Ceph can in some cases issue an async DIO request, in which case we can
end up calling ceph_end_io_direct before the I/O is actually complete.
That may allow buffered operations to proceed while DIO requests are
still in flight.
Fix this by incrementing the i_dio_count when issuing an async DIO
request, and decrement it when tearing down the aio_req.
Fixes: 321fe13c93 ("ceph: add buffered/direct exclusionary locking for reads and writes")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Most of the time, we (or the vfs layer) takes the inode_lock and then
acquires caps, but ceph_read_iter does the opposite, and that can lead
to a deadlock.
When there are multiple clients treading over the same data, we can end
up in a situation where a reader takes caps and then tries to acquire
the inode_lock. Another task holds the inode_lock and issues a request
to the MDS which needs to revoke the caps, but that can't happen until
the inode_lock is unwedged.
Fix this by having ceph_read_iter take the inode_lock earlier, before
attempting to acquire caps.
Fixes: 321fe13c93 ("ceph: add buffered/direct exclusionary locking for reads and writes")
Link: https://tracker.ceph.com/issues/36348
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Add a jump target so that a call of the function “clk_put”
can be better reused at the end of this function.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Tony Lindgren <tony@atomide.com>
The unit descriptor validation may lead to a probe error when the
device provides a buggy descriptor or the validator detected
incorrectly. For identifying such an error and band-aiding, give a
new module option, skip_validation. With this option, the driver
ignores the validation errors with the hexdump of the unit
descriptor, so we can check it in a bit more details.
Link: https://lore.kernel.org/r/20191114165613.7422-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The recently introduced unit descriptor validation had some bug for
processing and extension units, it counts a bControlSize byte twice so
it expected a bigger size than it should have been. This seems
resulting in a probe error on a few devices.
Fix the calculation for proper checks of PU and EU.
Fixes: 57f8770620 ("ALSA: usb-audio: More validations of descriptor units")
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20191114165613.7422-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
There used to be a bug in the video driver that caused the timings
for the LCD to calculate in a way on the DM3730 which made it hang.
The work around for this bug was to set
CONFIG_OMAP2_DSS_MIN_FCK_PER_PCK=4 in the kernel. This work around
is no longer needed as the video drivers have been corrected.
This patch removes the legacy note.
Signed-off-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
The baseboard of the Logic PD Torpedo development kit has a socket
for a rechargable battery. The battery is monitored by a charger
which can communicate of the the 1-wire HDQ pin.
This patch enables the pinmux for the HDQ pin.
Signed-off-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Pull Kbuild fixes from Masahiro Yamada:
- fix build error when compiling SPARC VDSO with CONFIG_COMPAT=y
- pass correct --arch option to Sparse
* tag 'kbuild-fixes-v5.4-3' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
kbuild: tell sparse about the $ARCH
sparc: vdso: fix build error of vdso32
Schema errors can cause make to exit before useful information is
printed. This leaves developers wondering what's wrong. It can be
overcome passing '-k' to make, but that's not an obvious solution.
There's 2 scenarios where this happens.
When using DT_SCHEMA_FILES to validate with a single schema, any error
in the schema results in processed-schema.yaml being empty causing a
make error. The result is the specific errors in the schema are never
shown because processed-schema.yaml is the first target built. Simply
making processed-schema.yaml last in extra-y ensures the full schema
validation with detailed error messages happen first.
The 2nd problem is while schema errors are ignored for
processed-schema.yaml, full validation of the schema still runs in
parallel and any schema validation errors will still stop the build when
running validation of dts files. The fix is to not add the schema
examples to extra-y in this case. This means 'dtbs_check' is no longer a
superset of 'dt_binding_check'. Update the documentation to make this
clear.
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Tested-by: Jeffrey Hugo <jhugo@codeaurora.org>
Signed-off-by: Rob Herring <robh@kernel.org>
Convert Samsung Exynos Soc Power Domain bindings to DT schema format using
json-schema.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Rob Herring <robh@kernel.org>
We have local timers on Cortex-A9, so using the gptimer option makes no
sense. Let's just drop it for omap4 to simplify the timer options a bit.
