Bjorn Andersson [Mon, 11 Jan 2016 18:33:27 +0000 (10:33 -0800)]
wcn36xx: Disable power save
Enabling power save results in extremely long ping times and stalling
transfers. So disable it for not until we can figure out how to work
around this.
Bjorn Andersson [Thu, 25 Feb 2016 16:09:15 +0000 (08:09 -0800)]
soc: qcom: wcnss_ctrl: Make wcnss_ctrl parent the other components
We need the signal from wcnss_ctrl indicating that the firmware is up
and running before we can communicate with the other components of the
chip. So make these other components children of the wcnss_ctrl device,
so they can be probed in order.
The process seems to take between 1/2-5 seconds, so this is done in a
worker, instead of holding up the probe.
Bluetooth: Introduce Qualcomm WCNSS SMD based HCI driver
The Qualcomm WCNSS chip provides two SMD channels to the BT core; one
for command and one for event packets. This driver exposes the two
channels as a hci device.
With the introduction of the upstream wcnss remoteproc driver we do iris
xo calibration before we boot the wcnss - as it should have been done
from the beginning.
This patch cleans out the iris xo calibration code and simplifies the
platform integration implementation.
soc: qcom: wcnss_ctrl: Expose the cbc status to others
The prima driver needs a way to wait for the wcnss firmware to become
available, it does this by repeatedly querying the wcnss control driver
if we've reached cbc (cold boot completed).
qcom-scm: Align PAS api with "upstream" implementation
This aligns the PAS implementation with the patches heading upstream,
in particular tying the dma allocation in pas_init_image to the scm
platform_device rather then the caller.
Bjorn Andersson [Thu, 25 Feb 2016 16:09:15 +0000 (08:09 -0800)]
soc: qcom: wcnss_ctrl: Make wcnss_ctrl parent the other components
This is a hacked up version of the patch sent to LKML, as I don't want
to change the smd API in the release branch.
We need the signal from wcnss_ctrl indicating that the firmware is up
and running before we can communicate with the other components of the
chip. So make these other components children of the wcnss_ctrl device,
so they can be probed in order.
The process seems to take between 1/2-5 seconds, so this is done in a
worker, instead of holding up the probe.
This introduces the peripheral image loader, for loading WCNSS firmware
and boot the core on e.g. MSM8974. The firmware is verified and booted
with the help of the Peripheral Authentication System (PAS) in
TrustZone.
Lee Jones [Tue, 12 Jan 2016 12:46:17 +0000 (12:46 +0000)]
remoteproc: debugfs: Add ability to boot remote processor using debugfs
This functionality is especially useful during the testing phase. When
used in conjunction with Mailbox's Test Framework we can trivially conduct
end-to-end testing i.e. boot co-processor, send and receive messages to
the co-processor, then shut it down again (repeat as required).
Signed-off-by: Ludovic Barre <ludovic.barre@st.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Parth Pancholi [Fri, 10 Jun 2016 18:19:55 +0000 (19:19 +0100)]
ARM: dts: sd_600eval: Fix eMMC lockup issue
This board locks up if we stress test the eMMC, as the regulator s4 is
unable to supply enough current for all the peripherials attached to it.
As this supply is wired up to most of the peripherials including DDR,
it resulted in such lockup.
This patch fixes this issue by setting s4 regulator correctly with
Auto power mode.
Arnd Bergmann [Fri, 13 May 2016 13:09:58 +0000 (15:09 +0200)]
net: qrtr: fix build problems
Having multiple loadable modules with the same name cannot work
with modprobe, and having both net/qrtr/smd.ko and drivers/soc/qcom/smd.ko
results in a (somewhat cryptic) build error:
Also, the qrtr driver uses the SMD interface and has a Kconfig dependency,
but also allows for compile-testing when SMD is disabled. However, if
with QCOM_SMD=m and COMPILE_TEST=y we can end up with QRTR_SMD=y and
that fails with a related link error.
The changes the dependency so we can still compile-test the driver but
not have it built-in if SMD is a module, to avoid running in the broken
configuration, and changes the Makefile to provide the driver under
a different module name.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Fixes: bdabad3e363d ("net: Add Qualcomm IPC router") Signed-off-by: David S. Miller <davem@davemloft.net>
Courtney Cavin [Fri, 6 May 2016 14:09:08 +0000 (07:09 -0700)]
net: Add Qualcomm IPC router
Add an implementation of Qualcomm's IPC router protocol, used to
communicate with service providing remote processors.
Signed-off-by: Courtney Cavin <courtney.cavin@sonymobile.com> Signed-off-by: Bjorn Andersson <bjorn.andersson@sonymobile.com>
[bjorn: Cope with 0 being a valid node id and implement RTM_NEWADDR] Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Tom Herbert [Mon, 7 Mar 2016 22:11:06 +0000 (14:11 -0800)]
kcm: Kernel Connection Multiplexor module
This module implements the Kernel Connection Multiplexor.
Kernel Connection Multiplexor (KCM) is a facility that provides a
message based interface over TCP for generic application protocols.
With KCM an application can efficiently send and receive application
protocol messages over TCP using datagram sockets.
For more information see the included Documentation/networking/kcm.txt
Signed-off-by: Tom Herbert <tom@herbertland.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Bjorn Andersson [Fri, 3 Jun 2016 00:48:28 +0000 (17:48 -0700)]
tty: serial: msm: Don't read off end of tx fifo
For dm uarts in pio mode tx data is transferred to the fifo register 4
bytes at a time, but care is not taken when these 4 bytes spans the end
of the xmit buffer so the loop might read up to 3 bytes past the buffer
and then skip the actual data at the beginning of the buffer.
Fix this by, analogous to the DMA case, make sure the chunk doesn't
wrap the xmit buffer.
