Move this out of the common header and include it only where needed. In
a number of cases this requires adding "struct udevice;" to avoid adding
another large header or in other cases replacing / adding missing header
files that had been pulled in, very indirectly. Finally, we have a few
cases where we did not need to include <asm/global_data.h> at all, so
remove that include.
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Rini <trini@konsulko.com>
The R-Car3 SDHI should set these two bits in DMA_MODE register according
to the specification, to indicate 64bit bus width. No other bus width
options are permitted and the default value is 0, which is incorrect.
Set the bits accordingly.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
When you enable CONFIG_OF_LIVE, you will end up with a lot of
conversions.
To generate this commit, I used coccinelle excluding drivers/core/,
include/dm/, and test/
The semantic patch that makes this change is as follows:
<smpl>
@@
expression dev;
@@
-devfdt_get_addr(dev)
+dev_read_addr(dev)
</smpl>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
When you enable CONFIG_OF_LIVE, you will end up with a lot of
conversions.
To generate this commit, I used coccinelle excluding drivers/core/,
include/dm/, and test/
The semantic patch that makes this change is as follows:
<smpl>
@@
expression dev;
@@
-devfdt_get_addr(dev)
+dev_read_addr(dev)
</smpl>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
In R-Car Gen 3, there is a DMA controller restriction of SDHI.
When the transfer exceeding the 4 kByte boundary is performed while
the DRAM address is not 128 byte aligned, the bus is occupied.
This patch avoids this.
Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@renesas.com>
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
The implementation of dma_map_single() and dma_unmap_single() is
exactly the same for all the architectures that support them.
Factor them out to <linux/dma-mapping.h>, and make all drivers to
include <linux/dma-mapping.h> instead of <asm/dma-mapping.h>.
If we need to differentiate them for some architectures, we can
move the generic definitions to <asm-generic/dma-mapping.h>.
Add some comments to the helpers. The concept is quite similar to
the DMA-API of Linux kernel. Drivers are agnostic about what is
going on behind the scene. Just call dma_map_single() before the
DMA, and dma_unmap_single() after it.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
At present dm/device.h includes the linux-compatible features. This
requires including linux/compat.h which in turn includes a lot of headers.
One of these is malloc.h which we thus end up including in every file in
U-Boot. Apart from the inefficiency of this, it is problematic for sandbox
which needs to use the system malloc() in some files.
Move the compatibility features into a separate header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Drop local dma_map_single() and dma_unmap_single() and use arch specific
common implementation
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Acked-by: Masahiro Yamada <yamada.masahiro@socionext.com>
These functions are CPU-related and do not use driver model. Move them to
cpu_func.h
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
On 16bit variants of the TMIO SD IP, the SECCNT register can only be
programmed to 16bit values, while on the 32bit and 64bit variants it
can be programmed to 32bit values. The SECCNT register indicates the
maximum number of blocks in a continuous transfer. Hence, limit the
maximum continuous transfer block count to 65535 blocks on 16bit
variants of the TMIO IP and to BIT(32)-1 blocks on 32bit and 64bit
variants.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
According to latest specification rev.0026, when HOST_MODE bit 0
(WMODE) is not set, HOST_MODE bit 8 (BUSWIDTH) is ignored. Clear
HOST_MODE bit 8 in such case and align the code with Linux and
avoid possible unforeseen issues.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Set the HOST_MODE register WMODE bit according to the SDHI bus width,
that is 0 for 64bit bus and 1 for 16/32bit bus.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
The internal DMAC asserts DMA transfer end bit too early on older
version of the TMIO IPs which use bit 17 for DTRAEND. Add 1uS
delay after the completion of DMA transfer and before invalidating
the cache to let the DMAC fully complete the transfer. Otherwise,
it could happen that the last few bytes of a transferred data are
not available.
