mm, oom, compaction: prevent from should_compact_retry looping for ever for costly orders

BugLink: https://bugs.launchpad.net/bugs/1655842
"mm: consider compaction feedback also for costly allocation" has
removed the upper bound for the reclaim/compaction retries based on the
number of reclaimed pages for costly orders.  While this is desirable
the patch did miss a mis interaction between reclaim, compaction and the
retry logic.  The direct reclaim tries to get zones over min watermark
while compaction backs off and returns COMPACT_SKIPPED when all zones
are below low watermark + 1<<order gap.  If we are getting really close
to OOM then __compaction_suitable can keep returning COMPACT_SKIPPED a
high order request (e.g.  hugetlb order-9) while the reclaim is not able
to release enough pages to get us over low watermark.  The reclaim is
still able to make some progress (usually trashing over few remaining
pages) so we are not able to break out from the loop.

I have seen this happening with the same test described in "mm: consider
compaction feedback also for costly allocation" on a swapless system.
The original problem got resolved by "vmscan: consider classzone_idx in
compaction_ready" but it shows how things might go wrong when we
approach the oom event horizont.

The reason why compaction requires being over low rather than min
watermark is not clear to me.  This check was there essentially since
56de7263fcf3 ("mm: compaction: direct compact when a high-order
allocation fails").  It is clearly an implementation detail though and
we shouldn't pull it into the generic retry logic while we should be
able to cope with such eventuality.  The only place in
should_compact_retry where we retry without any upper bound is for
compaction_withdrawn() case.

Introduce compaction_zonelist_suitable function which checks the given
zonelist and returns true only if there is at least one zone which would
would unblock __compaction_suitable if more memory got reclaimed.  In
this implementation it checks __compaction_suitable with NR_FREE_PAGES
plus part of the reclaimable memory as the target for the watermark
check.  The reclaimable memory is reduced linearly by the allocation
order.  The idea is that we do not want to reclaim all the remaining
memory for a single allocation request just unblock
__compaction_suitable which doesn't guarantee we will make a further
progress.

The new helper is then used if compaction_withdrawn() feedback was
provided so we do not retry if there is no outlook for a further
progress.  !costly requests shouldn't be affected much - e.g.  order-2
pages would require to have at least 64kB on the reclaimable LRUs while
order-9 would need at least 32M which should be enough to not lock up.

[vbabka@suse.cz: fix classzone_idx vs. high_zoneidx usage in compaction_zonelist_suitable]
[akpm@linux-foundation.org: fix it for Mel's mm-page_alloc-remove-field-from-alloc_context.patch]
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(backported from commit 86a294a81f93d6f36d00ec3ff779d36d218f852d)
[cascardo: fixup ac_classzone_idx to ac->classzone_idx]
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm: consider compaction feedback also for costly allocation

BugLink: https://bugs.launchpad.net/bugs/1655842
PAGE_ALLOC_COSTLY_ORDER retry logic is mostly handled inside
should_reclaim_retry currently where we decide to not retry after at
least order worth of pages were reclaimed or the watermark check for at
least one zone would succeed after reclaiming all pages if the reclaim
hasn't made any progress.  Compaction feedback is mostly ignored and we
just try to make sure that the compaction did at least something before
giving up.

The first condition was added by a41f24ea9fd6 ("page allocator: smarter
retry of costly-order allocations) and it assumed that lumpy reclaim
could have created a page of the sufficient order.  Lumpy reclaim, has
been removed quite some time ago so the assumption doesn't hold anymore.
Remove the check for the number of reclaimed pages and rely on the
compaction feedback solely.  should_reclaim_retry now only makes sure
that we keep retrying reclaim for high order pages only if they are
hidden by watermaks so order-0 reclaim makes really sense.

should_compact_retry now keeps retrying even for the costly allocations.
The number of retries is reduced wrt.  !costly requests because they are
less important and harder to grant and so their pressure shouldn't cause
contention for other requests or cause an over reclaim.  We also do not
reset no_progress_loops for costly request to make sure we do not keep
reclaiming too agressively.

This has been tested by running a process which fragments memory:
- compact memory
- mmap large portion of the memory (1920M on 2GRAM machine with 2G
  of swapspace)
- MADV_DONTNEED single page in PAGE_SIZE*((1UL<<MAX_ORDER)-1)
  steps until certain amount of memory is freed (250M in my test)
  and reduce the step to (step / 2) + 1 after reaching the end of
  the mapping
- then run a script which populates the page cache 2G (MemTotal)
  from /dev/zero to a new file
And then tries to allocate
nr_hugepages=$(awk '/MemAvailable/{printf "%d\n", $2/(2*1024)}' /proc/meminfo)
huge pages.

root@test1:~# echo 1 > /proc/sys/vm/overcommit_memory;echo 1 > /proc/sys/vm/compact_memory; ./fragment-mem-and-run /root/alloc_hugepages.sh 1920M 250M
Node 0, zone      DMA     31     28     31     10      2      0      2      1      2      3      1
Node 0, zone    DMA32    437    319    171     50     28     25     20     16     16     14    437

* This is the /proc/buddyinfo after the compaction

Done fragmenting. size=2013265920 freed=262144000
Node 0, zone      DMA    165     48      3      1      2      0      2      2      2      2      0
Node 0, zone    DMA32  35109  14575    185     51     41     12      6      0      0      0      0

* /proc/buddyinfo after memory got fragmented

Executing "/root/alloc_hugepages.sh"
Eating some pagecache
508623+0 records in
508623+0 records out
2083319808 bytes (2.1 GB) copied, 11.7292 s, 178 MB/s
Node 0, zone      DMA      3      5      3      1      2      0      2      2      2      2      0
Node 0, zone    DMA32    111    344    153     20     24     10      3      0      0      0      0

* /proc/buddyinfo after page cache got eaten

Trying to allocate 129
129

* 129 hugepages requested and all of them granted.

Node 0, zone      DMA      3      5      3      1      2      0      2      2      2      2      0
Node 0, zone    DMA32    127     97     30     99     11      6      2      1      4      0      0

* /proc/buddyinfo after hugetlb allocation.

10 runs will behave as follows:
Trying to allocate 130
130
--
Trying to allocate 129
129
--
Trying to allocate 128
128
--
Trying to allocate 129
129
--
Trying to allocate 128
128
--
Trying to allocate 129
129
--
Trying to allocate 132
132
--
Trying to allocate 129
129
--
Trying to allocate 128
128
--
Trying to allocate 129
129

So basically 100% success for all 10 attempts.
Without the patch numbers looked much worse:
Trying to allocate 128
12
--
Trying to allocate 129
14
--
Trying to allocate 129
7
--
Trying to allocate 129
16
--
Trying to allocate 129
30
--
Trying to allocate 129
38
--
Trying to allocate 129
19
--
Trying to allocate 129
37
--
Trying to allocate 129
28
--
Trying to allocate 129
37

Just for completness the base kernel without oom detection rework looks
as follows:
Trying to allocate 127
30
--
Trying to allocate 129
12
--
Trying to allocate 129
52
--
Trying to allocate 128
32
--
Trying to allocate 129
12
--
Trying to allocate 129
10
--
Trying to allocate 129
32
--
Trying to allocate 128
14
--
Trying to allocate 128
16
--
Trying to allocate 129
8

As we can see the success rate is much more volatile and smaller without
this patch. So the patch not only makes the retry logic for costly
requests more sensible the success rate is even higher.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 7854ea6c28c6076050e24773eeb78e2925bd7411)
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, oom: protect !costly allocations some more

BugLink: https://bugs.launchpad.net/bugs/1655842
should_reclaim_retry will give up retries for higher order allocations
if none of the eligible zones has any requested or higher order pages
available even if we pass the watermak check for order-0.  This is done
because there is no guarantee that the reclaimable and currently free
pages will form the required order.

This can, however, lead to situations where the high-order request (e.g.
order-2 required for the stack allocation during fork) will trigger OOM
too early - e.g.  after the first reclaim/compaction round.  Such a
system would have to be highly fragmented and there is no guarantee
further reclaim/compaction attempts would help but at least make sure
that the compaction was active before we go OOM and keep retrying even
if should_reclaim_retry tells us to oom if

- the last compaction round backed off or
- we haven't completed at least MAX_COMPACT_RETRIES active
  compaction rounds.

The first rule ensures that the very last attempt for compaction was not
ignored while the second guarantees that the compaction has done some
work.  Multiple retries might be needed to prevent occasional pigggy
backing of other contexts to steal the compacted pages before the
current context manages to retry to allocate them.

compaction_failed() is taken as a final word from the compaction that
the retry doesn't make much sense.  We have to be careful though because
the first compaction round is MIGRATE_ASYNC which is rather weak as it
ignores pages under writeback and gives up too easily in other
situations.  We therefore have to make sure that MIGRATE_SYNC_LIGHT mode
has been used before we give up.  With this logic in place we do not
have to increase the migration mode unconditionally and rather do it
only if the compaction failed for the weaker mode.  A nice side effect
is that the stronger migration mode is used only when really needed so
this has a potential of smaller latencies in some cases.

Please note that the compaction doesn't tell us much about how
successful it was when returning compaction_made_progress so we just
have to blindly trust that another retry is worthwhile and cap the
number to something reasonable to guarantee a convergence.

If the given number of successful retries is not sufficient for a
reasonable workloads we should focus on the collected compaction
tracepoints data and try to address the issue in the compaction code.
If this is not feasible we can increase the retries limit.

[mhocko@suse.com: fix warning]
Link: http://lkml.kernel.org/r/20160512061636.GA4200@dhcp22.suse.cz
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 33c2d21438daea807947923377995c73ee8ed3fc)
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, compaction: abstract compaction feedback to helpers

BugLink: https://bugs.launchpad.net/bugs/1655842
Compaction can provide a wild variation of feedback to the caller.  Many
of them are implementation specific and the caller of the compaction
(especially the page allocator) shouldn't be bound to specifics of the
current implementation.

This patch abstracts the feedback into three basic types:
- compaction_made_progress - compaction was active and made some
  progress.
- compaction_failed - compaction failed and further attempts to
  invoke it would most probably fail and therefore it is not
  worth retrying
- compaction_withdrawn - compaction wasn't invoked for an
          implementation specific reasons. In the current implementation
          it means that the compaction was deferred, contended or the
          page scanners met too early without any progress. Retrying is
          still worthwhile.

[vbabka@suse.cz: do not change thp back off behavior]
[akpm@linux-foundation.org: fix typo in comment, per Hillf]
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(backported from commit cab1802b5f0dddea30547a7451fda8c7e4c593f0)
[cascardo: fixup as we don't have kcompactd]
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, compaction: distinguish between full and partial COMPACT_COMPLETE

BugLink: https://bugs.launchpad.net/bugs/1655842
COMPACT_COMPLETE now means that compaction and free scanner met.  This
is not very useful information if somebody just wants to use this
feedback and make any decisions based on that.  The current caller might
be a poor guy who just happened to scan tiny portion of the zone and
that could be the reason no suitable pages were compacted.  Make sure we
distinguish the full and partial zone walks.

