The watermark levels in the DMA register are write only, the driver
should never have to read them back from the hardware. Replace the
current _MASK and _OFFSET defines with defines taking the watermark
level directly.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
This reverts patch 1476253cef (spi: imx: fix ecspi mode setup)
The patch tried to fix something by clearing bits in the cfg variable,
but cfg is initialized to zero on function entry. There are no bits to
clear.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
DMA transfer for SPI was limited to up to 8 bits word size until now.
Sync in SPI burst size and DMA bus width is necessary to correctly
support 16 and 32 BPW.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
There's no need for an extra dma_is_inited variable when we can
equally well check for the existence of a DMA channel.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
When the MX51_ECSPI_DMA is configured we control every single bit
of the register, so there's no need to read/modify/write it. Instead
just write the value we want to have in the register. Also, drop
unnecessary check if we are actually doing DMA. The values written
to the register have no effect in PIO mode and value written there
during the last DMA transfer is still in the register, so we can
equally well always write a value.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The SoC specific config function does not know if DMA will be used or
not. This information will be useful to configure the SPI controller
correctly for DMA in following patches, so initialize the usedma
variable before calling into the SoC specific config function.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
At the moment the driver decides to fallback to PIO mode the buffers
are already mapped for DMA. It's a bug to access them with the CPU
afterwards, so we cannot just fallback to PIO mode.
It should not be necessary anyway, since we only use DMA when we
verified that it's possible in the fist place, so when prep_slave_sg
fails it's a bug, either in the SDMA driver or in the can_dma
implementation.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Fixed timeout value can fire while transaction is ongoing. This may happen
because there are no strict requirements on SPI transaction duration.
Dynamic timeout value is generated based on SCLK and transaction size.
There is also 4 * SCLK delay between TX bursts related to HW internal CS change.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
A small clutch of driver specific fixes. The OMAP one is a bit worrying
since it seems to be triggered by some changes in the runtime PM core
code and I suspect there's other drivers across that are going to be
using the same pattern outside of OMAP but nothing seems to be coming up
in the testing people are doing.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQEcBAABAgAGBQJWwcN2AAoJECTWi3JdVIfQM9IH/2WWGIU4ByQeHLd7FnNHQwZk
wh4g7xpo/SRzob2rz2yqeKgkIg4Ew4+BsD5jslvGhBn9ph5HIFYHVcHt3vjVUNRE
EyOoM8X4lLc97LYog10ieXk5dFkvcCTDYoQDcTxVDmMPRsRrt2Z7htTB4jAbhlk/
kH5nJCir4Cwk5qDTw7+9W8d5hCXl3ESM7KoTCqtvpA8JmTD++vRcphp39NWZejAw
ukO/EAJcwqlTmds3zki1/lM6R4NBtbAE2BGy9j0eUQnL3/37Sl77hAxcsKjey1Zj
dVa8ow9ZRTEZtbCQjrK8WpD3V26TNuUGZs4qEQ+nizQnvAnrqgkfOOg6wBJ6LDg=
=imJW
-----END PGP SIGNATURE-----
Merge tag 'spi-fix-v4.5-rc4' into spi-imx
spi: Fixes for v4.5
A small clutch of driver specific fixes. The OMAP one is a bit worrying
since it seems to be triggered by some changes in the runtime PM core
code and I suspect there's other drivers across that are going to be
using the same pattern outside of OMAP but nothing seems to be coming up
in the testing people are doing.
Add a struct device * member to the private driver data and use
it to print messages using dev_* functions rather than pr_*.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
RX DMA tail data handling doesn't work correctly in many cases with current
implementation. It happens because SPI core was setup to generates both RX
and RX TAIL events. And RX TAIL event does not work correctly.
This can be easily verified by sending SPI transaction with size modulus
WML(32 in our case) not equal 0.
Also removing change introduced in f6ee9b582d
since this change only fix usecases with transfer size from 33 to 128 bytes
and doesn't fix 129 bytes and bigger.
