[ Upstream commit 0d4411408a ]
The current driver supports handling two vector0 interrupts, reset and
mailbox. When the hardware reports an interrupt of another type of
interrupt source, if the driver does not process the interrupt, but
enables the interrupt, the hardware will repeatedly report the unknown
interrupt.
Therefore, the driver enables the vector0 interrupt after clearing the
known type of interrupt source. Other conditions are not enabled.
Fixes: cd8c5c269b ("net: hns3: Fix for hclge_reset running repeatly problem")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 73b907a083 ]
When hns3_get_ring_config()/hns3_queue_to_ring()/
hns3_get_vector_ring_chain() failed during resetting, the allocated
memory has not been freed before these three functions return. So
this patch adds error handler in these functions to fix it.
Fixes: 76ad4f0ee7 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08 SoC")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e330af7889 ]
VF drivers can trigger PCIe completer aborts any time they read a queue
that they don't own. Even in nominal circumstances, it is not possible
to prevent the VF driver from reading queues it doesn't own. VF drivers
may attempt to read queues it previously owned, but which it no longer
does due to a PF reset.
Normally these completer aborts aren't an issue. However, on some
platforms these trigger machine check errors. This is true even if we
lower their severity from fatal to non-fatal. Indeed, we already have
code for lowering the severity.
We could attempt to mask these errors conditionally around resets, which
is the most common time they would occur. However this would essentially
be a race between the PF and VF drivers, and we may still occasionally
see machine check exceptions on these strictly configured platforms.
Instead, mask the errors entirely any time we resume VFs. By doing so,
we prevent the completer aborts from being sent to the parent PCIe
device, and thus these strict platforms will not upgrade them into
machine check errors.
Additionally, we don't lose any information by masking these errors,
because we'll still report VFs which attempt to access queues via the
FUM_BAD_VF_QACCESS errors.
Without this change, on platforms where completer aborts cause machine
check exceptions, the VF reading queues it doesn't own could crash the
host system. Masking the completer abort prevents this, so we should
mask it for good, and not just around a PCIe reset. Otherwise malicious
or misconfigured VFs could cause the host system to crash.
Because we are masking the error entirely, there is little reason to
also keep setting the severity bit, so that code is also removed.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 094bf4d0e9 ]
The timecounter needs to be updated at least once per ~550 seconds in
order to avoid a 40-bit SYSTIM timestamp to be misinterpreted as an old
timestamp.
Since commit 500462a9d ("timers: Switch to a non-cascading wheel"),
scheduling of delayed work seems to be less accurate and a requested
delay of 540 seconds may actually be longer than 550 seconds. Shorten
the delay to 480 seconds to be sure the timecounter is updated in time.
This fixes an issue with HW timestamps on 82580/I350/I354 being off by
~1100 seconds for few seconds every ~9 minutes.
Cc: Jacob Keller <jacob.e.keller@intel.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 381eab4a6e ]
There seem to be some problems as result of 30467e0b3b ("mm, hotplug:
fix concurrent memory hot-add deadlock"), which tried to fix a possible
lock inversion reported and discussed in [1] due to the two locks
a) device_lock()
b) mem_hotplug_lock
While add_memory() first takes b), followed by a) during
bus_probe_device(), onlining of memory from user space first took a),
followed by b), exposing a possible deadlock.
In [1], and it was decided to not make use of device_hotplug_lock, but
rather to enforce a locking order.
The problems I spotted related to this:
1. Memory block device attributes: While .state first calls
mem_hotplug_begin() and the calls device_online() - which takes
device_lock() - .online does no longer call mem_hotplug_begin(), so
effectively calls online_pages() without mem_hotplug_lock.
2. device_online() should be called under device_hotplug_lock, however
onlining memory during add_memory() does not take care of that.
In addition, I think there is also something wrong about the locking in
3. arch/powerpc/platforms/powernv/memtrace.c calls offline_pages()
without locks. This was introduced after 30467e0b3b. And skimming over
the code, I assume it could need some more care in regards to locking
(e.g. device_online() called without device_hotplug_lock. This will
be addressed in the following patches.
Now that we hold the device_hotplug_lock when
- adding memory (e.g. via add_memory()/add_memory_resource())
- removing memory (e.g. via remove_memory())
- device_online()/device_offline()
We can move mem_hotplug_lock usage back into
online_pages()/offline_pages().
