The driver currently hard-codes the max freqency for the sdhci
controllers. If the controller is unable to operate at the max
frequency, the mmc card will not be available on the first scan.
Subsequent scans will eventually find a working combination.
Fix this by allowing the driver to check for the max-frequency
property and default to the original value if it doesn't exist.
Signed-off-by: Peter Geis <pgwipeout@gmail.com>
Signed-off-by: Svyatoslav Ryhel <clamor95@gmail.com>
The block count limit on MMC based devices should be set according to
CONFIG_SYS_MMC_MAX_BLK_COUNT instead of hardcoding value.
Signed-off-by: Julien Masson <jmasson@baylibre.com>
Now that we have time conversion defines from in time.h there is no need
for each driver to define their own version.
Signed-off-by: Igor Prusov <ivprusov@salutedevices.com>
Reviewed-by: Svyatoslav Ryhel <clamor95@gmail.com> # tegra
Reviewed-by: Eugen Hristev <eugen.hristev@collabora.com> #at91
Reviewed-by: Caleb Connolly <caleb.connolly@linaro.org> #qcom geni
Reviewed-by: Stefan Bosch <stefan_b@posteo.net> #nanopi2
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Currently the renesas_sdhi_reset_tuning() unconditionally leaves SDHI
clock enabled after the tuning reset. This is not always necessary.
After the driver performed tuning reset at the end of probe function,
or in the unlikely case that tuning failed during regular operation,
the SDHI clock can be disabled after the tuning reset. The following
set_ios call would reconfigure the clock as needed.
In case of regular set_ios call which requires a tuning reset, keep
the clock enabled or disabled according to the mmc->clk_disable state.
With this in place, the controllers which have not been accessed via
block subsystem after boot are left in quiescent state. However, if an
MMC device is used e.g. for environment storage, that controller would
be accessed during the environment load and left active, including its
clock which would still be generated. This is due to the design of the
MMC subsystem, which does not deinit a controller after it was started
once, the controller is only deinited in case of mmc rescan, or before
OS boot.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Tested-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Reviewed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Tested-by: Thuan Nguyen Hong <thuan.nguyen-hong@banvien.com.vn>
commit 95168d77d3 ("sunxi: add Allwinner R528/T113 SoC support") added
the new entry out of order.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
The MMC controller driver is (and ought to be) the only user of these
register definitions. Put them in a header next to the driver to remove
the dependency on a specific ARM platform's headers.
Due to the sunxi_mmc_init() prototype, the file was not renamed. None of
the register definitions were changed.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Default-tap and default-trim values are used for eMMC setup
mostly on T114+ devices. As for now, those values are hardcoded
for T210 and ignored for all other Tegra generations. Fix this
by passing tap and trim values from dts.
Signed-off-by: Svyatoslav Ryhel <clamor95@gmail.com>
In line with changes elsewhere, drop inclusion of the common header.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
As noted in commit 3a6383207b ("mmc: sdhci: add the quirk for broken
r1b response"), some MMC controllers don't always set the transfer
complete bit with R1b responses.
According to the SD Host Controller Simplified Specification v4.20,
> In the case of a command pairing with response-with-busy[, Transfer
> Complete] is set when busy is de-asserted. Refer to DAT Line Active
> and Command Inhibit (DAT) in the Present State register.
By polling the DAT Line Active bit in the present state register, we can
detect when we are no longer busy, without waiting for a long timeout.
This results in much faster reads/writes on buggy controllers.
Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Tested-by: Henrik Grimler <henrik@grimler.se>
dm_pci_map_bar() return a value of (void *) already, hence no need
to cast it again before assigning to host->ioaddr.
Signed-off-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
mmc_bind() in mmc-uclass.c calls blk_create_devicef() which is
defined in blk-uclass.c, so SPL_BLK is required by SPL_DM_MMC.
Implicitly select SPL_BLK for SPL_DM_MMC.
