project/bcm63xx/u-boot.git
9 years agoDrop references to MAKEALL in the documentation
Simon Glass [Thu, 28 Jul 2016 02:33:08 +0000 (20:33 -0600)]
Drop references to MAKEALL in the documentation

It is confusing to mention MAKEALL when it is not the normal way of building
U-Boot anymore. Update the documentation to suit.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agoDrop the MAKEALL tool
Simon Glass [Thu, 28 Jul 2016 02:33:07 +0000 (20:33 -0600)]
Drop the MAKEALL tool

Buildman has been around for 3 years now. It has had a lot of use and
testing. Perhaps it is time to remove MAKEALL.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agobuildman: Add a quick-start note
Simon Glass [Thu, 28 Jul 2016 02:33:06 +0000 (20:33 -0600)]
buildman: Add a quick-start note

For those who just want to build a board, it is useful to see a quick hint
right at the start of the documentation. Add a few commands showing how to
download toolchains and build a board.

Signed-off-by: Simon Glass <[email protected]>
9 years agobuildman: Avoid overwriting existing toolchain entries
Simon Glass [Thu, 28 Jul 2016 02:33:05 +0000 (20:33 -0600)]
buildman: Avoid overwriting existing toolchain entries

The current code for setting up the toolchain config always writes the new
paths to an item called 'toolchain'. This means that it will overwrite any
existing toolchain item with the same name. In practice, this means that:

   buildman --fetch-arch all

will fetch all toolchains, but only the path of the final one will be added
to the config. This normally works out OK, since most toolchains are the
same version (e.g. gcc 4.9) and will be found on the same path. But it is
not correct and toolchains for archs which don't use the same version will
not function as expected.

Adjust the code to use a complete glob of the toolchain path.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agobuildman: Drop the toolchain error when downloading toolchains
Simon Glass [Thu, 28 Jul 2016 02:33:04 +0000 (20:33 -0600)]
buildman: Drop the toolchain error when downloading toolchains

It doesn't make sense to complain about missing toolchains when the
--fetch-arch option is being used. The user is presumably aware that there
is a toolchain problem and is actively correcting it by running with this
option.

Refactor the code to avoid printing this confusing message.

Signed-off-by: Simon Glass <[email protected]>
9 years agobuildman: Fix a typo in TestSettingsHasPath()
Simon Glass [Thu, 28 Jul 2016 02:33:03 +0000 (20:33 -0600)]
buildman: Fix a typo in TestSettingsHasPath()

The function comment should say 'buildman'. Fix it.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agobuildman: Improve the toolchain progress/error output
Simon Glass [Thu, 28 Jul 2016 02:33:02 +0000 (20:33 -0600)]
buildman: Improve the toolchain progress/error output

Use colour to make it easier to see what is going on. Also print a message
before downloading a new toolchain. Mention --fetch-arch in the message that
is shown when there are no available toolchains, since this is the quickest
way to resolve the problem.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agobuildman: Allow the toolchain error to be suppressed
Simon Glass [Thu, 28 Jul 2016 02:33:01 +0000 (20:33 -0600)]
buildman: Allow the toolchain error to be suppressed

When there are no toolchains a warning is printed. But in some cases this is
confusing, such as when the user is fetching new toolchains.

Adjust the function to supress the warning in this case.

Signed-off-by: Simon Glass <[email protected]>
9 years agobuildman: Fix the 'help' test to use the correct path
Simon Glass [Thu, 28 Jul 2016 02:33:00 +0000 (20:33 -0600)]
buildman: Fix the 'help' test to use the correct path

When buildman is run via a symlink, this test fails. Fix it to work the same
way as buildman itself.

Signed-off-by: Simon Glass <[email protected]>
9 years agobuildman: Automatically create a config file if needed
Simon Glass [Thu, 28 Jul 2016 02:32:59 +0000 (20:32 -0600)]
buildman: Automatically create a config file if needed

If there is no ~/.buildman file, buildman currently complains and exists. To
make things a little more friendly, create an empty one automatically. This
will not allow things to be built, but --fetch-arch can be used to handle
that.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agobuildman: Tidy up the README a little
Simon Glass [Thu, 28 Jul 2016 02:32:58 +0000 (20:32 -0600)]
buildman: Tidy up the README a little

Tidy up some problems found by a recent review.

Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
9 years agolibfdt: Introduce new ARCH_FIXUP_FDT option
Michal Simek [Thu, 28 Jul 2016 07:06:41 +0000 (09:06 +0200)]
libfdt: Introduce new ARCH_FIXUP_FDT option

Add new Kconfig option to disable arch_fixup_fdt() calls for cases where
U-Boot shouldn't update memory setup in DTB file.
One example of usage of this option is to boot OS with different memory
setup than U-Boot use.

Signed-off-by: Michal Simek <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agoMerge git://git.denx.de/u-boot-rockchip
Tom Rini [Mon, 1 Aug 2016 00:31:13 +0000 (20:31 -0400)]
Merge git://git.denx.de/u-boot-rockchip

9 years agosunxi: Re-enable h3 emac support
Hans de Goede [Wed, 27 Jul 2016 16:06:35 +0000 (18:06 +0200)]
sunxi: Re-enable h3 emac support

With the recent bug fixes for the sun8i_emac driver all known issues
are resolved, so we can re-enable the driver.

While at it, also enable the emac on the Orange Pi One.

