSUNG: Fix build without SDHCI controllers for S3C64XX

This patch fixes the following compilation problem if only NCP machine
is selected:

arch/arm/mach-s3c64xx/s3c6410.c: In function 's3c6410_map_io':
arch/arm/mach-s3c64xx/s3c6410.c:51: error: implicit declaration of function 's3c6410_default_sdhci2'

And also adds missed 's3c6400_default_sdhci2'.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
[kgene.kim@samsung.com: minor title fix and added comments]
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
13 years agoARM: S5PV210: Correct clock register properties
MyungJoo Ham [Sat, 26 Jun 2010 08:21:50 +0000 (17:21 +0900)]
ARM: S5PV210: Correct clock register properties

1. Corrected shift values of I2S and UART clocks (CLK_GATE_IP3), which were
defined incorrectly.

2. Corrected shift values of sclk_audio, uclk1, sclk_fimd, sclk_mmc,
sclk_spi, sclk_pwm, which had duplicated .enable/.ctrlbit with their
twins defined in struct clk init_clocks_disable[] and struct clk
init_clocks[]. We've changed their .enable/.ctrlbit to use CLK_SRC_MASK
register to avoid the duplicated clock problem described below.

NOTE: Duplicated Clock Problem
Please note that each clock definition should access different control
register; otherwise, the system may suffer lockups. For example, if we
have two clock definitions "a" and "b" which access the same register
(and the shift value). Then, when we do:

module A
clk = clk_get("a");
clk->clk_enable(clk);

module B (context switch)
clk = clk_get("b");
clk->clk_enable(clk);
do something with clk.
clk->clk_disable(clk);

module A (context switch)
do something with clk
* At this point, the system may hang.

Therefore, there should be no clock definitions with the same contol
register/shift. If we need to create "aliases", then, creating child
clocks sharing the clock should be fine.

3. Corrected other sclk_* shift values and access registers.

Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
[kgene.kim@samsung.com: minor title and message fix]
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
13 years agoARM: S5P: Bug fix on external interrupt for S5P SoCs
Boojin Kim [Sat, 26 Jun 2010 07:36:43 +0000 (16:36 +0900)]
ARM: S5P: Bug fix on external interrupt for S5P SoCs

This patch fixes bug on eint type set function, s5p_irq_eint_set_type().
In the IRQ_TYPE_EDGE_FALLING case, S5P_EXTINT_FALLEDGE is right
instead of S5P_EXTINT_RISEEDGE

Signed-off-by: Boojin Kim <boojin.kim@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
13 years agoLinux 2.6.35-rc4 v2.6.35-rc4
Linus Torvalds [Mon, 5 Jul 2010 03:22:50 +0000 (20:22 -0700)]
Linux 2.6.35-rc4

13 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 5 Jul 2010 03:21:39 +0000 (20:21 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  ARM: 6205/1: perf: ensure counter delta is treated as unsigned
  ARM: 6202/1: Do not ARM_DMA_MEM_BUFFERABLE on RealView boards with L210/L220
  ARM: 6201/1: RealView: Do not use outer_sync() on ARM11MPCore boards with L220
  ARM: 6195/1: OMAP3: pmu: make CPU_HAS_PMU dependent on OMAP3_EMU
  ARM: 6194/1: change definition of cpu_relax() for ARM11MPCore
  ARM: 6193/1: RealView: Align the machine_desc.phys_io to 1MB section
  ARM: 6192/1: VExpress: Align the machine_desc.phys_io to 1MB section
  ARM: 6188/1: Add a config option for the ARM11MPCore DMA cache maintenance workaround
  ARM: 6187/1: The v6_dma_inv_range() function must preserve data on SMP
  ARM: 6186/1: Avoid the CONSISTENT_DMA_SIZE warning on noMMU builds
  ARM: mx3: mx31lilly: fix build error for !CONFIG_USB_ULPI
  [ARM] mmp: fix build failure due to IRQ_PMU depends on ARCH_PXA
  [ARM] pxa/mioa701: fix camera regression
  [ARM] pxa/z2: fix flash layout to final version
  [ARM] pxa/z2: fix missing include in battery driver
  [ARM] pxa: fix incorrect gpio type in udc_pxa2xx.h