Why is mem_hotplug_lock still needed? Essentially to make
get_online_mems()/put_online_mems() be very fast (relying on
device_hotplug_lock would be very slow), and to serialize against
addition of memory that does not create memory block devices (hmm).
[1] http://driverdev.linuxdriverproject.org/pipermail/ driverdev-devel/
2015-February/065324.html
This patch is partly based on a patch by Vitaly Kuznetsov.
Link: http://lkml.kernel.org/r/20180925091457.28651-4-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Pavel Tatashin <pavel.tatashin@microsoft.com>
Reviewed-by: Rashmica Gupta <rashmica.g@gmail.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Len Brown <lenb@kernel.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Stephen Hemminger <sthemmin@microsoft.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Rashmica Gupta <rashmica.g@gmail.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Kate Stewart <kstewart@linuxfoundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Philippe Ombredanne <pombredanne@nexb.com>
Cc: Pavel Tatashin <pavel.tatashin@microsoft.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: YASUAKI ISHIMATSU <yasu.isimatu@gmail.com>
Cc: Mathieu Malaterre <malat@debian.org>
Cc: John Allen <jallen@linux.vnet.ibm.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6c9a3f843a ]
Currently extent and index i are both being incremented causing an array
out of bounds read on extent[i]. Fix this by removing the extraneous
increment of extent.
Ernesto said:
: This is only triggered when deleting a file with a resource fork. I
: may be wrong because the documentation isn't clear, but I don't think
: you can create those under linux. So I guess nobody was testing them.
:
: > A disk space leak, perhaps?
:
: That's what it looks like in general. hfs_free_extents() won't do
: anything if the block count doesn't add up, and the error will be
: ignored. Now, if the block count randomly does add up, we could see
: some corruption.
Detected by CoverityScan, CID#711541 ("Out of bounds read")
Link: http://lkml.kernel.org/r/20180831140538.31566-1-colin.king@canonical.com
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Ernesto A. Fernndez <ernesto.mnd.fernandez@gmail.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hin-Tak Leung <htl10@users.sourceforge.net>
Cc: Vyacheslav Dubeyko <slava@dubeyko.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 19a9d0f1ac ]
Creating, renaming or deleting a file may hit BUG_ON() if the first
record of both a leaf node and its parent are changed, and if this
forces the parent to be split. This bug is triggered by xfstests
generic/027, somewhat rarely; here is a more reliable reproducer:
truncate -s 50M fs.iso
mkfs.hfsplus fs.iso
mount fs.iso /mnt
i=1000
while [ $i -le 2400 ]; do
touch /mnt/$i &>/dev/null
((++i))
done
i=2400
while [ $i -ge 1000 ]; do
mv /mnt/$i /mnt/$(perl -e "print $i x61") &>/dev/null
((--i))
done
The issue is that a newly created bnode is being put twice. Reset
new_node to NULL in hfs_brec_update_parent() before reaching goto again.
Link: http://lkml.kernel.org/r/5ee1db09b60373a15890f6a7c835d00e76bf601d.1535682461.git.ernesto.mnd.fernandez@gmail.com
Signed-off-by: Ernesto A. Fernández <ernesto.mnd.fernandez@gmail.com>
Cc: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit ce1091d471 ]
For various alignments of buf, the current expression computes
4096 ok
4095 ok
8190
8189
...
4097
i.e., if the caller has already written two bytes into the page buffer,
len is 8190 rather than 4094, because PTR_ALIGN aligns up to the next
boundary. So if the printed version of the bitmap is huge, scnprintf()
ends up writing beyond the page boundary.
I don't think any current callers actually write anything before
bitmap_print_to_pagebuf, but the API seems to be designed to allow it.
[akpm@linux-foundation.org: use offset_in_page(), per Andy]
[akpm@linux-foundation.org: include mm.h for offset_in_page()]
Link: http://lkml.kernel.org/r/20180818131623.8755-7-linux@rasmusvillemoes.dk
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Yury Norov <ynorov@caviumnetworks.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c57cdf7a9e ]
rq_qos_exit() removes the current q->rq_qos, this action has to be
done after queue is frozen, otherwise the IO queue path may never
be waken up, then IO hang is caused.
So fixes this issue by moving rq_qos_exit() after queue is frozen.