Signed-off-by: Oleksandr Suvorov <oleksandr.suvorov@foundries.io>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
The R-Car Gen3 SD_CLK_CTRL register does not use BIT(16) to implement
divider 1/1024, instead BIT(16) is reserved bit and divider 1/1024 is
not supported. The divider 1/1024 is specific to Socionext variant of
the IP, as is indicated by Linux commit:
0196c8db8363 ("mmc: tmio: move tmio_mmc_set_clock() to platform hook")
"
Socionext (and Panasonic) uses bit 10 (CLKSEL) for 1/1. Also, newer
versions of UniPhier SoC variants use bit 16 for 1/1024.
"
Do not set the TMIO_SD_CAP_DIV1024 on Renesas R-Car platforms even if
the IP VERSION register does exist, and indicates IP version is newer
or equal to version 1.0 . The IP version 1.0 or newer does not imply
presence of the 1/1024 divider.
Since the TMIO driver is used exactly by two supported platforms, that
is Renesas R-Car and Socionext UniPhier, it is OK to check whether the
TMIO_SD_CAP_RCAR capability is not set to identify the UniPhier platform
and add the capability only on that platform.
Fixes: 58c35b17aa ("mmc: matsushita-common: Always check controller version")
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Acked-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Reviewed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Tested-by: Thuan Nguyen Hong <thuan.nguyen-hong@banvien.com.vn>
Unconditionally check for errors even after successful reception
of IRQ flag, since the hardware may set both an IRQ completion
flag and an error flag at the same time.
This mode of failure happens in case of an error during transfer,
in which case the hardware may set the expected IRQ completion
flag as well as error flags. The later is currently not checked
by the driver and such an error is not detected. Improve the
error detection.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Tested-by: Thuan Nguyen Hong <thuan.nguyen-hong@banvien.com.vn>
This driver is long superseded by renesas-sdhi.c and unused.
Drop the driver.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Acked-by: Paul Barker <paul.barker.ct@bp.renesas.com>
We should ensure that reset_free() is called in the error path.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Fixes: 93c811b733 ("mmc: renesas-sdhi: Initialize module on RZ/G2L")
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
This adds the remaining code bits to teach U-Boot about Allwinner's
newest SoC generation. This was introduced with the RISC-V based
Allwinner D1 SoC, which actually shares a die with the ARM cores versions
called R528 (BGA, without DRAM) and T113s (QFP, with embedded DRAM).
This adds the new Kconfig stanza, using the two newly introduced symbols
for the new SoC generation and pincontroller. It also adds the new symbols
to the relavent code places, to set all the hardcoded bits directly.
We need one DT override:
The ARM core version of the DT specifies the CPUX watchdog as
"reserved", which means it won't be recognised by U-Boot. Override this
in our generic sunxi-u-boot.dtsi, to let U-Boot pick up this watchdog,
so that the generic reset driver will work.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Allwinner seems to typically stick to a common MMIO memory map for
several SoCs, but from time to time does some breaking changes, which
also introduce new generations of some peripherals. The last time this
happened with the H6, which apart from re-organising the base addresses
also changed the clock controller significantly. We added a
CONFIG_SUN50I_GEN_H6 symbol back then to mark SoCs sharing those traits.
Now the Allwinner D1 changes the memory map again, and also extends the
pincontroller, among other peripherals.
To mark this generation of SoCs, add a CONFIG_SUNXI_GEN_NCAT2 symbol,
this name is reportedly used in the Allwinner BSP code, and prevents us
from inventing our own name.
Add this new symbol to some guards that were already checking for the H6
generation, since many features are shared between the two (like the
renovated clock controller).
This paves the way to introduce a first user of this generation.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Tested-by: Samuel Holland <samuel@sholland.org>
U-Boot's generic GPIO_EXTRA_HEADER is a convenience symbol to allow code
to more easily include platform specific GPIO headers. This should not
be needed in a DM world anymore, since the generic GPIO framework
handles that nicely.
