In order to avoid blocking for an excessive amount of time, eventually
impacting on system responsiveness, interrupt handlers should finish
executing in as little time as possible.
Use threaded interrupt and move the SPI transfer handling (both
CPU and DMA) for the non-spimem case to an interrupt thread instead.
For SPI-MEM (IPM) controllers, handling is kept in the blocking
interrupt as it simply consists in signalling completion.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://patch.msgid.link/20240726114721.142196-1-angelogioacchino.delregno@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Replace a comma between expression statements by a semicolon.
Fixes: 3e0cf4d3fc ("spi: meson-spicc: add a linear clock divider support")
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patch.msgid.link/20240716091151.1434450-1-nichen@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
Zero and negative number is not a valid IRQ for in-kernel code and the
irq_of_parse_and_map() function returns zero on error. So this check for
valid IRQs should only accept values > 0.
Fixes: 44dab88e7c ("spi: add spi_ppc4xx driver")
Signed-off-by: Ma Ke <make24@iscas.ac.cn>
Link: https://patch.msgid.link/20240724084047.1506084-1-make24@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
MT7981 has SPI controllers based on IPM design
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240727114828.29558-1-zajec5@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This adds a check on xfer->len to avoid emitting an XFER_BITS
instruction for empty transfers in the AXI SPI Engine driver. This
avoids unnecessary delays caused by executing an instruction that has
no effect on the actual SPI transfer.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Link: https://patch.msgid.link/20240723-spi-axi-spi-engine-opt-bpw-v1-1-2625ba4c4387@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Implement MOSI idle low and MOSI idle high to better support peripherals
that request specific MOSI behavior.
Acked-by: Nuno Sa <nuno.sa@analog.com>
Reviewed-by: David Lechner <dlechner@baylibre.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Marcelo Schmitt <marcelo.schmitt@analog.com>
Link: https://patch.msgid.link/629c55a10005ba26825c3a6a19184372ef81b3e1.1720810545.git.marcelo.schmitt@analog.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Some SPI peripherals may require strict MOSI line state when the controller
is not clocking out data. Implement support for MOSI idle state
configuration (low or high) by setting the data output line level on
controller setup and after transfers. Bitbang operations now call
controller specific set_mosi_idle() callback to set MOSI to its idle state.
The MOSI line is kept at its idle state if no tx buffer is provided.
Acked-by: Nuno Sa <nuno.sa@analog.com>
Reviewed-by: David Lechner <dlechner@baylibre.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Marcelo Schmitt <marcelo.schmitt@analog.com>
Link: https://patch.msgid.link/de61a600b56ed9cb714d5ea87afa88948e70041e.1720810545.git.marcelo.schmitt@analog.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The behavior of an SPI controller data output line (SDO or MOSI or COPI
(Controller Output Peripheral Input) for disambiguation) is usually not
specified when the controller is not clocking out data on SCLK edges.
However, there do exist SPI peripherals that require specific MOSI line
state when data is not being clocked out of the controller.
Conventional SPI controllers may set the MOSI line on SCLK edges then bring
it low when no data is going out or leave the line the state of the last
transfer bit. More elaborated controllers are capable to set the MOSI idle
state according to different configurable levels and thus are more suitable
for interfacing with demanding peripherals.
Add SPI mode bits to allow peripherals to request explicit MOSI idle state
when needed.
When supporting a particular MOSI idle configuration, the data output line
state is expected to remain at the configured level when the controller is
not clocking out data. When a device that needs a specific MOSI idle state
is identified, its driver should request the MOSI idle configuration by
setting the proper SPI mode bit.
Acked-by: Nuno Sa <nuno.sa@analog.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: David Lechner <dlechner@baylibre.com>
Tested-by: David Lechner <dlechner@baylibre.com>
Signed-off-by: Marcelo Schmitt <marcelo.schmitt@analog.com>
Link: https://patch.msgid.link/9802160b5e5baed7f83ee43ac819cb757a19be55.1720810545.git.marcelo.schmitt@analog.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Try to read bus width property using acpi_dev_get_property function, do
not rely on spi_mem_default_supports_op function only.
If of_device_get_match_data() will fail, retry with
acpi_device_get_match_data() to handle ACPI properly.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Signed-off-by: Piyush Malgujar <pmalgujar@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-10-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
In current implementation cs property can be read only from
device-tree(for_each_available_child_of_node_scoped). Change it to fwnode
based read to allow property reading in ACPI case too.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Signed-off-by: Piyush Malgujar <pmalgujar@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-9-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
If mapping resource by name will fail try to map resource by number.