Cc: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 69f8117f17 ]
Use TEST_GEN_PROGS and don't redefine all, this makes the out-of-tree
build work. We need to move the extra dependencies below the include
of lib.mk, because it adds the $(OUTPUT) prefix if it's defined.
We can also drop the clean rule, lib.mk does it for us.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 266bac361d ]
For the out-of-tree build to work we need to tell switch_endian_test
to look for check-reversed.S in $(OUTPUT).
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 27825349d7 ]
We should use TEST_GEN_PROGS, not TEST_PROGS. That tells the selftests
makefile (lib.mk) that those tests are generated (built), and so it
adds the $(OUTPUT) prefix for us, making the out-of-tree build work
correctly.
It also means we don't need our own clean rule, lib.mk does it.
We also have to update the signal_tm rule to use $(OUTPUT).
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c39b79082a ]
We should use TEST_GEN_PROGS, not TEST_PROGS. That tells the selftests
makefile (lib.mk) that those tests are generated (built), and so it
adds the $(OUTPUT) prefix for us, making the out-of-tree build work
correctly.
It also means we don't need our own clean rule, lib.mk does it.
We also have to update the ptrace-pkey and core-pkey rules to use
$(OUTPUT).
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 9c87156cce ]
When building with clang (8 trunk, 7.0 release) the frame size limit is
hit:
arch/powerpc/xmon/xmon.c:452:12: warning: stack frame size of 2576
bytes in function 'xmon_core' [-Wframe-larger-than=]
Some investigation by Naveen indicates this is due to clang saving the
addresses to printf format strings on the stack.
While this issue is investigated, bump up the frame size limit for xmon
when building with clang.
Link: https://github.com/ClangBuiltLinux/linux/issues/252
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 966c37f2d7 ]
Similiar with ipv6 mcast commit 89225d1ce6 ("net: ipv6: mld: fix v1/v2
switchback timeout to rfc3810, 9.12.")
i) RFC3376 8.12. Older Version Querier Present Timeout says:
The Older Version Querier Interval is the time-out for transitioning
a host back to IGMPv3 mode once an older version query is heard.
When an older version query is received, hosts set their Older
Version Querier Present Timer to Older Version Querier Interval.
This value MUST be ((the Robustness Variable) times (the Query
Interval in the last Query received)) plus (one Query Response
Interval).
Currently we only use a hardcode value IGMP_V1/v2_ROUTER_PRESENT_TIMEOUT.
Fix it by adding two new items mr_qi(Query Interval) and mr_qri(Query Response
Interval) in struct in_device.
Now we can calculate the switchback time via (mr_qrv * mr_qi) + mr_qri.
We need update these values when receive IGMPv3 queries.
Reported-by: Ying Xu <yinxu@redhat.com>
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 07d19dc9fb ]
A deduplication data corruption is exposed in XFS and btrfs. It is
caused by extending the block match range to include the partial EOF
block, but then allowing unknown data beyond EOF to be considered a
"match" to data in the destination file because the comparison is only
made to the end of the source file. This corrupts the destination file
when the source extent is shared with it.
The VFS remapping prep functions only support whole block dedupe, but
we still need to appear to support whole file dedupe correctly. Hence
if the dedupe request includes the last block of the souce file, don't
include it in the actual dedupe operation. If the rest of the range
dedupes successfully, then reject the entire request. A subsequent
patch will enable us to shorten dedupe requests correctly.
When reflinking sub-file ranges, a data corruption can occur when the
source file range includes a partial EOF block. This shares the unknown
data beyond EOF into the second file at a position inside EOF, exposing
stale data in the second file.
If the reflink request includes the last block of the souce file, only
proceed with the reflink operation if it lands at or past the
destination file's current EOF. If it lands within the destination file
EOF, reject the entire request with -EINVAL and make the caller go the
hard way. A subsequent patch will enable us to shorten reflink requests
correctly.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 917e2fd2c5 ]
This fixes a long standing bug where large amounts of output
could freeze the tty (most commonly seen on stdio console).
While the bug has always been there it became more pronounced
after moving to the new interrupt controller.
