When using RPCRDMA_MTHCAFMR memory registration, after a few
transport disconnect / reconnect cycles, ib_map_phys_fmr() starts to
return EINVAL because the provider has exhausted its map pool.
Make sure that all FMRs are unmapped during transport disconnect,
and that ->send_request remarshals them during an RPC retransmit.
This resets the transport's MRs to ensure that none are leaked
during a disconnect.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Recent work made FRMR registration and invalidation completions
unsignaled. This greatly reduces the adapter interrupt rate.
Every so often, however, a posted send Work Request is allowed to
signal. Otherwise, the provider's Work Queue will wrap and the
workload will hang.
The number of Work Requests that are allowed to remain unsignaled is
determined by the value of req_cqinit. Currently, this is set to the
size of the send Work Queue divided by two, minus 1.
For FRMR, the send Work Queue is the maximum number of concurrent
RPCs (currently 32) times the maximum number of Work Requests an
RPC might use (currently 7, though some adapters may need more).
For mlx4, this is 224 entries. This leaves completion signaling
disabled for 111 send Work Requests.
Some providers hold back dispatching Work Requests until a CQE is
generated. If completions are disabled, then no CQEs are generated
for quite some time, and that can stall the Work Queue.
I've seen this occur running xfstests generic/113 over NFSv4, where
eventually, posting a FAST_REG_MR Work Request fails with -ENOMEM
because the Work Queue has overflowed. The connection is dropped
and re-established.
Cap the rep_cqinit setting so completions are not left turned off
for too long.
BugLink: https://bugzilla.linux-nfs.org/show_bug.cgi?id=269
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
The RPC/RDMA send_request method and the chunk registration code
expects an errno from the registration function. This allows
the upper layers to distinguish between a recoverable failure
(for example, temporary memory exhaustion) and a hard failure
(for example, a bug in the registration logic).
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
We must use GFP_NOFS instead GFP_KERNEL inside ext4_mb_add_groupinfo
and ext4_calculate_overhead() because they are called from inside a
journal transaction. Call trace:
ioctl
->ext4_group_add
->journal_start
->ext4_setup_new_descs
->ext4_mb_add_groupinfo -> GFP_KERNEL
->ext4_flex_group_add
->ext4_update_super
->ext4_calculate_overhead -> GFP_KERNEL
->journal_stop
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Felipe writes:
usb: patches for v3.19 merge window
This time, a very pull request with 216 non-merge
commits. Most of the commits contained here are
sparse or coccinelle fixes ranging from missing
'static' to returning 0 in case of errors.
More importantly, we have the removal the now
unnecessary 'driver' argument to ->udc_stop().
DWC2 learned about Dual-Role builds. Users of
this IP can now have a single driver built for
host and device roles.
DWC3 got support for two new HW platforms: Exynos7
and AMD.
The Broadcom USB 3.0 Device Controller IP is now
supported and so is PLX USB338x, which means DWC3
has lost is badge as the only USB 3.0 peripheral
IP supported on Linux.
Thanks for Tony Lindgren's work, we can now have
a distro-like kernel where all MUSB glue layers
can be built into the same kernel (statically
or dynamically linked) and it'll work in PIO (DMA
will come probably on v3.20).
Other than these, the usual set of cleanups and
non-critical fixes.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Introduce a simple aging to extent status tree. Each extent has a
REFERENCED bit which gets set when the extent is used. Shrinker then
skips entries with referenced bit set and clears the bit. Thus
frequently used extents have higher chances of staying in memory.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Currently flags for extent status tree are defined twice, once shifted
and once without a being shifted. Consolidate these definitions into one
place and make some computations automatic to make adding flags less
error prone. Compiler should be clever enough to figure out these are
constants and generate the same code.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Currently we scan extent status trees of inodes until we reclaim nr_to_scan
extents. This can however require a lot of scanning when there are lots
of delayed extents (as those cannot be reclaimed).
Change shrinker to work as shrinkers are supposed to and *scan* only
nr_to_scan extents regardless of how many extents did we actually
reclaim. We however need to be careful and avoid scanning each status
tree from the beginning - that could lead to a situation where we would
not be able to reclaim anything at all when first nr_to_scan extents in
the tree are always unreclaimable. We remember with each inode offset
where we stopped scanning and continue from there when we next come
across the inode.
Note that we also need to update places calling __es_shrink() manually
to pass reasonable nr_to_scan to have a chance of reclaiming anything and
not just 1.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Currently callers adding extents to extent status tree were responsible
for adding the inode to the list of inodes with freeable extents. This
is error prone and puts list handling in unnecessarily many places.