Cc: Chen-Yu Tsai <[email protected]>
Cc: Corentin LABBE <[email protected]>
Cc: Amit Singh Tomar <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
Acked-by: Ian Campbell <[email protected]>
Acked-by: Jagan Teki <[email protected]>
9 years agonet: sun8i_emac: Fix DMA alignment issues with the rx / tx buffers
Hans de Goede [Wed, 27 Jul 2016 15:31:17 +0000 (17:31 +0200)]
net: sun8i_emac: Fix DMA alignment issues with the rx / tx buffers

This fixes the following CACHE warnings when using sun8i_emac:

=> dhcp
BOOTP broadcast 1
BOOTP broadcast 2
CACHE: Misaligned operation at range [7bf594a87bf59628]
BOOTP broadcast 3
CACHE: Misaligned operation at range [7bf59c907bf59e10]
CACHE: Misaligned operation at range [7bf5a4787bf5a5f8]
DHCP client bound to address 10.42.43.80 (1009 ms)

Note this commit also changes the max rx size from 2024 to 2044,
matching what the kernel driver uses.

Cc: Chen-Yu Tsai <[email protected]>
Cc: Corentin LABBE <[email protected]>
Cc: Amit Singh Tomar <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
Acked-by: Ian Campbell <[email protected]>
9 years agosunxi: On newer SoCs use words 1-3 instead of just word 3 from the SID
Hans de Goede [Fri, 29 Jul 2016 09:47:03 +0000 (11:47 +0200)]
sunxi: On newer SoCs use words 1-3 instead of just word 3 from the SID

It seems that bytes 13-14 of the SID / bytes 1-2 from word 3 of the SID
are always 0 on H3 making it a poor candidate to use as source for the
serialnr / mac-address, and the other non constant words (1 and 2) also
have quite a few bits which are the same for some boards,

This commits switches to using the crc32 of words 1 - 3 to get a
more unique value for the mac-address / serialnr.

Cc: Chen-Yu Tsai <[email protected]>
Cc: Corentin LABBE <[email protected]>
Cc: Amit Singh Tomar <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
Acked-by: Ian Campbell <[email protected]>
9 years agosunxi: Ensure that the NIC specific bytes of the mac are not all 0
Hans de Goede [Wed, 27 Jul 2016 15:58:06 +0000 (17:58 +0200)]
sunxi: Ensure that the NIC specific bytes of the mac are not all 0

On 2 of my H3 boards bytes 13-15 of the SID are all 0 leading to
the NIC specific bytes of the mac all being 0, which leads to the
boards not getting an ipv6 address from the dhcp server.

This commits adds a check to ensure this does not happen.

Cc: Chen-Yu Tsai <[email protected]>
Cc: Corentin LABBE <[email protected]>
Cc: Amit Singh Tomar <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
Acked-by: Ian Campbell <[email protected]>
9 years agosunxi: Hummingbird_A31_defconfig: Drop MACPWR option
Chen-Yu Tsai [Fri, 22 Jul 2016 08:13:00 +0000 (16:13 +0800)]
sunxi: Hummingbird_A31_defconfig: Drop MACPWR option

MACPWR was used to bring the Ethernet PHY out of reset. The designware
driver now supports the phy reset gpio binding, so this is no longer
needed. In fact in requesting the same GPIO, it makes the designware
driver fail to probe.

Signed-off-by: Chen-Yu Tsai <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
9 years agosunxi: gpio: Add .xlate function for gpio phandle resolution
Chen-Yu Tsai [Fri, 22 Jul 2016 08:12:59 +0000 (16:12 +0800)]
sunxi: gpio: Add .xlate function for gpio phandle resolution

sunxi uses a 2 cell phandle for gpio bindings. Also there are no
seperate nodes for each pin bank.

Add a custom .xlate function to map gpio phandles to the correct
pin bank device. This fixes gpio_request_by_name usage.

Fixes: 7aa974858422 ("dm: sunxi: Modify the GPIO driver to support driver
      model")
Signed-off-by: Chen-Yu Tsai <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
9 years agorockchip: add support for rk3288 PopMetal board
[email protected] [Tue, 26 Jul 2016 10:28:30 +0000 (18:28 +0800)]
rockchip: add support for rk3288 PopMetal board

PopMetal is a rockchip rk3288 based board made by ChipSpark, which has
many interface such as HDMI, VGA, USB, micro-SD card, WiFi, Audio and
Gigabit Ethernet.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agorockchip: add basic support for fennec-rk3288 board
[email protected] [Tue, 26 Jul 2016 10:28:29 +0000 (18:28 +0800)]
rockchip: add basic support for fennec-rk3288 board

Fennec is a RK3288-based development board with 2 USB ports, HDMI,
micro-SD card, audio and WiFi and Gigabit Ethernet. It also includes
on-board 8GB eMMC and 2GB of SDRAM. Expansion connectors provides access
to display pins, I2C, SPI, UART and GPIOs.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agorockchip: rk3288: move evb board to rockchip folder
[email protected] [Tue, 26 Jul 2016 10:28:27 +0000 (18:28 +0800)]
rockchip: rk3288: move evb board to rockchip folder

The 'evb-rk3288' is not a vendor name, change it to 'rockchip' which is
the real vendor name.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agorockchip: rk3288: revise CONFIG_FASTBOOT_BUF_ADDR
[email protected] [Tue, 26 Jul 2016 10:28:26 +0000 (18:28 +0800)]
rockchip: rk3288: revise CONFIG_FASTBOOT_BUF_ADDR

CONFIG_SYS_LOAD_ADDR is absolutely safe to store image for
fastboot.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agorockchip: remove the duplicated macro config
[email protected] [Tue, 26 Jul 2016 10:28:25 +0000 (18:28 +0800)]
rockchip: remove the duplicated macro config

CONFIG_DOS_PARTITION and CONFIG_EFI_PARTITION are already included in
config_distro_defaults.h, and we don't need them in SPL stage.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agorockchip: rk3288: disable fastboot in SPL stage
[email protected] [Tue, 26 Jul 2016 10:28:24 +0000 (18:28 +0800)]
rockchip: rk3288: disable fastboot in SPL stage