13 years agoMerge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 5 Jul 2010 03:20:53 +0000 (20:20 -0700)]
Merge branch 'perf-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  perf, x86: Fix incorrect branches event on AMD CPUs
  perf tools: Fix find tids routine by excluding "." and ".."
  x86: Send a SIGTRAP for user icebp traps

13 years agomodule: initialize module dynamic debug later
Yehuda Sadeh [Sat, 3 Jul 2010 03:07:35 +0000 (13:07 +1000)]
module: initialize module dynamic debug later

We should initialize the module dynamic debug datastructures
only after determining that the module is not loaded yet. This
fixes a bug that introduced in 2.6.35-rc2, where when a trying
to load a module twice, we also load it's dynamic printing data
twice which causes all sorts of nasty issues. Also handle
the dynamic debug cleanup later on failure.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (removed a #ifdef)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Mon, 5 Jul 2010 03:13:31 +0000 (20:13 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: remove block number from inode lookup code
  xfs: rename XFS_IGET_BULKSTAT to XFS_IGET_UNTRUSTED
  xfs: validate untrusted inode numbers during lookup
  xfs: always use iget in bulkstat
  xfs: prevent swapext from operating on write-only files

13 years agoMerge branch 'merge-devicetree' of git://git.secretlab.ca/git/linux-2.6
Linus Torvalds [Mon, 5 Jul 2010 03:12:31 +0000 (20:12 -0700)]
Merge branch 'merge-devicetree' of git://git.secretlab.ca/git/linux-2.6

* 'merge-devicetree' of git://git.secretlab.ca/git/linux-2.6:
  of/dma: fix build breakage in ppc4xx adma driver

13 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Mon, 5 Jul 2010 03:12:06 +0000 (20:12 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/mchehab/i7core

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/i7core:
  MAINTAINERS: Add an entry for i7core_edac
  i7core_edac: Avoid doing multiple probes for the same card
  i7core_edac: Properly discover the first QPI device

13 years agoMerge branch 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuil...
Linus Torvalds [Mon, 5 Jul 2010 02:55:23 +0000 (19:55 -0700)]
Merge branch 'rc-fixes' of git://git./linux/kernel/git/mmarek/kbuild-2.6

* 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6:
  kbuild: Propagate LOCALVERSION= down to scripts/setlocalversion
  kbuild: Clean up and speed up the localversion logic

13 years agoARM: 6205/1: perf: ensure counter delta is treated as unsigned
Will Deacon [Fri, 2 Jul 2010 15:41:52 +0000 (16:41 +0100)]
ARM: 6205/1: perf: ensure counter delta is treated as unsigned

Hardware performance counters on ARM are 32-bits wide but atomic64_t
variables are used to represent counter data in the hw_perf_event structure.

The armpmu_event_update function right-shifts a signed 64-bit delta variable
and adds the result to the event count. This can lead to shifting in sign-bits
if the MSB of the 32-bit counter value is set. This results in perf output
such as:

 Performance counter stats for 'sleep 20':

 18446744073460670464  cycles             <-- 0xFFFFFFFFF12A6000
        7783773  instructions             #      0.000 IPC
            465  context-switches
            161  page-faults
        1172393  branches

   20.154242147  seconds time elapsed

This patch ensures that the delta value is treated as unsigned so that the
right shift sets the upper bits to zero.

Cc: <stable@kernel.org>
Acked-by: Jamie Iles <jamie.iles@picochip.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
13 years agoxfrm: fix xfrm by MARK logic
Peter Kosyh [Fri, 2 Jul 2010 07:47:55 +0000 (07:47 +0000)]
xfrm: fix xfrm by MARK logic

While using xfrm by MARK feature in
2.6.34 - 2.6.35 kernels, the mark
is always cleared in flowi structure via memset in
_decode_session4 (net/ipv4/xfrm4_policy.c), so
the policy lookup fails.
IPv6 code is affected by this bug too.

Signed-off-by: Peter Kosyh <p.kosyh@gmail.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoperf, x86: Fix incorrect branches event on AMD CPUs
Vince Weaver [Thu, 1 Jul 2010 19:30:16 +0000 (15:30 -0400)]
perf, x86: Fix incorrect branches event on AMD CPUs

While doing some performance counter validation tests on some
assembly language programs I noticed that the "branches:u"
count was very wrong on AMD machines.

It looks like the wrong event was selected.