Just add inode to the list automatically when the first non-delay extent
is added to the tree and remove inode from the list when the last
non-delay extent is removed.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
In this commit we discard the lru algorithm for inodes with extent
status tree because it takes significant effort to maintain a lru list
in extent status tree shrinker and the shrinker can take a long time to
scan this lru list in order to reclaim some objects.
We replace the lru ordering with a simple round-robin. After that we
never need to keep a lru list. That means that the list needn't be
sorted if the shrinker can not reclaim any objects in the first round.
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Currently extent status tree doesn't cache extent hole when a write
looks up in extent tree to make sure whether a block has been allocated
or not. In this case, we don't put extent hole in extent cache because
later this extent might be removed and a new delayed extent might be
added back. But it will cause a defect when we do a lot of writes. If
we don't put extent hole in extent cache, the following writes also need
to access extent tree to look at whether or not a block has been
allocated. It brings a cache miss. This commit fixes this defect.
Also if the inode doesn't have any extent, this extent hole will be
cached as well.
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
For bigalloc filesystems we have to check whether newly requested inode
block isn't already part of a cluster for which we already have delayed
allocation reservation. This check happens in ext4_ext_map_blocks() and
that function sets EXT4_MAP_FROM_CLUSTER if that's the case. However if
ext4_da_map_blocks() finds in extent cache information about the block,
we don't call into ext4_ext_map_blocks() and thus we always end up
getting new reservation even if the space for cluster is already
reserved. This results in overreservation and premature ENOSPC reports.
Fix the problem by checking for existing cluster reservation already in
ext4_da_map_blocks(). That simplifies the logic and actually allows us
to get rid of the EXT4_MAP_FROM_CLUSTER flag completely.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
After registering mfd device with proper irq_base
platform_get_irq_byname() calls will return VIRQ instead of local IRQ.
This fixes da9063 rtc registration issue:
da9063-rtc da9063-rtc: Failed to request ALARM IRQ 1: -22
Signed-off-by: Dmitry Lavnikevich <d.lavnikevich@sam-solutions.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
When the extra 4 channels were added to AIF2 the necessary frame control
registers were not given defaults and marked readable. This patch fixes
this.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
The TC3589x driver is now a device tree-only driver, so we want
only dynamic IRQs and GPIO numbers from the tc3589x, no static
assignments.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
These registers are documented in the datasheet and used as part of the
extcon driver. Expose them properly through regmap as the datasheet
notes they should be treated as volatile do so.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Technically this register is not used on wm8997 however the regmap core
requires a continuous block of IRQs. The simplest solution is just to
add the register.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Change clk_enable/disable() calls to clk_prepare_enable() and
clk_disable_unrepapre().
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
First spotted pointless (incorrect) indent of 'if (ret)', then double
indentations of a struct attribute 'mask'. Decided to go through the
whole file and make amendments instead and this is the result.
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Some boards with TC6393XB chip require full state restore during system
resume thanks to chip's VCC being cut off during suspend (Sharp SL-6000
tosa is one of them). Failing to do so would result in ohci Oops on
resume due to internal memory contentes being changed. Fail ohci suspend
on tc6393xb is full state restore is required.
Recommended workaround is to unbind tmio-ohci driver before suspend and
rebind it after resume.
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
MFD core already cares about failing registration. It will remove successfully
registered devices in case of error. Thus, no need to repeatedly call
mfd_remove_devices().
Fixes: 5829e9b64e (mfd: lpc_sch: Accomodate partial population of the MFD devices)
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Precedence of & and >> is not the same and is not left to right.
shift has higher precedence and should be done after the mask.
Add parentheses around the mask.
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Commit e7cd1d1eb1 ("mfd: twl4030-power: Add generic reset
configuration") accidentally removed the compatible flag for
"ti,twl4030-power" that should be there as documented in the
binding.
If "ti,twl4030-power" only the poweroff configuration is done
by the driver.
Fixes: e7cd1d1eb1 ("mfd: twl4030-power: Add generic reset configuration")
Cc: stable@vger.kernel.org # v3.16+
Reported-by: "Dr. H. Nikolaus Schaller" <hns@goldelico.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Fix rts5227&5249 failed send buffer cmd after suspend,
PM_CTRL3 should reset before send any buffer cmd after suspend.
Otherwise, buffer cmd will failed, this will lead resume fail.