Fixes: 3a878c430fd6 ("tty: serial: msm: Add TX DMA support") Cc: Andy Gross <andy.gross@linaro.org> Cc: Ivan Ivanov <iivanov.xz@gmail.com> Cc: Stephen Boyd <sboyd@codeaurora.org> Cc: stable@vger.kernel.org Reported-by: Frank Rowand <frowand.list@gmail.com> Reported-by: Nicolas Dechesne <nicolas.dechesne@linaro.org> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Rob Herring [Mon, 11 Apr 2016 23:23:51 +0000 (18:23 -0500)]
drm/msm/mdp: Add support for more RGBX formats
Android needs XBGR8888 format. Add all the missing 32-bpp formats
without alpha for completeness.
Cc: Archit Taneja <architt@codeaurora.org> Cc: Rob Clark <robdclark@gmail.com> Signed-off-by: Rob Herring <robh@kernel.org> Reviewed-by: Archit Taneja <architt@codeaurora.org> Signed-off-by: Rob Clark <robdclark@gmail.com>
Nicolas Dechesne [Fri, 27 May 2016 13:52:25 +0000 (15:52 +0200)]
configs: move DEBUG configs in their own config file
* remove DEBUG options from distro.config and defconfig since some of them had
some drastic impact on system performance
* add debug.config which can be convenient to build a debug flavor of the
kernel:
make defconfig distro.config debug.config
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>
Andy Gross [Tue, 2 Feb 2016 23:00:53 +0000 (17:00 -0600)]
spi: qup: Fix DMA mode to work correctly
This patch fixes a few issues with the DMA mode. The QUP needs to be
placed in the run mode before the DMA transactions are executed. The
conditions for being able to DMA vary between revisions of the QUP.
This is due to v1.1.1 using ADM DMA and later revisions using BAM.
Andy Gross [Sat, 30 Jan 2016 04:06:50 +0000 (22:06 -0600)]
spi: qup: Fix transaction done signaling
Wait to signal done until we get all of the interrupts we are expecting
to get for a transaction. If we don't wait for the input done flag, we
can be inbetween transactions when the done flag comes in and this can
mess up the next transaction.
CC: Grant Grundler <grundler@chromium.org> CC: Sarthak Kukreti <skukreti@codeaurora.org> Signed-off-by: Andy Gross <andy.gross@linaro.org>
The newly added STM code uses SRCU, but does not ensure that
this code is part of the kernel:
drivers/built-in.o: In function `stm_source_link_show':
include/linux/srcu.h:221: undefined reference to `__srcu_read_lock'
include/linux/srcu.h:238: undefined reference to `__srcu_read_unlock'
drivers/built-in.o: In function `stm_source_link_drop':
include/linux/srcu.h:221: undefined reference to `__srcu_read_lock'
include/linux/srcu.h:238: undefined reference to `__srcu_read_unlock'
This adds a Kconfig 'select' statement like all the other SRCU using
drivers have.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Fixes: 7bd1d4093c2f ("stm class: Introduce an abstraction for System Trace Module devices") Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc-6 found a dubious indentation in the megasas_mgmt_fw_ioctl
function:
drivers/scsi/megaraid/megaraid_sas_base.c: In function 'megasas_mgmt_fw_ioctl':
drivers/scsi/megaraid/megaraid_sas_base.c:6658:4: warning: statement is indented as if it were guarded by... [-Wmisleading-indentation]
kbuff_arr[i] = NULL;
^~~~~~~~~
drivers/scsi/megaraid/megaraid_sas_base.c:6653:3: note: ...this 'if' clause, but it is not
if (kbuff_arr[i])
^~
The code is actually correct, as there is no downside in clearing a NULL
pointer again.
This clarifies the code and avoids the warning by adding extra curly
braces.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Fixes: 90dc9d98f01b ("megaraid_sas : MFI MPT linked list corruption fix") Reviewed-by: Hannes Reinecke <hare@suse.com> Acked-by: Sumit Saxena <sumit.saxena@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
sunrpc_cache_pipe_upcall() can detect a race if CACHE_PENDING is no longer
set. In this case it aborts the queuing of the upcall.
However it has already taken a new counted reference on "h" and
doesn't "put" it, even though it frees the data structure holding the reference.
So let's delay the "cache_get" until we know we need it.
Fixes: f9e1aedc6c79 ("sunrpc/cache: remove races with queuing an upcall.") Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As the Dan report the smatch check the thermal driver warning:
drivers/thermal/rockchip_thermal.c:551 rockchip_configure_from_dt()
warn: impossible condition '(thermal->tshut_temp > ((~0 >> 1))) =>
(s32min-s32max > s32max)'
Although The shut_temp read from DT is u32,the temperature is currently
represented as int not long in the thermal driver.
Let's change to make shut_temp instead of the thermal->tshut_temp for
the condition.
Fixes: commit 437df2172e8d
("thermal: rockchip: consistently use int for temperatures")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Caesar Wang <wxt@rock-chips.com> Signed-off-by: Eduardo Valentin <edubezval@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Prior to 3.13 make allmodconfig KCONFIG_ALLCONFIG=/dev/null used
to be equivalent to make allmodconfig; these days it hardwires MODULES to n.
In fact, any KCONFIG_ALLCONFIG that doesn't set MODULES explicitly is
treated as if it set it to n.
Regression had been introduced by commit cfa98f ("kconfig: do not
override symbols already set"); what happens is that conf_read_simple()
does sym_calc_value(modules_sym) on exit, which leaves SYMBOL_VALID set and
has conf_set_all_new_symbols() skip modules_sym.
It's pretty easy to fix - simply move that call of sym_calc_value()
into the callers, except for the ones in KCONFIG_ALLCONFIG handling.