The line semantics are now changed to have true IRQ write
semantics which should further improve the tty/line subsystem
stability and performance
Signed-off-by: Anton Ivanov <anton.ivanov@cambridgegreys.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit eaba68785c ]
The current IRQ handler clears all the IRQ status bits when it bails
out. This is dangerous because it might clear away the status bits
that have just been set while processing the current handler. If this
happens, the IRQ event for the latest transfer is lost forever.
The IRQ status bits must be cleared *before* the next transfer is
kicked.
Fixes: 6a62974b66 ("i2c: uniphier_f: add UniPhier FIFO-builtin I2C driver")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 39226aaa85 ]
Currently, a timeout error could happen at a repeated START condition.
For a (non-repeated) START condition, the controller starts sending
data when the UNIPHIER_FI2C_CR_STA bit is set. However, for a repeated
START condition, the hardware starts running when the slave address is
written to the TX FIFO - the write to the UNIPHIER_FI2C_CR register is
actually unneeded.
Because the hardware is already running before the IRQ is enabled for
a repeated START, the driver may miss the IRQ event. In most cases,
this problem does not show up since modern CPUs are much faster than
the I2C transfer. However, it is still possible that a context switch
happens after the controller starts, but before the IRQ register is
set up.
To fix this,
- Do not write UNIPHIER_FI2C_CR for repeated START conditions.
- Enable IRQ *before* writing the slave address to the TX FIFO.
- Disable IRQ for the current CPU while queuing up the TX FIFO;
If the CPU is interrupted by some task, the interrupt handler
might be invoked due to the empty TX FIFO before completing the
setup.
Fixes: 6a62974b66 ("i2c: uniphier_f: add UniPhier FIFO-builtin I2C driver")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f1fdcbbdf4 ]
This is unlikely to happen, but it is possible for a CPU to enter
the interrupt handler just after wait_for_completion_timeout() has
expired. If this happens, the hardware is accessed from multiple
contexts concurrently.
Disable the IRQ after wait_for_completion_timeout(), and do nothing
from the handler when the IRQ is disabled.
Fixes: 6a62974b66 ("i2c: uniphier_f: add UniPhier FIFO-builtin I2C driver")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6984046608 ]
There are two cases when handle DISCARD merge.
If max_discard_segments == 1, the bios/requests need to be contiguous
to merge. If max_discard_segments > 1, it takes every bio as a range
and different range needn't to be contiguous.
But now, attempt_merge screws this up. It always consider contiguity
for DISCARD for the case max_discard_segments > 1 and cannot merge
contiguous DISCARD for the case max_discard_segments == 1, because
rq_attempt_discard_merge always returns false in this case.
This patch fixes both of the two cases above.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jianchao Wang <jianchao.w.wang@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 07bddef983 ]
Currently, the kernel doesn't let the administrator set a macsec device
up unless its lower device is currently up. This is inconsistent, as a
macsec device that is up won't automatically go down when its lower
device goes down.
Now that linkstate propagation works, there's really no reason for this
limitation, so let's remove it.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Reported-by: Radu Rendec <radu.rendec@gmail.com>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e6ac075882 ]
Like all other virtual devices (macvlan, vlan), the operstate of a
macsec device should match the state of its lower device. This is done
by calling netif_stacked_transfer_operstate from its netdevice notifier.
We also need to call netif_stacked_transfer_operstate when a new macsec
device is created, so that its operstate is set properly. This is only
relevant when we try to bring the device up directly when we create it.
Radu Rendec proposed a similar patch, inspired from the 802.1q driver,
that included changing the administrative state of the macsec device,
instead of just the operstate. This version is similar to what the
macvlan driver does, and updates only the operstate.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Reported-by: Radu Rendec <radu.rendec@gmail.com>
Reported-by: Patrick Talbert <ptalbert@redhat.com>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit d7c3393413 ]
Patch series "migrate_misplaced_transhuge_page race conditions".
Aaron found a new instance of the THP MADV_DONTNEED race against
pmdp_clear_flush* variants, that was apparently left unfixed.
While looking into the race found by Aaron, I may have found two more
issues in migrate_misplaced_transhuge_page.
These race conditions would not cause kernel instability, but they'd
corrupt userland data or leave data non zero after MADV_DONTNEED.
I did only minor testing, and I don't expect to be able to reproduce this
(especially the lack of ->invalidate_range before migrate_page_copy,
requires the latest iommu hardware or infiniband to reproduce). The last
patch is noop for x86 and it needs further review from maintainers of
archs that implement flush_cache_range() (not in CC yet).
