tracing/mm: don't trace mm_page_pcpu_drain on offline cpus
Since tracepoints use RCU for protection, they must not be called on
offline cpus. trace_mm_page_pcpu_drain can be called on an offline cpu
in this scenario caught by LOCKDEP:
tracing/mm: don't trace mm_page_free on offline cpus
Since tracepoints use RCU for protection, they must not be called on
offline cpus. trace_mm_page_free can be called on an offline cpu in this
scenario caught by LOCKDEP:
tracing/mm: don't trace kmem_cache_free on offline cpus
Since tracepoints use RCU for protection, they must not be called on
offline cpus. trace_kmem_cache_free can be called on an offline cpu in
this scenario caught by LOCKDEP:
Dave Airlie [Fri, 29 May 2015 01:13:52 +0000 (11:13 +1000)]
Merge branch 'linux-4.1' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-fixes
Regression fix for Fermi acceleration, and fixes important to bringing
up display-less Maxwell boards.
* 'linux-4.1' of git://anongit.freedesktop.org/git/nouveau/linux-2.6:
drm/nouveau/gr/gm204: remove a stray printk
drm/nouveau/devinit/gm100-: force devinit table execution on boards without PDISP
drm/nouveau/devinit/gf100: make the force-post condition more obvious
drm/nouveau/gr/gf100-: fix wrong constant definition
Commit 3740c82590d8 ("drm/nouveau/gr/gf100-: add symbolic names for
classes") introduced a wrong macro definition causing acceleration setup
to fail. Fix it.
Signed-off-by: Lars Seipel <[email protected]> Fixes: 3740c82590d8 ("drm/nouveau/gr/gf100-: add symbolic names for classes") Signed-off-by: Ben Skeggs <[email protected]>
Brian Foster [Thu, 28 May 2015 22:14:55 +0000 (08:14 +1000)]
xfs: fix broken i_nlink accounting for whiteout tmpfile inode
XFS uses the internal tmpfile() infrastructure for the whiteout inode
used for RENAME_WHITEOUT operations. For tmpfile inodes, XFS allocates
the inode, drops di_nlink, adds the inode to the agi unlinked list,
calls d_tmpfile() which correspondingly drops i_nlink of the vfs inode,
and then finishes the common inode setup (e.g., clear I_NEW and unlock).
The d_tmpfile() call was originally made inxfs_create_tmpfile(), but was
pulled up out of that function as part of the following commit to
resolve a deadlock issue:
330033d6 xfs: fix tmpfile/selinux deadlock and initialize security
As a result, callers of xfs_create_tmpfile() are responsible for either
calling d_tmpfile() or fixing up i_nlink appropriately. The whiteout
tmpfile allocation helper does neither. As a result, the vfs ->i_nlink
becomes inconsistent with the on-disk ->di_nlink once xfs_rename() links
it back into the source dentry and calls xfs_bumplink().
Update the assert in xfs_rename() to help detect this problem in the
future and update xfs_rename_alloc_whiteout() to decrement the link
count as part of the manual tmpfile inode setup.
Dave Chinner [Thu, 28 May 2015 21:40:32 +0000 (07:40 +1000)]
xfs: xfs_iozero can return positive errno
It was missed when we converted everything in XFs to use negative error
numbers, so fix it now. Bug introduced in 3.17 by commit 2451337 ("xfs: global
error sign conversion"), and should go back to stable kernels.
Dave Chinner [Thu, 28 May 2015 21:40:08 +0000 (07:40 +1000)]
xfs: xfs_attr_inactive leaves inconsistent attr fork state behind
xfs_attr_inactive() is supposed to clean up the attribute fork when
the inode is being freed. While it removes attribute fork extents,
it completely ignores attributes in local format, which means that
there can still be active attributes on the inode after
xfs_attr_inactive() has run.
This leads to problems with concurrent inode writeback - the in-core
inode attribute fork is removed without locking on the assumption
that nothing will be attempting to access the attribute fork after a
call to xfs_attr_inactive() because it isn't supposed to exist on
disk any more.
To fix this, make xfs_attr_inactive() completely remove all traces
of the attribute fork from the inode, regardless of it's state.
Further, also remove the in-core attribute fork structure safely so
that there is nothing further that needs to be done by callers to
clean up the attribute fork. This means we can remove the in-core
and on-disk attribute forks atomically.
Also, on error simply remove the in-memory attribute fork. There's
nothing that can be done with it once we have failed to remove the
on-disk attribute fork, so we may as well just blow it away here
anyway.
Clearly indicating that the extent length is greater than MAXEXTLEN,
which is 2097151. A prior trace point shows the allocation was an
exact size match and that a length greater than MAXEXTLEN was asked
for:
We don't see this problem with extent size hints through the IO path
because we can't do single IOs large enough to trigger MAXEXTLEN
allocation. fallocate(), OTOH, is not limited in it's allocation
sizes and so needs help here.
The issue is that the extent size hint alignment is rounding up the
extent size past MAXEXTLEN, because xfs_bmapi_write() is not taking
into account extent size hints when calculating the maximum extent
length to allocate. xfs_bmapi_reserve_delalloc() is already doing
this, but direct extent allocation is not.
Unfortunately, the calculation in xfs_bmapi_reserve_delalloc() is
wrong, and it works only because delayed allocation extents are not
limited in size to MAXEXTLEN in the in-core extent tree. hence this
calculation does not work for direct allocation, and the delalloc
code needs fixing. This may, in fact be the underlying bug that
occassionally causes transaction overruns in delayed allocation
extent conversion, so now we know it's wrong we should fix it, too.