If this is really needed, it can be still done by specifying dts properties
in the board specific file for assigned-clocks and assigned-clock-parents.
This gets us a bit closer to start dropping legacy platform data for
gptimers except for timer1 that is used for system clockevent.
Cc: Keerthy <j-keerthy@ti.com>
Cc: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Pull RDMA fixes from Jason Gunthorpe:
"Bug fixes for old bugs in the hns and hfi1 drivers:
- Calculate various values in hns properly to avoid over/underflows
in some cases
- Fix an oops, PCI negotiation on Gen4 systems, and bugs related to
retries"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
RDMA/hns: Correct the value of srq_desc_size
RDMA/hns: Correct the value of HNS_ROCE_HEM_CHUNK_LEN
IB/hfi1: TID RDMA WRITE should not return IB_WC_RNR_RETRY_EXC_ERR
IB/hfi1: Calculate flow weight based on QP MTU for TID RDMA
IB/hfi1: Ensure r_tid_ack is valid before building TID RDMA ACK packet
IB/hfi1: Ensure full Gen3 speed in a Gen4 system
We must set the autogating bit on enable for AESS (Audio Engine SubSystem)
when probed with ti-sysc interconnect target module driver. Otherwise it
won't idle properly.
Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
We cannot rely on the entry memcpy as we only copy the actual size of the
command, the rest of the bytes must be memset to zero.
Currently providing non-zero memory will not have any user visible impact.
However, since admin commands are extendable (in a backwards compatible
way) everything beyond the size of the command must be cleared to prevent
issues in the future.
Fixes: 0420e54256 ("RDMA/efa: Implement functions that submit and complete admin commands")
Link: https://lore.kernel.org/r/20191112092608.46964-1-galpress@amazon.com
Reviewed-by: Daniel Kranzdorf <dkkranzd@amazon.com>
Reviewed-by: Firas JahJah <firasj@amazon.com>
Signed-off-by: Gal Pressman <galpress@amazon.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Commit 9287c6452d fixed a situation in which gfs2 could use a glock
after it had been freed. To do that, it temporarily added a new glock
reference by calling gfs2_glock_hold in function gfs2_add_revoke.
However, if the bd element was removed by gfs2_trans_remove_revoke, it
failed to drop the additional reference.
This patch adds logic to gfs2_trans_remove_revoke to properly drop the
additional glock reference.
Fixes: 9287c6452d ("gfs2: Fix occasional glock use-after-free")
Cc: stable@vger.kernel.org # v5.2+
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Function gfs2_log_shutdown is only called from within log.c. This
patch removes the extern declaration and makes it static.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
According to the comment, board files used to specify 1 ms for the
debounce time. gpiod_set_debounce() needs the debounce time to be
specified in units of microseconds, so make sure to multiply the value
by 1000.
Note that, according to the git log, the board files actually did
specify 1 us for bounce times, but that seems really low. Device tree
bindings for this type of GPIO typically specify the debounce times in
milliseconds, so setting this default value to 1 ms seems like it would
be somewhat safer.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Pavel Machek <pavel@denx.de>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
It have turned out that it's not a good idea to unconditionally do a power
cycle and then to re-initialize the SDIO card, as currently done through
mmc_hw_reset() -> mmc_sdio_hw_reset(). This because there may be multiple
SDIO func drivers probed, who also shares the same SDIO card.
To address these scenarios, one may be tempted to use a notification
mechanism, as to allow the core to inform each of the probed func drivers,
about an ongoing HW reset. However, supporting such an operation from the
func driver point of view, may not be entirely trivial.
Therefore, let's use a more simplistic approach to solve the problem, by
instead forcing the card to be removed and re-detected, via scheduling a
rescan-work. In this way, we can rely on existing infrastructure, as the
func driver's ->remove() and ->probe() callbacks, becomes invoked to deal
with the cleanup and the re-initialization.
This solution may be considered as rather heavy, especially if a func
driver doesn't share its card with other func drivers. To address this,
let's keep the current immediate HW reset option as well, but run it only
when there is one func driver probed for the card.