Objections?
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Fixes: cfa98f2e0ae9 ("kconfig: do not override symbols already set") Signed-off-by: Michal Marek <mmarek@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver should not use device_set_wakeup_enable() which is the policy
for user to decide.
Using device_init_wakeup() to initialize dev->power.should_wakeup and
dev->power.can_wakeup on driver initialization.
And use device_may_wakeup() on suspend to decide if WoL function should
be enabled on NIC.
Reported-by: Diego Viola <diego.viola@gmail.com> Signed-off-by: Guo-Fu Tseng <cooldavid@cooldavid.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Otherwise it might be back on resume right after going to suspend in
some hardware.
Reported-by: Diego Viola <diego.viola@gmail.com> Signed-off-by: Guo-Fu Tseng <cooldavid@cooldavid.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since the switch from mmp_pdma to pxa_dma driver for pxa architectures,
the pxa_dma requires 2 arguments, namely the requestor line and the
requested priority.
Fix the only left device node which was still passing only one argument,
making the pxa3xx-nand driver misbehave in a device-tree configuration,
ie. failing all data transfers.
Fixes: c943646d1f49 ("ARM: dts: pxa: add dma engine node to pxa3xx-nand") Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Armada 375 has the same SATA IP as Armada 370 and Armada XP, which
requires the PHY speed to be set in the LP_PHY_CTL register for SATA
hotplug to work.
Therefore, this commit updates the compatible string used to describe
the SATA IP in Armada 375 from marvell,orion-sata to
marvell,armada-370-sata.
Fixes: 4de59085091f753d08c8429d756b46756ab94665 ("ARM: mvebu: add Device Tree description of the Armada 375 SoC") Signed-off-by: Lior Amsalem <alior@marvell.com> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We cannot select a symbol that has disabled dependencies, so
we get a warning if we ever enable EXYNOS_THERMAL without
also turning on THERMAL_OF:
warning: (ARCH_EXYNOS) selects EXYNOS_THERMAL which has unmet direct dependencies (THERMAL && (ARCH_EXYNOS || COMPILE_TEST) && THERMAL_OF)
This adds another 'select' in the platform code to avoid that
case. Alternatively, we could decide to not select EXYNOS_THERMAL
here and instead make it a user option.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Fixes: f87e6bd3f740 ("thermal: exynos: Add the dependency of CONFIG_THERMAL_OF instead of CONFIG_OF") Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The atlas7 clock controller driver registers a reset controller
for itself, which causes a link error when the subsystem is
disabled:
drivers/built-in.o: In function `atlas7_clk_init':
drivers/clk/sirf/clk-atlas7.c:1681: undefined reference to `reset_controller_register'
As the clk driver does not have a Kconfig symbol for itself
but it always built-in when the platform is enabled, we have
to ensure that the reset controller subsystem is also built-in
in this case.
Based on CPU type choose generic omap3 or omap3430 specific cpuidle
parameters. Parameters for omap3430 were measured on Nokia N900 device and
added by commit 5a1b1d3a9efa ("OMAP3: RX-51: Pass cpu idle parameters")
which were later removed by commit 231900afba52 ("ARM: OMAP3: cpuidle -
remove rx51 cpuidle parameters table") due to huge code complexity.
This patch brings cpuidle parameters for omap3430 devices again, but uses
simple condition based on CPU type.
When doing delayed allocation, update of on-disk inode size is postponed
until IO submission time. However hole punch or zero range fallocate
calls can end up discarding the tail page cache page and thus on-disk
inode size would never be properly updated.
Make sure the on-disk inode size is updated before truncating page
cache.
Signed-off-by: Jan Kara <jack@suse.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Current code implementing FALLOC_FL_COLLAPSE_RANGE and
FALLOC_FL_INSERT_RANGE is prone to races with buffered writes and page
faults. If buffered write or write via mmap manages to squeeze between
filemap_write_and_wait_range() and truncate_pagecache() in the fallocate
implementations, the written data is simply discarded by
truncate_pagecache() although it should have been shifted.
Fix the problem by moving filemap_write_and_wait_range() call inside
i_mutex and i_mmap_sem. That way we are protected against races with
both buffered writes and page faults.
Signed-off-by: Jan Kara <jack@suse.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently ext4_alloc_file_blocks() was handling protection against
unlocked DIO. However we now need to sometimes call it under i_mmap_sem
and sometimes not and DIO protection ranks above it (although strictly
speaking this cannot currently create any deadlocks). Also
ext4_zero_range() was actually getting & releasing unlocked DIO
protection twice in some cases. Luckily it didn't introduce any real bug
but it was a land mine waiting to be stepped on. So move DIO protection
out from ext4_alloc_file_blocks() into the two callsites.
Signed-off-by: Jan Kara <jack@suse.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently, page faults and hole punching are completely unsynchronized.
This can result in page fault faulting in a page into a range that we
are punching after truncate_pagecache_range() has been called and thus
we can end up with a page mapped to disk blocks that will be shortly
freed. Filesystem corruption will shortly follow. Note that the same
race is avoided for truncate by checking page fault offset against
i_size but there isn't similar mechanism available for punching holes.
Fix the problem by creating new rw semaphore i_mmap_sem in inode and
grab it for writing over truncate, hole punching, and other functions
removing blocks from extent tree and for read over page faults. We
cannot easily use i_data_sem for this since that ranks below transaction
start and we need something ranking above it so that it can be held over
the whole truncate / hole punching operation. Also remove various
workarounds we had in the code to reduce race window when page fault
could have created pages with stale mapping information.
