Instead of having a table right before the code implementing the length
setting for documentation, have inline comments for the if branches
actually implementing the length setting described table's entries
(which is readable thanks to the set_char_len function).
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
The variable "char_size" holds the number of bits to be transferred in
the current loop iteration. A better name would be "xfer_bitlen", which
we rename this variable to.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Introduce the to_prescale_mod and set_char_len inline functions to make
the code more readable.
Note that the added "if (bitlen > 16)" check does not change the
semantics of the current code, and hence only preserves the current
error (this will be fixed in a later patch in the series).
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
The transmission loop starts with setting some variables, which are only
used inside the loop. Reduce the scope to the loop to make the
declaration and initialization of these variables coincide.
In the case of char_size this also always initializes the variable
immediately with the final value actually used in the loop (instead of
the placeholder value 32).
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
We do nothing in the loop if the "not empty" event was not detected. To
simplify the logic, check if this is the case, and skip the execution of
the loop early to reduce the nesting level and flag checking.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Get rid of the is_read variable, and just keep the state of the "not
empty" and "not full" events in two boolean variables within the loop
body.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Instead of having a nested if block, just have two branches within the
overarching if block to eliminate one nesting level.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Accesses to the register map are currently done by directly reading and
writing the structure.
Switch to the appropriate IO accessors instead.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Replace pre-processor defines with proper enums, and use the BIT macro
where applicable.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Replace the function name with a "%s" format string and the __func__
variable in debug statements (as proposed by checkpatch).
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
There are three variables that have camel-case names, which is not the
preferred naming style.
Give those variables more compliant names instead.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
There are some comments on the same line as the code they document. Put
comments above the code lines they document, so the line length is not
unnecessarily increased.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
The function signatures in the driver are quite long as is. Use short
type names (uint etc.) to make them more readable.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Convert 'dw_spi_ofdata_to_platdata' to use 'dev_read_u32_default'
instead of 'fdtdec_get_int' and get rid of DECLARE_GLOBAL_DATA_PTR.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Convert 'cadence_spi_ofdata_to_platdata' to use dev_read_* functions to
read driver parameters and 'dev_read_first_subnode'/'ofnode_read_*' to
read flash (child node) parameters.
Tested on socfpga_socrates (socfpga gen5).
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Like clk_get_by_index, there is requirement for
clk_get_by_index_nodev. In this case to make common
code functionalities for dev and nodev, clk_get_by_index
is trying to get the index of clock by passing ofnode
instead of actual dev like current gpio uclass does.
In these scenarios with current order of include files
the atcspi200_spi driver is unable to find CONFIG_ENV_SIZE.
In file included from arch/nds32/include/asm/u-boot.h:24,
from include/dm/of.h:10,
from include/dm/ofnode.h:12,
from include/clk.h:11,
from drivers/spi/atcspi200_spi.c:9:
include/environment.h:145:19: error: 'CONFIG_ENV_SIZE'
undeclared here (not in a function); did you mean 'CONFIG_CMD_XIMG'?
#define ENV_SIZE (CONFIG_ENV_SIZE - ENV_HEADER_SIZE)
So, fix consists of changing the order of include files
in atcspi200_spi.c to include first common.h file.
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Add ifdeffery to allow operation without the clock framework
enabled. This is required on RZ/A1, as it does not have clock
driver yet.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Chris Brandt <chris.brandt@renesas.com>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
Ocelot
- mt7620: rename arch to more generic name mtmips
- mips: pass initrd addresses via DT as physical addresses
-----BEGIN PGP SIGNATURE-----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=SZ/W
-----END PGP SIGNATURE-----
Merge tag 'mips-pull-2019-05-03' of git://git.denx.de/u-boot-mips
- mscc: small fixes, enhance network support for Serval, Luton and Ocelot
- mt7620: rename arch to more generic name mtmips
- mips: pass initrd addresses via DT as physical addresses
Currently mach-mt7620 contains only support for mt7628. To avoid confusion,
rename mach-mt7620 to mach-mtmips, which means MediaTek MIPS platforms.
MT7620 and MT7628 should be distinguished by SOC_MT7620 and SOC_MT7628
because they do not share the same lowlevel codes.
Dependencies of four drivers are changed to SOC_MT7628 as these drivers
are only used by MT7628.
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
Rockchip use 'arch-rockchip' instead of arch-$(SOC) as common
header file path, so that we can get the correct path directly.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
To support unaligned output buffers (i.e. 'in' in the terminology of
the SPI framework), this change splits each 16bit FIFO element after
reading and writes them to memory in two 8bit transactions. With this
change, we can now always use the optimised mode for receive-only
transcations independent on the alignment of the target buffer.