Reduce compilation time for SPL.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agoRevert "rockchip: Move the MMC setup check earlier"
[email protected] [Tue, 26 Jul 2016 10:28:23 +0000 (18:28 +0800)]
Revert "rockchip: Move the MMC setup check earlier"

Boot Rom wouldn't initialize sdmmc while booting from eMMC. We need to
setup sdmmc gpio, otherwise we will hit an error below:

=>mmc info
blk_get_device: if_type=6, devnum=0: dwmmc@ff0c0000.blk, 6, 0
uclass_find_device_by_seq: 0 -1
uclass_find_device_by_seq: 0 0
   - -1 -1
   - -1 0
   - found
uclass_find_device_by_seq: 0 1
   - -1 -1
   - -1 0
   - not found
fdtdec_get_int_array: interrupts
get_prop_check_min_len: interrupts
Buswidth = 1, clock: 0
Buswidth = 1, clock: 400000
Sending CMD0
dwmci_send_cmd: Timeout on data busy
dwmci_send_cmd: Timeout on data busy
dwmci_send_cmd: Timeout on data busy
dwmci_send_cmd: Timeout on data busy

This reverts commit 6efeeea79c880d3dd262e0dca9da2687f0ab68c9.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
Tested-by: Simon Glass <[email protected]>
9 years agocosmetic: rockchip: rk3288: pinctrl: fix config symbol naming
[email protected] [Tue, 26 Jul 2016 10:28:22 +0000 (18:28 +0800)]
cosmetic: rockchip: rk3288: pinctrl: fix config symbol naming

Revise config to CONFIG_ROCKCHIP_RK3288_PINCTRL.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
Signed-off-by: Ziyuan Xu <[email protected]>
9 years agorockchip: add a dummy byte for the sdram-channel property
[email protected] [Tue, 26 Jul 2016 10:28:21 +0000 (18:28 +0800)]
rockchip: add a dummy byte for the sdram-channel property

Add an extra byte so that this data is not byteswapped.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
Signed-off-by: Ziyuan Xu <[email protected]>
9 years agorockchip: rk3288: Fix pinctrl for GPIO bank 0
John Keeping [Mon, 25 Jul 2016 09:02:05 +0000 (10:02 +0100)]
rockchip: rk3288: Fix pinctrl for GPIO bank 0

Bank 0 is the "PMU GPIO" bank which is controlled by the PMU registers
rather than the GRF registers.  In the GRF the top half of the register
is used as a mask so that some bits can be updated without affecting the
others, but in the PMU this feature is not provided and the top half of
the register is reserved.

Take the same approach as the Linux driver to update the value via
read-modify-write but setting the mask for only the bits that have
changed.  The PMU registers ignore the top 16 bits so this works for
both GRF and PMU iomux registers.

Signed-off-by: John Keeping <[email protected]>
Reviewed-by: Kever Yang <[email protected]>
9 years agork3399: Reserve space for ARM Trust Firmware
Kever Yang [Mon, 25 Jul 2016 03:45:54 +0000 (11:45 +0800)]
rk3399: Reserve space for ARM Trust Firmware

RK3399 needs reserve 0x200000 at the beginning of DRAM, for ATF bl31.

Signed-off-by: Kever Yang <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agorockchip: rk3036: update MAINTAINER file
Xu Ziyuan [Sat, 23 Jul 2016 07:09:28 +0000 (15:09 +0800)]
rockchip: rk3036: update MAINTAINER file

Update MAINTAINER files for kylin_rk3036, evb_rk3036.

Signed-off-by: Ziyuan Xu <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agoconfigs: rockchip: remove no use MACRO
Kever Yang [Fri, 22 Jul 2016 09:41:37 +0000 (17:41 +0800)]
configs: rockchip: remove no use MACRO

The CONFIG_ROCKCHIP_COMMON and CONFIG_SPL_ROCKCHIP_COMMON are no use now,
remove them.

Signed-off-by: Kever Yang <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agommc-uclass: correct the device number
Kever Yang [Fri, 22 Jul 2016 09:22:50 +0000 (17:22 +0800)]
mmc-uclass: correct the device number

Not like the mmc-legacy which the devnum starts from 1, it starts from 0
in mmc-uclass, so the device number should be (devnum + 1) in get_mmc_num().

Signed-off-by: Kever Yang <[email protected]>
Acked-by: Simon Glass <[email protected]>
Reviewed-by: Jaehoon Chung <[email protected]>
9 years agom68k: code reformatting for all start.S files
Angelo Dureghello [Sat, 21 May 2016 22:14:29 +0000 (00:14 +0200)]
m68k: code reformatting for all start.S files

This patch is style-related only, to reformat all the start.S code,
actually not following a coherent style inside single files and
between different cpu start.S files.

Linux format has been respected, as
  - max line width at 80 columns
  - one 8 cols tab between asm instructions and operands
  - inline comments, where any, fixed at col 41

Signed-off-by: Angelo Dureghello <[email protected]>
9 years agoARM: am57xx_evm: Enable QSPI support
Vignesh R [Fri, 29 Jul 2016 08:52:30 +0000 (14:22 +0530)]
ARM: am57xx_evm: Enable QSPI support

AM571x IDK and AM572x IDK EVMs have spansion s25fl256s QSPI flash on the
board connected to TI QSPI IP over CS0. Therefore enable QSPI support.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
9 years agoARM: dts: am57xx-idk-common: Enable support for QSPI
Vignesh R [Fri, 29 Jul 2016 08:52:31 +0000 (14:22 +0530)]
ARM: dts: am57xx-idk-common: Enable support for QSPI