A test case to trigger this behavior is the following command, ran
on the U-Boot command line, with Sandisk 16 GiB UHS-I card inserted
into SDHI slot 0 and with first partition being of type FAT:
=> while true ; do mmc rescan ; fstype mmc 0:1 ; done
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Different versions of the SDHI core use either bit 17 or bit 20 for the
DTRAEND indication, which can differ even between SoC revisions. Make
the DTRAEND bit position part of the driver private data, so that the
probe function can set this accordingly. Set this to 20 on Socionext
SoCs and either 17 or 20 on Renesas SoCs, depending on the SoC.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Reorder the tmio_sd_set_clk_rate() function such that it handles all
of the clock requiests correctly. Specifically, before this patch,
clock request with (mmc->clock == 0 && mmc->clk_disable) could leave
the clock enabled, as the function would exit on if (!mmc->clock)
condition on top and will not handle the mmc->clk_disable at all.
Rather than band-aid fixing just that particular problem, reorder
the entire function to make it easier to understand and verify that
all the cases are covered. The function has three sections now:
First, if mmc->clock != 0, we calculate divider for the SD block.
Second, if mmc->clock != 0 and SD block clock are enabled and
current divider is not equal to the new divider, then
stop the clock and update the divider.
Third, if mmc->clk_disable is set, disable the clock, otherwise
enable the clock. This happens independently of divider
update now.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
The TMIO core has a feature where it can automatically disable clock output
when the bus is not in use. While this is useful, it also interferes with
switching the bus to 1.8V and other background tasks of the SD/MMC cards,
which require clock to be enabled.
This patch respects the mmc->clk_disable and only disables the clock when
the MMC core requests it. Otherwise the clock are continuously generated
on the bus.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
The TMIO core has a quirk where divider == 1 must not be set in DDR modes.
Handle this by setting divider to 2, as suggested in the documentation.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Switch the driver to using clk_get_rate()/clk_set_rate() instead of
caching the mclk frequency in it's private data. This is required on
the SDHI variant of the controller, where the upstream mclk need to
be adjusted when using UHS modes.
Platforms which do not support clock framework or do not support it
in eg. SPL default to 100 MHz clock.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
---
V2: - Fix build on certain platforms using SPL without clock framework
V3: - Turn clk_get_rate into a callback and fill it as needed on both
renesas and socionext platforms
Preinitialize the SD card signals regulator to 3.3V, which is the
default post-reset setting, to be sure the regulator is set to a
valid value.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Configure the clock settings before reconfiguring any other IO settings.
This is required when the clock must be stopped before changing eg. the
pin configuration or any of the other properties of the bus. Running the
clock configuration first allows the MMC core to do just that.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
In case the controller performs card tuning, that is, sends MMC
command 19 or 21, silence possible CRC error warning prints. The
warnings are bound to happen, since the tuning will fail for some
settings while searching for the optimal configuration of the bus
and that is perfectly OK.
This patch passes around the MMC command structure and adds check
into tmio_sd_check_error() to avoid printing CRC error warning
when the tuning happens.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Properly handle return values and abort operations when they are
non-zero. This is a minor improvement, which fixes two remaining
unchecked return values.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
The SD UHS SDR12, SDR25, SDR50, SDR104, DDR50 and MMC HS200, HS400
modes all use 1.8V signaling, while all the legacy modes use 3.3V
signaling. While there are extra modes which use 1.2V signaling,
the existing hardware does not support those.
Simplify the pinmux such that 3.3V signaling implies legacy mode
pinmux and the rest implies UHS mode pinmux. This prevents the
massive case statement from growing further. Moreover, it fixes
an edge case where during SD 1.8V switch, the bus mode is still
set to default while the signaling is already set to 1.8V, which
results in an attempt to communicate with a 1.8V card using pins
in 3.3V mode and thus communication failure.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
The internal DMAC on Gen3 is 32bit only, limit the DMA address
range to 32bit.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
Pass the entire source data pointer to tmio_sd_addr_is_dmaable()
so we don't have to apply casts throughout the code.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
I need to differentiate the clock handling for uniphier-sd. Move it
to each driver's probe function from the tmio common code so that
renesas-sdhi will not be affected.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Synchronize the naming with Linux, call the common code TMIO.
No functional change.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Jaehoon Chung <jh80.chung@samsung.com>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
