Convert driver(s) to DEVICE_DEFINE instead of DEVICE_AND_API_INIT
so we can deprecate DEVICE_AND_API_INIT in the future.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
It raises an error in case the source data width differs
from the dest data width.
The dma_stm32_width_config function is no more useful: removed.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
This new offset value in the dma config is made to
build the table of dma mux_channels with a dmamux.
Range depends on the nb of channels for selected dma instance
Signed-off-by: Francois Ramu <francois.ramu@st.com>
1, Change the allocation of DMA descriptors for the case
when multiple buffers are used for DMA transfer. This was
tested during SPI DMA transfers
2. Add support for the case when source and destination
addresses should not be incremented.
Signed-off-by: Mahesh Mahadevan <mahesh.mahadevan@nxp.com>
Not possible to use device_get_binding() because we are
currently initializing DMA_1 device, so it is not ready and
device_get_binding() will fail.
Directly use string compare is more efficient.
Signed-off-by: Alexandre Bourdiol <alexandre.bourdiol@st.com>
The dmamux requires HEAP size definition, so that k_malloc
is valid. The HEAP size config is defined in the common for
any stm32 soc instead of specific to dma Kconfig
Signed-off-by: Francois Ramu <francois.ramu@st.com>
The stm32_dma_disable_stream first checked if the stream was
disabled and if so returned OK. If it wasn't disabled it
tried to disable it and returned -EAGAIN.
The function is used in loops that try to disable the stream
by calling this function and if it fails wait for 1ms and
retry.
Becuase this function the first time (if the stream wasn't
disabled already) fails there is always a 1 ms delay. For
the SPI driver, that has a RX and TX stream, this means
a 2ms delay between the last data and CS going high.
By first trying to disable the stream and than checking
if it succeded most of the time the first call disables
the stream and the 1ms delay isn't needed.
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
The general DMA driver doesn't use kmalloc anymore so it doesn't
need a memory pool. The DMAMUX_SMT32 driver still uses kmalloc,
so move the HEAP_MEM_POOL_SIZE config under DMAMUX_STM32.
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
Use dma_stm32_disable_stream instead of stm32_dma_disable_stream
to check if the disabling of the stream really worked and if
not return an error.
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
Convert global conversion tables for id-to-stream and slot-to-channel
to functions that use local const static conversion tables.
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
- Remove the need for kmalloc
- On hardware that supports it use 1 IRQ handler per stream to
determine the stream ID, so the ISR does not have to loop
over all ID's to see which one is active. On hardware (like
STM32L0 and STM32F0) where up to 7 streams share 3 IRQ's use
the loop to check which stream is active.
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
Use volatile for flag shared between normal code and ISR to
prevent the compiler from possibly optimizing it away.
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
These are all the case that coccinelle cannot find as they are inside
macro declarations.
Fixed via:
git grep -rlz -E "\(struct device \*" |
xargs -0 sed -i 's/(struct device/(const struct device/g'
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Passing the device's config was necessary, and wrapping device inside
it.
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Now that device_api attribute is unmodified at runtime, as well as all
the other attributes, it is possible to switch all device driver
instance to be constant.
A coccinelle rule is used for this:
@r_const_dev_1
disable optional_qualifier
@
@@
-struct device *
+const struct device *
@r_const_dev_2
disable optional_qualifier
@
@@
-struct device * const
+const struct device *
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Add PL330 dma driver support, this driver
- supports 32 bit memory to memory operation
- supports 36 bit memory to memory operation
under a config flag DMA_64BIT
- supports secure channel
Signed-off-by: Raveendra Padasalagi <raveendra.padasalagi@broadcom.com>
Add 64bit source_address/dest_address fields support.
This is needed in situation where DMA controller is capable
of handling more than 32bit source and destination addresses.
Signed-off-by: Raveendra Padasalagi <raveendra.padasalagi@broadcom.com>
Add support for the NXP KE1xF SoC to the eDMA driver. The KE1xF supports
m2m DMA transfers from sources 60 to 63.
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
DMA driver shall be initialized before any of its
clients. Right now, its initialization priority is
lower than that of UARTs. So, increase its init
priority to support future async uart drivers.
Signed-off-by: Jun Li <jun.r.li@intel.com>
tested on mimxrt1060_evt
MEMORY_NOCACHE is needed
test on frdmk64f
special test slot need configure with
CONFIG_DMA_TEST_SLOT_START
Signed-off-by: Hake Huang <hake.huang@oss.nxp.com>
Replace individual device instance definitions with the macro that
expands to the equivalent change.