Signed-off-by: Jan Kara <jack@suse.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When this feature was introduced a check was made if there was a
resolved symbol under the cursor, it got lost in commit ea7cd5923309
("perf hists browser: Split popup menu actions - part 2"), reinstate it.
Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org>, Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Wang Nan <wangnan0@huawei.com> Fixes: ea7cd5923309 ("perf hists browser: Split popup menu actions - part 2") Link: http://lkml.kernel.org/r/1452960197-5323-9-git-send-email-namhyung@kernel.org
[ Carved out from a larger patch ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit 807f16d4db95 ("mtd: core: set some defaults
when dev.parent is set"), it's now legal for drivers
to call nand_scan and nand_scan_ident without setting
mtd.owner.
Drop the check and while at it remove the BUG() abuse.
Fixes: 807f16d4db95 ("mtd: core: set some defaults when dev.parent is set") Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar> Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Brian: editorial note - while commit 807f16d4db95 wasn't explicitly
broken, some follow-up commits in the v4.4 release broke a few
drivers, since they would hit this BUG() if they used nand_scan()
and were built as modules] Signed-off-by: Brian Norris <computersforpeace@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The BRCMNAND controller revision 7.1 is almost 100% compatible with the
previous v6.0 register offset layout, except for the Correctable Error
Reporting Threshold registers. Fix this by adding another table with the
correct offsets for CORR_THRESHOLD and CORR_THRESHOLD_EXT.
This patch remove the micron_quad_enable() function which force the Quad
SPI mode. However, once this mode is enabled, the Micron memory expect ALL
commands to use the SPI 4-4-4 protocol. Hence a failure does occur when
calling spi_nor_wait_till_ready() right after the update of the Enhanced
Volatile Configuration Register (EVCR) in the micron_quad_enable() as
the SPI controller driver is not aware about the protocol change.
Since there is almost no performance increase using Fast Read 4-4-4
commands instead of Fast Read 1-1-4 commands, we rather keep on using the
Extended SPI mode than enabling the Quad SPI mode.
Let's take the example of the pretty standard use of 8 dummy cycles during
Fast Read operations on 64KB erase sectors:
Fast Read 1-1-4 requires 8 cycles for the command, then 24 cycles for the
3byte address followed by 8 dummy clock cycles and finally 65536*2 cycles
for the read data; so 131112 clock cycles.
On the other hand the Fast Read 4-4-4 would require 2 cycles for the
command, then 6 cycles for the 3byte address followed by 8 dummy clock
cycles and finally 65536*2 cycles for the read data. So 131088 clock
cycles. The theorical bandwidth increase is 0.0%.
Now using Fast Read operations on 512byte pages:
Fast Read 1-1-4 needs 8+24+8+(512*2) = 1064 clock cycles whereas Fast
Read 4-4-4 would requires 2+6+8+(512*2) = 1040 clock cycles. Hence the
theorical bandwidth increase is 2.3%.
Consecutive reads for non sequential pages is not a relevant use case so
The Quad SPI mode is not worth it.
mtd_speedtest seems to confirm these figures.
Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Fixes: 548cd3ab54da ("mtd: spi-nor: Add quad I/O support for Micron SPI NOR") Signed-off-by: Brian Norris <computersforpeace@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The BSP team noticed that there is spin/mutex lock issue on sh-sci when
CPUFREQ is used. The issue is that the notifier function may call
mutex_lock() while the spinlock is held, which can lead to a BUG().
This may happen if CPUFREQ is changed while another CPU calls
clk_get_rate().
Taking the spinlock was added to the notifier function in commit e552de2413edad1a ("sh-sci: add platform device private data"), to
protect the list of serial ports against modification during traversal.
At that time the Common Clock Framework didn't exist yet, and
clk_get_rate() just returned clk->rate without taking a mutex.
Note that since commit d535a2305facf9b4 ("serial: sh-sci: Require a
device per port mapping."), there's no longer a list of serial ports to
traverse, and taking the spinlock became superfluous.
To fix the issue, just remove the cpufreq notifier:
1. The notifier doesn't work correctly: all it does is update stored
clock rates; it does not update the divider in the hardware.
The divider will only be updated when calling sci_set_termios().
I believe this was broken back in 2004, when the old
drivers/char/sh-sci.c driver (where the notifier did update the
divider) was replaced by drivers/serial/sh-sci.c (where the
notifier just updated port->uartclk).
Cfr. full-history-linux commits 6f8deaef2e9675d9 ("[PATCH] sh: port
sh-sci driver to the new API") and 3f73fe878dc9210a ("[PATCH]
Remove old sh-sci driver").
2. On modern SoCs, the sh-sci parent clock rate is no longer related
to the CPU clock rate anyway, so using a cpufreq notifier is
futile.
ext4_reserve_inode_write() in ext4_mark_inode_dirty() could fail on
error (e.g. EIO) and iloc.bh can be NULL in this case. But the error is
ignored in the following "if" condition and ext4_expand_extra_isize()
might be called with NULL iloc.bh set, which triggers NULL pointer
dereference.
This is uncovered by commit 8b4953e13f4c ("ext4: reserve code points for
the project quota feature"), which enlarges the ext4_inode size, and
run the following script on new kernel but with old mke2fs:
Because Linux might use bigger pages than the 4K pages to handle those mmio
ioremaps, the kmmio code shouldn't rely on the pade id as it currently does.
Using the memory address instead of the page id lets us look up how big the
page is and what its base address is, so that we won't get a page fault
within the same page twice anymore.