Given that we'll run with caches on, the impact should be negligible:
as expected, this has no adverse impact on throughput if running with
a 960MHz LPLL configuration.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
The SPI controller's documentation (I only had access to the RK3399,
RK3368 and PX30 TRMs) specifies that, when operating in master-mode,
the controller will stop the SCLK to avoid RXFIFO overruns and TXFIFO
underruns. Looks like my worries that we'd need to support DMA-330
(aka PL330) to make any further progress were unfounded.
This adds a driver-data structure to capture hardware-specific
settings of individual controller instances (after all, we don't know
if all versions are well-behaved) and adds a 'master_manages_fifo'
flag to it. The first use of said flag is in the optimised
receive-only transfer-handler, which can now request 64Kframe
(i.e. 128KByte) bursts of data on each reprogramming of CTRLR1
(i.e. every time through the loop).
This improves throughput to 46.85MBit/s (a 94.65% bus-utilisation).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
For the RK3399-Q7 we recommend storing SPL and u-boot.itb in the
on-module 32MBit (and sometimes even larger, if requested as part of a
configure-to-order configuration) SPI-NOR flash that is clocked for a
bitrate of 49.5MBit/s and connected in a single-IO configuration (the
RK3399 only supports single-IO for SPI).
Unfortunately, the existing SPI driver is excruciatingly slow at
reading out large chunks of data (in fact it is just as slow for small
chunks of data, but the overheads of the driver-framework make it less
noticeable): before this change, the throughput on a 4MB read from
SPI-NOR is 8.47MBit/s which equates a 17.11% bus-utilisation.
To improve on this, this commit adds an optimised receive-only
transfer (i.e.: out == NULL) handler that hooks into the main transfer
function and processes data in 16bit frames (utilising the full with
of each FIFO element). As of now, the receive-only handler requires
the in-buffer to be 16bit aligned. Any lingering data (i.e. either if
the in-buffer was not 16-bit aligned or if an odd number of bytes are
to be received) will be handled by the original 8bit reader/wirter.
Given that the SPI controller's documentation does not guarantuee any
interlocking between the RXFIFO and the master SCLK, the transfer loop
will be restarted for each chunk of 32 frames (i.e. 64 bytes).
With this new receive-only transfer handler, the throughput for a 4MB
read increases to 36.28MBit/s (i.e. 73.29% bus-utilisation): this is a
4x improvement over the baseline.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reported-by: Klaus Goger <klaus.goger@theobroma-systems.com>
Series-Cc: Klaus Goger <klaus.goger@theobroma-systems.com>
Series-Cc: Christoph Muellner <christoph.muellner@theobroma-systems.com>
The logic in the main transmit loop took a bit of reading the TRM to
fully understand (due to silent assumptions based in internal logic):
the "wait until idle" at the end of each iteration through the loop is
required for the transmit-path as each clearing of the ENA register
(to update run-length in the CTRLR1 register) will implicitly flush
the FIFOs... transmisson can therefore not overlap loop iterations.
This change adds a comment to clarify the reason/need for waiting
until the controller becomes idle and wraps the entire check into an
'if (out)' to make it clear that this is required for transfers with a
transmit-component only (for transfers having a receive-component,
completion of the transmit-side is trivially ensured by having
received the correct number of bytes).
The change does not increase execution time measurably in any of my
tests.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
While rkspi_enable_chip is called with true/false everywhere else in
the file, one call site uses '0' to denot 'false'.
This change this one parameter to 'false' and effects consistency.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
The maximum transfer length (in a single transaction) for the Rockchip
SPI controller is 64Kframes (i.e. 0x10000 frames) of 8bit or 16bit
frames and is encoded as (num_frames - 1) in CTRLR1. The existing
code subtracted the "minus 1" twice for a maximum transfer length of
0xffff (64K - 1) frames.
While this is not strictly an error (the existing code is correct, but
leads to a bit of head-scrating), fix this off-by-one situation.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Even though the priv-structure and the claim-bus function contain
logic for 16bit frames and for unidirectional transfer modes, neither
of these is used anywhere in the driver.
This removes the unused (as in "has no effect") logic and fields.
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
In analysing delays introduced for large SPI reads, the absence of any
indication when a delay was inserted (to ensure the CS toggling is
observed by devices) became apparent.
Add an additional debug-only debug message to record the insertion and
duration of any delay (note that the debug-message will cause a delay
on-top of the delay-duration).
