oweals/u-boot.git
6 years agoarm: imx: hab: Optimise flow of authenticate_image on hab_entry fail
Bryan O'Donoghue [Fri, 12 Jan 2018 12:39:58 +0000 (12:39 +0000)]
arm: imx: hab: Optimise flow of authenticate_image on hab_entry fail

The current code disjoins an entire block of code on hab_entry pass/fail
resulting in a large chunk of authenticate_image being offset to the right.

Fix this by checking hab_entry() pass/failure and exiting the function
directly if in an error state.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Sven Ebenfeld <sven.ebenfeld@gmail.com>
Cc: George McCollister <george.mccollister@gmail.com>
Cc: Breno Matheus Lima <brenomatheus@gmail.com>
Tested-by: Breno Lima <breno.lima@nxp.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoarm: imx: hab: Optimise flow of authenticate_image on is_enabled fail
Bryan O'Donoghue [Fri, 12 Jan 2018 12:39:57 +0000 (12:39 +0000)]
arm: imx: hab: Optimise flow of authenticate_image on is_enabled fail

There is no need to call is_enabled() twice in authenticate_image - it does
nothing but add an additional layer of indentation.

We can check for is_enabled() at the start of the function and return the
result code directly.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Sven Ebenfeld <sven.ebenfeld@gmail.com>
Cc: George McCollister <george.mccollister@gmail.com>
Cc: Breno Matheus Lima <brenomatheus@gmail.com>
Tested-by: Breno Lima <breno.lima@nxp.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoarm: imx: hab: Fix authenticate_image result code
Bryan O'Donoghue [Fri, 12 Jan 2018 12:39:56 +0000 (12:39 +0000)]
arm: imx: hab: Fix authenticate_image result code

authenticate_image returns 1 for success and 0 for failure. That result
code is mapped directly to the result code for the command line function
hab_auth_img - which means when hab_auth_img succeeds it is returning
CMD_RET_FAILURE (1) instead of CMD_RET_SUCCESS (0).

This patch fixes this behaviour by making authenticate_image() return 0 for
success and 1 for failure. Both users of authenticate_image() as a result
have some minimal churn. The upshot is once done when hab_auth_img is
called from the command line we set $? in the standard way for scripting
functions to act on.

Fixes: 36c1ca4d46ef ("imx: Support i.MX6 High Assurance Boot
authentication")

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Sven Ebenfeld <sven.ebenfeld@gmail.com>
Cc: George McCollister <george.mccollister@gmail.com>
Cc: Breno Matheus Lima <brenomatheus@gmail.com>
Tested-by: Breno Lima <breno.lima@nxp.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoarm: imx: hab: Make authenticate_image return int
Bryan O'Donoghue [Fri, 12 Jan 2018 12:39:55 +0000 (12:39 +0000)]
arm: imx: hab: Make authenticate_image return int

Both usages of authenticate_image treat the result code as a simple binary.
The command line usage of authenticate_image directly returns the result
code of authenticate_image as a success/failure code.

Right now when calling hab_auth_img and test the result code in a shell a
passing hab_auth_img will appear to the shell as a fail.

The first step in fixing this behaviour is to fix-up the result code return
by authenticate_image() itself, subsequent patches fix the interpretation
of authenticate_image so that zero will return CMD_RET_SUCCESS and non-zero
will return CMD_RET_FAILURE.

The first step is fixing the return type in authenticate_image() so do that
now.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Cc: Peng Fan <peng.fan@nxp.com>
Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Sven Ebenfeld <sven.ebenfeld@gmail.com>
Cc: George McCollister <george.mccollister@gmail.com>
Cc: Breno Matheus Lima <brenomatheus@gmail.com>
Tested-by: Breno Lima <breno.lima@nxp.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoMerge git://git.denx.de/u-boot-tegra
Tom Rini [Fri, 12 Jan 2018 19:18:34 +0000 (14:18 -0500)]
Merge git://git.denx.de/u-boot-tegra

6 years agoMakefile: ensure DTB doesn't overflow into initial stack
Stephen Warren [Tue, 9 Jan 2018 19:52:14 +0000 (12:52 -0700)]
Makefile: ensure DTB doesn't overflow into initial stack

With CONFIG_SYS_INIT_SP_BSS_OFFSET enabled, the initial (pre-relocation)
stack is placed some distance after bss_start. The control DTB is appended
to the U-Boot binary at bss_start. If the DTB is too large, or the SP BSS
offset too small, then the initial stack could corrupt the DTB. Enhance
the Makefile to check whether this is likely to occur.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: Tegra: p2771-0000: use calculate env var feature
Stephen Warren [Tue, 9 Jan 2018 00:41:25 +0000 (17:41 -0700)]
ARM: Tegra: p2771-0000: use calculate env var feature

Request that all environment variables containing hard-coded address be
calculated at boot time instead.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: Tegra186: calculate load addresses at boot
Stephen Warren [Tue, 9 Jan 2018 00:41:24 +0000 (17:41 -0700)]
ARM: Tegra186: calculate load addresses at boot

In the presence of potentially fragemented memory, we cannot hard-code
addresses into environment variables such as kernel_addr_r. Instead, we
must calculate those addresses at run-time based on available memory
locations. Implement the code to perform such runtime calculation, based
on requirements described in environment variables, to allow the user
full control over the allocation.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: Tegra186: don't map memory not in RAM banks
Stephen Warren [Thu, 4 Jan 2018 18:07:14 +0000 (11:07 -0700)]
ARM: Tegra186: don't map memory not in RAM banks

Tegra186 currently restricts its DRAM usage to entries in the /memory node
in the DTB passed to it. However, the MMU configuration always maps the
entire first 2GB of RAM. This could allow the CPU to speculatively access
RAM that isn't part of the in-use banks. This patch switches to runtime
construction of the table that's used to construct the MMU translation
tables, and thus prevents access to RAM that's not part of a valid bank.