Signed-off-by: Vince Weaver <vweaver1@eecs.utk.edu>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Robert Richter <robert.richter@amd.com>
Cc: Borislav Petkov <borislav.petkov@amd.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: <stable@kernel.org>
LKML-Reference: <alpine.DEB.2.00.1007011526010.23160@cl320.eecs.utk.edu>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
13 years agoMerge branch 'vhost-net' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
David S. Miller [Sat, 3 Jul 2010 05:29:22 +0000 (22:29 -0700)]
Merge branch 'vhost-net' of git://git./linux/kernel/git/mst/vhost

13 years agovirtio_net: fix oom handling on tx
Rusty Russell [Fri, 2 Jul 2010 16:34:01 +0000 (16:34 +0000)]
virtio_net: fix oom handling on tx

virtio net will never try to overflow the TX ring, so the only reason
add_buf may fail is out of memory. Thus, we can not stop the
device until some request completes - there's no guarantee anything
at all is outstanding.

Make the error message clearer as well: error here does not
indicate queue full.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (...and avoid TX_BUSY)
Cc: stable@kernel.org # .34.x (s/virtqueue_/vi->svq->vq_ops->/)
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agovirtio_net: do not reschedule rx refill forever
Michael S. Tsirkin [Fri, 2 Jul 2010 16:32:55 +0000 (16:32 +0000)]
virtio_net: do not reschedule rx refill forever

We currently fill all of RX ring, then add_buf
returns ENOSPC, which gets mis-detected as an out of
memory condition and causes us to reschedule the work,
and so on forever. Fix this by oom = err == -ENOMEM;

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: stable@kernel.org # .34.x
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agos2io: resolve statistics issues
Jon Mason [Fri, 2 Jul 2010 09:13:49 +0000 (09:13 +0000)]
s2io: resolve statistics issues

This patch resolves a number of issues in the statistics gathering of
the s2io driver.

On Xframe adapters, the received multicast statistics counter includes
pause frames which are not indicated to the driver.  This can cause
issues where the multicast packet count is higher than what has actually
been received, possibly higher than the number of packets received.

The driver software counters are replaced with the adapter hardware
statistics for rx_packets, rx_bytes, and tx_bytes.  It also uses the
overflow registers to determine if the statistics wrapped the 32bit
register (removing the window of having a statistic value less than the
previous call).  rx_length_errors statistic now includes undersized
packets in addition to oversized packets in its counting.  Finally,
rx_crc_errors are now being counted.

Signed-off-by: Jon Mason <jon.mason@exar.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/kaber/nf-2.6
David S. Miller [Sat, 3 Jul 2010 05:04:49 +0000 (22:04 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/kaber/nf-2.6

13 years agolinux/net.h: fix kernel-doc warnings
Randy Dunlap [Thu, 1 Jul 2010 13:18:58 +0000 (13:18 +0000)]
linux/net.h: fix kernel-doc warnings

Fix kernel-doc warnings in linux/net.h:

Warning(include/linux/net.h:151): No description found for parameter 'wq'
Warning(include/linux/net.h:151): Excess struct/union/enum/typedef member 'fasync_list' description in 'socket'
Warning(include/linux/net.h:151): Excess struct/union/enum/typedef member 'wait' description in 'socket'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: decreasing real_num_tx_queues needs to flush qdisc
John Fastabend [Thu, 1 Jul 2010 13:21:57 +0000 (13:21 +0000)]
net: decreasing real_num_tx_queues needs to flush qdisc

Reducing real_num_queues needs to flush the qdisc otherwise
skbs with queue_mappings greater then real_num_tx_queues can
be sent to the underlying driver.

The flow for this is,

dev_queue_xmit()
dev_pick_tx()
skb_tx_hash()  => hash using real_num_tx_queues
skb_set_queue_mapping()
...
qdisc_enqueue_root() => enqueue skb on txq from hash
...
dev->real_num_tx_queues -= n
...
sch_direct_xmit()
dev_hard_start_xmit()
ndo_start_xmit(skb,dev) => skb queue set with old hash

skbs are enqueued on the qdisc with skb->queue_mapping set
0 < queue_mappings < real_num_tx_queues.  When the driver
decreases real_num_tx_queues skb's may be dequeued from the
qdisc with a queue_mapping greater then real_num_tx_queues.

This fixes a case in ixgbe where this was occurring with DCB
and FCoE. Because the driver is using queue_mapping to map
skbs to tx descriptor rings we can potentially map skbs to
rings that no longer exist.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>