Consumers should treat COMPACT_PARTIAL_SKIPPED as a potential success
and be optimistic in retrying.

The existing users of COMPACT_COMPLETE are conservatively changed to use
COMPACT_PARTIAL_SKIPPED as well but some of them should be probably
reconsidered and only defer the compaction only for COMPACT_COMPLETE
with the new semantic.

This patch shouldn't introduce any functional changes.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit c8f7de0bfae36e8532e5e25a39d15407f02aca78)
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, compaction: simplify __alloc_pages_direct_compact feedback interface

BugLink: https://bugs.launchpad.net/bugs/1655842
__alloc_pages_direct_compact communicates potential back off by two
variables:
- deferred_compaction tells that the compaction returned
  COMPACT_DEFERRED
- contended_compaction is set when there is a contention on
  zone->lock resp. zone->lru_lock locks

__alloc_pages_slowpath then backs of for THP allocation requests to
prevent from long stalls. This is rather messy and it would be much
cleaner to return a single compact result value and hide all the nasty
details into __alloc_pages_direct_compact.

This patch shouldn't introduce any functional changes.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(backported from commit c5d01d0d18e2ab7a21f0371b00e4d1a06f79cdf5)
[cascardo: small cleanup as we backported 0a0337e0d1d1 before]
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, compaction: distinguish COMPACT_DEFERRED from COMPACT_SKIPPED

BugLink: https://bugs.launchpad.net/bugs/1655842
try_to_compact_pages() can currently return COMPACT_SKIPPED even when
the compaction is defered for some zone just because zone DMA is skipped
in 99% of cases due to watermark checks.  This makes COMPACT_DEFERRED
basically unusable for the page allocator as a feedback mechanism.

Make sure we distinguish those two states properly and switch their
ordering in the enum.  This would mean that the COMPACT_SKIPPED will be
returned only when all eligible zones are skipped.

As a result COMPACT_DEFERRED handling for THP in __alloc_pages_slowpath
will be more precise and we would bail out rather than reclaim.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit 1d4746d395975e0ff5103e20ab169d1a95b4ef9e)
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, compaction: change COMPACT_ constants into enum

BugLink: https://bugs.launchpad.net/bugs/1655842
Compaction code is doing weird dances between COMPACT_FOO -> int ->
unsigned long

But there doesn't seem to be any reason for that.  All functions which
return/use one of those constants are not expecting any other value so it
really makes sense to define an enum for them and make it clear that no
other values are expected.

This is a pure cleanup and shouldn't introduce any functional changes.

Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(cherry picked from commit ea7ab982b6bdb7ce218fd3a7850bb2e2b414fdd0)
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

mm, page_alloc: convert alloc_flags to unsigned

BugLink: https://bugs.launchpad.net/bugs/1655842
alloc_flags is a bitmask of flags but it is signed which does not
necessarily generate the best code depending on the compiler.  Even
without an impact, it makes more sense that this be unsigned.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
(backported from commit c603844bdcb5238980de8d58b393f52d7729d651)
[cascardo: keep ALLOC_CPUSET at __alloc_pages_nodemask]
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>

UBUNTU: SAUCE: ibmvscsis: Fix srp_transfer_data fail return code

BugLink: http://bugs.launchpad.net/bugs/1657194
If srp_transfer_data fails within ibmvscsis_write_pending, then
the most likely scenario is that the client timed out the op and
removed the TCE mapping. Thus it will loop forever retrying the
op that is pretty much guaranteed to fail forever. A better return
code would be EIO instead of EAGAIN.

Cc: stable@vger.kernel.org
Reported-by: Steven Royer <seroyer@linux.vnet.ibm.com>
Tested-by: Steven Royer <seroyer@linux.vnet.ibm.com>
Signed-off-by: Bryant G. Ly <bgly@us.ibm.com>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Stefan Bader <stefan.bader@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

UBUNTU: SAUCE: ibmvscsis: fix sleeping in interrupt context

BugLink: http://bugs.launchpad.net/bugs/1657194
Currently, dma_alloc_coherent is being called with a GFP_KERNEL
flag which allows it to sleep in an interrupt context, need to
change to GFP_ATOMIC.

Cc: stable@vger.kernel.org
Tested-by: Steven Royer <seroyer@linux.vnet.ibm.com>
Reviewed-by: Michael Cyr <mikecyr@linux.vnet.ibm.com>
Signed-off-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Stefan Bader <stefan.bader@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

UBUNTU: SAUCE: ibmvscsis: Fix max transfer length

BugLink: http://bugs.launchpad.net/bugs/1657194
Current code incorrectly calculates the max transfer length, since
it is assuming a 4k page table, but ppc64 all run on 64k page tables.

Cc: stable@vger.kernel.org
Reported-by: Steven Royer <seroyer@linux.vnet.ibm.com>
Tested-by: Steven Royer <seroyer@linux.vnet.ibm.com>
Signed-off-by: Bryant G. Ly <bryantly@linux.vnet.ibm.com>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Stefan Bader <stefan.bader@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

nvme: apply DELAY_BEFORE_CHK_RDY quirk at probe time too

BugLink: http://bugs.launchpad.net/bugs/1656913
Commit 54adc01055b7 ("nvme/quirk: Add a delay before checking for adapter
readiness") introduced a quirk to adapters that cannot read the bit
NVME_CSTS_RDY right after register NVME_REG_CC is set; these adapters
need a delay or else the action of reading the bit NVME_CSTS_RDY could
somehow corrupt adapter's registers state and it never recovers.

When this quirk was added, we checked ctrl->tagset in order to avoid
quirking in probe time, supposing we would never require such delay
during probe. Well, it was too optimistic; we in fact need this quirk
at probe time in some cases, like after a kexec.

In some experiments, after abnormal shutdown of machine (aka power cord
unplug), we booted into our bootloader in Power, which is a Linux kernel,
and kexec'ed into another distro. If this kexec is too quick, we end up
reaching the probe of NVMe adapter in that distro when adapter is in
bad state (not fully initialized on our bootloader). What happens next
is that nvme_wait_ready() is unable to complete, except if the quirk is
enabled.

So, this patch removes the original ctrl->tagset verification in order
to enable the quirk even on probe time.

Fixes: 54adc01055b7 ("nvme/quirk: Add a delay before checking for adapter readiness")
Reported-by: Andrew Byrne <byrneadw@ie.ibm.com>
Reported-by: Jaime A. H. Gomez <jahgomez@mx1.ibm.com>
Reported-by: Zachary D. Myers <zdmyers@us.ibm.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@linux.vnet.ibm.com>
Acked-by: Jeffrey Lien <Jeff.Lien@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
(cherry picked from commit b5a10c5f7532b7473776da87e67f8301bbc32693)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Add comments about ROM BAR updating

BugLink: http://bugs.launchpad.net/bugs/1625318
pci_update_resource() updates a hardware BAR so its address matches the
kernel's struct resource UNLESS it's a disabled ROM BAR.  We only update
those when we enable the ROM.

It's not obvious from the code why ROM BARs should be handled specially.
Apparently there are Matrox devices with defective ROM BARs that read as
zero when disabled.  That means that if pci_enable_rom() reads the disabled
BAR, sets PCI_ROM_ADDRESS_ENABLE (without re-inserting the address), and
writes it back, it would enable the ROM at address zero.

Add comments and references to explain why we can't make the code look more
rational.

The code changes are from 755528c860b0 ("Ignore disabled ROM resources at
setup") and 8085ce084c0f ("[PATCH] Fix PCI ROM mapping").

Link: https://lkml.org/lkml/2005/8/30/138
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
(back ported from commit 0b457dde3cf8b7c76a60f8e960f21bbd4abdc416)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Conflicts:
drivers/pci/rom.c

PCI: Decouple IORESOURCE_ROM_ENABLE and PCI_ROM_ADDRESS_ENABLE

BugLink: http://bugs.launchpad.net/bugs/1625318
Remove the assumption that IORESOURCE_ROM_ENABLE == PCI_ROM_ADDRESS_ENABLE.
PCI_ROM_ADDRESS_ENABLE is the ROM enable bit defined by the PCI spec, so if
we're reading or writing a BAR register value, that's what we should use.
IORESOURCE_ROM_ENABLE is a corresponding bit in struct resource flags.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
(cherry picked from commit 7a6d312b50e63f598f5b5914c4fd21878ac2b595)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Remove pci_resource_bar() and pci_iov_resource_bar()

BugLink: http://bugs.launchpad.net/bugs/1625318
pci_std_update_resource() only deals with standard BARs, so we don't have
to worry about the complications of VF BARs in an SR-IOV capability.

Compute the BAR address inline and remove pci_resource_bar().  That makes
pci_iov_resource_bar() unused, so remove that as well.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
(cherry picked from commit 286c2378aaccc7343ebf17ec6cd86567659caf70)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Don't update VF BARs while VF memory space is enabled

BugLink: http://bugs.launchpad.net/bugs/1625318
If we update a VF BAR while it's enabled, there are two potential problems:

  1) Any driver that's using the VF has a cached BAR value that is stale
     after the update, and

  2) We can't update 64-bit BARs atomically, so the intermediate state
     (new lower dword with old upper dword) may conflict with another
     device, and an access by a driver unrelated to the VF may cause a bus
     error.

Warn about attempts to update VF BARs while they are enabled.  This is a
programming error, so use dev_WARN() to get a backtrace.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
(cherry picked from commit 546ba9f8f22f71b0202b6ba8967be5cc6dae4e21)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Separate VF BAR updates from standard BAR updates

BugLink: http://bugs.launchpad.net/bugs/1625318
Previously pci_update_resource() used the same code path for updating
standard BARs and VF BARs in SR-IOV capabilities.

Split the VF BAR update into a new pci_iov_update_resource() internal
interface, which makes it simpler to compute the BAR address (we can get
rid of pci_resource_bar() and pci_iov_resource_bar()).

This patch:

  - Renames pci_update_resource() to pci_std_update_resource(),
  - Adds pci_iov_update_resource(),
  - Makes pci_update_resource() a wrapper that calls the appropriate one,

No functional change intended.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
(cherry picked from commit 6ffa2489c51da77564a0881a73765ea2169f955d)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Update BARs using property bits appropriate for type

BugLink: http://bugs.launchpad.net/bugs/1625318
The BAR property bits (0-3 for memory BARs, 0-1 for I/O BARs) are supposed
to be read-only, but we do save them in res->flags and include them when
updating the BAR.