Many thanks to Brian Foster for finding this problem during review
of this patch.
Hence the fix, after much code reading, is to allow
xfs_bmap_extsize_align() to align partial extents when full
alignment would extend the alignment past MAXEXTLEN. We can safely
do this because all callers have higher layer allocation loops that
already handle short allocations, and so will simply run another
allocation to cover the remainder of the requested allocation range
that we ignored during alignment. The advantage of this approach is
that it also removes the need for callers to do anything other than
limit their requests to MAXEXTLEN - they don't really need to be
aware of extent size hints at all.
Dave Chinner [Thu, 28 May 2015 21:39:34 +0000 (07:39 +1000)]
xfs: inode and free block counters need to use __percpu_counter_compare
Because the counters use a custom batch size, the comparison
functions need to be aware of that batch size otherwise the
comparison does not work correctly. This leads to ASSERT failures
on generic/027 like this:
XFS uses non-stanard batch sizes for avoiding frequent global
counter updates on it's allocated inode counters, as they increment
or decrement in batches of 64 inodes. Hence the standard percpu
counter batch of 32 means that the counter is effectively a global
counter. Currently Xfs uses a batch size of 128 so that it doesn't
take the global lock on every single modification.
However, Xfs also needs to compare accurately against zero, which
means we need to use percpu_counter_compare(), and that has a
hard-coded batch size of 32, and hence will spuriously fail to
detect when it is supposed to use precise comparisons and hence
the accounting goes wrong.
Add __percpu_counter_compare() to take a custom batch size so we can
use it sanely in XFS and factor percpu_counter_compare() to use it.
George Wang [Thu, 28 May 2015 21:39:34 +0000 (07:39 +1000)]
xfs: use percpu_counter_read_positive for mp->m_icount
Function percpu_counter_read just return the current counter, which can be
negative. This will cause the checking of "allocated inode
counts <= m_maxicount" false positive. Use percpu_counter_read_positive can
solve this problem, and be consistent with the purpose to introduce percpu
mechanism to xfs.
ALSA: hda - Disable Headphone Mic boost for ALC662
When headphone mic boost is above zero, some 10 - 20 second delay
might occur before the headphone mic is operational.
Therefore disable the headphone mic boost control (recording gain is
sufficient even without it).
(Note: this patch is not about the headset mic, it's about the less
common mic-in only mode.)
NeilBrown [Thu, 28 May 2015 07:53:29 +0000 (17:53 +1000)]
md: fix race when unfreezing sync_action
A recent change removed the need for locking around writing
to "sync_action" (and various other places), but introduced a
subtle race.
When e.g. setting 'reshape' on a 'frozen' array, the 'frozen'
flag is cleared before 'reshape' is set, so the md thread can
get in and start trying recovery - which isn't wanted.
So instead of clearing MD_RECOVERY_FROZEN for any command
except 'frozen', only clear it when each specific command
is parsed. This allows the handling of 'reshape' to clear
the bit while a lock is held.
Also remove some places where we set MD_RECOVERY_NEEDED,
as it is always set on non-error exit of the function.
Signed-off-by: NeilBrown <[email protected]> Fixes: 6791875e2e53 ("md: make reconfig_mutex optional for writes to md sysfs files.")
NeilBrown [Thu, 21 May 2015 02:56:41 +0000 (12:56 +1000)]
md/raid5: call break_stripe_batch_list from handle_stripe_clean_event
Now that the code in break_stripe_batch_list() is nearly identical
to the end of handle_stripe_clean_event, replace the later
with a function call.
The only remaining difference of any interest is the masking that is
applieds to dev[i].flags copied from head_sh.
R5_WriteError certainly isn't wanted as it is set per-stripe, not
per-patch. R5_Overlap isn't wanted as it is explicitly handled.
NeilBrown [Thu, 21 May 2015 02:40:26 +0000 (12:40 +1000)]
md/raid5: be more selective about distributing flags across batch.
When a batch of stripes is broken up, we keep some of the flags
that were per-stripe, and copy other flags from the head to all
others.
This only happens while a stripe is being handled, so many of the
flags are irrelevant.
The "SYNC_FLAGS" (which I've renamed to make it clear there are
several) and STRIPE_DEGRADED are set per-stripe and so need to be
preserved. STRIPE_INSYNC is the only flag that is set on the head
that needs to be propagated to all others.
For safety, add a WARN_ON if others are set, except:
STRIPE_HANDLE - this is safe and per-stripe and we are going to set
in several cases anyway
STRIPE_INSYNC
STRIPE_IO_STARTED - this is just a hint and doesn't hurt.
STRIPE_ON_PLUG_LIST
STRIPE_ON_RELEASE_LIST - It is a point pointless for a batched
stripe to be on one of these lists, but it can happen
as can be safely ignored.
NeilBrown [Thu, 21 May 2015 02:20:36 +0000 (12:20 +1000)]
md/raid5: add handle_flags arg to break_stripe_batch_list.
When we break a stripe_batch_list we sometimes want to set
STRIPE_HANDLE on the individual stripes, and sometimes not.
So pass a 'handle_flags' arg. If it is zero, always set STRIPE_HANDLE
(on non-head stripes). If not zero, only set it if any of the given
flags are present.