Finally, to allow the caller of mmc_hw_reset(), to understand if the reset
is being asynchronously managed from a scheduled work, it returns 1
(propagated from mmc_sdio_hw_reset()). If the HW reset is executed
successfully and synchronously it returns 0, which maintains the existing
behaviour.
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Cc: stable@vger.kernel.org # v5.4+
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Upfront in mmc_rescan() we use the host->rescan_entered flag, to allow
scanning only once for non-removable cards. Therefore, it's also not
possible that we can have a corresponding card bus attached (host->bus_ops
is NULL), when we are scanning non-removable cards.
For this reason, let' drop the check for mmc_card_is_removable() as it's
redundant.
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Cc: stable@vger.kernel.org # v5.4+
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The SDIO HW reset procedure in mwifiex_sdio_card_reset_work() is broken,
when the SDIO card is shared with another SDIO func driver. This is the
case when the Bluetooth btmrvl driver is being used in combination with
mwifiex. More precisely, when mwifiex_sdio_card_reset_work() runs to resets
the SDIO card, the btmrvl driver doesn't get notified about it. Beyond that
point, the btmrvl driver will fail to communicate with the SDIO card.
This is a generic problem for SDIO func drivers sharing an SDIO card, which
are about to be addressed in subsequent changes to the mmc core and the
mmc_hw_reset() interface. In principle, these changes means the
mmc_hw_reset() interface starts to return 1 if the are multiple drivers for
the SDIO card, as to indicate to the caller that the reset needed to be
scheduled asynchronously through a hotplug mechanism of the SDIO card.
Let's prepare the mwifiex driver to support the upcoming new behaviour of
mmc_hw_reset(), which means extending the mwifiex_sdio_card_reset_work() to
support the asynchronous SDIO HW reset path. This also means, we need to
allow the ->remove() callback to run, without waiting for the FW to be
loaded. Additionally, during system suspend, mwifiex_sdio_suspend() may be
called when a reset has been scheduled, but waiting to be executed. In this
scenario let's simply return -EBUSY to abort the suspend process, as to
allow the reset to be completed first.
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Cc: stable@vger.kernel.org # v5.4+
Acked-by: Kalle Valo <kvalo@codeaurora.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Add new command for getting/setting current transmit power
and propagate requests from user space to firmware.
Signed-off-by: Mikhail Karpenko <mkarpenko@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Report MIC failure from firmware to cfg80211 subsystem
using dedicated callback cfg80211_michael_mic_failure.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Send EAPOL frames via control path so they can be treated in a different
way rather than normal data frames. In this case EAPOLs are sent with
higher priority and with disabled aggregation and encryption. Besides,
all devices benefit from sending EAPOL frames via high priority path,
so move the functionality from chip specific to common code.
Signed-off-by: Igor Mitsyanko <igor.mitsyanko.os@quantenna.com>
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Rx descriptors queue length is hardware specific. Current common default
value is no more than an accident. So move Rx descriptor queue setup to
platform PCIe backend in the same way as it is already done for Tx.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Do not attempt to print frequency for an invalid channel
provided by firmware. That channel may simply not exist.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Fix merge artifact for commit 0b68fe10b8 ("qtnfmac: modify debugfs
to support multiple cards") and finally add debugfs support
for multiple qtnfmac wireless cards.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
KASAN reported use-after-free error:
[ 995.220767] BUG: KASAN: use-after-free in qtnf_cmd_send_with_reply+0x169/0x3e0 [qtnfmac]
[ 995.221098] Read of size 2 at addr ffff888213d1ded0 by task kworker/1:1/71
The issue in qtnf_cmd_send_with_reply impacts all the commands that do
not need response other then return code. For such commands, consume_skb
is used for response skb and right after that return code in response
skb is accessed.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
A number of people have reported the Edimax EW-7611ULB works fine.
Signed-off-by: Jes Sorensen <Jes.Sorensen@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Currently, some sanity checks for uapi headers are done by
scripts/headers_check.pl, which is wired up to the 'headers_check'
target in the top Makefile.