This is output from transaction with len 138 bytes in loopback mode at 10Mhz:
TX0000: a3 97 a2 55 53 be f1 fc f9 79 6b 52 14 13 e9 e2
TX0010: 2d 51 8e 1f 56 08 57 27 a7 05 d4 d0 52 82 77 75
TX0020: 1b 99 4a ed 58 3d 6a 52 36 d5 24 4a 68 8e ad 95
TX0030: 5f 3c 35 b5 c4 8c dd 6c 11 32 3d e2 b4 b4 59 cf
TX0040: ce 23 3d 27 df a7 f9 96 fc 1e e0 66 2c 0e 7b 8c
TX0050: ca 30 42 8f bc 9f 7b ce d1 b8 b1 87 ec 8a d6 bb
TX0060: 2e 15 63 0e 3c dc a4 3a 7a 06 20 a7 93 1b 34 dd
TX0070: 4c f5 ec 88 96 68 d6 68 a0 09 6f 8e 93 47 c9 41
TX0080: db ac cf 97 89 f3 51 05 79 71
RX0000: a3 97 a2 55 53 be f1 fc f9 79 6b 52 14 13 e9 e2
RX0010: 2d 51 8e 1f 56 08 57 27 a7 05 d4 d0 52 82 77 75
RX0020: 1b 99 4a ed 58 3d 6a 52 36 d5 24 4a 68 8e ad 95
RX0030: 5f 3c 35 00 00 b5 00 00 00 c4 00 00 8c 00 00 dd
RX0040: 6c 11 32 3d e2 b4 b4 59 cf ce 23 3d 27 df a7 f9
RX0050: 96 fc 1e e0 66 2c 0e 7b 8c ca 30 42 8f 1f 1f bc
RX0060: 9f 7b ce d1 b8 b1 87 ec 8a d6 bb 2e 15 63 0e ed
RX0070: ed 3c 58 58 58 dc 3d 3d a4 6a 6a 3a 52 52 7a 36
RX0080: 06 20 a7 93 1b 34 dd 4c f5 ec
Zeros at offset 33 and 34 caused by reading empty RX FIFO which not possible
if DMA RX read was triggered by RX event. This mean DMA was triggered
by RX TAIL event.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
In spi_imx_dma_transfer(), when desc_rx = dmaengine_prep_slave_sg()
fails, the context goes to label no_dma and then return. However,
the memory allocated for desc_tx has not been freed yet, which leads
to resource leak.
Signed-off-by: Gao Pan <pandy.gao@nxp.com>
Reviewed-by: Fugang Duan <B38611@freescale.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
If SPI device supports DMA mode, but DMA controller is not yet
available due to e.g. a delay in the corresponding kernel module
initialization, retry to initialize SPI driver later on instead of
falling back into PIO only mode.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
On SDMA initialization return exactly the same error, which is
reported by dma_request_slave_channel_reason(), it is a preceding
change to defer SPI DMA initialization, if SDMA module is not yet
available.
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Limit SPI_LOOP mode to ECSPI controller (iMX.51, iMX53 and i.MX6) only since
there is no support in other families specific code for now.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
If controller hold in reset it's not possible to write any
register except CTRL. So all other registers must be updated
only after controller bring out from reset.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Similar to other controller type checks add check function for
IMX51. It includes IMX53 and IMX6.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
There is no need to have different watermarks levels since they are the same.
Merge them into one WML parameter.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The overflow may happen due to rescheduling for another task and/or interrupt
if we enable SPI HW before starting RX DMA. So RX DMA enabled first to make
sure data would be read out from FIFO ASAP. TX DMA enabled next to start
filling TX FIFO with new data. And finaly SPI HW enabled to start actual
data transfer.
The risk rise in case of heavy system load and high SPI clock.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Not only TX DMA should be terminated, but RX DMA also. It's required
to avoid accidential DMA memory writes from RX DMA channel and properly
terminate transaction.