Note: This patch is intended to prevent access to RAM regions which U-Boot
does not need to access, with the primary purpose of avoiding theoretical
speculative access to physical regions for which the HW will throw errors
(e.g. carve-outs that the CPU has no permission to access at a bus level,
bad ECC pages, etc.). In particular, this patch is not deliberately
related to the speculation-related security issues that were recently
announced. The apparent similarity is a coincidence.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: tegra: p2771-000: increase max DRAM bank count
Stephen Warren [Wed, 3 Jan 2018 21:32:35 +0000 (14:32 -0700)]
ARM: tegra: p2771-000: increase max DRAM bank count

On this platform, there may be up to 1024 unusable chunks of memory.
Increase CONFIG_NR_DRAM_BANKS so that U-Boot can remember all the banks
required to represent such fragmented memory.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: Tegra186: search for best RAM bank
Stephen Warren [Wed, 3 Jan 2018 21:32:34 +0000 (14:32 -0700)]
ARM: Tegra186: search for best RAM bank

In the future, the list of DRAM regions passed to U-Boot in the DTB may
be quite long and fragmented. Due to this, U-Boot must search through the
regions to find the best region to relocate into, rather than relying on
the current assumption that the top of bank 0 is a reasonable relocation
target. This change implements such searching.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: bootm: don't assume sp is in DRAM bank 0
Stephen Warren [Fri, 5 Jan 2018 20:04:54 +0000 (13:04 -0700)]
ARM: bootm: don't assume sp is in DRAM bank 0

arch_lmb_reserve() currently assumes that the stack pointer is within DRAM
bank 0. This is not necessarily true. Enhance the code to search through
DRAM banks until the bank that does contain SP is found, and then reserve
the tail of that bank.

Fixes: 2d1916e48bd8 ("ARM: add flat device tree support")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: Tegra186: mem parsing fixes from downstream
Stephen Warren [Wed, 3 Jan 2018 21:32:33 +0000 (14:32 -0700)]
ARM: Tegra186: mem parsing fixes from downstream

Apply a few small fixes for the DTB /memory node parsing from NVIDIA's
downstream U-Boot:

- Allow arbitrary number of DRAM banks.
- Correctly calculate the number of DRAM banks.
- Clip PCIe memory in the same way as U-Boot CPU memory use.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: tegra: use LINUX_KERNEL_IMAGE_HEADER
Stephen Warren [Wed, 3 Jan 2018 21:31:52 +0000 (14:31 -0700)]
ARM: tegra: use LINUX_KERNEL_IMAGE_HEADER

Enable CONFIG_LINUX_KERNEL_IMAGE_HEADER for all 64-bit Tegra boards.
cboot (the boot SW that runs before U-Boot) will eventually use this
information.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARMv8: add optional Linux kernel image header
Stephen Warren [Wed, 3 Jan 2018 21:31:51 +0000 (14:31 -0700)]
ARMv8: add optional Linux kernel image header

Allow placing a Linux kernel image header at the start of the U-Boot
binary. This is useful since the image header reports the amount of memory
(BSS and similar) that U-Boot needs to use, but that isn't part of the
binary size. This can be used by the code that loads U-Boot into memory to
determine where to load U-Boot, based on other users of memory.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: tegra: use CONFIG_SYS_INIT_SP_BSS_OFFSET
Stephen Warren [Wed, 20 Dec 2017 01:30:37 +0000 (18:30 -0700)]
ARM: tegra: use CONFIG_SYS_INIT_SP_BSS_OFFSET

Enable CONFIG_SYS_INIT_SP_BSS_OFFSET for all 64-bit Tegra boards. Place
the stack/... 512KiB from the end of the U-Boot binary. This should be
plenty to accommodate the current DTBs (max 64 KiB), early malloc region
(6KiB), stack usage, and plenty of slack, while still not placing it too
far away from the U-Boot binary.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARMv8: Allow dynamic early stack pointer
Stephen Warren [Wed, 20 Dec 2017 01:30:36 +0000 (18:30 -0700)]
ARMv8: Allow dynamic early stack pointer

U-Boot typically uses a hard-coded value for the stack pointer before
relocation. Implement option SYS_INIT_SP_BSS_OFFSET to instead calculate
the initial SP at run-time. This is useful to avoid hard-coding addresses
into U-Boot, so that can be loaded and executed at arbitrary addresses and
thus avoid using arbitrary addresses at runtime. This option's value is
the offset added to &_bss_start in order to calculate the stack pointer.
This offset should be large enough so that the early malloc region, global
data (gd), and early stack usage do not overlap any appended DTB.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: tegra: remove SPL config for non-SPL SoCs
Stephen Warren [Wed, 20 Dec 2017 01:30:35 +0000 (18:30 -0700)]
ARM: tegra: remove SPL config for non-SPL SoCs

No 64-bit Tegra uses SPL. Remove various unused definitions from config
headers.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agoARM: tegra: don't use CONFIG_SPL_TEXT_BASE when no SPL
Stephen Warren [Wed, 20 Dec 2017 01:30:34 +0000 (18:30 -0700)]
ARM: tegra: don't use CONFIG_SPL_TEXT_BASE when no SPL

64-bit Tegra don't use SPL, and soon won't define CONFIG_SPL_TEXT_BASE
when building. Fix the binman .dts file so that it doesn't use undefined
values.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
6 years agommc: fsl_esdhc: Fix eMMC 1.8v setting issue
Peng Fan [Tue, 2 Jan 2018 08:51:22 +0000 (16:51 +0800)]
mmc: fsl_esdhc: Fix eMMC 1.8v setting issue

Current USDHC driver will reset VSELECT to 0 (3.3v) during mmc init,
then set to 1 for 1.8v eMMC I/O. When booting from eMMC, since ROM has
already set VSELECT to 1.8v before running the u-boot. This reset in
USDHC driver causes a short 2.2v pulse on CMD pin.