Mask the I/O property bits with ~PCI_BASE_ADDRESS_IO_MASK (0x3) instead of
PCI_REGION_FLAG_MASK (0xf) to make it obvious that we can't corrupt bits
2-3 of I/O addresses.

Use PCI_ROM_ADDRESS_MASK for ROM BARs.  This means we'll only check the top
21 bits (instead of the 28 bits we used to check) of a ROM BAR to see if
the update was successful.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
(cherry picked from commit 45d004f4afefdd8d79916ee6d97a9ecd94bb1ffe)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Ignore BAR updates on virtual functions

BugLink: http://bugs.launchpad.net/bugs/1625318
VF BARs are read-only zero, so updating VF BARs will not have any effect.
See the SR-IOV spec r1.1, sec 3.4.1.11.

We already ignore these updates because of 70675e0b6a1a ("PCI: Don't try to
restore VF BARs"); this merely restructures it slightly to make it easier
to split updates for standard and SR-IOV BARs.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
(cherry picked from commit 63880b230a4af502c56dde3d4588634c70c66006)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

PCI: Do any VF BAR updates before enabling the BARs

BugLink: http://bugs.launchpad.net/bugs/1625318
Previously we enabled VFs and enable their memory space before calling
pcibios_sriov_enable().  But pcibios_sriov_enable() may update the VF BARs:
for example, on PPC PowerNV we may change them to manage the association of
VFs to PEs.

Because 64-bit BARs cannot be updated atomically, it's unsafe to update
them while they're enabled.  The half-updated state may conflict with other
devices in the system.

Call pcibios_sriov_enable() before enabling the VFs so any BAR updates
happen while the VF BARs are disabled.

[bhelgaas: changelog]
Tested-by: Carol Soto <clsoto@us.ibm.com>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
(cherry picked from commit f40ec3c748c6912f6266c56a7f7992de61b255ed)

Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

kvm: x86: Check dest_map->vector to match eoi signals for rtc

BugLink: http://bugs.launchpad.net/bugs/1649718
Using the vector stored at interrupt delivery makes the eoi
matching safe agains irq migration in the ioapic.

Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 4d99ba898dd0c521ca6cdfdde55c9b58aea3cb3d)
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Andy Whitcroft <apw@canonical.com>
Acked-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

kvm: x86: Track irq vectors in ioapic->rtc_status.dest_map

BugLink: http://bugs.launchpad.net/bugs/1649718
This allows backtracking later in case the rtc irq has been
moved to another vcpu/vector.

Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 9daa50076f585854f0040aa8403eac020d6f5d64)
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Andy Whitcroft <apw@canonical.com>
Acked-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

kvm: x86: Convert ioapic->rtc_status.dest_map to a struct

BugLink: http://bugs.launchpad.net/bugs/1649718
Currently this is a bitmap which tracks which CPUs we expect
an EOI from. Move this bitmap to a struct so that we can
track additional information there.

Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 9e4aabe2bb3454c83dac8139cf9974503ee044db)
Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Andy Whitcroft <apw@canonical.com>
Acked-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>

Linux 4.4.44

BugLink: http://bugs.launchpad.net/bugs/1658091
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

pinctrl: sh-pfc: Do not unconditionally support PIN_CONFIG_BIAS_DISABLE

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 5d7400c4acbf7fe633a976a89ee845f7333de3e4 upstream.

Always stating PIN_CONFIG_BIAS_DISABLE is supported gives untrue output
when examining /sys/kernel/debug/pinctrl/e6060000.pfc/pinconf-pins if
the operation get_bias() is implemented but the pin is not handled by
the get_bias() implementation. In that case the output will state that
"input bias disabled" indicating that this pin has bias control
support.

Make support for PIN_CONFIG_BIAS_DISABLE depend on that the pin either
supports SH_PFC_PIN_CFG_PULL_UP or SH_PFC_PIN_CFG_PULL_DOWN. This also
solves the issue where SoC specific implementations print error messages
if their particular implementation of {set,get}_bias() is called with a
pin it does not know about.

Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

powerpc/ibmebus: Fix device reference leaks in sysfs interface

BugLink: http://bugs.launchpad.net/bugs/1658091
commit fe0f3168169f7c34c29b0cf0c489f126a7f29643 upstream.

Make sure to drop any reference taken by bus_find_device() in the sysfs
callbacks that are used to create and destroy devices based on
device-tree entries.

Fixes: 6bccf755ff53 ("[POWERPC] ibmebus: dynamic addition/removal of adapters, some code cleanup")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

powerpc/ibmebus: Fix further device reference leaks

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 815a7141c4d1b11610dccb7fcbb38633759824f2 upstream.

Make sure to drop any reference taken by bus_find_device() when creating
devices during init and driver registration.

Fixes: 55347cc9962f ("[POWERPC] ibmebus: Add device creation and bus probing based on of_device")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

bus: vexpress-config: fix device reference leak

BugLink: http://bugs.launchpad.net/bugs/1658091
commit c090959b9dd8c87703e275079aa4b4a824ba3f8e upstream.

Make sure to drop the reference to the parent device taken by
class_find_device() after populating the bus.

Fixes: 3b9334ac835b ("mfd: vexpress: Convert custom func API to regmap")
Acked-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

blk-mq: Always schedule hctx->next_cpu

BugLink: http://bugs.launchpad.net/bugs/1658091
commit c02ebfdddbafa9a6a0f52fbd715e6bfa229af9d3 upstream.

Commit 0e87e58bf60e ("blk-mq: improve warning for running a queue on the
wrong CPU") attempts to avoid triggering the WARN_ON in
__blk_mq_run_hw_queue when the expected CPU is dead.  Problem is, in the
last batch execution before round robin, blk_mq_hctx_next_cpu can
schedule a dead CPU and also update next_cpu to the next alive CPU in
the mask, which will trigger the WARN_ON despite the previous
workaround.

The following patch fixes this scenario by always scheduling the value
in hctx->next_cpu.  This changes the moment when we round-robin the CPU
running the hctx, but it really doesn't matter, since it still executes
BLK_MQ_CPU_WORK_BATCH times in a row before switching to another CPU.

Fixes: 0e87e58bf60e ("blk-mq: improve warning for running a queue on the wrong CPU")
Signed-off-by: Gabriel Krisman Bertazi <krisman@linux.vnet.ibm.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

ACPI / APEI: Fix NMI notification handling

BugLink: http://bugs.launchpad.net/bugs/1658091
commit a545715d2dae8d071c5b06af947b07ffa846b288 upstream.

When removing and adding cpu 0 on a system with GHES NMI the following stack
trace is seen when re-adding the cpu:

WARNING: CPU: 0 PID: 0 at arch/x86/kernel/apic/apic.c:1349 setup_local_APIC+
Modules linked in: nfsv3 rpcsec_gss_krb5 nfsv4 nfs fscache coretemp intel_ra
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.9.0-rc6+ #2
Call Trace:
 dump_stack+0x63/0x8e
 __warn+0xd1/0xf0
 warn_slowpath_null+0x1d/0x20
 setup_local_APIC+0x275/0x370
 apic_ap_setup+0xe/0x20
 start_secondary+0x48/0x180
 set_init_arg+0x55/0x55
 early_idt_handler_array+0x120/0x120
 x86_64_start_reservations+0x2a/0x2c
 x86_64_start_kernel+0x13d/0x14c

During the cpu bringup, wakeup_cpu_via_init_nmi() is called and issues an
NMI on CPU 0.  The GHES NMI handler, ghes_notify_nmi() runs the
ghes_proc_irq_work work queue which ends up setting IRQ_WORK_VECTOR
(0xf6).  The "faulty" IR line set at arch/x86/kernel/apic/apic.c:1349 is  also
0xf6 (specifically APIC IRR for irqs 255 to 224 is 0x400000) which confirms
that something has set the IRQ_WORK_VECTOR line prior to the APIC being
initialized.

Commit 2383844d4850 ("GHES: Elliminate double-loop in the NMI handler")
incorrectly modified the behavior such that the handler returns
NMI_HANDLED only if an error was processed, and incorrectly runs the ghes
work queue for every NMI.

This patch modifies the ghes_proc_irq_work() to run as it did prior to
2383844d4850 ("GHES: Elliminate double-loop in the NMI handler") by
properly returning NMI_HANDLED and only calling the work queue if
NMI_HANDLED has been set.

Fixes: 2383844d4850 (GHES: Elliminate double-loop in the NMI handler)
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Reviewed-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

block: cfq_cpd_alloc() should use @gfp

BugLink: http://bugs.launchpad.net/bugs/1658091
commit ebc4ff661fbe76781c6b16dfb7b754a5d5073f8e upstream.

cfq_cpd_alloc() which is the cpd_alloc_fn implementation for cfq was
incorrectly hard coding GFP_KERNEL instead of using the mask specified
through the @gfp parameter.  This currently doesn't cause any actual
issues because all current callers specify GFP_KERNEL.  Fix it.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: e4a9bde9589f ("blkcg: replace blkcg_policy->cpd_size with ->cpd_alloc/free_fn() methods")
Signed-off-by: Jens Axboe <axboe@fb.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

cpufreq: powernv: Disable preemption while checking CPU throttling state

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 8a10c06a20ec8097a68fd7a4a1c0e285095b4d2f upstream.

With preemption turned on we can read incorrect throttling state
while being switched to CPU on a different chip.

 BUG: using smp_processor_id() in preemptible [00000000] code: cat/7343
 caller is .powernv_cpufreq_throttle_check+0x2c/0x710
 CPU: 13 PID: 7343 Comm: cat Not tainted 4.8.0-rc5-dirty #1
 Call Trace:
 [c0000007d25b75b0] [c000000000971378] .dump_stack+0xe4/0x150 (unreliable)
 [c0000007d25b7640] [c0000000005162e4] .check_preemption_disabled+0x134/0x150
 [c0000007d25b76e0] [c0000000007b63ac] .powernv_cpufreq_throttle_check+0x2c/0x710
 [c0000007d25b7790] [c0000000007b6d18] .powernv_cpufreq_target_index+0x288/0x360
 [c0000007d25b7870] [c0000000007acee4] .__cpufreq_driver_target+0x394/0x8c0
 [c0000007d25b7920] [c0000000007b22ac] .cpufreq_set+0x7c/0xd0
 [c0000007d25b79b0] [c0000000007adf50] .store_scaling_setspeed+0x80/0xc0
 [c0000007d25b7a40] [c0000000007ae270] .store+0xa0/0x100
 [c0000007d25b7ae0] [c0000000003566e8] .sysfs_kf_write+0x88/0xb0
 [c0000007d25b7b70] [c0000000003553b8] .kernfs_fop_write+0x178/0x260
 [c0000007d25b7c10] [c0000000002ac3cc] .__vfs_write+0x3c/0x1c0
 [c0000007d25b7cf0] [c0000000002ad584] .vfs_write+0xc4/0x230
 [c0000007d25b7d90] [c0000000002aeef8] .SyS_write+0x58/0x100
 [c0000007d25b7e30] [c00000000000bfec] system_call+0x38/0xfc

Fixes: 09a972d16209 (cpufreq: powernv: Report cpu frequency throttling)
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Denis Kirjanov <kda@linux-powerpc.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

NFSv4.1: nfs4_fl_prepare_ds must be careful about reporting success.