NeilBrown [Thu, 21 May 2015 02:00:47 +0000 (12:00 +1000)]
md/raid5: duplicate some more handle_stripe_clean_event code in break_stripe_batch_list
break_stripe_batch list didn't clear head_sh->batch_head.
This was probably a bug.
Also clear all R5_Overlap flags and if any were cleared, wake up
'wait_for_overlap'.
This isn't always necessary but the worst effect is a little
extra checking for code that is waiting on wait_for_overlap.
Also, don't use wake_up_nr() because that does the wrong thing
if 'nr' is zero, and it number of flags cleared doesn't
strongly correlate with the number of threads to wake.
NeilBrown [Thu, 21 May 2015 01:50:16 +0000 (11:50 +1000)]
md/raid5: remove condition test from check_break_stripe_batch_list.
handle_stripe_clean_event() contains a chunk of code very
similar to check_break_stripe_batch_list().
If we make the latter more like the former, we can end up
with just one copy of this code.
This first step removed the condition (and the 'check_') part
of the name. This has the added advantage of making it clear
what check is being performed at the point where the function is
called.
NeilBrown [Fri, 22 May 2015 05:20:04 +0000 (15:20 +1000)]
md/raid5: Ensure a batch member is not handled prematurely.
If a stripe is a member of a batch, but not the head, it must
not be handled separately from the rest of the batch.
'clear_batch_ready()' handles this requirement to some
extent but not completely. If a member is passed to handle_stripe()
a second time it returns '0' indicating the stripe can be handled,
which is wrong.
So add an extra test.
da91309e0a7e (cpumask: Utility function to set n'th cpu...) created a
genuinely weird function. I never saw it before, it went through DaveM.
(He only does this to make us other maintainers feel better about our own
mistakes.)
cpumask_set_cpu_local_first's purpose is say "I need to spread things
across N online cpus, choose the ones on this numa node first"; you call
it in a loop.
It can fail. One of the two callers ignores this, the other aborts and
fails the device open.
It can fail in two ways: allocating the off-stack cpumask, or through a
convoluted codepath which AFAICT can only occur if cpu_online_mask
changes. Which shouldn't happen, because if cpu_online_mask can change
while you call this, it could return a now-offline cpu anyway.
It contains a nonsensical test "!cpumask_of_node(numa_node)". This was
drawn to my attention by Geert, who said this causes a warning on Sparc.
It sets a single bit in a cpumask instead of returning a cpu number,
because that's what the callers want.
It could be made more efficient by passing the previous cpu rather than
an index, but that would be more invasive to the callers.
NeilBrown [Tue, 26 May 2015 22:43:45 +0000 (08:43 +1000)]
md/raid5: close race between STRIPE_BIT_DELAY and batching.
When we add a write to a stripe we need to make sure the bitmap
bit is set. While doing that the stripe is not locked so it could
be added to a batch after which further changes to STRIPE_BIT_DELAY
and ->bm_seq are ineffective.
So we need to hold off adding to a stripe until bitmap_startwrite has
completed at least once, and we need to avoid further changes to
STRIPE_BIT_DELAY once the stripe has been added to a batch.
If a bitmap_startwrite() completes after the stripe was added to a
batch, it will not have set the bit, only incremented a counter, so no
extra delay of the stripe is needed.
Dave Airlie [Thu, 28 May 2015 00:37:35 +0000 (10:37 +1000)]
Merge branch 'drm-fixes-4.1' of git://people.freedesktop.org/~agd5f/linux into drm-fixes
one revert, and two regression fixes for audio/hdmi
* 'drm-fixes-4.1' of git://people.freedesktop.org/~agd5f/linux:
drm/radeon/audio: make sure connector is valid in hotplug case
Revert "drm/radeon: only mark audio as connected if the monitor supports it (v3)"
drm/radeon: don't share plls if monitors differ in audio support
PCI / ACPI: Do not set ACPI companions for host bridges with parents
Commit 97badf873ab6 (device property: Make it possible to use
secondary firmware nodes) uncovered a bug in the x86 (and ia64) PCI
host bridge initialization code that assumes bridge->bus->sysdata
to always point to a struct pci_sysdata object which need not be
the case (in particular, the Xen PCI frontend driver sets it to point
to a different data type). If it is not the case, an incorrect
pointer (or a piece of data that is not a pointer at all) will be
passed to ACPI_COMPANION_SET() and that may cause interesting
breakage to happen going forward.
To work around this problem use the observation that the ACPI
host bridge initialization always passes NULL as parent to
pci_create_root_bus(), so if pcibios_root_bridge_prepare() sees
a non-NULL parent of the bridge, it should not attempt to set
an ACPI companion for it, because that means that
pci_create_root_bus() has been called by someone else.
Fixes: 97badf873ab6 (device property: Make it possible to use secondary firmware nodes) Reported-and-tested-by: Sander Eikelenboom <[email protected]> Signed-off-by: Rafael J. Wysocki <[email protected]> Acked-by: Bjorn Helgaas <[email protected]>
tools/power turbostat: enable turbostat to support Knights Landing (KNL)
Changes mainly to account for minor differences in Knights Landing(KNL):
1. KNL supports C1 and C6 core states.
2. KNL supports PC2, PC3 and PC6 package states.
3. KNL has a different encoding of the TURBO_RATIO_LIMIT MSR
Mike Snitzer [Wed, 27 May 2015 19:25:27 +0000 (15:25 -0400)]
dm: requeue from blk-mq dm_mq_queue_rq() using BLK_MQ_RQ_QUEUE_BUSY
Use BLK_MQ_RQ_QUEUE_BUSY to requeue a blk-mq request directly from the
DM blk-mq device's .queue_rq. This cleans up the previous convoluted
handling of request requeueing that would return BLK_MQ_RQ_QUEUE_OK
(even though it wasn't) and then run blk_mq_requeue_request() followed
by blk_mq_kick_requeue_list().