To avoid confusion, it's not the first patch that introduces the bug fixed
in the second patch, even before removing the
pmdp_huge_clear_flush_notify, that _notify suffix was called after
migrate_page_copy already run.
This patch (of 3):
This is a corollary of ced108037c ("thp: fix MADV_DONTNEED vs. numa
balancing race"), 58ceeb6bec ("thp: fix MADV_DONTNEED vs. MADV_FREE
race") and 5b7abeae3a ("thp: fix MADV_DONTNEED vs clear soft dirty
race).
When the above three fixes where posted Dave asked
https://lkml.kernel.org/r/929b3844-aec2-0111-fef7-8002f9d4e2b9@intel.com
but apparently this was missed.
The pmdp_clear_flush* in migrate_misplaced_transhuge_page() was introduced
in a54a407fbf ("mm: Close races between THP migration and PMD numa
clearing").
The important part of such commit is only the part where the page lock is
not released until the first do_huge_pmd_numa_page() finished disarming
the pagenuma/protnone.
The addition of pmdp_clear_flush() wasn't beneficial to such commit and
there's no commentary about such an addition either.
I guess the pmdp_clear_flush() in such commit was added just in case for
safety, but it ended up introducing the MADV_DONTNEED race condition found
by Aaron.
At that point in time nobody thought of such kind of MADV_DONTNEED race
conditions yet (they were fixed later) so the code may have looked more
robust by adding the pmdp_clear_flush().
This specific race condition won't destabilize the kernel, but it can
confuse userland because after MADV_DONTNEED the memory won't be zeroed
out.
This also optimizes the code and removes a superfluous TLB flush.
[akpm@linux-foundation.org: reflow comment to 80 cols, fix grammar and typo (beacuse)]
Link: http://lkml.kernel.org/r/20181013002430.698-2-aarcange@redhat.com
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Aaron Tomlin <atomlin@redhat.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 64081362e8 ]
We've recently seen a workload on XFS filesystems with a repeatable
deadlock between background writeback and a multi-process application
doing concurrent writes and fsyncs to a small range of a file.
range_cyclic
writeback Process 1 Process 2
xfs_vm_writepages
write_cache_pages
writeback_index = 2
cycled = 0
....
find page 2 dirty
lock Page 2
->writepage
page 2 writeback
page 2 clean
page 2 added to bio
no more pages
write()
locks page 1
dirties page 1
locks page 2
dirties page 1
fsync()
....
xfs_vm_writepages
write_cache_pages
start index 0
find page 1 towrite
lock Page 1
->writepage
page 1 writeback
page 1 clean
page 1 added to bio
find page 2 towrite
lock Page 2
page 2 is writeback
<blocks>
write()
locks page 1
dirties page 1
fsync()
....
xfs_vm_writepages
write_cache_pages
start index 0
!done && !cycled
sets index to 0, restarts lookup
find page 1 dirty
find page 1 towrite
lock Page 1
page 1 is writeback
<blocks>
lock Page 1
<blocks>
DEADLOCK because:
- process 1 needs page 2 writeback to complete to make
enough progress to issue IO pending for page 1
- writeback needs page 1 writeback to complete so process 2
can progress and unlock the page it is blocked on, then it
can issue the IO pending for page 2
- process 2 can't make progress until process 1 issues IO
for page 1
The underlying cause of the problem here is that range_cyclic writeback is
processing pages in descending index order as we hold higher index pages
in a structure controlled from above write_cache_pages(). The
write_cache_pages() caller needs to be able to submit these pages for IO
before write_cache_pages restarts writeback at mapping index 0 to avoid
wcp inverting the page lock/writeback wait order.
generic_writepages() is not susceptible to this bug as it has no private
context held across write_cache_pages() - filesystems using this
infrastructure always submit pages in ->writepage immediately and so there
is no problem with range_cyclic going back to mapping index 0.
However:
mpage_writepages() has a private bio context,
exofs_writepages() has page_collect
fuse_writepages() has fuse_fill_wb_data
nfs_writepages() has nfs_pageio_descriptor
xfs_vm_writepages() has xfs_writepage_ctx
All of these ->writepages implementations can hold pages under writeback
in their private structures until write_cache_pages() returns, and hence
they are all susceptible to this deadlock.