Such situation can occur in ACPI case.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Signed-off-by: Piyush Malgujar <pmalgujar@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-8-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Marvell Xfer overlay extends xSPI capabilities to support non-memory SPI
operations. The Marvell overlay, combined with a generic command, allows
for full-duplex SPI transactions. It also enables transactions with
undetermined lengths using the cs_hold parameter and the ability to
extend CS signal assertion, even if the xSPI block requests CS signal
de-assertion.
Marvell overlay is using part of xSPI for writing data into device, and
additional hardware block to read data from the device. To do that xSPI
will trigger 1 byte generic command followed by data sequence. In same
time overlay block will monitor MISO pin to read data from the device.
Due to that SDMA data start will be shifted by 1 byte.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-7-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
It is possible that before enabling interrupt, interrupt bit will be
set. It might cause improper IRQ handler behaviour. To fix it, clear
interrupt bit before enabling interrupts. That behaviour is specific to
Marvell xSPI implementation.
In addition in Marvell xSPI interrupt must be cleared in two places -
xSPI itself, and Marvell overlay.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-6-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
In Marvell xSPI implementation any access to SDMA register will result
in 8 byte SPI data transfer. Reading less data(eg. 1B) will result in
losing remaining bytes. To avoid that read/write 8 bytes into temporary
buffer, and read/write whole temporary buffer into SDMA.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-5-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Add support for clock divider. Divider block can disable, enable and
divide clock signal. Only 14 different divide ratios are avalible, from
6.25 up to 200MHz. For calculations use default Marvell system clock
value(800MHz).
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-4-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This commit adds support for static PHY configuration of Cadence xSPI
block. Configuration will be applied only if Marvell overlay compatible
string will be detected. Configuration is static over the whole
frequency range.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Link: https://patch.msgid.link/20240724154739.582367-3-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Add new bindings for the v2 Marvell xSPI overlay: marvell,cn10-xspi-nor
compatible string. This new compatible string distinguishes between the
original and modified xSPI block.
Also add an optional base for the xfer register set with an additional
reg field to allocate the xSPI Marvell overlay XFER block.
Signed-off-by: Witold Sadowski <wsadowski@marvell.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240724154739.582367-2-wsadowski@marvell.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Replace a comma between expression statements by a semicolon.
Fixes: c0216c0e68 ("regulator: rt5120: Add PMIC regulator support")
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Reviewed-by: ChiYuan Huang <cy_huang@richtek.com>
Link: https://patch.msgid.link/20240716084659.1252690-1-nichen@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
Replace a comma between expression statements by a semicolon.
Fixes: d48acfd037 ("regulator: wm831x-isink: Convert to use regulator_set/get_current_limit_regmap")
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Acked-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://patch.msgid.link/20240716085115.1252817-1-nichen@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
Replace a comma between expression statements by a semicolon.
Fixes: 90f6a2a20b ("ASoC: cs42l42: Add SoundWire support")
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Reviewed-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Reviewed-by: Dragan Simic <dsimic@manjaro.org>
Link: https://patch.msgid.link/20240716025307.400156-1-nichen@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
Convert the Dialog Semiconductor DA7212/DA7213 Audio Codec Device Tree
binding documentation to json-schema.
Add missing properties.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/7645c9024a1762d281f4067504bc32a7a3d27caa.1721649741.git.geert+renesas@glider.be
Signed-off-by: Mark Brown <broonie@kernel.org>
The struct snd_soc_pcm_runtime has a flex array of snd_soc_component
objects at its end, and the size is kept in num_components field.
We can add __counted_by() annotation for compiler's assistance to
catch array overflows.
A slight additional change is the assignment of rtd->components[];
the array counter has to be incremented at first for avoiding
false-positive reports from compilers.
Also, the allocation size of snd_soc_pcm_runtime is cleaned up with
the standard struct_size() helper, too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20240726155237.21961-1-tiwai@suse.de
Signed-off-by: Mark Brown <broonie@kernel.org>
of_node_put() has taken the null pointer check into account. So it is safe
to remove the duplicated check before of_node_put().
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://patch.msgid.link/20240709085131.1436128-1-nichen@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
The wsa_macro_vi_feed_mixer_put() callback for setting VI feedback mixer
value could be used for different DAIs (planned in the future CPS DAI),
so make the code a bit more generic by using DAI ID from widget->shift,
instead of hard-coding it. The get() callback already follows such
convention.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240723144607.123240-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Return devm_of_clk_add_hw_provider() in order to transfer the error, if it
fails due to resource allocation failure or device tree clock provider
registration failure.