Also, document that DM blk-mq ontop of old request_fn devices cannot
fail in clone_rq() since the clone request is preallocated as part of
the pdu.
Linus Torvalds [Wed, 27 May 2015 21:09:16 +0000 (14:09 -0700)]
Merge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6
Pull cifs fixes from Steve French:
"Back from SambaXP - now have 8 small CIFS bug fixes to merge"
* 'for-next' of git://git.samba.org/sfrench/cifs-2.6:
CIFS: Fix race condition on RFC1002_NEGATIVE_SESSION_RESPONSE
Fix to convert SURROGATE PAIR
cifs: potential missing check for posix_lock_file_wait
Fix to check Unique id and FileType when client refer file directly.
CIFS: remove an unneeded NULL check
[cifs] fix null pointer check
Fix that several functions handle incorrect value of mapchars
cifs: Don't replace dentries for dfs mounts
1) Don't use MMIO on certain iwlwifi devices otherwise we get a
firmware crash.
2) Don't corrupt the GRO lists of mac80211 contexts by doing sends via
timer interrupt, from Johannes Berg.
3) SKB tailroom is miscalculated in AP_VLAN crypto code, from Michal
Kazior.
4) Fix fw_status memory leak in iwlwifi, from Haim Dreyfuss.
5) Fix use after free in iwl_mvm_d0i3_enable_tx(), from Eliad Peller.
6) JIT'ing of large BPF programs is broken on x86, from Alexei
Starovoitov.
7) EMAC driver ethtool register dump size is miscalculated, from Ivan
Mikhaylov.
8) Fix PHY initial link mode when autonegotiation is disabled in
amd-xgbe, from Tom Lendacky.
9) Fix NULL deref on SOCK_DEAD socket in AF_UNIX and CAIF protocols,
from Mark Salyzyn.
10) credit_bytes not initialized properly in xen-netback, from Ross
Lagerwall.
11) Fallback from MSI-X to INTx interrupts not handled properly in mlx4
driver, fix from Benjamin Poirier.
12) Perform ->attach() after binding dev->qdisc in packet scheduler,
otherwise we can crash. From Cong WANG.
13) Don't clobber data in sctp_v4_map_v6(). From Jason Gunthorpe.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (30 commits)
sctp: Fix mangled IPv4 addresses on a IPv6 listening socket
net_sched: invoke ->attach() after setting dev->qdisc
xen-netfront: properly destroy queues when removing device
mlx4_core: Fix fallback from MSI-X to INTx
xen/netback: Properly initialize credit_bytes
net: netxen: correct sysfs bin attribute return code
tools: bpf_jit_disasm: fix segfault on disabled debugging log output
unix/caif: sk_socket can disappear when state is unlocked
amd-xgbe-phy: Fix initial mode when autoneg is disabled
net: dp83640: fix improper double spin locking.
net: dp83640: reinforce locking rules.
net: dp83640: fix broken calibration routine.
net: stmmac: create one debugfs dir per net-device
net/ibm/emac: fix size of emac dump memory areas
x86: bpf_jit: fix compilation of large bpf programs
net: phy: bcm7xxx: Fix 7425 PHY ID and flags
iwlwifi: mvm: avoid use-after-free on iwl_mvm_d0i3_enable_tx()
iwlwifi: mvm: clean net-detect info if device was reset during suspend
iwlwifi: mvm: take the UCODE_DOWN reference when resuming
iwlwifi: mvm: BT Coex - duplicate the command if sent ASYNC
...
Takashi Iwai [Wed, 27 May 2015 19:20:12 +0000 (21:20 +0200)]
ALSA: hda - Disable power_save_node for IDT92HD71bxx
We've got a regression report that 4.1-rc causes noises on a Dell
laptop. Similar like Realtek codec, this seems also triggered by the
recent power_save_node feature. As this kind of issue is quite hard
to debug without actual hardware, disable the power_save_node flag for
this codec as a workaround.
Jason Gunthorpe [Tue, 26 May 2015 23:30:17 +0000 (17:30 -0600)]
sctp: Fix mangled IPv4 addresses on a IPv6 listening socket
sctp_v4_map_v6 was subtly writing and reading from members
of a union in a way the clobbered data it needed to read before
it read it.
Zeroing the v6 flowinfo overwrites the v4 sin_addr with 0, meaning
that every place that calls sctp_v4_map_v6 gets ::ffff:0.0.0.0 as the
result.
Reorder things to guarantee correct behaviour no matter what the
union layout is.
This impacts user space clients that open an IPv6 SCTP socket and
receive IPv4 connections. Prior to 299ee user space would see a
sockaddr with AF_INET and a correct address, after 299ee the sockaddr
is AF_INET6, but the address is wrong.
WANG Cong [Tue, 26 May 2015 23:08:48 +0000 (16:08 -0700)]
net_sched: invoke ->attach() after setting dev->qdisc
For mq qdisc, we add per tx queue qdisc to root qdisc
for display purpose, however, that happens too early,
before the new dev->qdisc is finally set, this causes
q->list points to an old root qdisc which is going to be
freed right before assigning with a new one.