It is true compiling headers has better test coverage, but there
are still several headers excluded from the compile test. I like
to keep headers_check.pl for a while, but we can delete a lot of
code by moving the build rule to usr/include/Makefile.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
There are both positive and negative options about this feature.
At first, I thought it was a good idea, but actually Linus stated a
negative opinion (https://lkml.org/lkml/2019/9/29/227). I admit it
is ugly and annoying.
The baseline I'd like to keep is the compile-test of uapi headers.
(Otherwise, kernel developers have no way to ensure the correctness
of the exported headers.)
I will maintain a small build rule in usr/include/Makefile.
Remove the other header test functionality.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acquire the per-VM slots_lock when zapping all shadow pages as part of
toggling nx_huge_pages. The fast zap algorithm relies on exclusivity
(via slots_lock) to identify obsolete vs. valid shadow pages, because it
uses a single bit for its generation number. Holding slots_lock also
obviates the need to acquire a read lock on the VM's srcu.
Failing to take slots_lock when toggling nx_huge_pages allows multiple
instances of kvm_mmu_zap_all_fast() to run concurrently, as the other
user, KVM_SET_USER_MEMORY_REGION, does not take the global kvm_lock.
(kvm_mmu_zap_all_fast() does take kvm->mmu_lock, but it can be
temporarily dropped by kvm_zap_obsolete_pages(), so it is not enough
to enforce exclusivity).
Concurrent fast zap instances causes obsolete shadow pages to be
incorrectly identified as valid due to the single bit generation number
wrapping, which results in stale shadow pages being left in KVM's MMU
and leads to all sorts of undesirable behavior.
The bug is easily confirmed by running with CONFIG_PROVE_LOCKING and
toggling nx_huge_pages via its module param.
Note, until commit 4ae5acbc4936 ("KVM: x86/mmu: Take slots_lock when
using kvm_mmu_zap_all_fast()", 2019-11-13) the fast zap algorithm used
an ulong-sized generation instead of relying on exclusivity for
correctness, but all callers except the recently added set_nx_huge_pages()
needed to hold slots_lock anyways. Therefore, this patch does not have
to be backported to stable kernels.
Given that toggling nx_huge_pages is by no means a fast path, force it
to conform to the current approach instead of reintroducing the previous
generation count.
Fixes: b8e8c8303f ("kvm: mmu: ITLB_MULTIHIT mitigation", but NOT FOR STABLE)
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Sparse uses the same executable for all archs and uses flags
like -m64, -mbig-endian or -D__arm__ for arch-specific parameters.
But Sparse also uses value from the host machine used to build
Sparse as default value for the target machine.
This works, of course, well for native build but can create
problems when cross-compiling, like defining both '__i386__'
and '__arm__' when cross-compiling for arm on a x86-64 machine.
Fix this by explicitely telling sparse the target architecture.
Reported-by: Ben Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Since commit 54b8ae66ae ("kbuild: change *FLAGS_<basetarget>.o to
take the path relative to $(obj)"), sparc allmodconfig fails to build
as follows:
CC arch/sparc/vdso/vdso32/vclock_gettime.o
unrecognized e_machine 18 arch/sparc/vdso/vdso32/vclock_gettime.o
arch/sparc/vdso/vdso32/vclock_gettime.o: failed
The cause of the breakage is that -pg flag not being dropped.
The vdso32 files are located in the vdso32/ subdirectory, but I missed
to update the Makefile.
I removed the meaningless CFLAGS_REMOVE_vdso-note.o since it is only
effective for C file.
vdso-note.o is compiled from assembly file:
arch/sparc/vdso/vdso-note.S
arch/sparc/vdso/vdso32/vdso-note.S
Fixes: 54b8ae66ae ("kbuild: change *FLAGS_<basetarget>.o to take the path relative to $(obj)")
Reported-by: Anatoly Pugachev <matorola@gmail.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Tested-by: Anatoly Pugachev <matorola@gmail.com>
Acked-by: David S. Miller <davem@davemloft.net>