Signed-off-by: Anton Bondarenko <anton.bondarenko.sama@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Loopback mode can be activated by setting bit LBC (LoopBack
Control) of register ECSPI_TESTREG.
Add support for it.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Fixed problem with setting spi mode 0 or 1 after setting mode 2 or 3
SPI_MODE_0 and SPI_MODE_1 requires clock low when inactive. SPI_MODE_2
and SPI_MODE_3 requires clk high when inactive.
Currently driver can just set bits in fields SCLK_PHA (SPI Clock/Data
Phase Control), SCLK_POL (SPI Clock Polarity Control),
SCLK_CTL (controls the inactive state of SCLK) ans SS_POL (SPI SS
Polarity Select) of ECSPIx_CONFIGREG register.
This patch allows driver to clear corresponding bits in these fields.
Signed-off-by: Andrew Y. Kuksov <qxovxp@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
DMA transfers must be greater than the watermark level size. spi_imx->rx_wml
and spi_imx->tx_wml contain the watermark level in 32bit words whereas struct
spi_transfer contains the transfer len in bytes. Fix the check if DMA is
possible for a transfer accordingly. This fixes transfers with sizes between
33 and 128 bytes for which previously was claimed that DMA is possible.
Fixes: f62caccd12 (spi: spi-imx: add DMA support)
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Cc: stable@vger.kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
The platform_device_id is not modified by the driver and core uses it as
const.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski.k@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
They are used to decide if the controller can do DMA on a buffer
of a specific length and thus are needed before any transfer is attempted.
This fixes a memory leak where the SPI core uses the drivers can_dma()
callback to determine if a buffer needs to be mapped. As the watermark
levels aren't correct at that point the driver falsely claims to be able to
DMA the buffer when it fact it isn't.
After the transfer has been done the core uses the same callback to
determine if it needs to unmap the buffers. As the driver now correctly
claims to not being able to DMA the buffer the core doesn't attempt to
unmap the buffer which leaves the SGT leaking.
Fixes: f62caccd12 (spi: spi-imx: add DMA support)
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
The fourth argument of dmaengine_prep_slave_sg needs to be of the
enumeration type dma_transfer_direction instead of dma_data_direction.
Since the used enumeration values actually stay the same, this is not
an actual issue at runtime.
Signed-off-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Mark Brown <broonie@kernel.org>
return type of wait_for_completion_timeout is unsigned long not int and
always returns >=0 , this patch adds a suitable return variable and
simplifies the return value checking as there is no < 0 case.
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
For TKT238285 hardware issue which may cause txfifo store data twice can only
be caught on i.mx6dl, we use pio mode instead of DMA mode on i.mx6dl.
Fixes: f62caccd12 (spi: spi-imx: add DMA support)
Signed-off-by: Robin Gong <b38343@freescale.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
The irq number is only used inside the probe function, so there is really no
need to store it in the private structure.
Use a local 'irq' variable to hold the the irq number instead.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Enable DMA support on i.mx6. The read speed can increase from 600KB/s
to 1.2MB/s on i.mx6q. You can disable or enable dma function in dts.
If not set "dma-names" in dts, spi will use PIO mode. This patch only
validate on i.mx6, not i.mx5, but encourage ones to apply this patch
on i.mx5 since they share the same IP.
Note:
Sometime, there is a weid data in rxfifo after one full tx/rx
transfer finish by DMA on i.mx6dl, so we disable dma functhion on
i.mx6dl.