Fix this by moving ->attach() after setting dev->qdisc.
David Vrabel [Wed, 27 May 2015 14:46:10 +0000 (15:46 +0100)]
xen-netfront: properly destroy queues when removing device
xennet_remove() freed the queues before freeing the netdevice which
results in a use-after-free when free_netdev() tries to delete the
napi instances that have already been freed.
Fix this by fully destroy the queues (which includes deleting the napi
instances) before freeing the netdevice.
Linus Torvalds [Wed, 27 May 2015 18:00:50 +0000 (11:00 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Ingo Molnar:
"This tree includes:
- a fix that disables the compacted FPU XSAVE format by disabling
XSAVES support: the fixes are too complex and the breakages
ABI-affecting, so we want this to be quirked off in a robust way
and backported, to make sure no broken kernel is exposed to the new
hardware (which exposure is still very limited).
- an MCE printk message fix
- a documentation fix"
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/fpu: Disable XSAVES* support for now
x86/Documentation: Update the contact email for L3 cache index disable functionality
x86/mce: Fix MCE severity messages
Linus Torvalds [Wed, 27 May 2015 17:52:13 +0000 (10:52 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mattst88/alpha
Pull alpha updates from Matt Turner:
"Please pull a small collection of patches that I've been neglecting.
Probably most importantly are the patches that wire up the new
syscalls needed by udev and the fix to the bootp{,z}file targets"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mattst88/alpha:
alpha: kernel: osf_sys: Set 'kts.tv_nsec' only when 'tv' has effect
alpha: Wire up all missing implemented syscalls
alpha: Fix bootpfile and bootpzfile make targets
alpha: copy_thread(): rename 'arg' argument to 'kthread_arg'
alpha: delete non-required instances of <linux/init.h>
alpha: don't use module_init for non-modular core code
smp, alpha: kill SMP single function call interrupt
alpha: Remove #include <uapi/asm/types.h> from <asm/types.h>
alpha: clean up unnecessary MSI/MSI-X capability find
Benjamin Poirier [Fri, 22 May 2015 23:12:26 +0000 (16:12 -0700)]
mlx4_core: Fix fallback from MSI-X to INTx
The test in mlx4_load_one() to remove MLX4_FLAG_MSI_X expects mlx4_NOP() to
fail with -EBUSY. It is also necessary to avoid the reset since the device
is not fully reinitialized before calling mlx4_start_hca() a second time.
Note that this will also affect mlx4_test_interrupts(), the only other user
of MLX4_CMD_NOP.
Fixes: f5aef5a ("net/mlx4_core: Activate reset flow upon fatal command cases") Signed-off-by: Benjamin Poirier <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Rusty Russell [Wed, 27 May 2015 01:29:26 +0000 (10:59 +0930)]
lguest: fix out-by-one error in address checking.
This bug has been there since day 1; addresses in the top guest physical
page weren't considered valid. You could map that page (the check in
check_gpte() is correct), but if a guest tried to put a pagetable there
we'd check that address manually when walking it, and kill the guest.
Ross Lagerwall [Wed, 27 May 2015 10:44:32 +0000 (11:44 +0100)]
xen/netback: Properly initialize credit_bytes
Commit e9ce7cb6b107 ("xen-netback: Factor queue-specific data into queue
struct") introduced a regression when moving queue-specific data into
the queue struct by failing to set the credit_bytes field. This
prevented bandwidth limiting from working. Initialize the field as it
was done before multiqueue support was added.
Linus Torvalds [Wed, 27 May 2015 16:47:57 +0000 (09:47 -0700)]
Merge branch 'overlayfs-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs
Pull two overlayfs fixes from Miklos Szeredi:
"Overlayfs rmdir() failed to check for emptiness in one case; this was
introduced in 4.0. The other bug was there since day one: failure to
mount if upper fs is full, which bit some OpenWRT folks"
* 'overlayfs-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
ovl: mount read-only if workdir can't be created
ovl: don't remove non-empty opaque directory
Linus Torvalds [Wed, 27 May 2015 15:56:21 +0000 (08:56 -0700)]
Merge tag 'backlight-fixes-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/backlight
Pull backlight fix from Lee Jones:
"One simple fix to correctly handle -EPROBE_DEFER"
* tag 'backlight-fixes-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/backlight:
backlight: pwm: Handle EPROBE_DEFER while requesting the PWM
Pull crypto fix from Herbert Xu:
"This disables the newly (4.1) added user-space AEAD interface so that
we can fix issues in the underlying kernel AEAD interface. Once the
new kernel AEAD interface is ready we can then reenable the user-space
AEAD interface"
* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: algif_aead - Disable AEAD user-space for now
Junichi Nomura [Wed, 27 May 2015 04:22:07 +0000 (04:22 +0000)]
dm: fix NULL pointer when clone_and_map_rq returns !DM_MAPIO_REMAPPED
When stacking request-based DM on blk_mq device, request cloning and
remapping are done in a single call to target's clone_and_map_rq().
The clone is allocated and valid only if clone_and_map_rq() returns
DM_MAPIO_REMAPPED.
The "IS_ERR(clone)" check in map_request() does not cover all the
!DM_MAPIO_REMAPPED cases that are possible (E.g. if underlying devices
are not ready or unavailable, clone_and_map_rq() may return
DM_MAPIO_REQUEUE without ever having established an ERR_PTR). Fix this
by explicitly checking for a return that is not DM_MAPIO_REMAPPED in
map_request().