BugLink: http://bugs.launchpad.net/bugs/1658091
commit cfd278c280f997cf2fe4662e0acab0fe465f637b upstream.

Various places assume that if nfs4_fl_prepare_ds() turns a non-NULL 'ds',
then ds->ds_clp will also be non-NULL.

This is not necessasrily true in the case when the process received a fatal signal
while nfs4_pnfs_ds_connect is waiting in nfs4_wait_ds_connect().
In that case ->ds_clp may not be set, and the devid may not recently have been marked
unavailable.

So add a test for ds_clp == NULL and return NULL in that case.

Fixes: c23266d532b4 ("NFS4.1 Fix data server connection race")
Signed-off-by: NeilBrown <neilb@suse.com>
Acked-by: Olga Kornievskaia <aglo@umich.edu>
Acked-by: Adamson, Andy <William.Adamson@netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

NFS: Fix a performance regression in readdir

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 79f687a3de9e3ba2518b4ea33f38ca6cbe9133eb upstream.

Ben Coddington reports that commit 311324ad1713, by adding the function
nfs_dir_mapping_need_revalidate() that checks page cache validity on
each call to nfs_readdir() causes a performance regression when
the directory is being modified.

If the directory is changing while we're iterating through the directory,
POSIX does not require us to invalidate the page cache unless the user
calls rewinddir(). However, we still do want to ensure that we use
readdirplus in order to avoid a load of stat() calls when the user
is doing an 'ls -l' workload.

The fix should be to invalidate the page cache immediately when we're
setting the NFS_INO_ADVISE_RDPLUS bit.

Reported-by: Benjamin Coddington <bcodding@redhat.com>
Fixes: 311324ad1713 ("NFS: Be more aggressive in using readdirplus...")
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Tested-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

pNFS: Fix race in pnfs_wait_on_layoutreturn

BugLink: http://bugs.launchpad.net/bugs/1658091
commit ee284e35d8c71bf5d4d807eaff6f67a17134b359 upstream.

We must put the task to sleep while holding the inode->i_lock in order
to ensure atomicity with the test for NFS_LAYOUT_RETURN.

Fixes: 500d701f336b ("NFS41: make close wait for layoutreturn")
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

pinctrl: meson: fix gpio request disabling other modes

BugLink: http://bugs.launchpad.net/bugs/1658091
commit f24d311f92b516a8aadef5056424ccabb4068e7b upstream.

The pinctrl_gpio_request is called with the "full" gpio number, already
containing the base, then meson_pmx_request_gpio is then called with the
final pin number.
Remove the base addition when calling meson_pmx_disable_other_groups.

Fixes: 6ac730951104 ("pinctrl: add driver for Amlogic Meson SoCs")
CC: Beniamino Galvani <b.galvani@gmail.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Kevin Hilman <khilman@baylibre.com>
Acked-by: Beniamino Galvani <b.galvani@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

btrfs: fix error handling when run_delayed_extent_op fails

BugLink: http://bugs.launchpad.net/bugs/1658091
commit aa7c8da35d1905d80e840d075f07d26ec90144b5 upstream.

In __btrfs_run_delayed_refs, the error path when run_delayed_extent_op
fails sets locked_ref->processing = 0 but doesn't re-increment
delayed_refs->num_heads_ready.  As a result, we end up triggering
the WARN_ON in btrfs_select_ref_head.

Fixes: d7df2c796d7 (Btrfs: attach delayed ref updates to delayed ref heads)
Reported-by: Jon Nelson <jnelson-suse@jamponi.net>
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

btrfs: fix locking when we put back a delayed ref that's too new

BugLink: http://bugs.launchpad.net/bugs/1658091
commit d0280996437081dd12ed1e982ac8aeaa62835ec4 upstream.

In __btrfs_run_delayed_refs, when we put back a delayed ref that's too
new, we have already dropped the lock on locked_ref when we set
->processing = 0.

This patch keeps the lock to cover that assignment.

Fixes: d7df2c796d7 (Btrfs: attach delayed ref updates to delayed ref heads)
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

x86/cpu: Fix bootup crashes by sanitizing the argument of the 'clearcpuid=' command-line option

BugLink: http://bugs.launchpad.net/bugs/1658091
commit dd853fd216d1485ed3045ff772079cc8689a9a4a upstream.

A negative number can be specified in the cmdline which will be used as
setup_clear_cpu_cap() argument. With that we can clear/set some bit in
memory predceeding boot_cpu_data/cpu_caps_cleared which may cause kernel
to misbehave. This patch adds lower bound check to setup_disablecpuid().

Boris Petkov reproduced a crash:

  [    1.234575] BUG: unable to handle kernel paging request at ffffffff858bd540
  [    1.236535] IP: memcpy_erms+0x6/0x10

Signed-off-by: Lukasz Odzioba <lukasz.odzioba@intel.com>
Acked-by: Borislav Petkov <bp@suse.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: andi.kleen@intel.com
Cc: bp@alien8.de
Cc: dave.hansen@linux.intel.com
Cc: luto@kernel.org
Cc: slaoub@gmail.com
Fixes: ac72e7888a61 ("x86: add generic clearcpuid=... option")
Link: http://lkml.kernel.org/r/1482933340-11857-1-git-send-email-lukasz.odzioba@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: ch341: fix modem-control and B0 handling

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 030ee7ae52a46a2be52ccc8242c4a330aba8d38e upstream.

The modem-control signals are managed by the tty-layer during open and
should not be asserted prematurely when set_termios is called from
driver open.

Also make sure that the signals are asserted only when changing speed
from B0.

Fixes: 664d5df92e88 ("USB: usb-serial ch341: support for DTR/RTS/CTS")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: ch341: fix resume after reset

BugLink: http://bugs.launchpad.net/bugs/1658091
commit ce5e292828117d1b71cbd3edf9e9137cf31acd30 upstream.

Fix reset-resume handling which failed to resubmit the read and
interrupt URBs, thereby leaving a port that was open before suspend in a
broken state until closed and reopened.

Fixes: 1ded7ea47b88 ("USB: ch341 serial: fix port number changed after
resume")
Fixes: 2bfd1c96a9fb ("USB: serial: ch341: remove reset_resume callback")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

drm/radeon: drop verde dpm quirks

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 8a08403bcb39f5d0e733bcf59a8a74f16b538f6e upstream.

fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=98897
https://bugs.launchpad.net/bugs/1651981

Acked-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Adrian Fiergolski <A.Fiergolski@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

sysctl: Drop reference added by grab_header in proc_sys_readdir

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 93362fa47fe98b62e4a34ab408c4a418432e7939 upstream.

Fixes CVE-2016-9191, proc_sys_readdir doesn't drop reference
added by grab_header when return from !dir_emit_dots path.
It can cause any path called unregister_sysctl_table will
wait forever.

The calltrace of CVE-2016-9191:

[ 5535.960522] Call Trace:
[ 5535.963265]  [<ffffffff817cdaaf>] schedule+0x3f/0xa0
[ 5535.968817]  [<ffffffff817d33fb>] schedule_timeout+0x3db/0x6f0
[ 5535.975346]  [<ffffffff817cf055>] ? wait_for_completion+0x45/0x130
[ 5535.982256]  [<ffffffff817cf0d3>] wait_for_completion+0xc3/0x130
[ 5535.988972]  [<ffffffff810d1fd0>] ? wake_up_q+0x80/0x80
[ 5535.994804]  [<ffffffff8130de64>] drop_sysctl_table+0xc4/0xe0
[ 5536.001227]  [<ffffffff8130de17>] drop_sysctl_table+0x77/0xe0
[ 5536.007648]  [<ffffffff8130decd>] unregister_sysctl_table+0x4d/0xa0
[ 5536.014654]  [<ffffffff8130deff>] unregister_sysctl_table+0x7f/0xa0
[ 5536.021657]  [<ffffffff810f57f5>] unregister_sched_domain_sysctl+0x15/0x40
[ 5536.029344]  [<ffffffff810d7704>] partition_sched_domains+0x44/0x450
[ 5536.036447]  [<ffffffff817d0761>] ? __mutex_unlock_slowpath+0x111/0x1f0
[ 5536.043844]  [<ffffffff81167684>] rebuild_sched_domains_locked+0x64/0xb0
[ 5536.051336]  [<ffffffff8116789d>] update_flag+0x11d/0x210
[ 5536.057373]  [<ffffffff817cf61f>] ? mutex_lock_nested+0x2df/0x450
[ 5536.064186]  [<ffffffff81167acb>] ? cpuset_css_offline+0x1b/0x60
[ 5536.070899]  [<ffffffff810fce3d>] ? trace_hardirqs_on+0xd/0x10
[ 5536.077420]  [<ffffffff817cf61f>] ? mutex_lock_nested+0x2df/0x450
[ 5536.084234]  [<ffffffff8115a9f5>] ? css_killed_work_fn+0x25/0x220
[ 5536.091049]  [<ffffffff81167ae5>] cpuset_css_offline+0x35/0x60
[ 5536.097571]  [<ffffffff8115aa2c>] css_killed_work_fn+0x5c/0x220
[ 5536.104207]  [<ffffffff810bc83f>] process_one_work+0x1df/0x710
[ 5536.110736]  [<ffffffff810bc7c0>] ? process_one_work+0x160/0x710
[ 5536.117461]  [<ffffffff810bce9b>] worker_thread+0x12b/0x4a0
[ 5536.123697]  [<ffffffff810bcd70>] ? process_one_work+0x710/0x710
[ 5536.130426]  [<ffffffff810c3f7e>] kthread+0xfe/0x120
[ 5536.135991]  [<ffffffff817d4baf>] ret_from_fork+0x1f/0x40
[ 5536.142041]  [<ffffffff810c3e80>] ? kthread_create_on_node+0x230/0x230

One cgroup maintainer mentioned that "cgroup is trying to offline
a cpuset css, which takes place under cgroup_mutex.  The offlining
ends up trying to drain active usages of a sysctl table which apprently
is not happening."
The real reason is that proc_sys_readdir doesn't drop reference added
by grab_header when return from !dir_emit_dots path. So this cpuset
offline path will wait here forever.

See here for details: http://www.openwall.com/lists/oss-security/2016/11/04/13

Fixes: f0c3b5093add ("[readdir] convert procfs")
Reported-by: CAI Qian <caiqian@redhat.com>
Tested-by: Yang Shukui <yangshukui@huawei.com>
Signed-off-by: Zhou Chengming <zhouchengming1@huawei.com>
Acked-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

sysrq: attach sysrq handler correctly for 32-bit kernel

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 802c03881f29844af0252b6e22be5d2f65f93fd0 upstream.