Fix this issue by not reset VSELECT to 0 when 1.8v flag is set.

Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
6 years agopower: Rearrange code to guard power command with CONFIG_SPL_BUILD guard
Tom Rini [Fri, 22 Dec 2017 17:27:14 +0000 (12:27 -0500)]
power: Rearrange code to guard power command with CONFIG_SPL_BUILD guard

In order to discard this code when unused in SPL we need to guard the
command with a check for CONFIG_SPL_BUILD and we rearrange the code
slightly to make this cleaner.

Cc: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
6 years agoARM: imx: cm_fx6: env: try to determine dtb to use
Christopher Spinrath [Tue, 9 Jan 2018 21:01:36 +0000 (22:01 +0100)]
ARM: imx: cm_fx6: env: try to determine dtb to use

Some distributions like Fedora expect U-Boot to select a proper
devicetree. Since there are several variants of the cm-fx6 module
featuring different SoC variants and the module can be paired with
several baseboards, it is not viable to hardcode a filename.

Instead, follow the lead of other i.MX6 based devices and try to
determine the devicetree to use with the help of the board name
and the SoC variant exported by the board code, before calling the
distro bootcommand.

For now, only for the Utilite Pro a proper devicetree filename is
known but further variants of the Utilite Computer or other devices
based on the cm-fx6 module may be added in the future.

Signed-off-by: Christopher Spinrath <christopher.spinrath@rwth-aachen.de>
6 years agoARM: imx: cm_fx6: export board and soc info to env
Christopher Spinrath [Tue, 9 Jan 2018 21:01:35 +0000 (22:01 +0100)]
ARM: imx: cm_fx6: export board and soc info to env

Like many other i.MX6 based boards, there are multiple variants of
the cm-fx6 module featuring different SoC variants. Furthermore, the
module can be paired with multiple baseboards.

At the same time modern distribution like Fedora require U-Boot to
select a proper devicetree which depends on the SoC variant and the
baseboard.

Thus, export the SoC variant and the actual board to the environment
following the conventions of other i.MX6 devices (e.g. the NXP boards)
such that the environment can select a devicetree file to load.

For now, we only know for sure that the cm-fx6 module and the SB-fx6m
baseboard amount to a Utilite Computer variant (depending on the SoC).
Further combinations may be added in the future; e.g. CompuLab's
evaluation board once someone can verify the identification string
stored in its eeprom.

Signed-off-by: Christopher Spinrath <christopher.spinrath@rwth-aachen.de>
Reviewed-by: Stefano Babic <sbabic@denx.de>
6 years agodrivers: pci: imx: fix enumeration logic error
Koen Vandeputte [Thu, 4 Jan 2018 13:54:34 +0000 (14:54 +0100)]
drivers: pci: imx: fix enumeration logic error

By default, the subordinate is set equally to the secondary bus (1) when
the RC boots, and does not alter afterwards.

This means that theoretically, the highest bus reachable downstream is
bus 1.

Force the PCIe RC subordinate to 0xff, otherwise no downstream
devices will be detected behind bus 1 if the booting OS does not allow
enumerating a higher busnr than the subordinate value of the primary
bus.

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
6 years agomx6memcal: spl: Disambiguate the error message
Fabio Estevam [Wed, 3 Jan 2018 14:55:36 +0000 (12:55 -0200)]
mx6memcal: spl: Disambiguate the error message

Currently mmdc_do_dqs_calibration() and mmdc_do_write_level_calibration()
show the same error message, which is confusing for debugging.

Disambiguate the mmdc_do_dqs_calibration() error message.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agomx6memcal: spl: Also take i.MX6ULL into account
Fabio Estevam [Wed, 3 Jan 2018 14:55:35 +0000 (12:55 -0200)]
mx6memcal: spl: Also take i.MX6ULL into account

i.MX6ULL also does not support 64-bit DDR bus, so add it to the
check logic.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agomx6memcal: Fix the UART ports for mx6sabresd/auto boards
Fabio Estevam [Wed, 3 Jan 2018 14:55:34 +0000 (12:55 -0200)]
mx6memcal: Fix the UART ports for mx6sabresd/auto boards

mx6sabresd board uses the following pins for console:

PAD_CSI0_DAT10__UART1_TX_DATA
PAD_CSI0_DAT11__UART1_RX_DATA

,so put it in the same config option as wandboard.

mx6sabreauto board uses the following pins for console:

PAD_KEY_COL0__UART4_TX_DATA
PAD_KEY_ROW0__UART4_RX_DATA

So do not mention sabreauto board as part of the UART1_SD3_DAT6_7 option.