Without this fix, DM core may call setup_clone() for a NULL clone
and oops like this:
Johan Hovold [Fri, 15 May 2015 14:27:40 +0000 (16:27 +0200)]
mfd: da9052: Fix broken regulator probe
Fix broken probe of da9052 regulators, which since commit b3f6c73db732
("mfd: da9052-core: Fix platform-device id collision") use a
non-deterministic platform-device id to retrieve static regulator
information. Fortunately, adequate error handling was in place so probe
would simply fail with an error message.
Update the mfd-cell ids to be zero-based and use those to identify the
cells when probing the regulator devices.
Lucas Stach [Wed, 27 May 2015 08:10:26 +0000 (10:10 +0200)]
ARM: imx6: gpc: don't register power domain if DT data is missing
If the devicetree is too old and does not provide the regulator and clocks
for the power domain, we need to avoid registering the power domain.
Otherwise runtime PM will try to control the domain, which will lead to
machine hangs without the proper DT configuration data.
This restores functionality to the kernel 4.0 level if an old DT is
detected, where the power domain is constantly powered on.
Lucas Stach [Tue, 26 May 2015 16:43:36 +0000 (18:43 +0200)]
ARM: imx6: allow booting with old DT
The GPC rewrite to IRQ domains has been on the premise that it may break
suspend/resume for new kernels on old DT, but otherwise keep things working
from a user perspective. This was an accepted compromise to be able to move
the GIC cleanup forward.
What actually happened was that booting a new kernel on an old DT crashes
before even the console is up, so the user does not even see the warning
that the DT is too old. The warning message suggests that this has been
known before, which is clearly unacceptable.
Fix the early crash by mapping the GPC memory space if the IRQ controller
doesn't claim it. This keeps at least CPUidle and the needed CPU wakeup
workarounds working. With this fixed the system is able to boot up
properly minus the expected suspend/resume breakage.
net: netxen: correct sysfs bin attribute return code
If read() syscall requests unexpected number of bytes from "dimm" binary
attribute file, return EINVAL instead of EPERM.
At the same time pin down sysfs file size to the fixed
sizeof(struct netxen_dimm_cfg), which allows to exploit some missing
sanity checks from kernfs (file boundary checks vs offset etc.)
Daniel Borkmann [Mon, 25 May 2015 12:08:03 +0000 (14:08 +0200)]
tools: bpf_jit_disasm: fix segfault on disabled debugging log output
With recent debugging, I noticed that bpf_jit_disasm segfaults when
there's no debugging output from the JIT compiler to the kernel log.
Reason is that when regexec(3) doesn't match on anything, start/end
offsets are not being filled out and contain some uninitialized garbage
from stack. Thus, we need zero out offsets first.
Mark Salyzyn [Tue, 26 May 2015 15:22:19 +0000 (08:22 -0700)]
unix/caif: sk_socket can disappear when state is unlocked
got a rare NULL pointer dereference in clear_bit
Signed-off-by: Mark Salyzyn <[email protected]> Acked-by: Hannes Frederic Sowa <[email protected]>
----
v2: switch to sock_flag(sk, SOCK_DEAD) and added net/caif/caif_socket.c
v3: return -ECONNRESET in upstream caller of wait function for SOCK_DEAD Signed-off-by: David S. Miller <[email protected]>
Chris Lesiak [Tue, 26 May 2015 20:40:44 +0000 (15:40 -0500)]
hwmon: (ntc_thermistor) Ensure iio channel is of type IIO_VOLTAGE
When configured via device tree, the associated iio device needs to be
measuring voltage for the conversion to resistance to be correct.
Return -EINVAL if that is not the case.
Lendacky, Thomas [Tue, 26 May 2015 14:51:49 +0000 (09:51 -0500)]
amd-xgbe-phy: Fix initial mode when autoneg is disabled
When the ethtool command is used to set the speed of the device while
the device is down, the check to set the initial mode may fail when
the device is brought up, causing failure to bring the device up.
Update the code to set the initial mode based on the desired speed if
auto-negotiation is disabled.
This patch fixes a bug introduced by: d9663c8c2149 ("amd-xgbe-phy: Use phydev advertising field vs supported")
David S. Miller [Tue, 26 May 2015 23:38:53 +0000 (19:38 -0400)]
Merge tag 'mac80211-for-davem-2015-05-26' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211
Johannes Berg says:
====================
We have three more fixes:
* AP_VLAN tailroom calculation fix, the bug leads to warnings
along with dropped packets
* NAPI context issue, calling napi_gro_receive() from a timer
(obviously) can lead to crashes
* remain-on-channel combining leads to dropped requests and not
being able to finish certain operations, so remove it
====================
Chen Gang [Mon, 18 May 2015 04:37:08 +0000 (12:37 +0800)]
alpha: kernel: osf_sys: Set 'kts.tv_nsec' only when 'tv' has effect
The related warning:
CC init/do_mounts.o
arch/alpha/kernel/osf_sys.c: In function 'SyS_osf_settimeofday':
arch/alpha/kernel/osf_sys.c:1028:14: warning: 'kts.tv_nsec' may be used uninitialized in this function [-Wmaybe-uninitialized]
kts.tv_nsec *= 1000;
^
arch/alpha/kernel/osf_sys.c:1016:18: note: 'kts' was declared here
struct timespec kts;
^
Helge Deller [Mon, 23 Mar 2015 21:47:21 +0000 (22:47 +0100)]
alpha: Fix bootpfile and bootpzfile make targets
Fix the bootpfile and bootpzfile make targets to creat BOOTP images.