Fixes: bdd229ab26 ("ASoC: rt5682s: Add driver for ALC5682I-VS codec")
Signed-off-by: Ma Ke <make24@iscas.ac.cn>
Link: https://patch.msgid.link/20240717115436.3449492-1-make24@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
The APQ8016 SBC and MSM8916 QDSP6 sound cards are a bit different from
others: they have additional IO muxing address space and pin control.
Move them to separate schema, so the original qcom,sm8250.yaml will be
easier to manage. New schema is going to grow for other platforms
having more of IO muxing address spaces.
Cc: Adam Skladowski <a39.skl@gmail.com>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240723083300.35605-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
We find that we need to disable micbias for the codec to enter suspend
So We modify the trigger conditions for enable_micbias and disable_micbias
Signed-off-by: Zhang Yi <zhangyi@everest-semi.com>
Link: https://patch.msgid.link/20240726031002.35055-1-zhangyi@everest-semi.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Convert the Freescale i.MX audio complex with ES8328 codec bindings to
DT schema format.
Cc: Daniel Baluta <daniel.baluta@nxp.com>
Signed-off-by: Animesh Agarwal <animeshagarwal28@gmail.com>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://patch.msgid.link/20240722063657.23018-1-animeshagarwal28@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Convert the PCM512x and TAS575x audio CODECs/amplifiers bindings to DT
schema format. Add missing sound-dai-cells property.
Cc: Daniel Baluta <daniel.baluta@nxp.com>
Signed-off-by: Animesh Agarwal <animeshagarwal28@gmail.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240717134729.51661-1-animeshagarwal28@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Driver does not use detected variant/version variables past the init
function, so do not store them in the state container structure.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240710-asoc-wsa88xx-version-v1-3-f1c54966ccde@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
If reading version and variant from registers fails (which is unlikely
but possible, because it is a read over bus), the driver will proceed
and perform device configuration based on uninitialized stack variables.
Handle it a bit better - bail out without doing any init and failing the
update status Soundwire callback.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20240710-asoc-wsa88xx-version-v1-2-f1c54966ccde@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Rename the function ata_dma_blacklisted() to ata_dev_nodma() as this new
name is more neutral. The function signature is also changed to return a
boolean instead of an int.
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Igor Pylypiv <ipylypiv@google.com>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
Change the function ata_dev_knobble() to return a boolean instead of a
u8.
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Igor Pylypiv <ipylypiv@google.com>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
While working on simplifying the minmax functions, and avoiding
excessive macro expansion, it turns out that the sr.c use of the
'clamp()' macro has the arguments the wrong way around.
The clamp logic is
val = clamp(in, low, high);
and it returns the input clamped to the low/high limits. But sr.c ddid
speed = clamp(0, speed, 0xffff / 177);
which clamps the value '0' to the range '[speed, 0xffff / 177]' and ends
up being nonsensical.
Happily, I don't think anybody ever cared.
Fixes: 9fad9d560a ("scsi: sr: Fix unintentional arithmetic wraparound")
Cc: Justin Stitt <justinstitt@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The Rockchip variant of the dwc-ahci controller does have and need power-
domains to work, though the binding does not mention them, making dtccheck
quite unhappy:
DTC_CHK arch/arm64/boot/dts/rockchip/rk3568-odroid-m1.dtb
/home/devel/hstuebner/00_git-repos/linux-rockchip/_build-arm64/arch/arm64/boot/dts/rockchip/rk3568-odroid-m1.dtb: sata@fc800000: Unevaluated properties are not allowed ('power-domains' was unexpected)
from schema $id: http://devicetree.org/schemas/ata/rockchip,dwc-ahci.yaml#
Fix that by adding the missing power-domain property to the binding.
Fixes: 85b0e13b19 ("dt-bindings: ata: dwc-ahci: add Rockchip RK3588")
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
The APQ8064 SATA AHCI controller is used in apq8064.dtsi, although it
was not documented in the old text bindings.
Add its compatible to ahci-platform.yaml.
Signed-off-by: Rayyan Ansari <rayyan.ansari@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Remove old text bindings and add ipq806x AHCI compatible to
ahci-common.yaml, as well as its required properties.
Signed-off-by: Rayyan Ansari <rayyan.ansari@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Use devm_clk_get_prepared() in order to remove a clk_unprepare() in an
error handling path of the probe and from the .remove() function.
This done, the whole .remove() function can also be axed because
'input_dev' is a managed resource allocated with
devm_input_allocate_device() and we can fully rely on devm for cleaning up.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/062986b0a5105cbc61330da0e55b22c00e2c1c4f.1722062145.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
This just standardizes the use of MIN() and MAX() macros, with the very
traditional semantics. The goal is to use these for C constant
expressions and for top-level / static initializers, and so be able to
simplify the min()/max() macros.