The config option for sabreauto can be added later when needed.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agotoradex: imx6: Rework PF0100 fuse programming commands to not be in SPL
Tom Rini [Wed, 3 Jan 2018 14:19:17 +0000 (09:19 -0500)]
toradex: imx6: Rework PF0100 fuse programming commands to not be in SPL

The code for programming the OTP fuses on the PMIC PF0100 can only be
used in full U-Boot, so do not build / link it into SPL.

Cc: Max Krummenacher <max.krummenacher@toradex.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
Tested-by: Max Krummenacher <max.krummenacher@toradex.com>
6 years agoimx: ventana: Rework CONFIG_CMD_GSC code to not be included in SPL
Tom Rini [Wed, 3 Jan 2018 14:16:29 +0000 (09:16 -0500)]
imx: ventana: Rework CONFIG_CMD_GSC code to not be included in SPL

The command can only be used from full U-Boot, so do not build it into
SPL.

Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Cc: Tim Harvey <tharvey@gateworks.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Acked-by: Tim Harvey <tharvey@gateworks.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoimx: ventana: Rework CONFIG_CMD_EECONFIG code to not be included in SPL
Tom Rini [Wed, 3 Jan 2018 14:15:22 +0000 (09:15 -0500)]
imx: ventana: Rework CONFIG_CMD_EECONFIG code to not be included in SPL

The command can only be used from full U-Boot, so do not build it into
SPL.

Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Cc: Tim Harvey <tharvey@gateworks.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Acked-by: Tim Harvey <tharvey@gateworks.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoarm: imx: Rework i.MX specific commands to be excluded from SPL
Tom Rini [Wed, 3 Jan 2018 13:52:39 +0000 (08:52 -0500)]
arm: imx: Rework i.MX specific commands to be excluded from SPL

The "clocks" and "bootaux" commands are only usable in full U-Boot, not
SPL, so do not link them inside of SPL.  Rework a little of the bootaux
related code to make use of __weak and declare parts of it static as
it's local to the file.

Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoimx7: spl: Add support for MMC3, SD3 and NAND boot devices
Eran Matityahu [Wed, 3 Jan 2018 10:56:24 +0000 (12:56 +0200)]
imx7: spl: Add support for MMC3, SD3 and NAND boot devices

Signed-off-by: Eran Matityahu <eran.m@variscite.com>
6 years agoimx7: spl: Use SPL boot device MMC1 for all of the SOCs MMC/SD boot devices
Eran Matityahu [Wed, 3 Jan 2018 10:53:08 +0000 (12:53 +0200)]
imx7: spl: Use SPL boot device MMC1 for all of the SOCs MMC/SD boot devices

Use only one SPL MMC device, similarly to the iMX6 code

Signed-off-by: Eran Matityahu <eran.m@variscite.com>
6 years agoimx: mx6ull-14x14-evk: enable DM QSPI driver
Peng Fan [Wed, 3 Jan 2018 00:52:03 +0000 (08:52 +0800)]
imx: mx6ull-14x14-evk: enable DM QSPI driver

To support QSPI DM driver
 - Add spi0 alias for qspi node. Which is used for bus number 0.
 - Modify the n25q256a@0 compatible property to "spi-flash".
 - Modify spi4 (gpio_spi) node to spi5
 - Define DM SPI/QSPI related config to enable QSPI

Signed-off-by: Peng Fan <peng.fan@nxp.com>
6 years agospi: fsl_qspi: support i.MX6UL/6ULLL/7D
Peng Fan [Wed, 3 Jan 2018 00:52:02 +0000 (08:52 +0800)]
spi: fsl_qspi: support i.MX6UL/6ULLL/7D

The QSPI module on i.MX7D is modified from i.MX6SX. The module used on
i.MX6UL/6ULL is reused from i.MX7D. They share same tx buffer size.

The endianness is not set at qspi driver initialization. So if we don't
boot from QSPI, we will get wrong endianness when accessing from AHB
address directly.

Add the compatible entry for 6ul/7d.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
6 years agomx6: ddr: Do not access MMDC_P1_BASE_ADDR on i.MX6ULL
Fabio Estevam [Tue, 2 Jan 2018 00:51:45 +0000 (22:51 -0200)]
mx6: ddr: Do not access MMDC_P1_BASE_ADDR on i.MX6ULL

i.MX6ULL also does not have a MMDC_P1_BASE_ADDR, so do not try to
access it.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Stefano Babic <ssbabic@denx.de>
6 years agopci: imx: request gpio before use
Peng Fan [Tue, 2 Jan 2018 10:27:29 +0000 (18:27 +0800)]
pci: imx: request gpio before use

Before use GPIO, we need to request gpio first. Free gpio after use.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
Reviewed-by: Stefano Babic <ssbabic@denx.de>
6 years agoimx: mx6sxsabresd: enlarge ENV offset
Peng Fan [Tue, 2 Jan 2018 01:32:10 +0000 (09:32 +0800)]
imx: mx6sxsabresd: enlarge ENV offset

The u-boot-dtb.imx size is about 519KB, so 8 * 64KB conflicts
with u-boot-dtb.imx. Enlarge the offset to 14 * 64KB to fix it.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoimx: mx6sxsabresd: config wdog pinmux
Peng Fan [Tue, 2 Jan 2018 01:32:09 +0000 (09:32 +0800)]
imx: mx6sxsabresd: config wdog pinmux

Because kernel set WDOG_B mux before pad with the common pinctrl
framwork now and wdog reset will be triggered once set WDOG_B mux
with default pad setting, we set pad setting here to workaround this.
Since imx_iomux_v3_setup_pad also set mux before pad setting, we set
as GPIO mux firstly here to workaround it.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoimx: mx6sxsabresd: Enable DM driver
Peng Fan [Tue, 2 Jan 2018 01:32:08 +0000 (09:32 +0800)]
imx: mx6sxsabresd: Enable DM driver