The sysrq input handler should be attached to the input device which has
a left alt key.

On 32-bit kernels, some input devices which has a left alt key cannot
attach sysrq handler.  Because the keybit bitmap in struct input_device_id
for sysrq is not correctly initialized.  KEY_LEFTALT is 56 which is
greater than BITS_PER_LONG on 32-bit kernels.

I found this problem when using a matrix keypad device which defines
a KEY_LEFTALT (56) but doesn't have a KEY_O (24 == 56%32).

Cc: Jiri Slaby <jslaby@suse.com>
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

tty/serial: atmel_serial: BUG: stop DMA from transmitting in stop_tx

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 89d8232411a85b9a6b12fd5da4d07d8a138a8e0c upstream.

If we don't disable the transmitter in atmel_stop_tx, the DMA buffer
continues to send data until it is emptied.
This cause problems with the flow control (CTS is asserted and data are
still sent).

So, disabling the transmitter in atmel_stop_tx is a sane thing to do.

Tested on at91sam9g35-cm(DMA)
Tested for regressions on sama5d2-xplained(Fifo) and at91sam9g20ek(PDC)

Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

mnt: Protect the mountpoint hashtable with mount_lock

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 3895dbf8985f656675b5bde610723a29cbce3fa7 upstream.

Protecting the mountpoint hashtable with namespace_sem was sufficient
until a call to umount_mnt was added to mntput_no_expire.  At which
point it became possible for multiple calls of put_mountpoint on
the same hash chain to happen on the same time.

Kristen Johansen <kjlx@templeofstupid.com> reported:
> This can cause a panic when simultaneous callers of put_mountpoint
> attempt to free the same mountpoint.  This occurs because some callers
> hold the mount_hash_lock, while others hold the namespace lock.  Some
> even hold both.
>
> In this submitter's case, the panic manifested itself as a GP fault in
> put_mountpoint() when it called hlist_del() and attempted to dereference
> a m_hash.pprev that had been poisioned by another thread.

Al Viro observed that the simple fix is to switch from using the namespace_sem
to the mount_lock to protect the mountpoint hash table.

I have taken Al's suggested patch moved put_mountpoint in pivot_root
(instead of taking mount_lock an additional time), and have replaced
new_mountpoint with get_mountpoint a function that does the hash table
lookup and addition under the mount_lock.   The introduction of get_mounptoint
ensures that only the mount_lock is needed to manipulate the mountpoint
hashtable.

d_set_mounted is modified to only set DCACHE_MOUNTED if it is not
already set.  This allows get_mountpoint to use the setting of
DCACHE_MOUNTED to ensure adding a struct mountpoint for a dentry
happens exactly once.

Fixes: ce07d891a089 ("mnt: Honor MNT_LOCKED when detaching mounts")
Reported-by: Krister Johansen <kjlx@templeofstupid.com>
Suggested-by: Al Viro <viro@ZenIV.linux.org.uk>
Acked-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

vme: Fix wrong pointer utilization in ca91cx42_slave_get

BugLink: http://bugs.launchpad.net/bugs/1658091
commit c8a6a09c1c617402cc9254b2bc8da359a0347d75 upstream.

In ca91cx42_slave_get function, the value pointed by vme_base pointer is
set through:

*vme_base = ioread32(bridge->base + CA91CX42_VSI_BS[i]);

So it must be dereferenced to be used in calculation of pci_base:

*pci_base = (dma_addr_t)*vme_base + pci_offset;

This bug was caught thanks to the following gcc warning:

drivers/vme/bridges/vme_ca91cx42.c: In function ‘ca91cx42_slave_get’:
drivers/vme/bridges/vme_ca91cx42.c:467:14: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
*pci_base = (dma_addr_t)vme_base + pci_offset;

Signed-off-by: Augusto Mecking Caringi <augustocaringi@gmail.com>
Acked-By: Martyn Welch <martyn@welchs.me.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

xhci: fix deadlock at host remove by running watchdog correctly

BugLink: http://bugs.launchpad.net/bugs/1658091
commit d6169d04097fd9ddf811e63eae4e5cd71e6666e2 upstream.

If a URB is killed while the host is removed we can end up in a situation
where the hub thread takes the roothub device lock, and waits for
the URB to be given back by xhci-hcd, blocking the host remove code.

xhci-hcd tries to stop the endpoint and give back the urb, but can't
as the host is removed from PCI bus at the same time, preventing the normal
way of giving back urb.

Instead we need to rely on the stop command timeout function to give back
the urb. This xhci_stop_endpoint_command_watchdog() timeout function
used a XHCI_STATE_DYING flag to indicate if the timeout function is already
running, but later this flag has been taking into use in other places to
mark that xhci is dying.

Remove checks for XHCI_STATE_DYING in xhci_urb_dequeue. We are still
checking that reading from pci state does not return 0xffffffff or that
host is not halted before trying to stop the endpoint.

This whole area of stopping endpoints, giving back URBs, and the wathdog
timeout need rework, this fix focuses on solving a specific deadlock
issue that we can then send to stable before any major rework.

Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

i2c: fix kernel memory disclosure in dev interface

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 30f939feaeee23e21391cfc7b484f012eb189c3c upstream.

i2c_smbus_xfer() does not always fill an entire block, allowing
kernel stack memory disclosure through the temp variable. Clear
it before it's read to.

Signed-off-by: Vlad Tsyrklevich <vlad@tsyrklevich.net>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

i2c: print correct device invalid address

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 6f724fb3039522486fce2e32e4c0fbe238a6ab02 upstream.

In of_i2c_register_device(), when the check for
device address validity fails we print the info.addr,
which has not been assigned properly.

Fix this by printing the actual invalid address.

Signed-off-by: John Garry <john.garry@huawei.com>
Reviewed-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Fixes: b4e2f6ac1281 ("i2c: apply DT flags when probing")
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Input: elants_i2c - avoid divide by 0 errors on bad touchscreen data

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 1c3415a06b1016a596bfe59e0cfee56c773aa958 upstream.

The following crash may be seen if bad data is received from the
touchscreen.

[ 2189.425150] elants_i2c i2c-ELAN0001:00: unknown packet ff ff ff ff
[ 2189.430738] divide error: 0000 [#1] PREEMPT SMP
[ 2189.434679] gsmi: Log Shutdown Reason 0x03
[ 2189.434689] Modules linked in: ip6t_REJECT nf_reject_ipv6 rfcomm evdi
uinput uvcvideo cmac videobuf2_vmalloc videobuf2_memops snd_hda_codec_hdmi
i2c_dev videobuf2_core snd_soc_sst_cht_bsw_rt5645 snd_hda_intel
snd_intel_sst_acpi btusb btrtl btbcm btintel bluetooth snd_soc_sst_acpi
snd_hda_codec snd_intel_sst_core snd_hwdep snd_soc_sst_mfld_platform
snd_hda_core snd_soc_rt5645 memconsole_x86_legacy memconsole zram snd_soc_rl6231
fuse ip6table_filter iwlmvm iwlwifi iwl7000_mac80211 cfg80211 iio_trig_sysfs
joydev cros_ec_sensors cros_ec_sensors_core industrialio_triggered_buffer
kfifo_buf industrialio snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq
snd_seq_device ppp_async ppp_generic slhc tun
[ 2189.434866] CPU: 0 PID: 106 Comm: irq/184-ELAN000 Tainted: G        W
3.18.0-13101-g57e8190 #1
[ 2189.434883] Hardware name: GOOGLE Ultima, BIOS Google_Ultima.7287.131.43 07/20/2016
[ 2189.434898] task: ffff88017a0b6d80 ti: ffff88017a2bc000 task.ti: ffff88017a2bc000
[ 2189.434913] RIP: 0010:[<ffffffffbecc48d5>]  [<ffffffffbecc48d5>] elants_i2c_irq+0x190/0x200
[ 2189.434937] RSP: 0018:ffff88017a2bfd98  EFLAGS: 00010293
[ 2189.434948] RAX: 0000000000000000 RBX: ffff88017a967828 RCX: ffff88017a9678e8
[ 2189.434962] RDX: 0000000000000000 RSI: 0000000000000246 RDI: 0000000000000000
[ 2189.434975] RBP: ffff88017a2bfdd8 R08: 00000000000003e8 R09: 0000000000000000
[ 2189.434989] R10: 0000000000000000 R11: 000000000044a2bd R12: ffff88017a991800
[ 2189.435001] R13: ffffffffbe8a2a53 R14: ffff88017a0b6d80 R15: ffff88017a0b6d80
[ 2189.435011] FS:  0000000000000000(0000) GS:ffff88017fc00000(0000) knlGS:0000000000000000
[ 2189.435022] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 2189.435030] CR2: 00007f678d94b000 CR3: 000000003f41a000 CR4: 00000000001007f0
[ 2189.435039] Stack:
[ 2189.435044]  ffff88017a2bfda8 ffff88017a9678e8 646464647a2bfdd8 0000000006e09574
[ 2189.435060]  0000000000000000 ffff88017a088b80 ffff88017a921000 ffffffffbe8a2a53
[ 2189.435074]  ffff88017a2bfe08 ffffffffbe8a2a73 ffff88017a0b6d80 0000000006e09574
[ 2189.435089] Call Trace:
[ 2189.435101]  [<ffffffffbe8a2a53>] ? irq_thread_dtor+0xa9/0xa9
[ 2189.435112]  [<ffffffffbe8a2a73>] irq_thread_fn+0x20/0x40
[ 2189.435123]  [<ffffffffbe8a2be1>] irq_thread+0x14e/0x222
[ 2189.435135]  [<ffffffffbee8cbeb>] ? __schedule+0x3b3/0x57a
[ 2189.435145]  [<ffffffffbe8a29aa>] ? wake_threads_waitq+0x2d/0x2d
[ 2189.435156]  [<ffffffffbe8a2a93>] ? irq_thread_fn+0x40/0x40
[ 2189.435168]  [<ffffffffbe87c385>] kthread+0x10e/0x116
[ 2189.435178]  [<ffffffffbe87c277>] ? __kthread_parkme+0x67/0x67
[ 2189.435189]  [<ffffffffbee900ac>] ret_from_fork+0x7c/0xb0
[ 2189.435199]  [<ffffffffbe87c277>] ? __kthread_parkme+0x67/0x67
[ 2189.435208] Code: ff ff eb 73 0f b6 bb c1 00 00 00 83 ff 03 7e 13 49 8d 7c
24 20 ba 04 00 00 00 48 c7 c6 8a cd 21 bf eb 4d 0f b6 83 c2 00 00 00 99 <f7> ff
83 f8 37 75 15 48 6b f7 37 4c 8d a3 c4 00 00 00 4c 8d ac
[ 2189.435312] RIP  [<ffffffffbecc48d5>] elants_i2c_irq+0x190/0x200
[ 2189.435323]  RSP <ffff88017a2bfd98>
[ 2189.435350] ---[ end trace f4945345a75d96dd ]---
[ 2189.443841] Kernel panic - not syncing: Fatal exception
[ 2189.444307] Kernel Offset: 0x3d800000 from 0xffffffff81000000
(relocation range: 0xffffffff80000000-0xffffffffbfffffff)
[ 2189.444519] gsmi: Log Shutdown Reason 0x02

The problem was seen with a 3.18 based kernel, but there is no reason
to believe that the upstream code is safe.