Tested-by: Pierre Moreau <pierre.morrow@free.fr> Signed-off-by: Karol Herbst <nouveau@karolherbst.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Borislav Petkov <bp@alien8.de> Cc: Brian Gerst <brgerst@gmail.com> Cc: Denys Vlasenko <dvlasenk@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Luis R. Rodriguez <mcgrof@suse.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Toshi Kani <toshi.kani@hp.com> Cc: linux-mm@kvack.org Cc: linux-x86_64@vger.kernel.org Cc: nouveau@lists.freedesktop.org Cc: pq@iki.fi Cc: rostedt@goodmis.org Link: http://lkml.kernel.org/r/1456966991-6861-1-git-send-email-nouveau@karolherbst.de Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The regmap_irq_get_virq() can return 0 or -EINVAL in error conditions
but driver checked only for value of 0.
This could lead to a cast of -EINVAL to an unsigned int used as a
interrupt number for devm_request_threaded_irq(). Although this is not
yet fatal (devm_request_threaded_irq() will just fail with -EINVAL) but
might be a misleading when diagnosing errors.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com> Fixes: 6f1c1e71d933 ("mfd: max77686: Convert to use regmap_irq") Reviewed-by: Javier Martinez Canillas <javier@osg.samsung.com> Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We call spin_lock_irqrestore with "flags" set to zero instead of to the
value from spin_lock_irqsave().
Fixes: aaaf5fbf56f1 ('rtc: add driver for DS1685 family of real time clocks') Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 30e7a65b3fdb (PM / Domains: Ensure subdomain is not in use
before removing) added a test to ensure that a subdomain is not a
master to another subdomain or if any devices are using the subdomain
before removing. This change incorrectly used the "slave_links" list to
determine if the subdomain is a master to another subdomain, where it
should have been using the "master_links" list instead. The
"slave_links" list will never be empty for a subdomain and so a
subdomain can never be removed. Fix this by testing if the
"master_links" list is empty instead.
Fixes: 30e7a65b3fdb (PM / Domains: Ensure subdomain is not in use before removing) Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Reviewed-by: Thierry Reding <treding@nvidia.com> Acked-by: Ulf Hansson <ulf.hansson@linaro.org> Acked-by: Kevin Hilman <khilman@baylibre.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We kept u_volt_min/max initialized to 0, when only the target voltage is
present in DT, instead of the target/min/max triplet.
This didn't go well with the regulator framework, as on few calls the
min voltage was set to target and max was set to 0 and so resulted in a
kernel crash like below:
kernel BUG at ../drivers/regulator/core.c:216!
[<c0684af4>] (regulator_check_voltage) from [<c06857ac>] (regulator_set_voltage_unlocked+0x58/0x230)
[<c06857ac>] (regulator_set_voltage_unlocked) from [<c06859ac>] (regulator_set_voltage+0x28/0x54)
[<c06859ac>] (regulator_set_voltage) from [<c0775b28>] (_set_opp_voltage+0x30/0x98)
[<c0775b28>] (_set_opp_voltage) from [<c0776630>] (dev_pm_opp_set_rate+0xf0/0x28c)
[<c0776630>] (dev_pm_opp_set_rate) from [<c096f784>] (__cpufreq_driver_target+0x184/0x2b4)
[<c096f784>] (__cpufreq_driver_target) from [<c0973760>] (dbs_check_cpu+0x1b0/0x1f4)
[<c0973760>] (dbs_check_cpu) from [<c0973f30>] (cpufreq_governor_dbs+0x324/0x5c4)
[<c0973f30>] (cpufreq_governor_dbs) from [<c0970958>] (__cpufreq_governor+0xe4/0x1ec)
[<c0970958>] (__cpufreq_governor) from [<c09711e0>] (cpufreq_init_policy+0x64/0x8c)
[<c09711e0>] (cpufreq_init_policy) from [<c09718cc>] (cpufreq_online+0x2fc/0x708)
[<c09718cc>] (cpufreq_online) from [<c0765ff0>] (subsys_interface_register+0x94/0xd8)
[<c0765ff0>] (subsys_interface_register) from [<c0970530>] (cpufreq_register_driver+0x14c/0x19c)
[<c0970530>] (cpufreq_register_driver) from [<c09746dc>] (dt_cpufreq_probe+0x70/0xec)
[<c09746dc>] (dt_cpufreq_probe) from [<c076907c>] (platform_drv_probe+0x4c/0xb0)
[<c076907c>] (platform_drv_probe) from [<c07678e0>] (driver_probe_device+0x214/0x2c0)
[<c07678e0>] (driver_probe_device) from [<c0767a18>] (__driver_attach+0x8c/0x90)
[<c0767a18>] (__driver_attach) from [<c0765c2c>] (bus_for_each_dev+0x68/0x9c)
[<c0765c2c>] (bus_for_each_dev) from [<c0766d78>] (bus_add_driver+0x1a0/0x218)
[<c0766d78>] (bus_add_driver) from [<c076810c>] (driver_register+0x78/0xf8)
[<c076810c>] (driver_register) from [<c0301d74>] (do_one_initcall+0x90/0x1d8)
[<c0301d74>] (do_one_initcall) from [<c1100e14>] (kernel_init_freeable+0x15c/0x1fc)
[<c1100e14>] (kernel_init_freeable) from [<c0b27a0c>] (kernel_init+0x8/0xf0)
[<c0b27a0c>] (kernel_init) from [<c0307d78>] (ret_from_fork+0x14/0x3c)
Code: e1550004baffffebe3a00000e8bd8070 (e7f001f2)
Fix that by initializing u_volt_min/max to the target voltage in such cases.
Reported-and-tested-by: Krzysztof Kozlowski <k.kozlowski@samsung.com> Fixes: 274659029c9d (PM / OPP: Add support to parse "operating-points-v2" bindings) Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed integer overflow is undefined. Also I added a check for
"(offset < 0)" in scif_unregister() because that makes it match the
other conditions and because I didn't want to subtract a negative.