Enable I2C/MMC/GPIO/REGUALTOR/PMIC/USB DM drivers.
There are some dependency, such as when DM MMC enabled, USB compile error.
Also the i.MX I2C MMC DM driver does not support legacy GPIO interface.
So enable them all together.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoboard: freescale: common: add pfuze dm code
Peng Fan [Tue, 2 Jan 2018 01:32:06 +0000 (09:32 +0800)]
board: freescale: common: add pfuze dm code

Add pfuze dm code, this code could be enabled with CONFIG_DM_PMIC_PFUZE100.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoARM: imx: Enable dts for i.MX6SX-SDB
Peng Fan [Tue, 2 Jan 2018 01:32:05 +0000 (09:32 +0800)]
ARM: imx: Enable dts for i.MX6SX-SDB

Enable DTS and OF_CONTROL for i.MX6SX-SDB.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoARM: imx: Introduce dts for i.MX6SX-SDB
Peng Fan [Tue, 2 Jan 2018 01:32:04 +0000 (09:32 +0800)]
ARM: imx: Introduce dts for i.MX6SX-SDB

Introduce dts from Kernel commit
commit 71ee203389f7cb1c("Merge tag 'scsi-fixes' of
git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi")

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <fabio.estevam@nxp.com>
6 years agomisc: mxc_ocotp: check fuse word before programming on i.MX7ULP
Peng Fan [Tue, 2 Jan 2018 07:51:20 +0000 (15:51 +0800)]
misc: mxc_ocotp: check fuse word before programming on i.MX7ULP

On i.MX7ULP, the fuse words (except bank 0 and 1) only supports to
write once, because they use ECC mode. Multiple writes may damage
the ECC value and cause a wrong fuse value decoded when reading.
This patch adds a checking before the fuse word programming, only
can write when the word value is 0.

Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
6 years agomx6: Support SKS-Kinkel sksimx6 Board
Stefano Babic [Wed, 3 Jan 2018 15:11:56 +0000 (16:11 +0100)]
mx6: Support SKS-Kinkel sksimx6 Board

Board  has 1GB RAM and boots from SD Card

U-Boot SPL 2018.01-rc3-00005-ga1898b8 (Jan 02 2018 - 13:48:54)
BT_FUSE_SEL already fused, will do nothing
Trying to boot from MMC1

U-Boot 2018.01-rc3-00005-ga1898b8 (Jan 02 2018 - 13:48:54 +0100)

CPU:   Freescale i.MX6DL rev1.2 996 MHz (running at 792 MHz)
CPU:   Commercial temperature grade (0C to 95C) at 40C
Reset cause: POR
I2C:   ready
DRAM:  1 GiB
MMC:   FSL_SDHC: 0
In:    serial
Out:   serial
Err:   serial
Net:   FEC [PRIME]

Signed-off-by: Stefano Babic <sbabic@denx.de>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agotravis.yml: Support RISC-V
Rick Chen [Fri, 12 Jan 2018 06:57:09 +0000 (14:57 +0800)]
travis.yml: Support RISC-V

Enable travis-ci support with a link having built.

Signed-off-by: Chih-Mao Chen <cmchen@andestech.com>
Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: doc: Add relative doc to describe RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:59 +0000 (13:55 +0800)]
riscv: doc: Add relative doc to describe RISC-V

Add documents to describe NX25 and AE250.
Also update other documents for RISC-V.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: Modify generic codes to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:58 +0000 (13:55 +0800)]
riscv: Modify generic codes to support RISC-V

Support common commands bdinfo and image format,
also modify common generic flow for RISC-V.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
6 years agoriscv: Support standalone
Rick Chen [Tue, 26 Dec 2017 05:55:57 +0000 (13:55 +0800)]
riscv: Support standalone

Run hello_world successfully.

U-Boot 2018.01-rc2-00033-gb265b91-dirty (Dec 22 2017 - 13:54:21 +0800)

DRAM:  1 GiB
MMC:   mmc@f0e00000: 0
SF: Detected mx25u1635e with page size 256 Bytes, erase size 4 KiB, total 2 MiB
In:    serial@f0300000
Out:   serial@f0300000
Err:   serial@f0300000
Net:
Warning: mac@e0100000 (eth0) using random MAC address - 0a:47:9b:f8:b4:f2
eth0: mac@e0100000
RISC-V # mmc rescan
RISC-V # fatls mmc 0:1
318907   u-boot-ae250-64.bin
1252   hello_world_ae250_32.bin
328787   u-boot-ae250-32.bin

3 file(s), 0 dir(s)

RISC-V # fatload mmc 0:1 0x600000 hello_world_ae250_32.bin
reading hello_world_ae250_32.bin
1252 bytes read in 23 ms (52.7 KiB/s)
RISC-V # go 0x600000
Example expects ABI version 9
Actual U-Boot ABI version 9
Hello World
argc = 1
argv[0] = "0x600000"
argv[1] = "$B@"
Hit any key to exit ...

RISC-V #

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: tools: Prelink u-boot
Rick Chen [Tue, 26 Dec 2017 05:55:56 +0000 (13:55 +0800)]
riscv: tools: Prelink u-boot

Add prelink-riscv to arrange .rela.dyn and .rela.got
in compile time. So that u-boot can be directly
executed without fixup.