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
This adds reset handling to the cadence qspi driver.
For backwards compatibility, only a warning is printed when failing to
get reset handles.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Convert driver to use spi-mem ops in order to support accelerated MMIO
flash interface in generic way and for better performance.
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Now that all boards using TI QSPI have moved to DM and DT, drop non DM
code completely.
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
[jagan: update MIGRATION.txt, rebase config_whitelist.txt]
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
This patch fixes the tap delay values to be set at 100MHz and 150MHz
as per TRM by fixing the if condition to use <= instead of <.
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
We are facing issues in the driver since SPI NOR framework has moved
on SPI MEM framework, and SPI NAND framework is not running properly
with the current driver.
To be able to solve issues met on SPI NOR Flashes and to be able to
support SPI NAND Flashes, the driver has been reworked. We are now using
exec_op ops instead of using xfer ops.
Thanks to this rework, the driver has been successfully tested with:
- mx66l51235l SPI NOR Flash on stm32f746 SOC
- n25q128a SPI NOR Flash on stm32f769 SOC
- mx66l51235l SPI NOR Flash on stm32mp1 SOC
- mt29f2g01abagd SPI NAND Flash on stm32mp1 SOC
Signed-off-by: Christophe Kerello <christophe.kerello@st.com>
Tested-by: Patrick DELAUNAY <patrick.delaunay@st.com>
Reviewed-by: Patrick DELAUNAY <patrick.delaunay@st.com>
This patch update the behavior introduced by
commit 96907c0fe5 ("dm: spi: Read default speed and mode values from DT")
In case of DT boot, don't read default speed and mode for SPI from
CONFIG_* but instead read from DT node. This will make sure that boards
with multiple SPI/QSPI controllers can be probed at different
bus frequencies and SPI modes.
Remove also use in boards of the value speed=0 (no more supported)
for ENV in SPI by using CONFIG_ENV_SPI_MAX_HZ=0.
DT values will be always used when available (full DM support of
SPI slave with available DT node) even if speed and mode are requested;
for example in splash screen support (in splash_sf_read_raw)
or in SPL boot (in spl_spi_load_image).
The caller of spi_get_bus_and_cs() no more need to force speed=0.
But the current behavior don't change if the SPI slave is not
present (device with generic driver is created automatically)
or if platdata is used (CONFIG_OF_PLATDATA).
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Update to prevent a byte of zeros being transmitted between
each byte in the tx buffer when providing both a tx buffer
and a rx buffer.
Signed-off-by: Jared Bents <jared.bents@rockwellcollins.com>
Cc: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Add check if exec_op is set before calling it.
At the moment it is called unconditionally, which leads to a crash if it
is not set correctly.
Signed-off-by: Bernhard Messerklinger <bernhard.messerklinger@br-automation.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Use quad write if SPI_TX_QUAD flag is set.
Tested quad write on Stratix 10 SoC board (Micron
serial NOR flash, mt25qu02g)
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Now the same SPI controller driver is reusable in all Allwinner
SoC variants, so rename the existing sun4i_spi.c into spi-sunxi.c
which eventually look like a common sunxi driver.
Also update the function, variable, structure names in driver from
sun4i into sunxi.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
- drop unused macros.
- use base instead of base_addr, for better code readability
- move .probe and .ofdata_to_platdata functions in required
places to add platdata support in future.
- use sentinel sun4i_spi_ids.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
The usual SPI transmission protocol in Allwinner A10 and A31
controllers share similar context with minimal changes in register
offsets along with few additional register bits on A31.
So, add A31 spi controller support in existing sun4i_spi with A31
specific register offsets and bits.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Add CLK support to enable AHB and MOD SPI clocks on sun4i_spi driver.
Clock disablement could be done while releasing the bus transfer, but
the existing code doesn't disable the clocks it only taken care of clock
enablement globally in probe.
So to make a proper clock handling, the clocks should enable it in claim
and disable it in release.
This patch would also do that change, by enable and disable clock in
proper order.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Support fifo_depth via drvdata instead of macro definition, this would
eventually reduce another macro definition for new SPI controller fifo
depth support addition.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Allwinner support two different SPI controllers one for A10 and
another for A31 with minimal changes in register offsets and
respective register bits, but the logic for accessing the SPI
master via SPI slave remains nearly similar.
Add enum offsets for register set and register bits, so-that
it can access both classes of SPI controllers.
Assign same control register for global, transfer and fifo control
registers to make the same code compatible with A31 SPI controller.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Tested-by: Stefan Mavrodiev <stefan@olimex.com> # A20-SOM204