Fixes: 66aee90088da2 ("Input: add support for Elan eKTH I2C touchscreens")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: ch341: fix open and resume after B0

BugLink: http://bugs.launchpad.net/bugs/1658091
commit a20047f36e2f6a1eea4f1fd261aaa55882369868 upstream.

The private baud_rate variable is used to configure the port at open and
reset-resume and must never be set to (and left at) zero or reset-resume
and all further open attempts will fail.

Fixes: aa91def41a7b ("USB: ch341: set tty baud speed according to tty struct")
Fixes: 664d5df92e88 ("USB: usb-serial ch341: support for DTR/RTS/CTS")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: ch341: fix control-message error handling

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 2d5a9c72d0c4ac73cf97f4b7814ed6c44b1e49ae upstream.

A short control transfer would currently fail to be detected, something
which could lead to stale buffer data being used as valid input.

Check for short transfers, and make sure to log any transfer errors.

Note that this also avoids leaking heap data to user space (TIOCMGET)
and the remote device (break control).

Fixes: 6ce76104781a ("USB: Driver for CH341 USB-serial adaptor")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: ch341: fix open error handling

BugLink: http://bugs.launchpad.net/bugs/1658091
commit f2950b78547ffb8475297ada6b92bc2d774d5461 upstream.

Make sure to stop the interrupt URB before returning on errors during
open.

Fixes: 664d5df92e88 ("USB: usb-serial ch341: support for DTR/RTS/CTS")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: ch341: fix initial modem-control state

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 4e2da44691cffbfffb1535f478d19bc2dca3e62b upstream.

DTR and RTS will be asserted by the tty-layer when the port is opened
and deasserted on close (if HUPCL is set). Make sure the initial state
is not-asserted before the port is first opened as well.

Fixes: 664d5df92e88 ("USB: usb-serial ch341: support for DTR/RTS/CTS")
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

USB: serial: kl5kusb105: fix line-state error handling

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 146cc8a17a3b4996f6805ee5c080e7101277c410 upstream.

The current implementation failed to detect short transfers when
attempting to read the line state, and also, to make things worse,
logged the content of the uninitialised heap transfer buffer.

Fixes: abf492e7b3ae ("USB: kl5kusb105: fix DMA buffers on stack")
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

nl80211: fix sched scan netlink socket owner destruction

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 753aacfd2e95df6a0caf23c03dc309020765bea9 upstream.

A single netlink socket might own multiple interfaces *and* a
scheduled scan request (which might belong to another interface),
so when it goes away both may need to be destroyed.

Remove the schedule_scan_stop indirection to fix this - it's only
needed for interface destruction because of the way this works
right now, with a single work taking care of all interfaces.

Fixes: 93a1e86ce10e4 ("nl80211: Stop scheduled scan if netlink client disappears")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: x86: Introduce segmented_write_std

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 129a72a0d3c8e139a04512325384fe5ac119e74d upstream.

Introduces segemented_write_std.

Switches from emulated reads/writes to standard read/writes in fxsave,
fxrstor, sgdt, and sidt.  This fixes CVE-2017-2584, a longstanding
kernel memory leak.

Since commit 283c95d0e389 ("KVM: x86: emulate FXSAVE and FXRSTOR",
2016-11-09), which is luckily not yet in any final release, this would
also be an exploitable kernel memory *write*!

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Fixes: 96051572c819194c37a8367624b285be10297eca
Fixes: 283c95d0e3891b64087706b344a4b545d04a6e62
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Steve Rutherford <srutherford@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: x86: emulate FXSAVE and FXRSTOR

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 283c95d0e3891b64087706b344a4b545d04a6e62 upstream.

Internal errors were reported on 16 bit fxsave and fxrstor with ipxe.
Old Intels don't have unrestricted_guest, so we have to emulate them.

The patch takes advantage of the hardware implementation.

AMD and Intel differ in saving and restoring other fields in first 32
bytes.  A test wrote 0xff to the fxsave area, 0 to upper bits of MCSXR
in the fxsave area, executed fxrstor, rewrote the fxsave area to 0xee,
and executed fxsave:

  Intel (Nehalem):
    7f 1f 7f 7f ff 00 ff 07 ff ff ff ff ff ff 00 00
    ff ff ff ff ff ff 00 00 ff ff 00 00 ff ff 00 00
  Intel (Haswell -- deprecated FPU CS and FPU DS):
    7f 1f 7f 7f ff 00 ff 07 ff ff ff ff 00 00 00 00
    ff ff ff ff 00 00 00 00 ff ff 00 00 ff ff 00 00
  AMD (Opteron 2300-series):
    7f 1f 7f 7f ff 00 ee ee ee ee ee ee ee ee ee ee
    ee ee ee ee ee ee ee ee ff ff 00 00 ff ff 02 00

fxsave/fxrstor will only be emulated on early Intels, so KVM can't do
much to improve the situation.

Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: x86: add asm_safe wrapper

BugLink: http://bugs.launchpad.net/bugs/1658091
commit aabba3c6abd50b05b1fc2c6ec44244aa6bcda576 upstream.

Move the existing exception handling for inline assembly into a macro
and switch its return values to X86EMUL type.

Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: x86: add Align16 instruction flag

BugLink: http://bugs.launchpad.net/bugs/1658091
commit d3fe959f81024072068e9ed86b39c2acfd7462a9 upstream.

Needed for FXSAVE and FXRSTOR.

Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: x86: flush pending lapic jump label updates on module unload

BugLink: http://bugs.launchpad.net/bugs/1658091
commit cef84c302fe051744b983a92764d3fcca933415d upstream.

KVM's lapic emulation uses static_key_deferred (apic_{hw,sw}_disabled).
These are implemented with delayed_work structs which can still be
pending when the KVM module is unloaded. We've seen this cause kernel
panics when the kvm_intel module is quickly reloaded.

Use the new static_key_deferred_flush() API to flush pending updates on
module unload.

Signed-off-by: David Matlack <dmatlack@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

jump_labels: API for flushing deferred jump label updates

BugLink: http://bugs.launchpad.net/bugs/1658091
commit b6416e61012429e0277bd15a229222fd17afc1c1 upstream.

Modules that use static_key_deferred need a way to synchronize with
any delayed work that is still pending when the module is unloaded.
Introduce static_key_deferred_flush() which flushes any pending
jump label updates.

Signed-off-by: David Matlack <dmatlack@google.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: eventfd: fix NULL deref irqbypass consumer

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 4f3dbdf47e150016aacd734e663347fcaa768303 upstream.

Reported syzkaller:

    BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
    IP: irq_bypass_unregister_consumer+0x9d/0xb70 [irqbypass]
    PGD 0

    Oops: 0002 [#1] SMP
    CPU: 1 PID: 125 Comm: kworker/1:1 Not tainted 4.9.0+ #1
    Workqueue: kvm-irqfd-cleanup irqfd_shutdown [kvm]
    task: ffff9bbe0dfbb900 task.stack: ffffb61802014000
    RIP: 0010:irq_bypass_unregister_consumer+0x9d/0xb70 [irqbypass]
    Call Trace:
     irqfd_shutdown+0x66/0xa0 [kvm]
     process_one_work+0x16b/0x480
     worker_thread+0x4b/0x500
     kthread+0x101/0x140
     ? process_one_work+0x480/0x480
     ? kthread_create_on_node+0x60/0x60
     ret_from_fork+0x25/0x30
    RIP: irq_bypass_unregister_consumer+0x9d/0xb70 [irqbypass] RSP: ffffb61802017e20
    CR2: 0000000000000008

The syzkaller folks reported a NULL pointer dereference that due to
unregister an consumer which fails registration before. The syzkaller
creates two VMs w/ an equal eventfd occasionally. So the second VM
fails to register an irqbypass consumer. It will make irqfd as inactive
and queue an workqueue work to shutdown irqfd and unregister the irqbypass
consumer when eventfd is closed. However, the second consumer has been
initialized though it fails registration. So the token(same as the first
VM's) is taken to unregister the consumer through the workqueue, the
consumer of the first VM is found and unregistered, then NULL deref incurred
in the path of deleting consumer from the consumers list.

This patch fixes it by making irq_bypass_register/unregister_consumer()
looks for the consumer entry based on consumer pointer itself instead of
token matching.

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

KVM: x86: fix emulation of "MOV SS, null selector"

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 33ab91103b3415e12457e3104f0e4517ce12d0f3 upstream.

This is CVE-2017-2583.  On Intel this causes a failed vmentry because
SS's type is neither 3 nor 7 (even though the manual says this check is
only done for usable SS, and the dmesg splat says that SS is unusable!).
On AMD it's worse: svm.c is confused and sets CPL to 0 in the vmcb.

The fix fabricates a data segment descriptor when SS is set to a null
selector, so that CPL and SS.DPL are set correctly in the VMCS/vmcb.
Furthermore, only allow setting SS to a NULL selector if SS.RPL < 3;
this in turn ensures CPL < 3 because RPL must be equal to CPL.

Thanks to Andy Lutomirski and Willy Tarreau for help in analyzing
the bug and deciphering the manuals.

Reported-by: Xiaohan Zhang <zhangxiaohan1@huawei.com>
Fixes: 79d5b4c3cd809c770d4bf9812635647016c56011
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

mm/hugetlb.c: fix reservation race when freeing surplus pages

BugLink: http://bugs.launchpad.net/bugs/1658091
commit e5bbc8a6c992901058bc09e2ce01d16c111ff047 upstream.

return_unused_surplus_pages() decrements the global reservation count,
and frees any unused surplus pages that were backing the reservation.

Commit 7848a4bf51b3 ("mm/hugetlb.c: add cond_resched_lock() in
return_unused_surplus_pages()") added a call to cond_resched_lock in the
loop freeing the pages.

As a result, the hugetlb_lock could be dropped, and someone else could
use the pages that will be freed in subsequent iterations of the loop.
This could result in inconsistent global hugetlb page state, application
api failures (such as mmap) failures or application crashes.

When dropping the lock in return_unused_surplus_pages, make sure that
the global reservation count (resv_huge_pages) remains sufficiently
large to prevent someone else from claiming pages about to be freed.

