If DMA(SDMA or ADMA) is not used, the cache invalidation
after reading is no need, should be skipped. Otherwise
U-Boot may hang at the cache invalidation.
Found this issue and tested this fix on DragonBoard 410c.
Fixes: commit 4155ad9aac ("mmc: sdhci: fix missing cache invalidation after reading by DMA")
Signed-off-by: Yuezhang.Mo <Yuezhang.Mo@sony.com>
Reviewed-by: Andy Wu <Andy.Wu@sony.com>
This will allow us to use DM variants of phys_to_bus()/bus_to_phys()
when relevant.
Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Peter Robinson <pbrobinson@gmail.com>
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
Add a set_voltage() function which handles the switch from 3.3V to 1.8V
for SD card UHS modes.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Signed-off-by: Aswath Govindraju <a-govindraju@ti.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
This reverts commit 942b5fc032.
This is partial revert of the above commit.
mmc_of_parse() is reading no-1-8-v from device tree and if set,
it is clearing the UHS speed capabilities of cfg->host_caps.
cfg->host_caps &= ~(UHS_CAPS | MMC_MODE_HS200 |
MMC_MODE_HS400 | MMC_MODE_HS400_ES);
This is still missing to clear UHS speeds like SDHCI_SUPPORT_SDR104,
SDHCI_SUPPORT_SDR50 and SDHCI_SUPPORT_DDR50.
Even if we clear the flags SDHCI_SUPPORT_XXX in mmc_of_parse(),
these speed flags are getting set again in cfg->host_caps in
sdhci_setup_cfg().
The reason for this is, SDHCI_SUPPORT_XXX flags are cleared
only if controller is not capable of supporting MMC_VDD_165_195 volts.
if (caps & SDHCI_CAN_VDD_180)
cfg->voltages |= MMC_VDD_165_195;
if (!(cfg->voltages & MMC_VDD_165_195))
caps_1 &= ~(SDHCI_SUPPORT_SDR104 | SDHCI_SUPPORT_SDR50 |
SDHCI_SUPPORT_DDR50);
It means "no-1-8-v", which is read from DT is not coming in to effect.
So it is better we keep the host quirks(SDHCI_QUIRK_NO_1_8_V) to
clear UHS speeds based on no-1-8-v from device tree.
Hence revert the functionality related to no-1-8-v only, rest is fine
in the patch.
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
There are other (non-SDHCI) controllers which supports ADMA2 descriptor
tables, namely the Freescale eSDHC. Instead of copying the code, move it
into an own module.
Signed-off-by: Michael Walle <michael@walle.cc>
Instead of recasting and shifting use macros which are designed for taking
upper/lower 32bit value from 64bit variable.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
The capabilities register has a field to indicate whether the host
supports high speed mode or not. Add high speed host_caps based on
this bit instead of enabling it by default.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Reviewed-by: Michal Simek <michal.simek@xilnx.com>
Tested-by: Michal Simek <michal.simek@xilnx.com>
(zcu104 with sdhci-caps-mask = <0 0x200000>;)
SDHCI HISPD bits need to be configured based on desired mmc
timings mode and some HISPD quirks.
So, handle the HISPD bit based on the mmc computed selected
mode(timing parameter) rather than fixed mmc card clock
frequency.
Linux handle the HISPD similar like this in below commit but no
SDHCI_QUIRK_BROKEN_HISPD_MODE,
commit <501639bf2173> ("mmc: sdhci: fix SDHCI_QUIRK_NO_HISPD_BIT handling")
This eventually fixed the mmc write issue observed in
rk3399 sdhci controller.
Bug log for refernece,
=> gpt write mmc 0 $partitions
Writing GPT: mmc write failed
** Can't write to device 0 **
** Can't write to device 0 **
error!
Cc: Kever Yang <kever.yang@rock-chips.com>
Cc: Peng Fan <peng.fan@nxp.com>
Peng Fan: added back "ctrl &= ~SDHCI_CTRL_HISPD;" per Jaehoon's suggestion
Tested-by: Suniel Mahesh <sunil@amarulasolutions.com> # roc-rk3399-pc
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Move this header out of the common header. Network support is used in
quite a few places but it still does not warrant blanket inclusion.
Note that this net.h header itself has quite a lot in it. It could be
split into the driver-mode support, functions, structures, checksumming,
etc.
Signed-off-by: Simon Glass <sjg@chromium.org>
If CONFIG_MMC_SDHCI_SDMA is enabled but the HW could not support it,
we no longer error out. Instead we do not enable it in the host.
Change the output from printf to debug as this isn't an error but only
additional information now.
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
If Host controller doesn't support SDMA, it doesn't need to return
error. Because it can be worked with PIO mode.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
Use phys2bus macro when dma address is accessed.
Some targets need to use pyhs2bus macro. (e.g, RPI4)
After applied it, SDMA mode can be used.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
Currently, the entry "bus-width = <8>" in the ZynqMP's sdhci nodes
is not evaluated. This results in the bus width staying at its default
value (4 bit in HS200 mode).
Fix this by calling mmc_of_parse. This function also checks for the
"no-1-8-v" and "max-frequency" entries. Remove the handling of those
nodes from this driver.
Signed-off-by: Benedikt Grassl <Benedikt.Grassl@rohde-schwarz.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Add the sdhci_deferred_probe() function to register as the
deferred_probe() callback to the mmc core. It will in turn call the
deferred_probe() callback of the platform drivers as declared in the
sdhci_ops.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
This driver currently performs cache operation before the DMA start,
but does nothing after the DMA completion.
When reading data by DMA, the cache invalidation is needed also after
finishing the DMA transfer. Otherwise, the CPU might read data from
the cache instead of from the main memory when speculative memory read
or memory prefetch occurs.