Signed-off-by: Frank Li <Frank.Li@freescale.com>
Signed-off-by: Robin Gong <b38343@freescale.com>
Acked-by: Marek Vasut <marex@denx.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently, at module removal, one gets the following warnings:
------------[ cut here ]------------
WARNING: at drivers/clk/clk.c:780 clk_disable+0x18/0x24()
Modules linked in: spi_imx(-) [last unloaded: ev76c560]
CPU: 1 PID: 16337 Comm: rmmod Tainted: G W 3.10.17-80548-g90191eb-dirty #33
[<80013b4c>] (unwind_backtrace+0x0/0xf8) from [<800115dc>] (show_stack+0x10/0x14)
[<800115dc>] (show_stack+0x10/0x14) from [<800257b8>] (warn_slowpath_common+0x4c/0x68)
[<800257b8>] (warn_slowpath_common+0x4c/0x68) from [<800257f0>] (warn_slowpath_null+0x1c/0x24)
[<800257f0>] (warn_slowpath_null+0x1c/0x24) from [<803f60ec>] (clk_disable+0x18/0x24)
[<803f60ec>] (clk_disable+0x18/0x24) from [<7f02c9cc>] (spi_imx_remove+0x54/0x9c [spi_imx])
[<7f02c9cc>] (spi_imx_remove+0x54/0x9c [spi_imx]) from [<8025868c>] (platform_drv_remove+0x18/0x1c)
[<8025868c>] (platform_drv_remove+0x18/0x1c) from [<80256f60>] (__device_release_driver+0x70/0xcc)
[<80256f60>] (__device_release_driver+0x70/0xcc) from [<80257770>] (driver_detach+0xcc/0xd0)
[<80257770>] (driver_detach+0xcc/0xd0) from [<80256d90>] (bus_remove_driver+0x7c/0xc0)
[<80256d90>] (bus_remove_driver+0x7c/0xc0) from [<80068668>] (SyS_delete_module+0x144/0x1f8)
[<80068668>] (SyS_delete_module+0x144/0x1f8) from [<8000e080>] (ret_fast_syscall+0x0/0x30)
---[ end trace 1f5df9ad54996300 ]---
------------[ cut here ]------------
WARNING: at drivers/clk/clk.c:780 clk_disable+0x18/0x24()
Modules linked in: spi_imx(-) [last unloaded: ev76c560]
CPU: 1 PID: 16337 Comm: rmmod Tainted: G W 3.10.17-80548-g90191eb-dirty #33
[<80013b4c>] (unwind_backtrace+0x0/0xf8) from [<800115dc>] (show_stack+0x10/0x14)
[<800115dc>] (show_stack+0x10/0x14) from [<800257b8>] (warn_slowpath_common+0x4c/0x68)
[<800257b8>] (warn_slowpath_common+0x4c/0x68) from [<800257f0>] (warn_slowpath_null+0x1c/0x24)
[<800257f0>] (warn_slowpath_null+0x1c/0x24) from [<803f60ec>] (clk_disable+0x18/0x24)
[<803f60ec>] (clk_disable+0x18/0x24) from [<7f02c9e8>] (spi_imx_remove+0x70/0x9c [spi_imx])
[<7f02c9e8>] (spi_imx_remove+0x70/0x9c [spi_imx]) from [<8025868c>] (platform_drv_remove+0x18/0x1c)
[<8025868c>] (platform_drv_remove+0x18/0x1c) from [<80256f60>] (__device_release_driver+0x70/0xcc)
[<80256f60>] (__device_release_driver+0x70/0xcc) from [<80257770>] (driver_detach+0xcc/0xd0)
[<80257770>] (driver_detach+0xcc/0xd0) from [<80256d90>] (bus_remove_driver+0x7c/0xc0)
[<80256d90>] (bus_remove_driver+0x7c/0xc0) from [<80068668>] (SyS_delete_module+0x144/0x1f8)
[<80068668>] (SyS_delete_module+0x144/0x1f8) from [<8000e080>] (ret_fast_syscall+0x0/0x30)
---[ end trace 1f5df9ad54996301 ]---
Since commit 9e556dcc55, "spi: spi-imx: only
enable the clocks when we start to transfer a message", clocks are always
disabled except when transmitting messages. There is thus no need to
disable them at module removal.
Fixes: 9e556dcc55 (spi: spi-imx: only enable the clocks when we start to transfer a message)
Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Acked-by: Huang Shijie <b32955@freescale.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Cc: stable@vger.kernel.org
No need to return a 'fake' return value on platform_get_irq() failure.