Also worth noting is that ext4 has it's own bastardised version of
write_cache_pages() and so it /may/ have an equivalent deadlock. I looked
at the code long enough to understand that it has a similar retry loop for
range_cyclic writeback reaching the end of the file and then promptly ran
away before my eyes bled too much. I'll leave it for the ext4 developers
to determine if their code is actually has this deadlock and how to fix it
if it has.
There's a few ways I can see avoid this deadlock. There's probably more,
but these are the first I've though of:
1. get rid of range_cyclic altogether
2. range_cyclic always stops at EOF, and we start again from
writeback index 0 on the next call into write_cache_pages()
2a. wcp also returns EAGAIN to ->writepages implementations to
indicate range cyclic has hit EOF. writepages implementations can
then flush the current context and call wpc again to continue. i.e.
lift the retry into the ->writepages implementation
3. range_cyclic uses trylock_page() rather than lock_page(), and it
skips pages it can't lock without blocking. It will already do this
for pages under writeback, so this seems like a no-brainer
3a. all non-WB_SYNC_ALL writeback uses trylock_page() to avoid
blocking as per pages under writeback.
I don't think #1 is an option - range_cyclic prevents frequently
dirtied lower file offset from starving background writeback of
rarely touched higher file offsets.
#2 is simple, and I don't think it will have any impact on
performance as going back to the start of the file implies an
immediate seek. We'll have exactly the same number of seeks if we
switch writeback to another inode, and then come back to this one
later and restart from index 0.
#2a is pretty much "status quo without the deadlock". Moving the
retry loop up into the wcp caller means we can issue IO on the
pending pages before calling wcp again, and so avoid locking or
waiting on pages in the wrong order. I'm not convinced we need to do
this given that we get the same thing from #2 on the next writeback
call from the writeback infrastructure.
#3 is really just a band-aid - it doesn't fix the access/wait
inversion problem, just prevents it from becoming a deadlock
situation. I'd prefer we fix the inversion, not sweep it under the
carpet like this.
#3a is really an optimisation that just so happens to include the
band-aid fix of #3.
So it seems that the simplest way to fix this issue is to implement
solution #2
Link: http://lkml.kernel.org/r/20181005054526.21507-1-david@fromorbit.com
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Jan Kara <jack@suse.de>
Cc: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 999865764f ]
The kernel module may sleep with holding a spinlock.
The function call paths (from bottom to top) in Linux-4.16 are:
[FUNC] get_zeroed_page(GFP_NOFS)
fs/ocfs2/dlm/dlmdebug.c, 332: get_zeroed_page in dlm_print_one_mle
fs/ocfs2/dlm/dlmmaster.c, 240: dlm_print_one_mle in __dlm_put_mle
fs/ocfs2/dlm/dlmmaster.c, 255: __dlm_put_mle in dlm_put_mle
fs/ocfs2/dlm/dlmmaster.c, 254: spin_lock in dlm_put_ml
[FUNC] get_zeroed_page(GFP_NOFS)
fs/ocfs2/dlm/dlmdebug.c, 332: get_zeroed_page in dlm_print_one_mle
fs/ocfs2/dlm/dlmmaster.c, 240: dlm_print_one_mle in __dlm_put_mle
fs/ocfs2/dlm/dlmmaster.c, 222: __dlm_put_mle in dlm_put_mle_inuse
fs/ocfs2/dlm/dlmmaster.c, 219: spin_lock in dlm_put_mle_inuse
To fix this bug, GFP_NOFS is replaced with GFP_ATOMIC.
This bug is found by my static analysis tool DSAC.
Link: http://lkml.kernel.org/r/20180901112528.27025-1-baijiaju1990@gmail.com
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 6c2fc9cddc ]
Such as:
fs/ocfs2/file.c: In function ‘ocfs2_file_write_iter’:
./arch/sparc/include/asm/cmpxchg_64.h:55:22: warning: value computed is not used [-Wunused-value]
#define xchg(ptr,x) ((__typeof__(*(ptr)))__xchg((unsigned long)(x),(ptr),sizeof(*(ptr))))
and
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c: In function ‘ixgbevf_xdp_setup’:
./arch/sparc/include/asm/cmpxchg_64.h:55:22: warning: value computed is not used [-Wunused-value]
#define xchg(ptr,x) ((__typeof__(*(ptr)))__xchg((unsigned long)(x),(ptr),sizeof(*(ptr))))
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e901378578 ]
Fix a bug introduced by the creation of flush_all_to_thread() for
processors that have SPE (Signal Processing Engine) and use it to
compute floating-point operations.