Signed-off-by: Chih-Mao Chen <cmchen@andestech.com>
Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: defconfig: Add nx25-ae250 defconfig to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:55 +0000 (13:55 +0800)]
riscv: defconfig: Add nx25-ae250 defconfig to support RISC-V

Add nx25-ae250 default configuration for RISC-V

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: configs: Add nx25-ae250.h to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:54 +0000 (13:55 +0800)]
riscv: configs: Add nx25-ae250.h to support RISC-V

Add nx25-ae250 board configuartion options for RISC-V

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: board: Add nx25-ae250 to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:53 +0000 (13:55 +0800)]
riscv: board: Add nx25-ae250 to support RISC-V

Add nx25-ae250 board to do platform initializations.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: Add Kconfig to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:52 +0000 (13:55 +0800)]
riscv: Add Kconfig to support RISC-V

Add Kconfig and makefile for RISC-V
Also modify MAINTAINERS for it.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
Cc: Padmarao Begari <Padmarao.Begari@microsemi.com>
6 years agoriscv: nx25: include: Add header files to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:51 +0000 (13:55 +0800)]
riscv: nx25: include: Add header files to support RISC-V

Add header files for RISC-V.
Cache, ptregs, data type and other definitions are included.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: nx25: dts: Add AE250 dts to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:50 +0000 (13:55 +0800)]
riscv: nx25: dts: Add AE250 dts to support RISC-V

AE250 is the Soc using NX25 cpu core base on RISC-V arch.
Details please see the doc/README.ae250.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
6 years agoriscv: nx25: lib: Add relative lib funcs to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:49 +0000 (13:55 +0800)]
riscv: nx25: lib: Add relative lib funcs to support RISC-V

Add makefile, interrupts.c and boot.c,... functions
to support RISC-V arch.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
Cc: Padmarao Begari <Padmarao.Begari@microsemi.com>
6 years agoriscv: cpu: Add nx25 to support RISC-V
Rick Chen [Tue, 26 Dec 2017 05:55:48 +0000 (13:55 +0800)]
riscv: cpu: Add nx25 to support RISC-V

Add Andes nx25 cpu core (called AndesStar V5) to support RISC-V arch

Verifications:
1. startup and relocation ok.
2. boot from rom or ram both ok.
2. timer driver ok.
3. uart driver ok
4. mmc driver ok
5. spi driver ok.
6. 32/64 bit both ok.

Detail verification message please see doc/README.ae250.

Signed-off-by: Rick Chen <rick@andestech.com>
Signed-off-by: Rick Chen <rickchen36@gmail.com>
Signed-off-by: Greentime Hu <green.hu@gmail.com>
Cc: Padmarao Begari <Padmarao.Begari@microsemi.com>
6 years agommc: remove hc_wp_grp_size from struct mmc if not needed
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:36 +0000 (15:23 +0100)]
mmc: remove hc_wp_grp_size from struct mmc if not needed

hc_wp_grp_size is needed only if hardware partitionning is used.
On ARM removing it saves about 30 bytes of code space.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: don't read the size of eMMC enhanced user data area in SPL
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:35 +0000 (15:23 +0100)]
mmc: don't read the size of eMMC enhanced user data area in SPL

This information is only used by the "mmc info" command.
On ARM removing this information from SPL saves about 140 of code space.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: compile out erase and write mmc commands if write operations are not enabled
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:34 +0000 (15:23 +0100)]
mmc: compile out erase and write mmc commands if write operations are not enabled

Also remove erase_grp_size and write_bl_len from struct mmc as they are
not used anymore. On ARM, removing them saves about 100 bytes of code
space in SPL.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: read ssr only if MMC write support is enabled
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:33 +0000 (15:23 +0100)]
mmc: read ssr only if MMC write support is enabled

The content of ssr is useful only for erase operations.
on ARM, removing sd_read_ssr() saves around 300 bytes.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: add a Kconfig option to enable the support for MMC write operations
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:32 +0000 (15:23 +0100)]
mmc: add a Kconfig option to enable the support for MMC write operations

This allows using CONFIG_IS_ENABLED(MMC_WRITE) to compile out code
needed only if write support is required.
The option is added for u-boot and for SPL

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: reworked version lookup in mmc_startup_v4
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:31 +0000 (15:23 +0100)]
mmc: reworked version lookup in mmc_startup_v4

Using a table versus a switch() structure saves a bit of space

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: compile out more code if support for UHS and HS200 is not enabled
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:30 +0000 (15:23 +0100)]
mmc: compile out more code if support for UHS and HS200 is not enabled

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: atmel: when sending a data command, use the provided block size
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:29 +0000 (15:23 +0100)]
mmc: atmel: when sending a data command, use the provided block size

struct mmc_data contains the block size to use for the data transfer.
Use this information instead of using the default value or the block length
information stored in struct mmc.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agocommon: do not compile common fastboot code when building the SPL
Jean-Jacques Hiblot [Thu, 4 Jan 2018 14:23:28 +0000 (15:23 +0100)]
common: do not compile common fastboot code when building the SPL

This is not required as fastboot can't be started from SPL.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agoam335x_hs_evm: Trim options in SPL to reduce binary size
Tom Rini [Sun, 17 Dec 2017 04:01:22 +0000 (23:01 -0500)]
am335x_hs_evm: Trim options in SPL to reduce binary size

The am335x_hs_evm runs into size constraint problems at times with
various toolchains as changes come in due to the config have a large
number of options in SPL (to showcase what is possible) while also
having rather constrained binary limits.  Gain some of this room back by
lowering the loglevel, disabling HW partition support and switching over
to the tiny FIT image support.

