On versal platform, enable apb linear mode for apb read and write
execute operations amd disable it when using dma reads. This is done by
xilinx_pm_request() secure calls when CONFIG_ZYNQMP_FIRMWARE is enabled,
else we use direct raw reads and writes in case of mini U-Boot.
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@xilinx.com>
Link: https://lore.kernel.org/r/20220512100535.16364-5-ashok.reddy.soma@xilinx.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Add support for cadence ospi driver for Versal platform. This driver
provides support for DMA read operation which utilizes cadence qspi
driver.
If "cdns,is-dma" DT property is specified use dma for read operation
from cadence_qspi driver. As cadence_qspi_apb_dma_read() is defined in
cadence_ospi_versal driver add a weak function defination in
cadence_qspi driver.
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@xilinx.com>
Link: https://lore.kernel.org/r/20220512100535.16364-3-ashok.reddy.soma@xilinx.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Move all the cadence macros from cadence_qspi_apb.c to cadence_qspi.h
file.
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@xilinx.com>
Link: https://lore.kernel.org/r/20220512100535.16364-2-ashok.reddy.soma@xilinx.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Wait for the read/write transfer finish bit get actually cleared,
this does not happen immediately on at least SoCFPGA Gen5.
Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Cc: Vignesh R <vigneshr@ti.com>
Cc: Pratyush Yadav <p.yadav@ti.com>
Both dummy.nbytes and dummy.buswidth may be zero. By not checking
the later, it is possible to trigger division by zero and a crash.
This does happen with tiny SPI NOR framework in SPL. Fix this by
adding the check and returning zero dummy bytes in such a case.
Fixes: 38b0852b0e ("spi: cadence-qspi: Add support for octal DTR flashes")
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Jagan Teki <jagan@amarulasolutions.com>
Cc: Vignesh R <vigneshr@ti.com>
Cc: Pratyush Yadav <p.yadav@ti.com>
[trini: Drop Pratyush's RB as his requested changes weren't made as
Marek disagreed]
Set up opcode extension and enable/disable DTR mode based on whether the
command is DTR or not.
xSPI flashes can have a 4-byte dummy address associated with some
commands like the Read Status Register command in octal DTR mode. Since
the flash does not support sending the dummy address, we can not use
automatic write completion polling in DTR mode. Further, no write
completion polling makes it impossible to use DAC mode for DTR writes.
In that mode, the controller does not know beforehand how long a write
will be and so it can de-assert Chip Select (CS#) at any time. Once CS#
is de-assert, the flash will go into burning phase. But since the
controller does not do write completion polling, it does not know when
the flash is busy and might send in writes while the flash is not ready.
So, disable write completion polling and make writes go through indirect
mode for DTR writes and let spi-mem take care of polling the SR.
Signed-off-by: Pratyush Yadav <p.yadav@ti.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Once the start bit is toggled it takes a small amount of time before it
is internally synchronized. This means we can't start writing during
that part. So add a small delay to allow the bit to be synchronized.
Signed-off-by: Pratyush Yadav <p.yadav@ti.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Cadence OSPI is similar to QSPI IP except that it supports Octal IO
(8 IO lines) flashes. Add support for Cadence OSPI IP with existing
driver using new compatible
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Add support for Direct Access Controller mode of Cadence QSPI. This
allows MMIO access to SPI NOR flash providing better read performance.
Direct mode is only exercised if AHB window size is greater than 8MB.
Support for flash address remapping is also not supported at the moment
and can be added in future.
For better performance, driver uses DMA to copy data from flash in
direct mode using dma_memcpy().
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Tested-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Current Cadence QSPI driver has few limitations. It assumes all read
operations to be in Quad mode and thus does not support SFDP parsing.
Also, adding support for new mode such as Octal mode would not be
possible with current configuration. Therefore move the driver over to spi-mem
framework. This has added advantage that driver can be used to support
SPI NAND memories too.