>From userspace perspective, the problem was seen in attempts of
computing floating-point operations which should generate exceptions.
For example:
fork();
float x = 0.0 / 0.0;
isnan(x); // forked process returns False (should be True)
The operation above also should always cause the SPEFSCR FINV bit to
be set. However, the SPE floating-point exceptions were turned off
after a fork().
Kernel versions prior to the bug used flush_spe_to_thread(), which
first saves SPEFSCR register values in tsk->thread and then calls
giveup_spe(tsk).
After commit 579e633e76, the save_all() function was called first
to giveup_spe(), and then the SPEFSCR register values were saved in
tsk->thread. This would save the SPEFSCR register values after
disabling SPE for that thread, causing the bug described above.
Fixes 579e633e76 ("powerpc: create flush_all_to_thread()")
Signed-off-by: Felipe Rechia <felipe.rechia@datacom.com.br>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 4a6998aff8 ]
Wenwen Wang reported:
In btf_parse(), the header of the user-space btf data 'btf_data'
is firstly parsed and verified through btf_parse_hdr().
In btf_parse_hdr(), the header is copied from user-space 'btf_data'
to kernel-space 'btf->hdr' and then verified. If no error happens
during the verification process, the whole data of 'btf_data',
including the header, is then copied to 'data' in btf_parse(). It
is obvious that the header is copied twice here. More importantly,
no check is enforced after the second copy to make sure the headers
obtained in these two copies are same. Given that 'btf_data' resides
in the user space, a malicious user can race to modify the header
between these two copies. By doing so, the user can inject
inconsistent data, which can cause undefined behavior of the
kernel and introduce potential security risk.
This issue is similar to the one fixed in commit 8af03d1ae2 ("bpf:
btf: Fix a missing check bug"). To fix it, this patch copies the user
'btf_data' *before* parsing / verifying the BTF header.
Fixes: 69b693f0ae ("bpf: btf: Introduce BPF Type Format (BTF)")
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Co-developed-by: Wenwen Wang <wang6495@umn.edu>
Acked-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f592f80483 ]
The dev_map_notification() removes interface in devmap if
unregistering interface's ifindex is same.
But only checking ifindex is not enough because other netns can have
same ifindex. so that wrong interface selection could occurred.
Hence netdev pointer comparison code is added.
v2: compare netdev pointer instead of using net_eq() (Daniel Borkmann)
v1: Initial patch
Fixes: 2ddf71e23c ("net: add notifier hooks for devmap bpf map")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Acked-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3a31386217 ]
On r8a7791/koelsch, sometimes the following message is printed during
system suspend:
rcar_thermal e61f0000.thermal: thermal sensor was broken
This happens if the workqueue runs while the device is already
suspended. Fix this by using the freezable system workqueue instead,
cfr. commit 51e20d0e3a ("thermal: Prevent polling from happening
during system suspend").
Fixes: e0a5172e9e ("thermal: rcar: add interrupt support")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit fb363e2d20 ]
Fixes the following warnings:
dirty_log_test.c: In function ‘help’:
dirty_log_test.c:216:9: warning: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 2 has type ‘int’ [-Wformat=]
printf(" -i: specify iteration counts (default: %"PRIu64")\n",
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from include/test_util.h:18:0,
from dirty_log_test.c:16:
/usr/include/inttypes.h:105:34: note: format string is defined here
# define PRIu64 __PRI64_PREFIX "u"
dirty_log_test.c:218:9: warning: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 2 has type ‘int’ [-Wformat=]
printf(" -I: specify interval in ms (default: %"PRIu64" ms)\n",
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from include/test_util.h:18:0,
from dirty_log_test.c:16:
/usr/include/inttypes.h:105:34: note: format string is defined here
# define PRIu64 __PRI64_PREFIX "u"
Signed-off-by: Andrea Parri <andrea.parri@amarulasolutions.com>
Signed-off-by: Shuah Khan (Samsung OSG) <shuah@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