For Allwinner boards we still need to deal with non-DM GPIO in the SPL,
but this should become the exception, not the rule.
Make this more obvious by removing the definition of GPIO_EXTRA_HEADER,
and just force every legacy user of platform specific GPIO to include
the new sunxi_gpio.h header explicitly. Everyone doing so should feel
ashamed and should find a way to avoid it from now on.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Tested-by: Samuel Holland <samuel@sholland.org>
On the Renesas RZ/G2L SoC family, we must ensure that the required clock
signals are enabled and the reset signal is de-asserted before we try to
communicate with the SDHI module.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Reviewed-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Move the assignment of priv->quirks earlier in the function. This allows
us to drop the quirks local variable and makes it easier to maintain
clean error handling when we add RZ/G2L support in the next patch.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
This old patch was marked as deferred. Bring it back to life, to continue
towards the removal of common.h
Move this out of the common header and include it only where needed.
Signed-off-by: Simon Glass <sjg@chromium.org>
With the commit 4fcba5d556 ("regulator: implement basic reference
counter"), keeping regulator enablement in balance become more important.
Disable vqmmc-supply before signal voltage is changed to keep regulator
enable counter in balance.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
With the commit 4fcba5d556 ("regulator: implement basic reference
counter") the return value of regulator_set_enable may be EALREADY or
EBUSY for fixed/gpio regulators.
Change to use the more relaxed regulator_set_enable_if_allowed to
continue if regulator already was enabled or disabled.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Tested-by: Svyatoslav Ryhel <clamor95@gmail.com> # P895 Tegra 3;
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org> # rockpro64-rk3399
This name is a little confusing since it suggests that it sets up the
sibling block device. In fact it sets up a bootdev for it. Rename the
function to make this clearer.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
If the device attached to the MMC bus is not removable, set force card-detect
bit to bypass card detection procedure, so card detection pin can be used for
other purposes.
It's also a workaround for SAMA5D2 who doesn't drive CMD if using GPIO for card
detection.
Signed-off-by: Zixun LI <zli@ogga.fr>
Reviewed-by: Eugen Hristev <eugen.hristev@collabora.com>
Dll reset is needed only for ZynqMP platforms, add condition in tuning
to call arasan_zynqmp_dll_reset() just for ZynqMP platforms.
On other platforms like Versal NET, If this condition is not added, we
see PLM error messages when dll reset smc is called.
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@amd.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/d673ff3bdc5c236a7f0403c920e719684abd6059.1688991117.git.michal.simek@amd.com
Set MMC clock when reverting to safe bus mode and speed
in case current MMC mode fails. Otherwise, trying out
the other modes may fail as well.
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Signed-off-by: Valentine Barshak <valentine.barshak@cogentembedded.com>
[hp: fallback to legacy_speed]
Signed-off-by: Hai Pham <hai.pham.ud@renesas.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
When tuning command (CMD21) fails with command error, call
mmc_send_stop_transmission() to send stop command (CMD12).
Reviewed-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Signed-off-by: Hai Pham <hai.pham.ud@renesas.com>
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
[Marek: Add dev_dbg() message in case tuning abort fails
Move tuning opcode check from mmc_abort_tuning()]
Reviewed-by: Peng Fan <peng.fan@nxp.com>
If a tuning command times out, the card could still be processing it,
which will cause problems for recovery. The eMMC specification section
6.6 Data transfer mode (cont’d) claims that CMD12 can be used to stop
CMD21:
"
The relationship between the various data transfer modes is summarized (see Figure 27):
- All data read commands can be aborted any time by the stop command (CMD12).
The data transfer will terminate and the Device will return to the Transfer State.
The read commands are: ... send tuning block (CMD21) ....
"
Add a function that does that.
Based on Linux commit [1] and [2].