Hence, move driver over to new spi-mem APIs.
Please note that this gets rid of mode bit setting done when
CONFIG_SPL_SPI_XIP is defined as there does not seem to be any user to
that config option.
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Tested-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Acked-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>
Use "%p" to print cmdbuf.
Compilation warning as below:
CC spl/drivers/spi/cadence_qspi_apb.o
LD spl/lib/built-in.o
drivers/spi/cadence_qspi_apb.c: In function ‘cadence_qspi_apb_indirect_write_setup’:
drivers/spi/cadence_qspi_apb.c:696:18: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
cmdlen, (unsigned int)cmdbuf);
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
Acked-by: Marek Vasut <marex@denx.de>
Make flash writes 32 bit aligned by using bounce buffers to deal with
non 32 bit aligned buffers.
This is required because as per TI K2G TRM[1], the external master is
only permitted to issue 32-bit data interface writes until the last word
of an indirect transfer. Otherwise indirect writes is known to fail
sometimes.
[1] http://www.ti.com/lit/ug/spruhy8g/spruhy8g.pdf
Signed-off-by: Vignesh R <vigneshr@ti.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Simon Goldschmidt <sgoldschmidt@de.pepperl-fuchs.com>
Reviewed-by: Jason Rush <jarush@gmail.com>
Acked-by: Jason Rush <jarush@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
This reverts commit 57897c13de.
Using bounce_buf.c to handle non-DMA alignment problems is bad as
bounce_buf.c does cache manipulations which is not required. Therefore
revert this patch in favour of local bounce buffer solution in the next
patch.
Signed-off-by: Vignesh R <vigneshr@ti.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Simon Goldschmidt <sgoldschmidt@de.pepperl-fuchs.com>
Reviewed-by: Jason Rush <jarush@gmail.com>
Acked-by: Jason Rush <jarush@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
This reverts commit b63b46313e.
This commit changed cadence_qspi_apb to use bouncebuf.c, which invalidates
the data cache after reading. This is meant for dma transfers only and
breaks the cadence_qspi driver which copies via cpu only: data that is
copied by the cpu is in cache only and the cache invalidation at the end
throws away this data.
Signed-off-by: Simon Goldschmidt <sgoldschmidt@de.pepperl-fuchs.com>
Signed-off-by: Vignesh R <vigneshr@ti.com>
Acked-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jason Rush <jarush@gmail.com>
Acked-by: Jason Rush <jarush@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Adopt the Linux DT bindings. This also fixes an issue
with the indaddrtrig register on the Cadence QSPI
device being programmed with the wrong value for the
socfpga arch.
Tested on TI K2G platform:
Tested-by: Vignesh R <vigneshr@ti.com>
Tested on a socfpga-cyclonev board:
Tested-by: Simon Goldschmidt <sgoldschmidt@de.pepperl-fuchs.com>
Signed-off-by: Jason Rush <jarush@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Acked-by: Simon Goldschmidt <sgoldschmidt@de.pepperl-fuchs.com>
Acked-by: Marek Vasut <marex@denx.de>
wait_for_bit callers use the 32 bit LE version
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
According to Section 11.15.4.9.1 Indirect Read Controller of K2G SoC
TRM SPRUHY8D[1], the external master is only permitted to issue 32-bit
data interface reads until the last word of an indirect transfer
So, make sure that QSPI indirect reads are 32 bit sized except for the
final read. If the rxbuf is unaligned then use bounce buffer, so that
readsl() can be used instead of readsb() to avoid non 32-bit accesses.
[1]www.ti.com/lit/ug/spruhy8d/spruhy8d.pdf
Signed-off-by: Vignesh R <vigneshr@ti.com>
Reviewed-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jagan Teki <jagan@openedev.com>
According to Section 11.15.4.9.2 Indirect Write Controller of K2G SoC
TRM SPRUHY8D[1], the external master is only permitted to issue 32-bit
data interface writes until the last word of an indirect transfer
otherwise indirect writes is known to fails sometimes. So, make sure
that QSPI indirect writes are 32 bit sized except for the last write. If
the txbuf is unaligned then use bounce buffer to avoid data aborts.