AM571x and AM572x IDK have a spansion s25fl256s QSPI flash on the board
connected to TI QSPI over CS0. Hence, add QSPI and flash slave
DT nodes.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
9 years agoconfigs: am43xx_evm_defconfig: Enable CONFIG_SPI_FLASH_BAR
Vignesh R [Thu, 28 Jul 2016 11:57:57 +0000 (17:27 +0530)]
configs: am43xx_evm_defconfig: Enable CONFIG_SPI_FLASH_BAR

AM437x SK and AM437x IDK EVMs have 64MB flash, therefore enable
CONFIG_SPI_FLASH_BAR to access flash regions above 16MB.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
9 years agoARM: dts: dra7xx: Update spi-max-frequency for QSPI
Vignesh R [Mon, 25 Jul 2016 10:15:47 +0000 (15:45 +0530)]
ARM: dts: dra7xx: Update spi-max-frequency for QSPI

According to AM572x DM SPRS953A, QSPI max bus speed is 76.8MHz.
Therefore update the spi-max-frequency value of QSPI node for DRA74 and
DRA72 evm. This increase flash read speed by ~2MB/s.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
9 years agoconfigs: dra7xx: Update QSPI speed to 76.8MHz
Vignesh R [Mon, 25 Jul 2016 10:15:46 +0000 (15:45 +0530)]
configs: dra7xx: Update QSPI speed to 76.8MHz

Now that QSPI driver can support 76.8MHz, update the
CONFIG_SF_DEFAULT_SPEED to the same value.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
9 years agospi: ti_qspi: dra7xx: Add support to use 76.8MHz clock
Vignesh R [Mon, 25 Jul 2016 10:15:45 +0000 (15:45 +0530)]
spi: ti_qspi: dra7xx: Add support to use 76.8MHz clock

According to AM572x DM SPRS953A, QSPI bus speed can be 76.8MHz, update
the driver to use the same.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
9 years agoARM: dra7xx: Change DPLL_PER_HS13 divider value
Lokesh Vutla [Mon, 25 Jul 2016 10:15:44 +0000 (15:45 +0530)]
ARM: dra7xx: Change DPLL_PER_HS13 divider value

According to AM572x DM SPRS953A, QSPI bus speed can be 76.8MHz, hence
update QSPI input clock divider value (DPLL_PER_HS13) to provide 76.8MHz
clock, so that driver can use the same.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
9 years agosf: sf_params: Add AT25DF321 flash support
Wenyou Yang [Tue, 5 Jul 2016 05:17:12 +0000 (13:17 +0800)]
sf: sf_params: Add AT25DF321 flash support

Add AT25DF321 flash support.
Fix AT25DF321A device name.

Signed-off-by: Wenyou Yang <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
9 years agospi: ti_qspi: Remove delay in read path for dra7xx
Vignesh R [Fri, 22 Jul 2016 05:25:50 +0000 (10:55 +0530)]
spi: ti_qspi: Remove delay in read path for dra7xx

As per commit b545a98f5dc563 ("spi: ti_qspi: Add delay
for successful bulk erase) says its added to meet bulk erase timing
constraints. But bulk erase is a cmd to flash and delay in read path
does not make sense. Morever, testing on DRA74/DRA72 evm has shown that
this delay is no longer required.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
9 years agospi: ti_qspi: Fix compiler warning when DEBUG macro is set
Vignesh R [Fri, 22 Jul 2016 05:25:49 +0000 (10:55 +0530)]
spi: ti_qspi: Fix compiler warning when DEBUG macro is set

clk_div is uninitialized at the beginning of ti_spi_set_speed(), move
debug() print after clk_div calculation to avoid compiler warning and to
have proper value of clk_div printed during debugging.

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
9 years agospi: ti_qspi: Fix failure on multiple READ_ID cmd
Vignesh R [Fri, 22 Jul 2016 05:25:48 +0000 (10:55 +0530)]
spi: ti_qspi: Fix failure on multiple READ_ID cmd

Populating QSPI_RD_SNGL bit(0x1) in priv->cmd means that value
QSPI_INVAL (0x4) is not written to CMD field of QSPI_SPI_CMD_REG in
ti_qspi_cs_deactivate(). Therefore CS is never deactivated between
successive READ ID which results in sf probe to fail.
Fix this by not populating priv->cmd with QSPI_RD_SNGL and OR it wih
priv->cmd as required (similar to the convention followed in the
driver).

Signed-off-by: Vignesh R <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
9 years agospi: Add support for N25Q016A
Moritz Fischer [Thu, 14 Jul 2016 23:22:39 +0000 (16:22 -0700)]
spi: Add support for N25Q016A

This commit adds support in the spi-nor driver for the
N25Q016A, a 16Mbit SPI NOR flash from Micron.

Signed-off-by: Moritz Fischer <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
9 years agoMerge branch 'master' of git://www.denx.de/git/u-boot-imx
Tom Rini [Thu, 28 Jul 2016 12:45:00 +0000 (08:45 -0400)]
Merge branch 'master' of git://denx.de/git/u-boot-imx

9 years agoMAINTAINERS: i.MX: Add board/freescale/*mx* path
Fabio Estevam [Wed, 27 Jul 2016 00:27:49 +0000 (21:27 -0300)]
MAINTAINERS: i.MX: Add board/freescale/*mx* path

Pass the board/freescale/*mx*/ path as files maintained by Stefano
Babic.

While this is not ideal and does not cover all the i.MX board cases,
it gives at least a better hint for the /scripts/get_maintainer.pl
tool.

Signed-off-by: Fabio Estevam <[email protected]>
9 years agomx7dsabresd: MAINTAINERS: Add mx7dsabresd_secure_defconfig
Fabio Estevam [Wed, 27 Jul 2016 00:08:58 +0000 (21:08 -0300)]
mx7dsabresd: MAINTAINERS: Add mx7dsabresd_secure_defconfig

Add an entry for the mx7dsabresd_secure_defconfig target.