[1] e711f0309109 ("mmc: mmc: Introduce mmc_abort_tuning()")
[2] 21adc2e45f4e ("mmc: Improve function name when aborting a tuning
cmd")
Reviewed-by: Takeshi Kihara <takeshi.kihara.df@renesas.com>
Reviewed-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Signed-off-by: Hai Pham <hai.pham.ud@renesas.com>
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
[Marek: Update commit message, quote relevant part of the specification.
Rename to mmc_send_stop_transmission().
Remove tuning opcode check, this is controller driver specific.
Deduplicate part of mmc_read_blocks() using this function.]
Reviewed-by: Peng Fan <peng.fan@nxp.com>
For MMC/eMMC, the MMC_CMD_STOP_TRANSMISSION response is R1 for read
transfers and R1b for write transfers per JEDEC Standard No. 84-B51
Page 126 . The response is R1b unconditionally per Physical Layer
Simplified Specification Version 9.00.
Correct the response type and add a comment about it.
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
It seems better to call this a 'bootdev' since this is name used in the
documentation. The older 'Bootdevice' name is no-longer used and may cause
confusion with the 'bootdevice' environment variable.
Update throughout to use bootdev.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Buffers created through DEFINE_(CACHE_)ALIGN_BUFFER are actually
pointers to the real underlying buffer. Using sizeof(...) is
not appropriate in this case.
Signed-off-by: Sam Edwards <CFSworks@gmail.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Loading part of TF-A into SRAM from eMMC using DMA fails on RK3399
similar to other Rockchip SoCs. Checksum validation fails with:
## Checking hash(es) for Image atf-2 ... sha256 error!
Bad hash value for 'hash' hash node in 'atf-2' image node
spl_load_simple_fit: can't load image loadables index 1 (ret = -1)
mmc_load_image_raw_sector: mmc block read error
SPL: failed to boot from all boot devices
### ERROR ### Please RESET the board ###
Add a device tree property, u-boot,spl-fifo-mode, to control when the
rockchip_sdhci driver should disable the use of DMA and fallback on PIO
mode. Same device tree property is used by the rockchip_dw_mmc driver.
In commit 2cc6cde647 ("mmc: rockchip_sdhci: Limit number of blocks
read in a single command") the DMA mode was disabled using a CONFIG
option on RK3588. Revert that and instead disable DMA using the device
tree property for all RK3588 boards, also apply similar workaround for
all RK3399 boards.
Fixes: 2cc6cde647 ("mmc: rockchip_sdhci: Limit number of blocks read in a single command")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Tested-by: Quentin Schulz <foss+uboot@0leil.net> # RK3399 Puma, RK3588 Tiger
The workaround to limit number of blocks to read in a single command
should only be applied to RK3568 and RK3588. Change to be more strict
when to apply the workaround.
Fixes: 2cc6cde647 ("mmc: rockchip_sdhci: Limit number of blocks read in a single command")
Suggested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Tested-by: Quentin Schulz <foss+uboot@0leil.net> # RK3399 Puma, RK3588 Tiger
This reverts commit a034ec06ff.
Commit 4a3ea75de4 ("Revert "mmc: sdhci: set to INT_DATA_END when
there are data"") reverted the alternative fix that was added for
Exynos 4 devices, causing an error when trying to boot from an sdcard:
<...>
Loading Environment from MMC... sdhci_send_command: Timeout for status update!
mmc fail to send stop cmd
<...>
Re-add the quirk to allow booting from sdcards again.
Signed-off-by: Henrik Grimler <henrik@grimler.se>
The fdt_addr_t and phys_addr_t size have been decoupled. A 32bit CPU
can expect 64-bit data from the device tree parser, so use
devfdt_get_addr_index_ptr instead of the devfdt_get_addr_index function
in the various files in the drivers directory that cast to a pointer.
As we are there also streamline the error response to -EINVAL on return.