Signed-off-by: Micky Ching <micky_ching@realsil.com.cn>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
This reverts commit b7cde7078d
("mfd: sec-core: Prepare regulators for suspend state to reduce power-consumption")
Commit b7cde7078d called regulator_suspend_prepare() to prepare the
regulators for a suspend state. But it did from the device pm suspend
handler while the regulator suspend prepare function iterates over all
regulators and not only the one managed by this device so it doesn't
seems to be correct to call it from within a device driver.
It is better to call the regulator suspend prepare/finish functions
from platform code instead so this patch reverts the mentioned commit.
Suggested-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Reviewed-by: Chanwoo Choi <cw00.choi@samsung.com>
Reviewed-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Add a "maxim,max77693-charger" of_compatible to the mfd_cell so the MFD
child device (the charger) will have its own of_node set. This will be
used by the max77693 charger driver in next patches to obtain battery
configuration from DTS.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
This was found whilst running checkpatch.pl on arizona-spi.
WARNING: Missing a blank line after declarations
+ struct arizona *arizona = spi_get_drvdata(spi);
+ arizona_dev_exit(arizona);
Signed-off-by: Will Sheppard <wsheppard@embedded-bits.co.uk>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Remove old MAX77693_NUM_IRQ_MUIC_REGS define. Not used anywhere.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
The HLCDC IP available on some Atmel SoCs (i.e. at91sam9n12, at91sam9x5
family or sama5d3 family) exposes 2 subdevices:
- a display controller (controlled by a DRM driver)
- a PWM chip
This patch adds documentation for atmel-hlcdc DT bindings.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Tested-by: Anthony Harivel <anthony.harivel@emtrion.de>
Tested-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
The HLCDC IP available on some Atmel SoCs (i.e. at91sam9n12, at91sam9x5
family or sama5d3 family) exposes 2 subdevices:
- a display controller (controlled by a DRM driver)
- a PWM chip
The MFD device provides a regmap and several clocks (those connected
to this hardware block) to its subdevices.
This way concurrent accesses to the iomem range are handled by the regmap
framework, and each subdevice can safely access HLCDC registers.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Tested-by: Anthony Harivel <anthony.harivel@emtrion.de>
Tested-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Make sure to always honour multi-function devices registered with
PLATFORM_DEVID_NONE (-1) or PLATFORM_DEVID_AUTO (-2) as id base. In this
case it does not make sense to append the cell id to the mfd-id base and
potentially change the requested behaviour.
Specifically this will allow multi-function devices to be registered
with PLATFORM_DEVID_AUTO while still having non-zero cell ids.
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Use mfd_add_hotplug_devices() helper to register the subdevices.
Compile-only tested.
Signed-off-by: Johan Hovold <johan@kernel.org>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Use mfd_add_hotplug_devices helper to register the subdevices.
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Currently a syscon entity can be only registered directly through a
platform device that binds to a dedicated syscon driver. However in
certain use cases it is desirable to make a device used with another
driver a syscon interface provider.
For example, certain SoCs (e.g. Exynos) contain system controller
blocks which perform various functions such as power domain control,
CPU power management, low power mode control, but in addition contain
certain IP integration glue, such as various signal masks,
coprocessor power control, etc. In such case, there is a need to have
a dedicated driver for such system controller but also share registers
with other drivers. The latter is where the syscon interface is helpful.
In case of DT based platforms, this patch decouples syscon object from
syscon platform driver, and allows to create syscon objects first time
when it is required by calling of syscon_regmap_lookup_by APIs and keep
a list of such syscon objects along with syscon provider device_nodes
and regmap handles.
For non-DT based platforms, this patch keeps syscon platform driver
structure so that syscon can be probed and such non-DT based drivers
can use syscon_regmap_lookup_by_pdev API and access regmap handles.
Once all users of "syscon_regmap_lookup_by_pdev" migrated to DT based,
we can completely remove platform driver of syscon, and keep only helper
functions to get regmap handles.
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Suggested-by: Tomasz Figa <tomasz.figa@gmail.com>
Tested-by: Vivek Gautam <gautam.vivek@samsung.com>
Tested-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: Pankaj Dubey <pankaj.dubey@samsung.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
This patch add haptic DT binding documentation and example
to support haptic driver in max77693 Multifunction device.
Signed-off-by: Jaewon Kim <jaewon02.kim@samsung.com>
Acked-by: Chanwoo Choi <cw00.choi@samsung.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