Fixes: ba612aa8b487 ('misc: mic: SCIF memory registration and unregistration') Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 985087dbcb02 'misc: add support for bmp18x chips to the bmp085
driver' changed the BMP085 config symbol to a boolean. I see no
reason why the shared code cannot be built as a module, so change it
back to tristate.
Fixes: 985087dbcb02 ("misc: add support for bmp18x chips to the bmp085 driver") Cc: Eric Andersson <eric.andersson@unixphere.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk> Acked-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The limbs are integers in the host endianness, so we can't simply
iterate over the individual bytes. The current code happens to work on
little-endian, because the order of the limbs in the MPI array is the
same as the order of the bytes in each limb, but it breaks on
big-endian.
Fixes: 0f74fbf77d45 ("MPI: Fix mpi_read_buffer") Signed-off-by: Michal Marek <mmarek@suse.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 028cd86b794f4a ("video: da8xx-fb: fix the polarities of the
hsync/vsync pulse") fixes polarities of HSYNC/VSYNC pulse but
forgot to update known_lcd_panels[] which had sync values
according to old logic. This breaks LCD at least on DA850 EVM.
This patch fixes this issue and I have tested this for panel
"Sharp_LK043T1DG01" using DA850 EVM board.
Fixes: 028cd86b794f4a ("video: da8xx-fb: fix the polarities of the hsync/vsync pulse") Signed-off-by: Sushaanth Srirangapathi <sushaanth.s@ti.com> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the scsi_dh core was moved into the scsi core module,
CONFIG_SCSI_DH became a 'bool' option, and now anything depending on it
can be built-in even when CONFIG_SCSI=m. This of course cannot link
successfully:
drivers/scsi/built-in.o: In function `rdac_init':
scsi_dh_alua.c:(.init.text+0x14): undefined reference to `scsi_register_device_handler'
scsi_dh_alua.c:(.init.text+0x64): undefined reference to `scsi_unregister_device_handler'
drivers/scsi/built-in.o: In function `alua_init':
scsi_dh_alua.c:(.init.text+0xb0): undefined reference to `scsi_register_device_handler'
As a workaround, this adds an extra dependency on CONFIG_SCSI, so
Kconfig can figure out whether built-in is allowed or not.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Fixes: 086b91d052eb ("scsi_dh: integrate into the core SCSI code") Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc-6.0 found an ancient bug in the paride driver, which had a
"module_param(verbose, bool, 0);" since before 2.6.12, but actually uses
it to accept '0', '1' or '2' as arguments:
drivers/block/paride/pd.c: In function 'pd_init_dev_parms':
drivers/block/paride/pd.c:298:29: warning: comparison of constant '1' with boolean expression is always false [-Wbool-compare]
#define DBMSG(msg) ((verbose>1)?(msg):NULL)
In 2012, Rusty did a cleanup patch that also changed the type of the
variable to 'bool', which introduced what is now a gcc warning.
This changes the type back to 'int' and adapts the module_param() line
instead, so it should work as documented in case anyone ever cares about
running the ancient driver with debugging.
The s5m8767_pmic_probe() function calls s5m8767_get_register() to
read data without checking the return code, which produces a compile-time
warning when that data is accessed:
drivers/regulator/s5m8767.c: In function 's5m8767_pmic_probe':
drivers/regulator/s5m8767.c:924:7: error: 'enable_reg' may be used uninitialized in this function [-Werror=maybe-uninitialized]
drivers/regulator/s5m8767.c:944:30: error: 'enable_val' may be used uninitialized in this function [-Werror=maybe-uninitialized]
This changes the s5m8767_get_register() function to return a -EINVAL
not just for an invalid register number but also for an invalid
regulator number, as both would result in returning uninitialized
data. The s5m8767_pmic_probe() function is then changed accordingly
to fail on a read error, as all the other callers of s5m8767_get_register()
already do.
In practice this probably cannot happen, as we don't call
s5m8767_get_register() with invalid arguments, but the gcc
warning seems valid in principle, in terms writing safe
error checking.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Fixes: 9c4c60554acf ("regulator: s5m8767: Convert to use regulator_[enable|disable|is_enabled]_regmap") Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The of_io_request_and_map() returns a valid pointer in iomem region or
ERR_PTR(), check for NULL always fails and may cause a NULL pointer
dereference on error path.
Fixes: 25e34b44313b ("irqchip/mxs: Prepare driver for hardware with different offsets") Signed-off-by: Vladimir Zapolskiy <vz@mleia.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Oleksij Rempel <linux@rempel-privat.de> Cc: Sascha Hauer <kernel@pengutronix.de> Cc: Shawn Guo <shawnguo@kernel.org> Cc: linux-arm-kernel@lists.infradead.org Link: http://lkml.kernel.org/r/1457486500-10237-1-git-send-email-vz@mleia.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The of_io_request_and_map() returns a valid pointer in iomem region or
ERR_PTR(), check for NULL always fails and may cause a NULL pointer
dereference on error path.
Fixes: 0e841b04c829 ("irqchip/sunxi-nmi: Switch to of_io_request_and_map() from of_iomap()") Signed-off-by: Vladimir Zapolskiy <vz@mleia.com> Cc: Jason Cooper <jason@lakedaemon.net> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Chen-Yu Tsai <wens@csie.org> Cc: Maxime Ripard <maxime.ripard@free-electrons.com> Cc: linux-arm-kernel@lists.infradead.org Link: http://lkml.kernel.org/r/1457486489-10189-1-git-send-email-vz@mleia.com Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Rockchip_spi_set_cs could be called by spi_setup, but
spi_setup may be called by device driver after runtime suspend.
Then the spi clock is closed, rockchip_spi_set_cs may access the
spi registers, which causes cpu block in some socs.