Signed-off-by: Johan Jonker <jbx6244@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The fdt_addr_t and phys_addr_t size have been decoupled. A 32bit CPU
can expect 64-bit data from the device tree parser, so use
dev_read_addr_ptr instead of the dev_read_addr function in the
various files in the drivers directory that cast to a pointer.
As we are there also streamline the error response to -EINVAL on return.
Signed-off-by: Johan Jonker <jbx6244@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
It adds compatible "hisilicon,hi3798mv200-dw-mshc" for HC2910 SoC
Hi3798MV200 to probe this mmc driver.
Signed-off-by: Yang Xiwen <forbidden405@outlook.com>
Using DMA to load TF-A into SRAM fails when booting from eMMC on RK3588.
## Checking hash(es) for Image atf-3 ... sha256 error!
Bad hash value for 'hash' hash node in 'atf-3' image node
spl_load_simple_fit: can't load image loadables index 2 (ret = -1)
mmc_load_image_raw_sector: mmc block read error
Fix this by using PIO mode in SPL and limit the number of blocks used in
a single read command to avoid triggering Data End Bit Error interrupt.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Rockchip emmc devices have a similar issue to Rockchip dwmmc devices,
where performing DMA to SRAM later causes issues with suspend/resume.
Allow us to toggle SDMA in SPL for sdhci similar to ADMA support, so we
can ensure DMA is not used when loading the SRAM code.
Signed-off-by: Peter Geis <pgwipeout@gmail.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
[jonas@kwiboo.se: add Kconfig default value and fix ADMA typo]
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Add support for RK3588 to the rockchip sdhci driver.
Use driver data to handle differences between RK3568 and RK3588:
- Set "Receive original clock source is auto gating" for RK3588.
- Set "Receive clock source is no-inverted" only on RK3568 and "Transmit
clock source is invertion of original clock input" for RK3588.
- Use different txclk_tapnum for HS400 modes on RK3588.
- Configure the CMDOUT reg for HS400 modes for RK3588.
This is based on the mainline linux and vendor kernel driver and have
successfully been tested with rock5b-rk3588_defconfig and
CONFIG_MMC_HS200_SUPPORT=y
CONFIG_MMC_HS400_SUPPORT=y
CONFIG_MMC_HS400_ES_SUPPORT=y
CONFIG_MMC_SPEED_MODE_SET=y
using the following command to switch mode and then read 512 MiB of data
from eMMC into memory,
=> mmc dev 0 0 <mode> && mmc info && mmc read 10000000 2000 10000
for each of the modes below.
0 = MMC legacy
1 = MMC High Speed (26MHz)
3 = MMC High Speed (52MHz)
4 = MMC DDR52 (52MHz)
10 = HS200 (200MHz)
11 = HS400 (200MHz)
12 = HS400ES (200MHz)
Signed-off-by: Yifeng Zhao <yifeng.zhao@rock-chips.com>
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Shawn Lin <shawn.lin@rock-chips.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Adjust tap number for transmit clock, tap number and delay number for
strobe input to fix HS400 modes on RK3568.
New values have been picked from vendor kernel and u-boot and have
successfully been tested with rock-3a-rk3568_defconfig and
CONFIG_MMC_HS200_SUPPORT=y
CONFIG_MMC_HS400_SUPPORT=y
CONFIG_MMC_HS400_ES_SUPPORT=y
CONFIG_MMC_SPEED_MODE_SET=y
using the following command to switch mode and then read 512 MiB of data
from eMMC into memory,
=> mmc dev 0 0 <mode> && mmc info && mmc read 10000000 2000 10000
for each of the modes below.
0 = MMC legacy
1 = MMC High Speed (26MHz)
3 = MMC High Speed (52MHz)
4 = MMC DDR52 (52MHz)
10 = HS200 (200MHz)
11 = HS400 (200MHz)
12 = HS400ES (200MHz)
Signed-off-by: Yifeng Zhao <yifeng.zhao@rock-chips.com>
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