Signed-off-by: Fabio Estevam <[email protected]>
9 years agomx7_common: initialize generic timer on all CPU's
Stefan Agner [Tue, 26 Jul 2016 05:57:47 +0000 (22:57 -0700)]
mx7_common: initialize generic timer on all CPU's

Use CONFIG_TIMER_CLK_FREQ to let the non-secure init code initialize
the generic timer on all CPU's. This allows to make use of the timer
freuquency register also on other CPU than the start CPU which is
important for KVM.

Signed-off-by: Stefan Agner <[email protected]>
9 years agomx6ul_14x14_evk: Remove unused define
Diego Dorta [Mon, 25 Jul 2016 16:45:30 +0000 (13:45 -0300)]
mx6ul_14x14_evk: Remove unused define

Remove unused define constant.

Signed-off-by: Diego Dorta <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
9 years agocgtqmx6eval: Remove uneeded PHYS_SDRAM_SIZE
Fabio Estevam [Sat, 23 Jul 2016 16:23:42 +0000 (13:23 -0300)]
cgtqmx6eval: Remove uneeded PHYS_SDRAM_SIZE

cgtqmx6eval uses the imx_ddr_size() function to calculate the DDR size in
runtime, so there is no need to define PHYS_SDRAM_SIZE.

Remove the unneeded definition.

Cc: Otavio Salvador <[email protected]>
Signed-off-by: Fabio Estevam <[email protected]>
Acked-by: Otavio Salvador <[email protected]>
9 years agonovena: Remove uneeded PHYS_SDRAM_SIZE
Fabio Estevam [Sat, 23 Jul 2016 16:23:41 +0000 (13:23 -0300)]
novena: Remove uneeded PHYS_SDRAM_SIZE

novena uses the imx_ddr_size() function to calculate the DDR size in
runtime, so there is no need to define PHYS_SDRAM_SIZE.

Remove the unneeded definition.

Cc: Marek Vasut <[email protected]>
Signed-off-by: Fabio Estevam <[email protected]>
Acked-by: Marek Vasut <[email protected]>
9 years agobx50v3: Use imx_ddr_size() for calculating the DDR size
Fabio Estevam [Sat, 23 Jul 2016 16:23:40 +0000 (13:23 -0300)]
bx50v3: Use imx_ddr_size() for calculating the DDR size

imx_ddr_size() can be used to calculate the DDR size in runtime.

By using this function we no longer need to define PHYS_SDRAM_SIZE.

Cc: Martin Donnelly <[email protected]>
Signed-off-by: Fabio Estevam <[email protected]>
9 years agoaristainetos: Use imx_ddr_size() for calculating the DDR size
Fabio Estevam [Sat, 23 Jul 2016 16:23:39 +0000 (13:23 -0300)]
aristainetos: Use imx_ddr_size() for calculating the DDR size

imx_ddr_size() can be used to calculate the DDR size in runtime.

By using this function we no longer need to define PHYS_SDRAM_SIZE.

Cc: Heiko Schocher <[email protected]>
Signed-off-by: Fabio Estevam <[email protected]>
Acked-by: Heiko Schocher <[email protected]>
9 years agowarp: Use imx_ddr_size() for calculating the DDR size
Fabio Estevam [Sat, 23 Jul 2016 16:23:38 +0000 (13:23 -0300)]
warp: Use imx_ddr_size() for calculating the DDR size

imx_ddr_size() can be used to calculate the DDR size in runtime.

By using this function we no longer need to define PHYS_SDRAM_SIZE.

Signed-off-by: Fabio Estevam <[email protected]>
9 years agowarp7: Move some USB configuration options to defconfig
Breno Lima [Fri, 22 Jul 2016 18:29:30 +0000 (15:29 -0300)]
warp7: Move some USB configuration options to defconfig

Currently it's recommended to move some configuration options to the
defconfig file.

Move some USB related options to the defconfig file.

Signed-off-by: Breno Lima <[email protected]>
9 years agocolibri_imx7: add Colibri iMX7S/iMX7D module support
Stefan Agner [Thu, 21 Jul 2016 04:27:49 +0000 (21:27 -0700)]
colibri_imx7: add Colibri iMX7S/iMX7D module support

This commit adds support for the Toradex Computer on Modules
Colibri iMX7S/iMX7D. The two modules/SoC's are very similar hence
can be easily supported by one board. The board code detects RAM
size at runtime which is one of the differences between the two
boards. The board also uses the UART's in DTE mode, hence making
use of the new DTE support via serial DM.

Signed-off-by: Stefan Agner <[email protected]>
9 years agocgtqmx6eval: Replace is_mx6q() for macro
Breno Lima [Fri, 22 Jul 2016 12:12:12 +0000 (09:12 -0300)]
cgtqmx6eval: Replace is_mx6q() for macro

It's not necessary to implement the is_mx6q function, there is a macro in
sys_proto.h already implemented.

Signed-off-by: Breno Lima <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
9 years agomx6cuboxi: Replace is_mx6q() for macro
Breno Lima [Fri, 22 Jul 2016 12:11:30 +0000 (09:11 -0300)]
mx6cuboxi: Replace is_mx6q() for macro

It's not necessary to implement the is_mx6q function, there is a macro in
sys_proto.h already implemented.

Signed-off-by: Breno Lima <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
9 years agowandboard: Replace is_cpu_type() for macro
Breno Lima [Fri, 22 Jul 2016 12:11:02 +0000 (09:11 -0300)]
wandboard: Replace is_cpu_type() for macro

It's not necessary to use the is_cpu_type function, there is a macro in
sys_proto.h already implemented.