Fixes: 64e36824b32 ("spi/rockchip: add driver for Rockchip RK3xxx") Signed-off-by: Huibin Hong <huibin.hong@rock-chips.com> Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Similar to commit b4b29f94856a ("locking/osq: Fix ordering of node
initialisation in osq_lock") the use of xchg_acquire() is
fundamentally broken with MCS like constructs.
Furthermore, it turns out we rely on the global transitivity of this
operation because the unlock path observes the pointer with a
READ_ONCE(), not an smp_load_acquire().
This is non-critical because the MCS code isn't actually used and
mostly serves as documentation, a stepping stone to the more complex
things we've build on top of the idea.
Reported-by: Andrea Parri <parri.andrea@gmail.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Will Deacon <will.deacon@arm.com> Fixes: 3552a07a9c4a ("locking/mcs: Use acquire/release semantics") Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit fa731ac7ea04 ("regulator: core: avoid unused variable warning")
introduced a subtle change in how supplies are locked. Where previously
code was always locking the regulator of the current iteration, the new
implementation only locks the regulator if it has a supply. For any
given power tree that means that the root will never get locked.
On the other hand the regulator_unlock_supply() will still release all
the locks, which in turn causes the lock debugging code to warn about a
mutex being unlocked which wasn't locked.
Cc: Mark Brown <broonie@kernel.org> Cc: Arnd Bergmann <arnd@arndb.de> Fixes: fa731ac7ea04 ("regulator: core: avoid unused variable warning") Signed-off-by: Thierry Reding <treding@nvidia.com> Signed-off-by: Mark Brown <broonie@kernel.org> Cc: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The latest workaround for the lockdep interface's not using the second
argument of mutex_lock_nested() changed the loop missed locking the last
regulator due to a thinko with the loop termination condition exiting
one regulator too soon.
Reported-by: Tyler Baker <tyler.baker@linaro.org> Signed-off-by: Mark Brown <broonie@kernel.org> Cc: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When a buffer is being dequeued using VIDIOC_DQBUF IOCTL, the exact buffer
which will be dequeued is not known until the buffer has been removed from
the queue. The number of planes is specific to a buffer, not to the queue.
This does lead to the situation where multi-plane buffers may be requested
and queued with n planes, but VIDIOC_DQBUF IOCTL may be passed an argument
struct with fewer planes.
__fill_v4l2_buffer() however uses the number of planes from the dequeued
videobuf2 buffer, overwriting kernel memory (the m.planes array allocated
in video_usercopy() in v4l2-ioctl.c) if the user provided fewer
planes than the dequeued buffer had. Oops!
Fixes: b0e0e1f83de3 ("[media] media: videobuf2: Prepare to divide videobuf2") Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Acked-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The number of planes in videobuf2 is specific to a buffer. In order to
verify that the planes array provided by the user is long enough, a new
vb2_buf_op is required.
Call __verify_planes_array() when the dequeued buffer is known. Return an
error to the caller if there was one, otherwise remove the buffer from the
done list.
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Acked-by: Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix potential out-of-bounds write to urb->transfer_buffer
usbip handles network communication directly in the kernel. When receiving a
packet from its peer, usbip code parses headers according to protocol. As
part of this parsing urb->actual_length is filled. Since the input for
urb->actual_length comes from the network, it should be treated as untrusted.
Any entity controlling the network may put any value in the input and the
preallocated urb->transfer_buffer may not be large enough to hold the data.
Thus, the malicious entity is able to write arbitrary data to kernel memory.
There are three subsystem callbacks in css shutdown path -
css_offline(), css_released() and css_free(). Except for
css_released(), cgroup core didn't guarantee the order of invocation.
css_offline() or css_free() could be called on a parent css before its
children. This behavior is unexpected and led to bugs in cpu and
memory controller.
The previous patch updated ordering for css_offline() which fixes the
cpu controller issue. While there currently isn't a known bug caused
by misordering of css_free() invocations, let's fix it too for
consistency.
css_free() ordering can be trivially fixed by moving putting of the
parent css below css_free() invocation.
Currently, migration code increses num_poisoned_pages on *failed*
migration page as well as successfully migrated one at the trial of
memory-failure. It will make the stat wrong. As well, it marks the
page as PG_HWPoison even if the migration trial failed. It would mean
we cannot recover the corrupted page using memory-failure facility.
We have been reclaimed highmem zone if buffer_heads is over limit but
commit 6b4f7799c6a5 ("mm: vmscan: invoke slab shrinkers from
shrink_zone()") changed the behavior so it doesn't reclaim highmem zone
although buffer_heads is over the limit. This patch restores the logic.
Fixes: 6b4f7799c6a5 ("mm: vmscan: invoke slab shrinkers from shrink_zone()") Signed-off-by: Minchan Kim <minchan@kernel.org> Cc: Johannes Weiner <hannes@cmpxchg.org> 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>
In gather_pte_stats() a THP pmd is cast into a pte, which is wrong
because the layouts may differ depending on the architecture. On s390
this will lead to inaccurate numa_maps accounting in /proc because of
misguided pte_present() and pte_dirty() checks on the fake pte.
On other architectures pte_present() and pte_dirty() may work by chance,
but there may be an issue with direct-access (dax) mappings w/o
underlying struct pages when HAVE_PTE_SPECIAL is set and THP is
available. In vm_normal_page() the fake pte will be checked with
pte_special() and because there is no "special" bit in a pmd, this will
always return false and the VM_PFNMAP | VM_MIXEDMAP checking will be
skipped. On dax mappings w/o struct pages, an invalid struct page
pointer would then be returned that can crash the kernel.
This patch fixes the numa_maps THP handling by introducing new "_pmd"
variants of the can_gather_numa_stats() and vm_normal_page() functions.