Cc: Andrew F. Davis <afd@ti.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
---
I'd really appreciate a run-time test of this patch if at all possible
as I'm a little worried about TINY_FIT being incompatible with all of
the security options.  Thanks!

6 years agodm: mmc: sandbox: Update SD card emulation
Jean-Jacques Hiblot [Thu, 14 Dec 2017 10:47:14 +0000 (11:47 +0100)]
dm: mmc: sandbox: Update SD card emulation

The SDcard initialization procedure does a few more things than it did earlier:
* switch the bus width even for 1-bit bus width
* check that speed has been properly set (in resp[4] of SD_CMD_SWITCH_FUNC)

Update the SD simulator to handle those requests gracefully.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
6 years agoconfigs: omapl138_lcdk: decrease the loglevel to reduce the size of the SPL
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:44:04 +0000 (17:44 +0100)]
configs: omapl138_lcdk: decrease the loglevel to reduce the size of the SPL

The changes in the MMC stack have increased its footprint up to the point
were its breaks the generation of the SPL for this platform.
Fix this by reducing the loglevel.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Peter Howard <phoward@gme.net.au>
6 years agoconfigs: openrd: removed support for eMMC hardware partitioning
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:44:03 +0000 (17:44 +0100)]
configs: openrd: removed support for eMMC hardware partitioning

builds are broken because the size of the binary exceeds the limit.
Make some space by removing support for hardware partitioning as those
boards don't have any eMMC.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: make optional the support for eMMC hardware partitioning
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:44:02 +0000 (17:44 +0100)]
mmc: make optional the support for eMMC hardware partitioning

Not all boards have an eMMC and not all users have a need for this.
Allow to compile it out. By default it is still included.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: make UHS and HS200 optional
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:44:01 +0000 (17:44 +0100)]
mmc: make UHS and HS200 optional

Supporting USH and HS200 increases the code size as it brings in IO voltage
control, tuning and fatter data structures.
Use Kconfig configuration to select which of those features should be
built in.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: convert most of printf() to pr_err() and pr_warn()
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:44:00 +0000 (17:44 +0100)]
mmc: convert most of printf() to pr_err() and pr_warn()

This allows to compile out the log message by tweaking the LOGLEVEL.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
6 years agommc: don't use malloc_cache_aligned()
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:43:59 +0000 (17:43 +0100)]
mmc: don't use malloc_cache_aligned()

Not using this function reduces the size of the binary. It's replaces by
a standard malloc() and the alignment requirement is handled by an
intermediate buffer on the stack.

Also make sure that the allocated buffer is freed in case of error.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: fix for old MMCs (below version 4)
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:43:58 +0000 (17:43 +0100)]
mmc: fix for old MMCs (below version 4)

The ext_csd is allocated only for MMC above version 4. The compare will
crash or fail for older MMCs.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: all hosts support 1-bit bus width and legacy timings
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:43:57 +0000 (17:43 +0100)]
mmc: all hosts support 1-bit bus width and legacy timings

Make sure that those basic capabilities are advertised by the host.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: Fixed a problem with old sd or mmc that do not support High speed
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:43:56 +0000 (17:43 +0100)]
mmc: Fixed a problem with old sd or mmc that do not support High speed

As the legacy modes were not added to the list of supported modes, old
cards that do not support other modes could not be used.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agodm: mmc: update mmc_of_parse()
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:43:55 +0000 (17:43 +0100)]
dm: mmc: update mmc_of_parse()

* convert to livetree API
* don't fail because of an invalid bus-width, instead default to 1-bit.
* recognize 1.2v DDR and 1.2v HS200 flags

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: dump card and host capabilities if debug is enabled
Jean-Jacques Hiblot [Thu, 30 Nov 2017 16:43:54 +0000 (17:43 +0100)]
mmc: dump card and host capabilities if debug is enabled

This is a useful information while debugging the initialization process or
performance issues.
Also dump this information with the other mmc info if the verbose option
is selected

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: meson_gx_mmc: fix the complie error
Jaehoon Chung [Mon, 27 Nov 2017 09:42:05 +0000 (18:42 +0900)]
mmc: meson_gx_mmc: fix the complie error

mmc_set_clock() is changed.
This patch is for fixing complie error.

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
6 years agodm: mmc: Add a library function to parse generic dt binding
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:13 +0000 (16:30 +0200)]
dm: mmc: Add a library function to parse generic dt binding

Add a new function to parse host controller dt node and
set mmc_config. This function can be used by mmc controller
drivers to set the generic mmc_config.
This function can be extended to set other UHS mode caps
once UHS mode support is added.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: add a library function to send tuning command
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:30:12 +0000 (16:30 +0200)]
mmc: add a library function to send tuning command

HS200/SDR104 requires tuning command to be sent to the card.
Add a simple function to send tuning command and to read and
compare the received data with the tuning block pattern.
This function can be used by platform driver to perform DLL
tuning.
This patch is similar to
commit 996903de92f0 ("mmc: core: add core-level function for
sending tuning commands") added in linux kernel.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: use the right voltage level for MMC DDR and HS200 modes
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:30:11 +0000 (16:30 +0200)]
mmc: use the right voltage level for MMC DDR and HS200 modes

HS200 only supports 1.2v and 1.8v signal voltages. DDR52 supports 3.3v/1.8v
or 1.2v signal voltages.
Select the lowest voltage available when using those modes.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: Retry some MMC cmds on failure
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:10 +0000 (16:30 +0200)]
mmc: Retry some MMC cmds on failure

With certain SD cards like Kingston 8GB/16GB UHS card, it is seen that
MMC_CMD_ALL_SEND_CID cmd fails on first attempt, but succeeds
subsequently. Therefore, retry MMC_CMD_ALL_SEND_CID cmd a few time
as done in Linux kernel.
Similarly, it is seen that MMC_CMD_SET_BLOCKLEN may fail on first
attempt, therefore retry this cmd a few times as done in kernel.