Just return the error code itself instead.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Calling init_completion() once is enough.
For the rest of the iterations, call reinit_completion() instead.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>. Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Problem:
--------
The problem this patch addresses has the following assumptions about the
SPI bus setup:
- The hardware used to find this is Freescale i.MX537 @ 1200MHz
- The SPI SCLK operate at very low speed, less than 200 kHz
- There are two SPI devices attached to the bus
- Each device uses different GPIO for chipselect
- Each device requires different SCLK signal polarity
The observation of the SCLK and GPIO chipselect lines with a logic analyzer
shows, that the SCLK polarity change does sometimes happen after the GPIO
chipselect is asserted. The SPI slave device reacts on that by counting the
SCLK polarity change as a clock pulse, which disrupts the communication with
the SPI slave device.
Explanation:
------------
We found an interesting correlation, that the maximum delay between the write
into the ECSPIx_CONFIGREG register and the change of SCLK polarity at each
SCLK frequency of 10 kHz, 20 kHz, 50 kHz and 100 kHz is 100 uS, 50 uS, 20 uS
and 10 uS respectively. This lead us to a theory, that at SCLK frequency of
1 Hz, the delay would be 1 S. Therefore, the time it takes for the write to
ECSPIx_CONFIGREG to take effect in the hardware is up to the duration of 1
tick of the SCLK clock.
During this delay period, if the SCLK frequency is too low, the execution of
the spi-imx.c driver can advance so much, that the GPIO chipselect will be
asserted. The GPIO chipselect is asserted almost immediatelly.
Solution:
---------
The solution this patch presents is simple. We calculate the resulting SCLK
clock first by dividing the ECSPI block clock by both dividers that are to be
programmed into the configuration register. Based on the resulting SCLK clock,
we derive the delay it will take for the changes to get really applied. We are
extra careful here so we delay twice as long as we should. Note that the patch
does not create additional overhead at high speeds as the delay will likely be
close to zero there.
Signed-off-by: Marek Vasut <marex@denx.de>
To: linux-spi@vger.kernel.org
Cc: Fabio Estevam <fabio.estevam@freescale.com>
Cc: Huang Shijie <b32955@freescale.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Mark Brown <broonie@linaro.org>
Current code keeps the clocks enabled all the time, it wastes the power
when there is no operaiton on the spi controller.
In order to save the power, this patch adds the two hooks:
spi_imx_prepare_message: enable the clocks for this message
spi_imx_unprepare_message: disable the clocks.
This patch also disables the clocks in the end of the probe.
Signed-off-by: Huang Shijie <b32955@freescale.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Many drivers that use bitbang library have a leak on probe error paths.
This is because once a spi_master_get() call succeeds, we need an additional
spi_master_put() call to free the memory.
Fix this issue by moving the code taking a reference to master to
spi_bitbang_start(), so spi_bitbang_start() will take a reference to master on
success. With this change, the caller is responsible for calling
spi_bitbang_stop() to decrement the reference and spi_master_put() as
counterpart of spi_alloc_master() to prevent a memory leak.
So now we have below patten for drivers using bitbang library:
probe:
spi_alloc_master -> Init reference count to 1
spi_bitbang_start -> Increment reference count
remove:
spi_bitbang_stop -> Decrement reference count
spi_master_put -> Decrement reference count (reference count reaches 0)
Fixup all users accordingly.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Suggested-by: Uwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Acked-by: Uwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
Enable auto loading by udev when spi-imx is compiled as a module.
Signed-off-by: Niels de Vos <ndevos@redhat.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
clk_prepare_enable() may fail, so let's check its return value and propagate it
in the case of error.
While at it, fix the order of clk_disable_unprepare calls: clk_ipg should be
disabled first, followed by clk_per.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Using devm functions can make the code smaller and cleaner.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@linaro.org>