Signed-off-by: Breno Lima <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
9 years agoimx: ventana: add dt fixup for watchdog external reset
Tim Harvey [Fri, 15 Jul 2016 14:16:29 +0000 (07:16 -0700)]
imx: ventana: add dt fixup for watchdog external reset

Added removal of the fsl,ext-reset-output property in the wdog node for board
revisions that pre-date the addition of the external watchdog reset signal.

This property is a recent addition to mainline linux kernel in order to
specify that the IMX watchdog external reset should be used instead of the
internal chip-level reset.

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: refactor board-specific dt fixups (no functional change)
Tim Harvey [Fri, 15 Jul 2016 14:16:28 +0000 (07:16 -0700)]
imx: ventana: refactor board-specific dt fixups (no functional change)

Re-factor the board-specific dt fixups so that they are easier to follow
and extend in the future:
 - use defines for DT paths
 - use switch/case per board
 - order models numerically

There is no functional change in the code

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: make hwconfig initialize based on board configuration
Tim Harvey [Fri, 15 Jul 2016 14:14:25 +0000 (07:14 -0700)]
imx: ventana: make hwconfig initialize based on board configuration

The hwconfig env var allows user to control hardware specific configuration
of board specific features but not all Ventana boards have the same features.

We will use the magic default value of "_UNKNOWN_" to signify that the
bootloader should create this based on detected board model.

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: add extra DIO's for GW5520
Tim Harvey [Fri, 15 Jul 2016 14:14:24 +0000 (07:14 -0700)]
imx: ventana: add extra DIO's for GW5520

The GW5520 has 10 DIO's instead of the typical 4 found on the Ventana
product family.

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: make number of digital I/O's dynamic
Tim Harvey [Fri, 15 Jul 2016 14:14:23 +0000 (07:14 -0700)]
imx: ventana: make number of digital I/O's dynamic

Replace the static list of board-specific digital I/O's with a dynamic list.

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: make RS232 enable board specific
Tim Harvey [Fri, 15 Jul 2016 14:14:22 +0000 (07:14 -0700)]
imx: ventana: make RS232 enable board specific

Not all Ventana boards have an RS232 transceiver, make it board specific.

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: re-enable late board info display
Tim Harvey [Wed, 29 Jun 2016 15:58:00 +0000 (08:58 -0700)]
imx: ventana: re-enable late board info display

3b1f681131149b5f62602f582a7e60b0185a2a49 caused a regression that removes
board info dispaly for Gateworks Ventana boards because it made the invalid
assumption that CONFIG_DISPLAY_BOARDINFO_LATE was the same thing as
CONFIG_DISPLAY_BOARDINFO.

Ventana needs to call show_board_info in late init because we need to have
the i2c eeprom based model info. Re-define CONFIG_DISPLAY_BOARDINFO_LATE
to allow that to happen.

Cc: Peter Robinson <[email protected]>
Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: default pci to disabled
Tim Harvey [Fri, 17 Jun 2016 13:20:26 +0000 (06:20 -0700)]
imx: ventana: default pci to disabled

The IMX6 PCIe host controller does not have a proper reset and as such there
are several issues that can arise if PCI is enabled in the bootloader follwed
by Linux trying to re-configure LTSSM and/or toggling PERST# to the devices.

For now, the best approach seems to default to disabling PCI by defaulting
pciedisable=1. This can be overridden by the user if they need PCI in the
bootloader, for example:
 - GW552x needing ethernet access in bootloader
 - GW16082 expansion board needing a device-tree fixup for irq mapping

Signed-off-by: Tim Harvey <[email protected]>
9 years agopci: allow disabling of pci init/enum via env
Tim Harvey [Fri, 17 Jun 2016 13:20:25 +0000 (06:20 -0700)]
pci: allow disabling of pci init/enum via env

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: add dt fixup for eth1 mac-address
Tim Harvey [Fri, 17 Jun 2016 13:10:42 +0000 (06:10 -0700)]
imx: ventana: add dt fixup for eth1 mac-address

Ventana boards with a PCI Marvell Sky2 GigE MAC require the MAC address to
be placed in a DT node in order for the mainline linux driver to obtain it.

Signed-off-by: Tim Harvey <[email protected]>
9 years agoimx: ventana: add dt fixup for GW16082 irq mapping
Tim Harvey [Fri, 17 Jun 2016 13:10:41 +0000 (06:10 -0700)]
imx: ventana: add dt fixup for GW16082 irq mapping

The GW16082 mini-PCI expansion mezzanine uses a TI XIO2001 PCIe-to-PCI
bridge with legacy INTA/B/C/D interrupts. These interrupts are assigned
in the reverse order according to the PCI spec.

If the TI bridge is found on the Ventana PCI bus, add device-tree nodes
according to bus enumeration explicitly defining the interrupt mapping
to override the default PCI mapping in the Linux kernel. This allows
the GW16082 to work with upstream kernels that support device-tree
irq parsing.

Signed-off-by: Tim Harvey <[email protected]>
9 years agomx7dsabresd_secure_defconfig: Use CONFIG_ARMV7_BOOT_SEC_DEFAULT
Fabio Estevam [Fri, 22 Jul 2016 18:21:12 +0000 (15:21 -0300)]
mx7dsabresd_secure_defconfig: Use CONFIG_ARMV7_BOOT_SEC_DEFAULT

There is no need for introducing MX7_SEC, as there is the
CONFIG_ARMV7_BOOT_SEC_DEFAULT option for this purpose.

Switch to CONFIG_ARMV7_BOOT_SEC_DEFAULT and get rid of
MX7_SEC.

Tested by booting a 4.1.15 NXP kernel with mx7dsabresd_secure_defconfig
target.