So, now that the driver uses bounce_buffer, enable CONFIG_BOUNCE_BUFFER
for all boards that use Cadence QSPI driver.
[1]www.ti.com/lit/ug/spruhy8d/spruhy8d.pdf
Signed-off-by: Vignesh R <vigneshr@ti.com>
Reviewed-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jagan Teki <jagan@openedev.com>
The Cadence QSPI controller has specified overheads for the various CS
times that are in addition to those programmed in to the Device Delay
register. The overheads are different for the delays.
In addition, the existing code does not handle the case when the delay
is less than a SCLK period.
This change accurately calculates the additional delays in Ref clocks.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Instead of extracting mode settings and passing them as separate
args to another function, just pass the SPI mode as an arg.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
A lot of the #defines are for single bits in a register, where the
name has _MASK on the end. Since this can be used for both a mask
and the value, remove _MASK from them.
Whilst doing so, also remove the unnecessary brackets around the
constants.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Acked-by: Marek Vasut <marek.vasut@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Most of the code already uses #defines for the bit value, rather
than the shift required to get the value. This changes the remaining
code over.
Whislt at it, fix the names of the "Rd Data Capture" register defs.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Acked-by: Marek Vasut <marek.vasut@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Show what the output clock rate actually is.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Acked-by: Marek Vasut <marek.vasut@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
With the existing code, when the requested SPI clock rate is near
to the lowest that can be achieved by the hardware (max divider
of the ref clock is 32), the generated clock rate is wrong.
For example, with a 50MHz ref clock, when asked for anything less
than a 1.5MHz SPI clock, the code sets up the divider to generate
25MHz.
This change fixes the calculation.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Or'ing together bit positions is clearly wrong.
Signed-off-by: Phil Edworthy <phil.edworthy@renesas.com>
Acked-by: Marek Vasut <marek.vasut@gmail.com>
Reviewed-by: Jagan Teki <jagan@openedev.com>
Now, arch/${ARCH}/include/asm/errno.h and include/linux/errno.h have
the same content. (both just wrap <asm-generic/errno.h>)
Replace all include directives for <asm/errno.h> with <linux/errno.h>.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
[trini: Fixup include/clk.]
Signed-off-by: Tom Rini <trini@konsulko.com>
Ensuring the baudrate divisor value doesn't exceed the max value
in the calculation.It will be capped at max value to ensure the
correct value being written into the register.
Example of the existing bug is when calculated div = 16. After and
with the mask, the value written to register is actually 0 (register
field for baudrate divisor). With this fix, the value written is now
15 which is max value for baudrate divisor.
Signed-off-by: Chin Liang See <clsee@altera.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Jagan Teki <jteki@openedev.com>
Cc: Dinh Nguyen <dinguyen@altera.com>
Instead of relying on CONFIG_SPI_FLASH_QUAD to be defined to enable QUAD
mode, make use of mode_rx field of dm_spi_slave_platdata to determine
whether to enable or disable QUAD mode. This is necessary to support
muliple SPI controllers where one of them may not support QUAD mode.
Signed-off-by: Vignesh R <vigneshr@ti.com>
Tested-by: Marek Vasut <marex@denx.de>
Acked-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jagan Teki <jteki@openedev.com>
AHB address can be as long as 32 bit, hence remove the
CQSPI_REG_INDIRECTRDSTARTADDR mask. Since AHB address is passed from DT
and read as u32 value, it anyway does not make sense to mask upper bits.