Both targets were broken due to some missing defines to re-map ELF
constants. In addition the old code used the generic vsprintf function
of the kernel which we now replace by a simple and much smaller
implementation for the bootloader.
Alex Dowad [Fri, 13 Mar 2015 18:04:17 +0000 (20:04 +0200)]
alpha: copy_thread(): rename 'arg' argument to 'kthread_arg'
The 'arg' argument to copy_thread() is only ever used when forking a new
kernel thread. Hence, rename it to 'kthread_arg' for clarity (and consistency
with do_fork() and other arch-specific implementations of copy_thread()).
Paul Gortmaker [Tue, 21 Jan 2014 21:22:40 +0000 (16:22 -0500)]
alpha: delete non-required instances of <linux/init.h>
None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>. Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Paul Gortmaker [Fri, 10 Jan 2014 22:02:02 +0000 (17:02 -0500)]
alpha: don't use module_init for non-modular core code
The srm console is always built in. It will never be modular,
so using module_init as an alias for __initcall is rather
misleading.
Fix this up now, so that we can relocate module_init from
init.h into module.h in the future. If we don't do this, we'd
have to add module.h to obviously non-modular code, and that
would be a worse thing.
Direct use of __initcall is discouraged, vs prioritized ones.
Use of device_initcall is consistent with what __initcall
maps onto, and hence does not change the init order, making the
impact of this change zero. Should someone with real hardware
for boot testing want to change it later to arch_initcall or
console_initcall, they can do that at a later date.
Jiang Liu [Sun, 15 Dec 2013 16:36:25 +0000 (00:36 +0800)]
smp, alpha: kill SMP single function call interrupt
Commit 9a46ad6d6df3b54 "smp: make smp_call_function_many() use logic
similar to smp_call_function_single()" has unified the way to handle
single and multiple cross-CPU function calls. Now only one interrupt
is needed for architecture specific code to support generic SMP function
call interfaces, so kill the redundant single function call interrupt.
Yijing Wang [Thu, 8 Aug 2013 13:13:54 +0000 (21:13 +0800)]
alpha: clean up unnecessary MSI/MSI-X capability find
PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev(). So device driver should use
pci_dev->msi_cap/msix_cap to determine whether the device
support MSI/MSI-X instead of using
pci_find_capability(pci_dev, PCI_CAP_ID_MSI/MSIX).
Access to PCIe device config space again will consume more time.
MIPS: Fuloong 2E: Replace CONFIG_USB_ISP1760_HCD by CONFIG_USB_ISP1760
Since commit 100832abf065bc18 ("usb: isp1760: Make HCD support
optional"), CONFIG_USB_ISP1760_HCD is automatically selected when
needed. Enabling that option in the defconfig is now a no-op, and no
longer enables ISP1760 HCD support.
Re-enable the ISP1760 driver in the defconfig by enabling
USB_ISP1760_HOST_ROLE instead.
Joe Perches [Wed, 20 May 2015 12:44:54 +0000 (05:44 -0700)]
MIPS: irq: Use DECLARE_BITMAP
Use the generic mechanism to declare a bitmap instead of unsigned long.
This could fix an overwrite defect of whatever follows irq_map.
Not all "#define NR_IRQS <value>" are a multiple of BITS_PER_LONG so
using DECLARE_BITMAP allocates the proper number of longs required
for the possible bits.
James Hogan [Tue, 28 Apr 2015 09:57:30 +0000 (10:57 +0100)]
ttyFDC: Fix to use native endian MMIO reads
The MIPS Common Device Memory Map (CDMM) is internal to the core and has
native endianness. There is therefore no need to byte swap the accesses
on big endian targets, so convert the Fast Debug Channel (FDC) TTY
driver to use __raw_readl()/__raw_writel() rather than
ioread32()/iowrite32().
James Hogan [Tue, 28 Apr 2015 09:57:29 +0000 (10:57 +0100)]
MIPS: Fix CDMM to use native endian MMIO reads
The MIPS Common Device Memory Map (CDMM) is internal to the core and has
native endianness. There is therefore no need to byte swap the accesses
on big endian targets, so convert the CDMM bus driver to use
__raw_readl() rather than readl().
Junichi Nomura [Tue, 26 May 2015 08:25:54 +0000 (08:25 +0000)]
dm: run queue on re-queue
Without kicking queue, requeued request may stay forever in
the queue if there are no other I/O activities to the device.
The original error had been in v2.6.39 with commit 7eaceaccab5f
("block: remove per-queue plugging"), which replaced conditional
plugging by periodic runqueue.
Commit 9d1deb83d489 in v4.1-rc1 removed the periodic runqueue
and the problem started to manifest.
Fixes: 9d1deb83d489 ("dm: don't schedule delayed run of the queue if nothing to do") Signed-off-by: Jun'ichi Nomura <[email protected]> Signed-off-by: Mike Snitzer <[email protected]>
Herbert Xu [Tue, 26 May 2015 07:32:42 +0000 (15:32 +0800)]
crypto: algif_aead - Disable AEAD user-space for now
The newly added AEAD user-space isn't quite ready for prime time
just yet. In particular it is conflicting with the AEAD single
SG list interface change so this patch disables it now.