Analyzed by Paul Cassella.

Fixes: 7848a4bf51b3 ("mm/hugetlb.c: add cond_resched_lock() in return_unused_surplus_pages()")
Link: http://lkml.kernel.org/r/1483991767-6879-1-git-send-email-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reported-by: Paul Cassella <cassella@cray.com>
Suggested-by: Michal Hocko <mhocko@kernel.org>
Cc: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

ocfs2: fix crash caused by stale lvb with fsdlm plugin

BugLink: http://bugs.launchpad.net/bugs/1658091
commit e7ee2c089e94067d68475990bdeed211c8852917 upstream.

The crash happens rather often when we reset some cluster nodes while
nodes contend fiercely to do truncate and append.

The crash backtrace is below:

   dlm: C21CBDA5E0774F4BA5A9D4F317717495: dlm_recover_grant 1 locks on 971 resources
   dlm: C21CBDA5E0774F4BA5A9D4F317717495: dlm_recover 9 generation 5 done: 4 ms
   ocfs2: Begin replay journal (node 318952601, slot 2) on device (253,18)
   ocfs2: End replay journal (node 318952601, slot 2) on device (253,18)
   ocfs2: Beginning quota recovery on device (253,18) for slot 2
   ocfs2: Finishing quota recovery on device (253,18) for slot 2
   (truncate,30154,1):ocfs2_truncate_file:470 ERROR: bug expression: le64_to_cpu(fe->i_size) != i_size_read(inode)
   (truncate,30154,1):ocfs2_truncate_file:470 ERROR: Inode 290321, inode i_size = 732 != di i_size = 937, i_flags = 0x1
   ------------[ cut here ]------------
   kernel BUG at /usr/src/linux/fs/ocfs2/file.c:470!
   invalid opcode: 0000 [#1] SMP
   Modules linked in: ocfs2_stack_user(OEN) ocfs2(OEN) ocfs2_nodemanager ocfs2_stackglue(OEN) quota_tree dlm(OEN) configfs fuse sd_mod    iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi af_packet iscsi_ibft iscsi_boot_sysfs softdog xfs libcrc32c ppdev parport_pc pcspkr parport      joydev virtio_balloon virtio_net i2c_piix4 acpi_cpufreq button processor ext4 crc16 jbd2 mbcache ata_generic cirrus virtio_blk ata_piix               drm_kms_helper ahci syscopyarea libahci sysfillrect sysimgblt fb_sys_fops ttm floppy libata drm virtio_pci virtio_ring uhci_hcd virtio ehci_hcd       usbcore serio_raw usb_common sg dm_multipath dm_mod scsi_dh_rdac scsi_dh_emc scsi_dh_alua scsi_mod autofs4
   Supported: No, Unsupported modules are loaded
   CPU: 1 PID: 30154 Comm: truncate Tainted: G           OE   N  4.4.21-69-default #1
   Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.1-0-g4adadbd-20151112_172657-sheep25 04/01/2014
   task: ffff88004ff6d240 ti: ffff880074e68000 task.ti: ffff880074e68000
   RIP: 0010:[<ffffffffa05c8c30>]  [<ffffffffa05c8c30>] ocfs2_truncate_file+0x640/0x6c0 [ocfs2]
   RSP: 0018:ffff880074e6bd50  EFLAGS: 00010282
   RAX: 0000000000000074 RBX: 000000000000029e RCX: 0000000000000000
   RDX: 0000000000000001 RSI: 0000000000000246 RDI: 0000000000000246
   RBP: ffff880074e6bda8 R08: 000000003675dc7a R09: ffffffff82013414
   R10: 0000000000034c50 R11: 0000000000000000 R12: ffff88003aab3448
   R13: 00000000000002dc R14: 0000000000046e11 R15: 0000000000000020
   FS:  00007f839f965700(0000) GS:ffff88007fc80000(0000) knlGS:0000000000000000
   CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
   CR2: 00007f839f97e000 CR3: 0000000036723000 CR4: 00000000000006e0
   Call Trace:
     ocfs2_setattr+0x698/0xa90 [ocfs2]
     notify_change+0x1ae/0x380
     do_truncate+0x5e/0x90
     do_sys_ftruncate.constprop.11+0x108/0x160
     entry_SYSCALL_64_fastpath+0x12/0x6d
   Code: 24 28 ba d6 01 00 00 48 c7 c6 30 43 62 a0 8b 41 2c 89 44 24 08 48 8b 41 20 48 c7 c1 78 a3 62 a0 48 89 04 24 31 c0 e8 a0 97 f9 ff <0f> 0b 3d 00 fe ff ff 0f 84 ab fd ff ff 83 f8 fc 0f 84 a2 fd ff
   RIP  [<ffffffffa05c8c30>] ocfs2_truncate_file+0x640/0x6c0 [ocfs2]

It's because ocfs2_inode_lock() get us stale LVB in which the i_size is
not equal to the disk i_size.  We mistakenly trust the LVB because the
underlaying fsdlm dlm_lock() doesn't set lkb_sbflags with
DLM_SBF_VALNOTVALID properly for us.  But, why?

The current code tries to downconvert lock without DLM_LKF_VALBLK flag
to tell o2cb don't update RSB's LVB if it's a PR->NULL conversion, even
if the lock resource type needs LVB.  This is not the right way for
fsdlm.

The fsdlm plugin behaves different on DLM_LKF_VALBLK, it depends on
DLM_LKF_VALBLK to decide if we care about the LVB in the LKB.  If
DLM_LKF_VALBLK is not set, fsdlm will skip recovering RSB's LVB from
this lkb and set the right DLM_SBF_VALNOTVALID appropriately when node
failure happens.

The following diagram briefly illustrates how this crash happens:

RSB1 is inode metadata lock resource with LOCK_TYPE_USES_LVB;

The 1st round:

             Node1                                    Node2
RSB1: PR
                                                  RSB1(master): NULL->EX
ocfs2_downconvert_lock(PR->NULL, set_lvb==0)
  ocfs2_dlm_lock(no DLM_LKF_VALBLK)

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

dlm_lock(no DLM_LKF_VALBLK)
  convert_lock(overwrite lkb->lkb_exflags
               with no DLM_LKF_VALBLK)

RSB1: NULL                                        RSB1: EX
                                                  reset Node2
dlm_recover_rsbs()
  recover_lvb()

/* The LVB is not trustable if the node with EX fails and
 * no lock >= PR is left. We should set RSB_VALNOTVALID for RSB1.
 */

 if(!(kb_exflags & DLM_LKF_VALBLK)) /* This means we miss the chance to
           return;                   * to invalid the LVB here.
                                     */

The 2nd round:

         Node 1                                Node2
RSB1(become master from recovery)

ocfs2_setattr()
  ocfs2_inode_lock(NULL->EX)
    /* dlm_lock() return the stale lvb without setting DLM_SBF_VALNOTVALID */
    ocfs2_meta_lvb_is_trustable() return 1 /* so we don't refresh inode from disk */
  ocfs2_truncate_file()
      mlog_bug_on_msg(disk isize != i_size_read(inode))  /* crash! */

The fix is quite straightforward.  We keep to set DLM_LKF_VALBLK flag
for dlm_lock() if the lock resource type needs LVB and the fsdlm plugin
is uesed.

Link: http://lkml.kernel.org/r/1481275846-6604-1-git-send-email-zren@suse.com
Signed-off-by: Eric Ren <zren@suse.com>
Reviewed-by: Joseph Qi <jiangqi903@gmail.com>
Cc: Mark Fasheh <mfasheh@versity.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

mm: fix devm_memremap_pages crash, use mem_hotplug_{begin, done}

BugLink: http://bugs.launchpad.net/bugs/1658091
commit f931ab479dd24cf7a2c6e2df19778406892591fb upstream.

Both arch_add_memory() and arch_remove_memory() expect a single threaded
context.

For example, arch/x86/mm/init_64.c::kernel_physical_mapping_init() does
not hold any locks over this check and branch:

    if (pgd_val(*pgd)) {
     pud = (pud_t *)pgd_page_vaddr(*pgd);
     paddr_last = phys_pud_init(pud, __pa(vaddr),
        __pa(vaddr_end),
        page_size_mask);
     continue;
    }

    pud = alloc_low_page();
    paddr_last = phys_pud_init(pud, __pa(vaddr), __pa(vaddr_end),
        page_size_mask);

The result is that two threads calling devm_memremap_pages()
simultaneously can end up colliding on pgd initialization.  This leads
to crash signatures like the following where the loser of the race
initializes the wrong pgd entry:

    BUG: unable to handle kernel paging request at ffff888ebfff0000
    IP: memcpy_erms+0x6/0x10
    PGD 2f8e8fc067 PUD 0 /* <---- Invalid PUD */
    Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
    CPU: 54 PID: 3818 Comm: systemd-udevd Not tainted 4.6.7+ #13
    task: ffff882fac290040 ti: ffff882f887a4000 task.ti: ffff882f887a4000
    RIP: memcpy_erms+0x6/0x10
    [..]
    Call Trace:
      ? pmem_do_bvec+0x205/0x370 [nd_pmem]
      ? blk_queue_enter+0x3a/0x280
      pmem_rw_page+0x38/0x80 [nd_pmem]
      bdev_read_page+0x84/0xb0

Hold the standard memory hotplug mutex over calls to
arch_{add,remove}_memory().

Fixes: 41e94a851304 ("add devm_memremap_pages")
Link: http://lkml.kernel.org/r/148357647831.9498.12606007370121652979.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

selftests: do not require bash for the generated test

BugLink: http://bugs.launchpad.net/bugs/1658091
commit a2b1e8a20c992b01eeb76de00d4f534cbe9f3822 upstream.

Nothing in this minimal script seems to require bash. We often run these
tests on embedded devices where the only shell available is the busybox
ash. Use sh instead.

Signed-off-by: Rolf Eike Beer <eb@emlix.com>
Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

selftests: do not require bash to run netsocktests testcase

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 3659f98b5375d195f1870c3e508fe51e52206839 upstream.

Nothing in this minimal script seems to require bash. We often run these
tests on embedded devices where the only shell available is the busybox
ash. Use sh instead.

Signed-off-by: Rolf Eike Beer <eb@emlix.com>
Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Input: i8042 - add Pegatron touchpad to noloop table

BugLink: http://bugs.launchpad.net/bugs/1658091
commit 41c567a5d7d1a986763e58c3394782813c3bcb03 upstream.

Avoid AUX loopback in Pegatron C15B touchpad, so input subsystem is able
to recognize a Synaptics touchpad in the AUX port.

Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=93791
(Touchpad is not detected on DNS 0801480 notebook (PEGATRON C15B))

Suggested-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Input: xpad - use correct product id for x360w controllers

BugLink: http://bugs.launchpad.net/bugs/1658091
commit b6fc513da50c5dbc457a8ad6b58b046a6a68fd9d upstream.

currently the controllers get the same product id as the wireless
receiver. However the controllers actually have their own product id.