Signed-off-by: Vignesh R <vigneshr@ti.com>
Tested-by: Marek Vasut <marex@denx.de>
Acked-by: Marek Vasut <marex@denx.de>
Reviewed-by: Jagan Teki <jteki@openedev.com>
The indirect read code is a pile of nastiness. This patch replaces
the whole unmaintainable indirect read implementation with the one
from upcoming Linux CQSPI driver, which went through multiple rounds
of thorough review and testing. All the patch does is it plucks out
duplicate ad-hoc code distributed across the driver and replaces it
with more compact code doing exactly the same thing. There is no
speed change of the read operation.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Anatolij Gustschin <agust@denx.de>
Cc: Chin Liang See <clsee@altera.com>
Cc: Dinh Nguyen <dinguyen@opensource.altera.com>
Cc: Jagan Teki <jteki@openedev.com>
Cc: Pavel Machek <pavel@denx.de>
Cc: Stefan Roese <sr@denx.de>
Cc: Vignesh R <vigneshr@ti.com>
The indirect write code is buggy pile of nastiness which fails horribly
when the system runs fast enough to saturate the controller. The failure
results in some pages (256B) not being written to the flash. This can be
observed on systems which run with Dcache enabled and L2 cache enabled,
like the Altera SoCFPGA.
This patch replaces the whole unmaintainable indirect write implementation
with the one from upcoming Linux CQSPI driver, which went through multiple
rounds of thorough review and testing. While this makes the patch look
terrifying and violates all best-practices of software development, all
the patch does is it plucks out duplicate ad-hoc code distributed across
the driver and replaces it with more compact code doing exactly the same
thing.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Anatolij Gustschin <agust@denx.de>
Cc: Chin Liang See <clsee@altera.com>
Cc: Dinh Nguyen <dinguyen@opensource.altera.com>
Cc: Jagan Teki <jteki@openedev.com>
Cc: Pavel Machek <pavel@denx.de>
Cc: Stefan Roese <sr@denx.de>
Cc: Vignesh R <vigneshr@ti.com>
Replace numerical bit shift with BIT macro
in cadence_qspi_apb
:%s/(1 << nr)/BIT(nr)/g
where nr = 0, 1, 2 .... 31
Cc: Stefan Roese <sr@denx.de>
Cc: Marek Vasut <marex@denx.de>
Acked-by: Vikas Manocha <vikas.manocha@st.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Jagan Teki <jteki@openedev.com>
This patch makes the code compatible with FIFO depths other than 4
bytes. It also simplify read/write FIFO loops.
Signed-off-by: Vikas Manocha <vikas.manocha@st.com>
Tested-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagannadh Teki <jteki@openedev.com>
sram size could be different on different socs, e.g. on stv0991 it is 256 while
on altera platform it is 128. It is better to receive it from device tree.
Signed-off-by: Vikas Manocha <vikas.manocha@st.com>
Tested-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagannadh Teki <jteki@openedev.com>
There is no need to re-configure sram partition for every read/write for
better full use of sram for read or write. This patch divides the half
sram for read & half for write once at initialization.
Signed-off-by: Vikas Manocha <vikas.manocha@st.com>
Tested-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagannadh Teki <jteki@openedev.com>
This driver is cloned from the Altera Rockerboard.org U-Boot
repository. I used this git tag: ACDS14.0.1_REL_GSRD_RC2. With Some
modification to support the U-Boot driver model (DM).
As mentioned above, in this new version I ported this driver to the
new driver model (DM). One big advantage of this move is that now
multiple SPI drivers can be enabled on one platform. And since the
SoCFPGA also has the Designware SPI master controller integrated,
this feature is really needed to support both controllers.
Because of this, this series needs the DT support for SoCFPGA
to be applied. For DT based probing in the SPI DM.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Chin Liang See <clsee@altera.com>
Cc: Dinh Nguyen <dinguyen@altera.com>
Cc: Vince Bridgers <vbridger@altera.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Pavel Machek <pavel@denx.de>
Cc: Simon Glass <sjg@chromium.org>
Cc: Jagannadha Sutradharudu Teki <jagannadh.teki@gmail.com>