Signed-off-by: Fabio Estevam <[email protected]>
Acked-by: Stefan Agner <[email protected]>
9 years agopico-imx6ul: drop warning due to redefined
Stefano Babic [Wed, 20 Jul 2016 15:54:07 +0000 (17:54 +0200)]
pico-imx6ul: drop warning due to redefined

USB gadget configuration is set in defconfig and
must be removed from pico-imx6ul.h.

Signed-off-by: Stefano Babic <[email protected]>
CC: Fabio Estevam <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
9 years agomx6: wandboard: fix warning due to missing prototype
Stefano Babic [Wed, 20 Jul 2016 15:53:56 +0000 (17:53 +0200)]
mx6: wandboard: fix warning due to missing prototype

Signed-off-by: Stefano Babic <[email protected]>
CC: Fabio Estevam <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
9 years agoFix build for mx7dsabresd (secure config)
Stefano Babic [Wed, 20 Jul 2016 15:52:53 +0000 (17:52 +0200)]
Fix build for mx7dsabresd (secure config)

After moving CONFIG_USB_EHCI_MX7 to Kconfig,
the flag must be set in defconfig for mx7dsabresd.
It is already for the not secure config, it is
missing in the secure configuration.

Signed-off-by: Stefano Babic <[email protected]>
CC: Fabio Estevam <[email protected]>
Tested-by: Fabio Estevam <[email protected]>
9 years agoMerge git://git.denx.de/u-boot-dm
Tom Rini [Thu, 28 Jul 2016 02:30:20 +0000 (22:30 -0400)]
Merge git://git.denx.de/u-boot-dm

9 years agommc: dw_mmc: reduce timeout detection cycle
Xu Ziyuan [Tue, 19 Jul 2016 01:38:22 +0000 (09:38 +0800)]
mmc: dw_mmc: reduce timeout detection cycle

It's no need to speed 10 seconds to wait the mmc device out from busy
status. 500 milliseconds enough.

Signed-off-by: Ziyuan Xu <[email protected]>
Reviewed-by: Jaehoon Chung <[email protected]>
Tested-by: Jaehoon Chung <[email protected]>
9 years agoAdd a power domain framework/uclass
Stephen Warren [Wed, 13 Jul 2016 19:45:31 +0000 (13:45 -0600)]
Add a power domain framework/uclass

Many SoCs allow power to be applied to or removed from portions of the SoC
(power domains). This may be used to save power. This API provides the
means to control such power management hardware.

Signed-off-by: Stephen Warren <[email protected]>
Acked-by: Simon Glass <[email protected]>
9 years agodm: spl: mmc: Support raw partitions with CONFIG_BLK
Simon Glass [Tue, 5 Jul 2016 23:10:18 +0000 (17:10 -0600)]
dm: spl: mmc: Support raw partitions with CONFIG_BLK

Fix up the call in mmc_load_image_raw_partition() to use the correct
function to obtain the MMC device, so that this code can support driver
model.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: usb: Use blk_dread/write() instead of direct calls
Simon Glass [Tue, 5 Jul 2016 23:10:17 +0000 (17:10 -0600)]
dm: usb: Use blk_dread/write() instead of direct calls

Update the USB mass storage code to allow it to work with driver model.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: socfpga: mmc: Support CONFIG_BLK
Simon Glass [Tue, 5 Jul 2016 23:10:16 +0000 (17:10 -0600)]
dm: socfpga: mmc: Support CONFIG_BLK

Update the driver to support using driver model for block devices.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: mmc: zynq: Convert zynq to use driver model for MMC
Simon Glass [Tue, 5 Jul 2016 23:10:15 +0000 (17:10 -0600)]
dm: mmc: zynq: Convert zynq to use driver model for MMC

Move zynq to the latest driver model support by enabling CONFIG_DM_MMC,
CONFIG_DM_MMC_OPS and CONFIG_BLK.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: zynq: usb: Convert to CONFIG_DM_USB
Simon Glass [Tue, 5 Jul 2016 23:10:14 +0000 (17:10 -0600)]
dm: zynq: usb: Convert to CONFIG_DM_USB

Convert zynq USB to driver model. Note this is tested on zynq-zybo only.

Signed-off-by: Simon Glass <[email protected]>
9 years agozynq: Increase the early malloc() size
Simon Glass [Tue, 5 Jul 2016 23:10:13 +0000 (17:10 -0600)]
zynq: Increase the early malloc() size

This is needed to support driver-model conversion of USB and block devices.

Signed-off-by: Simon Glass <[email protected]>
9 years agonet: phy: marvell: Add a missing errno.h header
Simon Glass [Tue, 5 Jul 2016 23:10:12 +0000 (17:10 -0600)]
net: phy: marvell: Add a missing errno.h header

This corrects a build error on zynqmp.

Signed-off-by: Simon Glass <[email protected]>
9 years agoarm: Show early-malloc() usage in bdinfo
Simon Glass [Tue, 5 Jul 2016 23:10:11 +0000 (17:10 -0600)]
arm: Show early-malloc() usage in bdinfo

This is useful information to show how close we are to the limit. At present
it is only available by enabling DEBUG in board_r.c.

Make it available with the 'bdinfo' command also.

Note that this affects ARM only. The bdinfo command is different for each
architecture. Rather than duplicating the code it would be better to
refactor it (as was done with global_data).