The patch makes the driver expose the same product id as the windows
driver.

This improves compatibility when running applications with WINE.

see https://github.com/paroj/xpad/issues/54

Signed-off-by: Pavel Rojtberg <rojtberg@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: [Config] CONFIG_SND_SOC_INTEL_BYTCR_RT5660_MACH=m, CONFIG_SND_SOC_RT5660=m

BugLink: http://bugs.launchpad.net/bugs/1657674
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: (no-up) ASoC: Intel: Support machine driver for RT5660 on Baytrail

BugLink: http://bugs.launchpad.net/bugs/1657674
Dell Caracalla IoT gateways sport an RT5660 codec with HID 10EC3277.
This patch adds the machine driver required to support RT5660 codec on
Baytrail based systems.

Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: (no-up) ASoC: rt5660: Add ACPI support

BugLink: http://bugs.launchpad.net/bugs/1657674
On Dell IoT Gateways, RT5660 codec is available with ACPI ID 10EC3277.
Also, GPIO's are only available by index, so we register mappings to allow
machine drivers to access them by name.

Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

ASoC: Intel: Atom: flip logic for gain Switch

BugLink: http://bugs.launchpad.net/bugs/1657674
The upstreamed code modified the control names from Mute to
Switch without changing the logic. To get audio working the Switch
needs to be off which isn't aligned with normal ALSA conventions.

Inverting the logic now so that Switch Off means mute and Switch On
means active audio using the specific volume setting.

Signed-off-by: Sebastien Guiriec <sebastien.guiriec@intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
(cherry picked from commit 940a5a014d50e15269b5d197ab571d1ca9971c43)
Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

ASoC: rt5660: enable MCLK detection

BugLink: http://bugs.launchpad.net/bugs/1657674
There is a power saving mechanism in rt5660. It will turn off some
unused power when MCLK is not present. We call that "MCLK detection"
and it should be enabled by default.

Signed-off-by: Bard Liao <bardliao@realtek.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
(cherry picked from commit d01580c394f58dff234a03f0f2f32c051a556f15)
Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

ASoC: rt5660: add rt5660 codec driver

BugLink: http://bugs.launchpad.net/bugs/1657674
This is the initial codec driver for rt5660

Signed-off-by: Oder Chiou <oder_chiou@realtek.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
(cherry picked from commit 2b26dd4c1fc5f83bc088f4a053120ca03817045e)
Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: [Config] Enable CONFIG_VEN_RSI_* configs

BugLink: http://bugs.launchpad.net/bugs/1657682
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: RS9113: Comment out IDs from upstream driver

BugLink: http://bugs.launchpad.net/bugs/1657682
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: RS9113: Use vendor driver to support WLAN/BT card on Caracalla HW only

BugLink: http://bugs.launchpad.net/bugs/1657682
This patch limits the RS9113 driver from vendor to be used for the WLAN/BT
card mounted on Dell Caracalla IoT gateways.

Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: Redpine RS9113 WLAN/BT driver ver. 0.9.7

BugLink: http://bugs.launchpad.net/bugs/1657682
This is the beta2 release for RS9113 driver from Redpine

Signed-off-by: Darren Wu <darren.wu@canonical.com>
Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: Separate Redpine RS9113 WLAN/BT vendor and kernel drivers

BugLink: http://bugs.launchpad.net/bugs/1657682
The driver submitted by vendor Redpine for RS9113 WLAN/BT exports same
symbols as their existing old driver in the kernel. This patch tries to
resolve this conflict.

Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

UBUNTU: SAUCE: Support Redpine RS9113 WLAN/BT

BugLink: http://bugs.launchpad.net/bugs/1657682
This patch adds support for RS9113 WLAN-BT cards found on Dell Caracalla
IoT gateways.
Vendor release: RS9113.NB0.NL.GNU.LNX.0.9.3 on 12/01/2016

Signed-off-by: Shrirang Bagul <shrirang.bagul@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

netvsc: add rcu_read locking to netvsc callback

BugLink: http://bugs.launchpad.net/bugs/1657540
The receive callback (in tasklet context) is using RCU to get reference
to associated VF network device but this is not safe. RCU read lock
needs to be held. Found by running with full lockdep debugging
enabled.

Fixes: f207c10d9823 ("hv_netvsc: use RCU to protect vf_netdev")
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit 0719e72ccb801829a3d735d187ca8417f0930459)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

tools: hv: Add a script to help bonding synthetic and VF NICs

BugLink: http://bugs.launchpad.net/bugs/1650059
This script helps to create bonding network devices based on synthetic NIC
(the virtual network adapter usually provided by Hyper-V) and the matching
VF NIC (SRIOV virtual function). So the synthetic NIC and VF NIC can
function as one network device, and fail over to the synthetic NIC if VF is
down.

Mayjor distros (RHEL, Ubuntu, SLES) supported by Hyper-V are supported by
this script.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from linux-next commit 178cd55f086629cf0bad9c66c793a7e2bcc3abb6)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Drivers: hv: vmbus: Base host signaling strictly on the ring state

BugLink: http://bugs.launchpad.net/bugs/1650059
One of the factors that can result in the host concluding that a given
guest in mounting a DOS attack is if the guest generates interrupts
to the host when the host is not expecting it. If these "spurious"
interrupts reach a certain rate, the host can throttle the guest to
minimize the impact. The host computation of the "expected number
of interrupts" is strictly based on the ring transitions. Until
the host logic is fixed, base the guest logic to interrupt solely
on the ring state.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit 74198eb4a42c4a3c4fbef08fa01a291a282f7c2e)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

netvsc: reduce maximum GSO size

BugLink: http://bugs.launchpad.net/bugs/1650059
Hyper-V (and Azure) support using NVGRE which requires some extra space
for encapsulation headers. Because of this the largest allowed TSO
packet is reduced.

For older releases, hard code a fixed reduced value.  For next release,
there is a better solution which uses result of host offload
negotiation.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit a50af86dd49ee1851d1ccf06dd0019c05b95e297)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

vmbus: make sysfs names consistent with PCI

BugLink: http://bugs.launchpad.net/bugs/1650059
In commit 9a56e5d6a0ba ("Drivers: hv: make VMBus bus ids persistent")
the name of vmbus devices in sysfs changed to be (in 4.9-rc1):
  /sys/bus/vmbus/vmbus-6aebe374-9ba0-11e6-933c-00259086b36b

The prefix ("vmbus-") is redundant and differs from how PCI is
represented in sysfs. Therefore simplify to:
  /sys/bus/vmbus/6aebe374-9ba0-11e6-933c-00259086b36b

Please merge this before 4.9 is released and the old format
has to live forever.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit f6b2db084b65b9dc0f910bc48d5f77c0e5166dc6)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Revert "hv_netvsc: report vmbus name in ethtool"

BugLink: http://bugs.launchpad.net/bugs/1650059
This reverts commit e3f74b841d48
("hv_netvsc: report vmbus name in ethtool")'
because of problem introduced by commit f9a56e5d6a0ba
("Drivers: hv: make VMBus bus ids persistent").
This changed the format of the vmbus name and this new format is too
long to fit in the bus_info field of ethtool.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit e934f684856393a0b2aecc7fdd8357a48b79c535)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

net/hyperv: avoid uninitialized variable

BugLink: http://bugs.launchpad.net/bugs/1650059
The hdr_offset variable is only if we deal with a TCP or UDP packet,
but as the check surrounding its usage tests for skb_is_gso()
instead, the compiler has no idea if the variable is initialized
or not at that point:

drivers/net/hyperv/netvsc_drv.c: In function ‘netvsc_start_xmit’:
drivers/net/hyperv/netvsc_drv.c:494:42: error: ‘hdr_offset’ may be used uninitialized in this function [-Werror=maybe-uninitialized]

This adds an additional check for the transport type, which
tells the compiler that this path cannot happen. Since the
get_net_transport_info() function should always be inlined
here, I don't expect this to result in additional runtime
checks.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit 52ccd6318481a467d8ad54ea40f60c61e957d58f)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

netvsc: Remove mistaken udp.h inclusion.

BugLink: http://bugs.launchpad.net/bugs/1650059
Based upon v2 of Stephen's patch.

Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit 3f2b0a5a3583a2c6c3aeb2d59e7f775d43faa89c)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

netvsc: fix checksum on UDP IPV6

BugLink: http://bugs.launchpad.net/bugs/1650059
The software calculation of UDP checksum in Netvsc driver was
only handling IPv4 case. By using skb_checksum_help() instead
all protocols can be handled. Rearrange code to eliminate goto
and look like other drivers.

This is a temporary solution; recent versions of Window Server etc
do support UDP checksum offload, just need to do the appropriate negotiation
with host to validate before using. This will be done in later patch.

Please queue this for -stable as well.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit ad19bc8a95baee4588e9ec4481297d97c0bec765)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Drivers: hv: get rid of id in struct vmbus_channel

BugLink: http://bugs.launchpad.net/bugs/1650059
The auto incremented counter is not being used anymore, get rid of it.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit e7fca5d860aeeb1e606448f5191cea8d925cc7a3)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

Drivers: hv: make VMBus bus ids persistent

BugLink: http://bugs.launchpad.net/bugs/1650059
Some tools use bus ids to identify devices and they count on the fact
that these ids are persistent across reboot. This may be not true for
VMBus as we use auto incremented counter from alloc_channel() as such
id. Switch to using if_instance from channel offer, this id is supposed
to be persistent.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
(cherry picked from commit b294809dbfa7e50229d00253d43f9a56e5d6a0ba)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

hv_netvsc: fix comments

BugLink: http://bugs.launchpad.net/bugs/1650059
Typo's and spelling errors. Also remove old comment from staging era.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit c6a77ff82fb849534748719f37f3f9086d78ed39)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

hv_netvsc: count multicast packets received

BugLink: http://bugs.launchpad.net/bugs/1650059
Useful for debugging issues with multicast and SR-IOV to keep track
of number of received multicast packets.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit f7ad75b753f386454f50044fd69edad767b69ce8)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

hv_netvsc: remove VF in flight counters

BugLink: http://bugs.launchpad.net/bugs/1650059
Since VF reference is now protected by RCU, no longer need the VF usage
counter and can use device flags to see whether to inject or not.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit 9cbcc4280645f0e7e19e6a0da443ec7e69cecf40)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>

hv_netvsc: use RCU to protect vf_netdev

BugLink: http://bugs.launchpad.net/bugs/1650059
The vf_netdev pointer in the netvsc device context can simply be protected
by RCU because network device destruction is already RCU synchronized.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit f207c10d982388fa42710922ad1c0c9d3ba9a87b)
Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Brad Figg <brad.figg@canonical.com>
Acked-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>