These macro names were used by various kernel code - they are very
traditional, after all - and all such users have been fixed up, with a
few different approaches:
- trivial duplicated macro definitions have been removed
Note that 'trivial' here means that it's obviously kernel code that
already included all the major kernel headers, and thus gets the new
generic MIN/MAX macros automatically.
- non-trivial duplicated macro definitions are guarded with #ifndef
This is the "yes, they define their own versions, but no, the include
situation is not entirely obvious, and maybe they don't get the
generic version automatically" case.
- strange use case #1
A couple of drivers decided that the way they want to describe their
versioning is with
#define MAJ 1
#define MIN 2
#define DRV_VERSION __stringify(MAJ) "." __stringify(MIN)
which adds zero value and I just did my Alexander the Great
impersonation, and rewrote that pointless Gordian knot as
#define DRV_VERSION "1.2"
instead.
- strange use case #2
A couple of drivers thought that it's a good idea to have a random
'MIN' or 'MAX' define for a value or index into a table, rather than
the traditional macro that takes arguments.
These values were re-written as C enum's instead. The new
function-line macros only expand when followed by an open
parenthesis, and thus don't clash with enum use.
Happily, there weren't really all that many of these cases, and a lot of
users already had the pattern of using '#ifndef' guarding (or in one
case just using '#undef MIN') before defining their own private version
that does the same thing. I left such cases alone.
Cc: David Laight <David.Laight@aculab.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Another oddity in these config entries is their default value can fall
back to 'n', which is a value for bool or tristate symbols.
The '|| echo n' is an incorrect workaround to avoid the syntax error.
This is not a big deal, as the entry is hidden by 'depends on RUST' in
situations where '$(RUSTC) --version' or '$(BINDGEN) --version' fails.
Anyway, it looks odd.
The default of a string type symbol should be a double-quoted string
literal. Turn it into an empty string when the version command fails.
Fixes: 2f7ab1267d ("Kbuild: add Rust support")
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Link: https://lore.kernel.org/r/20240727140302.1806011-2-masahiroy@kernel.org
[ Rebased on top of v6.11-rc1. - Miguel ]
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
While this is a somewhat unusual case, I encountered odd error messages
when I ran Kconfig in a foreign architecture chroot.
$ make allmodconfig
sh: 1: rustc: not found
sh: 1: bindgen: not found
#
# configuration written to .config
#
The successful execution of 'command -v rustc' does not necessarily mean
that 'rustc --version' will succeed.
$ sh -c 'command -v rustc'
/home/masahiro/.cargo/bin/rustc
$ sh -c 'rustc --version'
sh: 1: rustc: not found
Here, 'rustc' is built for x86, and I ran it in an arm64 system.
The current code:
command -v $(RUSTC) >/dev/null 2>&1 && $(RUSTC) --version || echo n
can be turned into:
command -v $(RUSTC) >/dev/null 2>&1 && $(RUSTC) --version 2>/dev/null || echo n
However, I did not understand the necessity of 'command -v $(RUSTC)'.
I simplified it to:
$(RUSTC) --version 2>/dev/null || echo n
Fixes: 2f7ab1267d ("Kbuild: add Rust support")
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Link: https://lore.kernel.org/r/20240727140302.1806011-1-masahiroy@kernel.org
[ Rebased on top of v6.11-rc1. - Miguel ]
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
The NBD protocol defines the flag NBD_FLAG_ROTATIONAL to flag that the
export in use should be treated as a rotational device.
Add support for that flag to the kernel driver.
Signed-off-by: Wouter Verhelst <w@uter.be>
Reviewed-by: Eric Blake <eblake@redhat.com>
Link: https://lore.kernel.org/r/20240725164536.1275851-1-w@uter.be
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Currently _drbd_send_page() use sendpage_ok() in order to enable
MSG_SPLICE_PAGES, it check the first page of the iterator, the iterator
may represent contiguous pages.
MSG_SPLICE_PAGES enables skb_splice_from_iter() which checks all the
pages it sends with sendpage_ok().
When _drbd_send_page() sends an iterator that the first page is
sendable, but one of the other pages isn't skb_splice_from_iter() warns
and aborts the data transfer.
Using the new helper sendpages_ok() in order to enable MSG_SPLICE_PAGES
solves the issue.
Acked-by: Christoph Böhmwalder <christoph.boehmwalder@linbit.com>
Signed-off-by: Ofir Gal <ofir.gal@volumez.com>
Link: https://lore.kernel.org/r/20240718084515.3833733-4-ofir.gal@volumez.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