F='struct device DEVICE_NAME_GET'
git grep -l "$F" \
| xargs sed -i -r \
-e "s@$F"'\(([^)]*)\);@DEVICE_DECLARE(\1);@'
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Several reviewers agreed that DT_HAS_NODE_STATUS_OKAY(...) was an
undesirable API for the following reasons:
- it's inconsistent with the rest of the DT_NODE_HAS_FOO names
- DT_NODE_HAS_FOO_BAR_BAZ(node) was agreed upon as a shorthand
for macros which are equivalent to
DT_NODE_HAS_FOO(node) && DT_NODE_HAS_BAR(node) &&
- DT_NODE_HAS_BAZ(node), and DT_HAS_NODE_STATUS_OKAY is an odd duck
- DT_NODE_HAS_STATUS(..., okay) was viewed as more readable anyway
- it is seen as a somewhat aesthetically challenged name
Replace all users with DT_NODE_HAS_STATUS(..., okay), which is
semantically equivalent.
This is mostly done with sed, but a few remaining cases were done by
hand, along with whitespace, docs, and comment changes. These special
cases include the Nordic SOC static assert files.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Usually, we want to operate only on "available" device
nodes ("available" means "status is okay and a matching binding is
found"), but that's not true in all cases.
Sometimes we want to operate on special nodes without matching
bindings, such as those describing memory.
To handle the distinction, change various additional devicetree APIs
making it clear that they operate only on available device nodes,
adjusting gen_defines and devicetree.h implementation details
accordingly:
- emit macros for all existing nodes in gen_defines.py, regardless
of status or matching binding
- rename DT_NUM_INST to DT_NUM_INST_STATUS_OKAY
- rename DT_NODE_HAS_COMPAT to DT_NODE_HAS_COMPAT_STATUS_OKAY
- rename DT_INST_FOREACH to DT_INST_FOREACH_STATUS_OKAY
- rename DT_ANY_INST_ON_BUS to DT_ANY_INST_ON_BUS_STATUS_OKAY
- rewrite DT_HAS_NODE_STATUS_OKAY in terms of a new DT_NODE_HAS_STATUS
- resurrect DT_HAS_NODE in the form of DT_NODE_EXISTS
- remove DT_COMPAT_ON_BUS as a public API
- use the new default_prop_types edtlib parameter
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Convert sam_xdmac driver to utilize devicetree. As part of the
controller binding we specify that dmas should contain a channel and the
perid for the DMA transaction.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Convert to using DT_INST_LABEL() in the dma driver and convert dma users
to use the DMA property macros to get the dma controller name. We make
the assumption in the drivers that there is a single DMA controller
instance.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Make drivers multi-instance wherever possible using DT_INST_FOREACH.
This allows removing DT_HAS_DRV_INST in favor of making drivers just
do the right thing regardless of how many instances there are.
There are a few exceptions:
- SoC drivers which use CMake input files (like i2c_dw.c) or otherwise
would require more time to convert than I have at the moment. For the
sake of expediency, just inline the DT_HAS_DRV_INST expansion for
now in these cases.
- SoC drivers which are explicitly single-instance (like the nRF SAADC
driver). Again for the sake of expediency, drop a BUILD_ASSERT in
those cases to make sure the assumption that all supported SoCs have
at most one available instance is valid, failing fast otherwise.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
This reverts commit 90cc723e65.
Using `DMAC_CHCTRLA_TRIGACT_BLOCK` breaks SERCOM UART peripheral DMA
operations (DMA-based asynchronous UART transmit operation only sends
the first byte and does not proceed any further).
The `DMAC_CHCTRLA_TRIGACT_BURST` with `DMAC_CHCTRLA_BURSTLEN(0)` is a
special case utilising the "internal FIFO", according to the datasheet
(see DS60001507E; 22.6.2.8), and should always be specified for
peripheral data transfer operations.
Also it is worth noting that Atmel and other third-party drivers use
the aforementioned "internal FIFO" configuration for peripheral data
transfers as well.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The channel direction for a V1 DMA is not allowed to be memory to memory
and there is a check in place for this. However, the check is being
performed on the stream prior to actually configuring the stream. This
results in the check always failing regardless of the channel direction.
The check has been modified to be performed on the incoming
configuration.
Signed-off-by: Abe Kohandel <abe@electronshepherds.com>
This patch prepares the dma and introduces the dmamux on soc series
which supports this feature for memory/periph transfers.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