Once the SG list stuff is completely done we can then renable
this interface.
Rob Clark [Thu, 21 May 2015 15:58:30 +0000 (11:58 -0400)]
drm/vgem: drop DRIVER_PRIME (v2)
For actual sharing of buffers with other drivers (ie. actual hardware)
we'll need to pimp things out a bit better to deal w/ caching, multiple
memory domains, etc. See thread:
But for the llvmpipe use-case this isn't a problem. Nor do we really
need prime/dri3 (dri2 is sufficient). So until the other issues are
sorted lets remove DRIVER_PRIME.
v2: also drop the dead code
[airlied:
Okay I'm convinced this API could have a lot of use cases
that are really really bad, yes the upload use case is valid
however that isn't the only use case enabled, and if we allow
all the other use cases, people will start to (ab)use them,
and then they'll be ABI and my life will get worse, so disable
PRIME for now]
Guenter Roeck [Mon, 4 May 2015 22:30:47 +0000 (15:30 -0700)]
xtensa: Provide dummy dma_alloc_attrs() and dma_free_attrs()
xtensa:allmodconfig fails to build with the following errors.
drivers/gpu/drm/nouveau/nvkm/subdev/instmem/gk20a.c:
In function ‘gk20a_instobj_dtor_dma’:
drivers/gpu/drm/nouveau/nvkm/subdev/instmem/gk20a.c:154:2: error:
implicit declaration of function ‘dma_free_attrs’
drivers/gpu/drm/nouveau/nvkm/subdev/instmem/gk20a.c:
In function ‘gk20a_instobj_ctor_dma’:
drivers/gpu/drm/nouveau/nvkm/subdev/instmem/gk20a.c:218:2: error:
implicit declaration of function ‘dma_alloc_attrs’
Xtensa does not provide those functions at this time.
Provide dummy implementations to avoid build errors.
David S. Miller [Mon, 25 May 2015 22:22:00 +0000 (18:22 -0400)]
Merge branch 'phyter-bug-fixes'
Richard Cochran says:
====================
phyter bug fixes
While working on a project using the phyter, I noticed some bugs that
have crept in over time. This series fixes those bugs. These patches
are also meant for stable.
====================
Richard Cochran [Mon, 25 May 2015 09:55:45 +0000 (11:55 +0200)]
net: dp83640: fix improper double spin locking.
A pair of nested spin locks was introduced in commit 63502b8d0
"dp83640: Fix receive timestamp race condition".
Unfortunately the 'flags' parameter was reused for the inner lock,
clobbering the originally saved IRQ state. This patch fixes the issue
by changing the inner lock to plain spin_lock without irqsave.
Richard Cochran [Mon, 25 May 2015 09:55:44 +0000 (11:55 +0200)]
net: dp83640: reinforce locking rules.
Callers of the ext_write function are supposed to hold a mutex that
protects the state of the dialed page, but one caller was missing the
lock from the very start, and over time the code has been changed
without following the rule. This patch cleans up the call sites in
violation of the rule.
Richard Cochran [Mon, 25 May 2015 09:55:43 +0000 (11:55 +0200)]
net: dp83640: fix broken calibration routine.
Currently, the calibration function that corrects the initial offsets
among multiple devices only works the first time. If the function is
called more than once, the calibration fails and bogus offsets will be
programmed into the devices.
In a well hidden spot, the device documentation tells that trigger indexes
0 and 1 are special in allowing the TRIG_IF_LATE flag to actually work.
This patch fixes the issue by using one of the special triggers during the
recalibration method.
Linus Torvalds [Mon, 25 May 2015 22:15:54 +0000 (15:15 -0700)]
Merge tag 'pinctrl-v4.1-4' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl
Pull pin control fixes from Linus Walleij:
"Here are some three pin control fixes for the v4.1 cycle, all
driver-specific. Business as usual and calm as it should be in this
portion of the merge window.
- IRQ trigger fix for the Intel Cherryview
- GPIO-to-pin mapping fix for the Cygnus driver
- GPIO-to-pin mapping fix for the Meson8b driver"
* tag 'pinctrl-v4.1-4' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
pinctrl: Fix gpio/pin mapping for Meson8b
pinctrl: cygnus: fixed incorrect GPIO-pin mapping
pinctrl: cherryview: Read triggering type from HW if not set when requested
Linus Torvalds [Mon, 25 May 2015 22:10:59 +0000 (15:10 -0700)]
Merge tag 'gpio-v4.1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio
Pull GPIO fixes from Linus Walleij:
"Here are two GPIO fixes targeted for stable:
- a leak in gpiochip_add path destined for stable
- a kempld driver bug destined for stable"
* tag 'gpio-v4.1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio:
gpio: gpio-kempld: Fix get_direction return value
gpio: fix gpio leak in gpiochip_add error path
Mathieu Olivari [Sat, 23 May 2015 02:03:29 +0000 (19:03 -0700)]
net: stmmac: create one debugfs dir per net-device
stmmac DebugFS entries are currently global to the driver. As a result,
having more than one stmmac device in the system creates the following
error:
* ERROR stmmaceth, debugfs create directory failed
* stmmac_hw_setup: failed debugFS registration
This also results in being able to access the debugfs information for
the first registered device only.
This patch changes the debugfs structure to have one sub-directory per
net-device. Files under "/sys/kernel/debug/stmmaceth" will now show-up
under /sys/kernel/debug/stmmaceth/ethN/.
projects / linux.git / log