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: Use dm_scan_fdt_dev() directly where possible
Simon Glass [Tue, 5 Jul 2016 23:10:10 +0000 (17:10 -0600)]
dm: Use dm_scan_fdt_dev() directly where possible

Quite a few places have a bind() method which just calls dm_scan_fdt_dev().
We may as well call dm_scan_fdt_dev() directly. Update the code to do this.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: Convert users from dm_scan_fdt_node() to dm_scan_fdt_dev()
Simon Glass [Tue, 5 Jul 2016 23:10:09 +0000 (17:10 -0600)]
dm: Convert users from dm_scan_fdt_node() to dm_scan_fdt_dev()

This new function is more convenient for callers, and handles pre-relocation
situations automatically.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: core: Add a function to bind child devices
Simon Glass [Tue, 5 Jul 2016 23:10:08 +0000 (17:10 -0600)]
dm: core: Add a function to bind child devices

We currently use dm_scan_fdt_node() to bind devices. It is an internal
function and it requires the caller to know whether we are pre- or post-
relocation.

This requirement has become quite common in drivers, so the current function
is not ideal.

Add a new function with fewer arguments, that does not require internal
headers. This can be used directly as a post_bind() method if needed.

Signed-off-by: Simon Glass <[email protected]>
9 years agodm: mmc: dwmmc: use the callback functions as static
Jaehoon Chung [Tue, 28 Jun 2016 06:52:21 +0000 (15:52 +0900)]
dm: mmc: dwmmc: use the callback functions as static

There are no places to call these functions.
It should be used the callback function.
Then it can be used as static functions.

Signed-off-by: Jaehoon Chung <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
9 years agodm: mmc: dwmmc: fix the wrong explanation for clock values
Jaehoon Chung [Tue, 28 Jun 2016 06:52:20 +0000 (15:52 +0900)]
dm: mmc: dwmmc: fix the wrong explanation for clock values

This e,g is wrong. Maximum/minimum e.g values are swapped each other.

Signed-off-by: Jaehoon Chung <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
9 years agoMerge branch 'master' of git://git.denx.de/u-boot-video
Tom Rini [Wed, 27 Jul 2016 19:22:21 +0000 (15:22 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-video

9 years agoMerge branch 'master' of http://git.denx.de/u-boot-sunxi
Tom Rini [Tue, 26 Jul 2016 22:33:04 +0000 (18:33 -0400)]
Merge branch 'master' of git.denx.de/u-boot-sunxi

9 years agosunxi: Disable sun8i emac driver
Hans de Goede [Tue, 26 Jul 2016 20:26:39 +0000 (22:26 +0200)]
sunxi: Disable sun8i emac driver

Disable the sun8i emac driver for now, there are 2 issues with it:

1) It is causing issues with network connectivity under the kernel driver,
when booting the kernel with v2 of Corentin's sun8i-h3 emac driver, I get
the connection status bouncing between connected at 100mbps full-duplex
and being down every second.

The second issue is that when trying to use it from u-boot
I get a number of unaligned cache flush errors:

=> dhcp
BOOTP broadcast 1
BOOTP broadcast 2
CACHE: Misaligned operation at range [7bf594a87bf59628]
BOOTP broadcast 3
CACHE: Misaligned operation at range [7bf59c907bf59e10]
CACHE: Misaligned operation at range [7bf5a4787bf5a5f8]
DHCP client bound to address 10.42.43.80 (1009 ms)

Cc: Chen-Yu Tsai <[email protected]>
Cc: Corentin LABBE <[email protected]>
Cc: Amit Singh Tomar <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
9 years agoARM: uniphier: move CONFIG_I2C_EEPROM to defconfig
Masahiro Yamada [Tue, 26 Jul 2016 18:47:58 +0000 (03:47 +0900)]
ARM: uniphier: move CONFIG_I2C_EEPROM to defconfig

We already have the entry for this option in Kconfig, so let's
migrate to it.

Signed-off-by: Masahiro Yamada <[email protected]>
9 years agoMerge git://git.denx.de/u-boot-mpc86xx
Tom Rini [Tue, 26 Jul 2016 21:34:51 +0000 (17:34 -0400)]
Merge git://git.denx.de/u-boot-mpc86xx

9 years agoMerge git://git.denx.de/u-boot-fsl-qoriq
Tom Rini [Tue, 26 Jul 2016 21:34:28 +0000 (17:34 -0400)]
Merge git://git.denx.de/u-boot-fsl-qoriq

9 years agonet: sun8i_emac: Drop redundant and incorrect setting of syscon register
Chen-Yu Tsai [Fri, 22 Jul 2016 10:16:10 +0000 (18:16 +0800)]
net: sun8i_emac: Drop redundant and incorrect setting of syscon register

In sun8i_emac_board_setup, the driver partially configures the syscon
register for H3 EPHY. However, the settings are incomplete, and
completely unusable. The correct settings are later set in
sun8i_emac_set_syscon, but the incorrect CLK_SEL setting persists.

It is incorrect to use CLK_SEL to select 25 MHz, as the SoC does not
have a 25 MHz clock the EPHY can use.

This patch removes the setting of the syscon register in board_setup,
and also moves set_syscon above mdio_init. While mdio_init does not
access the PHY, it is better to have the PHY parameters setup before
the MDIO bus is registered.

Fixes: a29710c525ff ("net: Add EMAC driver for H3/A83T/A64 SoCs.")
Signed-off-by: Chen-Yu Tsai <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
9 years agonet: sun8i_emac: Do not configure AHB2 clock
Chen-Yu Tsai [Fri, 22 Jul 2016 10:16:09 +0000 (18:16 +0800)]
net: sun8i_emac: Do not configure AHB2 clock

The sun8i_emac driver erroneously configures the AHB2 clock when it
assumes it is configuring the AXI gates, which is not even documented
or ever appeared in either the WiP kernel driver or Allwinner's original
driver.

As a result, AHB2 clock mux is set to an invalid setting, making the
EPHY unusable.

Fixes: a29710c525ff ("net: Add EMAC driver for H3/A83T/A64 SoCs.")
Signed-off-by: Chen-Yu Tsai <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>