To make it clear that those are optionnal workarounds, a new Kconfig
option 'MMC_QUIRKS' is added (enabled by default).

Signed-off-by: Vignesh R <vigneshr@ti.com>
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: Change mode when switching to a boot partition
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:30:09 +0000 (16:30 +0200)]
mmc: Change mode when switching to a boot partition

Boot partitions do not support HS200. Changing to a lower performance mode
is required to access them.
mmc_select_mode_and_width() and sd_select_mode_and_width() are modified to
make it easier to call them outside of the initialization context.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: disable UHS modes if Vcc cannot be switched on and off
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:30:08 +0000 (16:30 +0200)]
mmc: disable UHS modes if Vcc cannot be switched on and off

If a power cycle cannot be done on Vcc, it is safer not to try the UHS
modes because we wouldn't be able to recover from an error occurring
during the UHS initialization.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: Add support for UHS modes
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:30:07 +0000 (16:30 +0200)]
mmc: Add support for UHS modes

Add UHS modes to the list of supported modes, get the UHS capabilites of
the SDcard and implement the procedure to switch the voltage (UHS modes
use 1v8 IO lines)
During the voltage switch procedure, DAT0 is used by the card to signal
when it's ready. The optional card_busy() callback can be used to get this
information from the host driver.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: add HS200 support in MMC core
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:06 +0000 (16:30 +0200)]
mmc: add HS200 support in MMC core

Add HS200 to the list of supported modes and introduce tuning in the MMC
startup process.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: Add a execute_tuning() callback to the mmc operations.
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:05 +0000 (16:30 +0200)]
mmc: Add a execute_tuning() callback to the mmc operations.

Tuning is a mandatory step in the initialization of SDR104 and HS200 modes.
This callback execute the tuning process.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: disable the mmc clock during power off
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:04 +0000 (16:30 +0200)]
mmc: disable the mmc clock during power off

There is no point in having the mmc clock enabled during
power off. Disable the mmc clock. This is similar to how it's
programmed in Linux Kernel.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Vignesh R <vigneshr@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: add a new mmc parameter to disable mmc clock
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:03 +0000 (16:30 +0200)]
mmc: add a new mmc parameter to disable mmc clock

mmc clock has to be disabled in certain cases like during
the voltage switch sequence. Modify mmc_set_clock function
to take disable as an argument that signifies if the
clock has to be enabled or disabled.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: add power cyle support in mmc core
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:02 +0000 (16:30 +0200)]
mmc: add power cyle support in mmc core

mmc/sd specification requires vdd to be disabled for 1 ms
and then enabled again during power cycle. Add a
function in mmc core to perform power cycle and set
the io signal to it's initial state.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: Add a new callback function to perform the 74 clocks cycle sequence
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:30:01 +0000 (16:30 +0200)]
mmc: Add a new callback function to perform the 74 clocks cycle sequence

Add a new callback function *send_init_stream* which start a sequence of
at least 74 clock cycles.
The mmc core uses *mmc_send_init_stream* in order to invoke the callback
function. This will be used during power cycle where the specification
requires such a sequence after power up.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: Enable signal voltage to be selected from mmc core
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:30:00 +0000 (16:30 +0200)]
mmc: Enable signal voltage to be selected from mmc core

Add a new function *mmc_set_signal_voltage* in mmc core
which can be used during mmc initialization to select the
signal voltage. Platform driver should use the set_ios
callback function to select the signal voltage.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
6 years agommc: make mmc_set_ios() return status
Kishon Vijay Abraham I [Thu, 21 Sep 2017 14:29:59 +0000 (16:29 +0200)]
mmc: make mmc_set_ios() return status

set_ios callback has a return value of 'int' but the mmc_set_ios()
function ignore this. Modify mmc_set_ios() and the callers of mmc_set_ios() to
to return the error status.

Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: refactor MMC startup to make it easier to support new modes
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:29:58 +0000 (16:29 +0200)]
mmc: refactor MMC startup to make it easier to support new modes

The MMC startup process currently handles 4 modes. To make it easier to
add support for more modes, let's make the process more generic and use a
list of the modes to try.
The major functional change is that when a mode fails we try the next one.
Not all modes are tried, only those supported by the card and the host.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: refactor SD startup to make it easier to support new modes
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:29:57 +0000 (16:29 +0200)]
mmc: refactor SD startup to make it easier to support new modes

The SDcard startup process currently handles only 2 modes. To make it
easier to add support for more modes, let's make the process more generic
and use a list of the modes to try.
The major functional change is that when a mode fails we try the next one.
Not all modes are tried, only those supported by the card and the host.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agocmd: mmc: display the mode name and current bus speed in the mmc info
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:29:56 +0000 (16:29 +0200)]
cmd: mmc: display the mode name and current bus speed in the mmc info

Display the mode name when the user execute 'mmc info'. Also instead of
displaying tran_speed, display the actual bus speed.

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agommc: use mmc modes to select the correct bus speed
Jean-Jacques Hiblot [Thu, 21 Sep 2017 14:29:55 +0000 (16:29 +0200)]
mmc: use mmc modes to select the correct bus speed

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>