Instead of calling the cache operation directly, this commit adds
dma_unmap_single(), which performs cache invalidation internally,
but drivers do not need which operation is being run.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, sdhci_prepare_dma() calls flush_cache() regardless of the
DMA direction.
Actually, cache invalidation is enough when reading data from the device.
This is correctly handled by dma_map_single(), which mimics the DMA-API
in Linux kernel. Drivers can be agnostic which cache operation occurs
behind the scene.
This commit also sanitizes the difference between the virtual address
and the dma address.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
host->mmc is already (struct mmc *).
memalign() returns an opaque pointer, so there is no need for casting.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Use {lower,upper}_32_bits() instead of the combination of cast
and shift.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
The same code is run for both SDHCI_QUIRK_32BIT_DMA_ADDR and
define(CONFIG_FIXED_SDHCI_ALIGNED_BUFFER).
Unify the code.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Using the global variable does not look nice.
Add a new field sthci::align_buffer to point to the bounce buffer.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.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>
The sdhci capabilities registers can be incorrect. The
sdhci-caps-mask and sdhci-caps dt properties specify which bits of
the registers are incorrect and what their values should be. This
patch makes the sdhci driver use those properties to correct the caps.
Also use "dev_read_u64_default" instead of "dev_read_u32_array" for
caps mask.
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
The standard property name for chip-detect gpio is "cd-gpios". All
in-tree DT files use only this name.
Fixes: 451931ea70 ("mmc: sdhci: Read cd-gpio from devicetree")
Cc: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
The HOST_CONTROL2 register is a part of SDHC v3.00 and not just specific
to arasan/zynq controllers. Add the same to sdhci.h.
Also create a common API to set UHS timings in HOST_CONTROL2.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
The am654_sdhci driver needs to switch the clock off
before disabling its phy dll and needs to re-enable
the clock before enabling the phy again.
Therefore, make the sdhci_set_clock() function accessible
in the am654_sdhci driver.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add Support for masking some bits in the capabilities
register of a host controller.
Also remove the redundant readl() into caps1.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Card detect function implemented for SDHCI framework.
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
This patch reads cd-gpio property from devicetree
Signed-off-by: T Karthik Reddy <t.karthik.reddy@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
The Standard Host Controller Interface (SDHCI) specification version
3.00 adds support for Advanced DMA (ADMA) for both 64 and 32 bit widths
of DMA. ADMA2 uses a table of descriptors for aggregating DMA requests.
This significantly improves read and write throughput.
Add Support for the same.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
In preparation for addition of ADMA2 support, cleanup SDMA handling by
moving it to a new sdhci_prepare_dma() function. Also add a flags field
in sdhci_host to indicate if DMA is enabled.
Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
This patch fixes the mmc tuning command failures
when tuning pattern data needs to read back for
comparision against the expected bit pattern.
Reported-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Tested-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The check is necessary to avoid NULL pointer dereference.
Signed-off-by: Ramon Fried <ramon.fried@gmail.com>
Reviewed-by: Michal Simek <michal.simek@xilinx.com>
This patch reads the capabilities register1 and update the host
caps accordingly for mmc layer usage. This patch mainly reads
for UHS capabilities inorder to support SD3.0.
Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
This patch adds support to invoke any platform specific tuning
and delay routines if available.
Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
This patch adds support to disable clock if clk_disable
was set and then enable or set clock if the clock was changed
or clock was disabled when clock needs to be enabled.
Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
This patch changed the datatype of variable "start" from uint to ulong
to work properly on 64-bit machines as well. Also the return type of
get_timer() function is ulong.
Signed-off-by: Vipul Kumar <vipul.kumar@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.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>
Some IP-core implementations of the SDHCI have different troubles on the
silicon where they are placed.
On ZYNQ platform for example Xilinx doesn't accept the hold timing of an
eMMC chip which operates in High-Speed mode and must be forced to
operate in non high-speed mode. To get rid of this
"SDHCI_QUIRK_BROKEN_HISPD_MODE" is introduced.
For more details about this refer to the Xilinx answer-recor #59999https://www.xilinx.com/support/answers/59999.html
This commit:
- doesn't set HISPD bit on the host-conroller
- reflects this fact within the host-controller capabilities
Upon this the layer above (mmc-driver) can setup the card correctly.
Otherwise the MMC card will be switched into high-speed mode and causes
possible timing violation on the host-controller side.
Signed-off-by: Hannes Schmelzer <oe5hpm@oevsv.at>
Signed-off-by: Hannes Schmelzer <hannes.schmelzer@br-automation.com>
During the tuning, drivers repeat data transfer, changing timing
parameters in the controller hardware. So, the tuning commands
(CMD19 for SD, CMD21 for eMMC) fail, and this is not a problem
at all.
Showing "Error detected..." in normal operation just make users
upset. This should not be shown.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This line overwrites host_cap that has been set by drivers and/or
helpers like mmc_of_parse(). Accumulate capabilities flags.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
data is defined as struct mmc_data *data.
So it should not be compared to 0.
Problem identified with Coccinelle.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Fixes emmc initialization regression on the db410c platform.
Clearing this register while SDHCI_PRESENT_STATE reports
SDHCI_CMD_INHIBIT leads to undefined behaviour on the db410c.
When commit 7dde50 was merged (mmc: sdhci: Wait for SDHCI_INT_DATA_END
when transferring), SDHCI transfers transitioned to wait for bit
SDHCI_INT_DATA_END before flagging transfers done.
Without this patch, the db410 platform fails to initialize its eMMC
due to all of its transfers timing out (SDHCI_INT_DATA_END is never
raised after all the blocks have been transferred).
Signed-off-by: Jorge Ramirez-Ortiz <jorge.ramirez-ortiz@linaro.org>