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Jerome Marchand <jmarchan@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Michael Holzheu <holzheu@linux.vnet.ibm.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>
Khugepaged detects own VMAs by checking vm_file and vm_ops but this way
it cannot distinguish private /dev/zero mappings from other special
mappings like /dev/hpet which has no vm_ops and popultes PTEs in mmap.
This fixes false-positive VM_BUG_ON and prevents installing THP where
they are not expected.
Link: http://lkml.kernel.org/r/CACT4Y+ZmuZMV5CjSFOeXviwQdABAgT7T+StKfTqan9YDtgEi5g@mail.gmail.com Fixes: 78f11a255749 ("mm: thp: fix /dev/zero MAP_PRIVATE and vm_flags cleanups") Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Andrea Arcangeli <aarcange@redhat.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>
So, this ended up a lot simpler than I originally expected. I tested
it lightly and it seems to work fine. Petr, can you please test these
two patches w/o the lru drain drop patch and see whether the problem
is gone?
Thanks.
------ 8< ------
If charge moving is used, memcg performs relabeling of the affected
pages from its ->attach callback which is called under both
cgroup_threadgroup_rwsem and thus can't create new kthreads. This is
fragile as various operations may depend on workqueues making forward
progress which relies on the ability to create new kthreads.
There's no reason to perform charge moving from ->attach which is deep
in the task migration path. Move it to ->post_attach which is called
after the actual migration is finished and cgroup_threadgroup_rwsem is
dropped.
* move_charge_struct->mm is added and ->can_attach is now responsible
for pinning and recording the target mm. mem_cgroup_clear_mc() is
updated accordingly. This also simplifies mem_cgroup_move_task().
* mem_cgroup_move_task() is now called from ->post_attach instead of
->attach.
Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@kernel.org> Debugged-and-tested-by: Petr Mladek <pmladek@suse.com> Reported-by: Cyril Hrubis <chrubis@suse.cz> Reported-by: Johannes Weiner <hannes@cmpxchg.org> Fixes: 1ed1328792ff ("sched, cgroup: replace signal_struct->group_rwsem with a global percpu_rwsem") Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since e93ad19d0564 ("cpuset: make mm migration asynchronous"), cpuset
kicks off asynchronous NUMA node migration if necessary during task
migration and flushes it from cpuset_post_attach_flush() which is
called at the end of __cgroup_procs_write(). This is to avoid
performing migration with cgroup_threadgroup_rwsem write-locked which
can lead to deadlock through dependency on kworker creation.
memcg has a similar issue with charge moving, so let's convert it to
an official callback rather than the current one-off cpuset specific
function. This patch adds cgroup_subsys->post_attach callback and
makes cpuset register cpuset_post_attach_flush() as its ->post_attach.
The conversion is mostly one-to-one except that the new callback is
called under cgroup_mutex. This is to guarantee that no other
migration operations are started before ->post_attach callbacks are
finished. cgroup_mutex is one of the outermost mutex in the system
and has never been and shouldn't be a problem. We can add specialized
synchronization around __cgroup_procs_write() but I don't think
there's any noticeable benefit.
Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Li Zefan <lizefan@huawei.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This change is primarily an attempt to make it easier to realize the
optimizations the compiler performs in-case CONFIG_MEMCG_KMEM is not
enabled.
Performance wise, even when CONFIG_MEMCG_KMEM is compiled in, the
overhead is zero. This is because, as long as no process have enabled
kmem cgroups accounting, the assignment is replaced by asm-NOP
operations. This is possible because memcg_kmem_enabled() uses a
static_key_false() construct.
It also helps readability as it avoid accessing the p[] array like:
p[size - 1] which "expose" that the array is processed backwards inside
helper function build_detached_freelist().
Lastly this also makes the code more robust, in error case like passing
NULL pointers in the array. Which were previously handled before commit 033745189b1b ("slub: add missing kmem cgroup support to
kmem_cache_free_bulk").
Fixes: 033745189b1b ("slub: add missing kmem cgroup support to kmem_cache_free_bulk") Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Vladimir Davydov <vdavydov@virtuozzo.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>
So blk_flush_plug_list() was called with from_schedule == true.
If from_schedule is true, that means that finally blk_mq_insert_requests()
offloads execution of __blk_mq_run_hw_queue() and uses kblockd workqueue,
i.e. it calls kblockd_schedule_delayed_work_on().
That means, that we race with another CPU, which is about to execute
__blk_mq_run_hw_queue() work.
Further debugging shows the following traces from different CPUs:
CPU#0 CPU#1
---------------------------------- -------------------------------
reqeust A inserted
STORE hctx->ctx_map[0] bit marked
kblockd_schedule...() returns 1
<schedule to kblockd workqueue>
request B inserted
STORE hctx->ctx_map[1] bit marked
kblockd_schedule...() returns 0
*** WORK PENDING bit is cleared ***
flush_busy_ctxs() is executed, but
bit 1, set by CPU#1, is not observed
As a result request B pended forever.
This behaviour can be explained by speculative LOAD of hctx->ctx_map on
CPU#0, which is reordered with clear of PENDING bit and executed _before_
actual STORE of bit 1 on CPU#1.
The proper fix is an explicit full barrier <mfence>, which guarantees
that clear of PENDING bit is to be executed before all possible
speculative LOADS or STORES inside actual work function.
Signed-off-by: Roman Pen <roman.penyaev@profitbricks.com> Cc: Gioh Kim <gi-oh.kim@profitbricks.com> Cc: Michael Wang <yun.wang@profitbricks.com> Cc: Tejun Heo <tj@kernel.org> Cc: Jens Axboe <axboe@kernel.dk> Cc: linux-block@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
projects / karo-tx-linux.git / log