Tom Rini [Tue, 29 Nov 2016 17:41:19 +0000 (12:41 -0500)]
travis.yml: Add samsung and rockchip builds
The catch-all job is failing due to time limits depending on factors out
of our control, so move Samsung and Rockchip boards into their own jobs
and then exclude them from the general ARM and AArch64 jobs.
Peng Fan [Tue, 22 Nov 2016 11:41:09 +0000 (19:41 +0800)]
armv7: psci: cpu_off: flush D-Cache before disable D-Cache
Before disable cache, need to first flush cache.
There maybe dirty data in D-Cache before disable D-Cache.
After disable D-Cache, the first store instructions in
psci_v7_flush_dcache_all will directly store registers
{r4-r5, r7, r9-r11, lr} to memory.
If there is dirty data before disable D-Cache,
psci_v7_flush_dcache_all will flush data to memory,
and may overwrite the memory that hold the registers
{r4-r5, r7, r9-r11, lr}.
So before disable cache, first flush D-Cache.
Signed-off-by: Peng Fan <peng.fan@nxp.com> Cc: Albert Aribaud <albert.u.boot@aribaud.net> Cc: Chen-Yu Tsai <wens@csie.org> Cc: Hans de Goede <hdegoede@redhat.com> Cc: Hongbo Zhang <hongbo.zhang@nxp.com> Cc: York Sun <york.sun@nxp.com> Cc: Marc Zyngier <marc.zyngier@arm.com> Cc: Tom Rini <trini@konsulko.com>
Liviu Dudau [Tue, 22 Nov 2016 11:19:18 +0000 (11:19 +0000)]
vexpress64: Juno: Change PCI buss addresses for IO to start from zero.
Juno uses a 1:1 mapping between CPU and PCI addresses for IO. First,
that will trip devices that cannot use more than 16 bits of addresses
for IO, second it is un-necessary as the system can handle zero-based
PCI addresses just fine.
Change the mapping to start IO bus addresses from zero.
Mugunthan V N [Thu, 11 Aug 2016 14:34:03 +0000 (20:04 +0530)]
drivers: net: keystone_net: add rgmii link type support when parsing dt
Add support to detect RGMII link interface from link-interface
device tree entry. Also rename the existing link type enums so
that it provides meaningful interface like SGMII.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com> Reported-by: Sekhar Nori <nsekhar@ti.com> Tested-by: Sekhar Nori <nsekhar@ti.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When using OF_CONTROL, the disabled value of the mini UART platdata
gets reset after board_early_init_f. So move detection and disabling
to board_init and remove board_early_init_f.
This uses the first device using the mini uart driver, as this method
works reliably with different device trees or even no device tree at all.
Signed-off-by: Fabian Vogt <fvogt@suse.com> Reviewed-by: Simon Glass <sjg@chromium.org>
serial: bcm283x_mu: support disabling after initialization
For the Raspberry Pi 3 it needs to be possible to disable the serial
device after initialization happens, as only after the GPIO device is available
it is known whether the mini uart is usable.
Signed-off-by: Fabian Vogt <fvogt@suse.com> Reviewed-by: Simon Glass <sjg@chromium.org>
The information currently set via platdata has to be represented in the
device tree now. bcm283x-uboot.dtsi adds the u-boot specific "skip-init"
property to the serial nodes and enables initialization in the pre-reloc phase.
Cc: Albert Aribaud <albert.u.boot@aribaud.net> Signed-off-by: Fabian Vogt <fvogt@suse.com> Reviewed-by: Simon Glass <sjg@chromium.org>
serial: pl01x: expose skip_init platdata option in DT
To be able to represent the skip-init platdata element with OF_CONTROL,
it needs to be read from the device tree as well and put into the platform data.
Cc: Eric Anholt <eric@anholt.net> Signed-off-by: Fabian Vogt <fvogt@suse.com> Reviewed-by: Simon Glass <sjg@chromium.org>
fs/fat/fatwrite: Local variable as buffer to store dir_slot entries
fill_dir_slot use get_contents_vfatname_block as a temporary buffer for
constructing a list of dir_slot entries. To save the memory and providing
correct type of memory for above usage, a local buffer with accurate size
declaration is introduced.
The local array size 640 is used because for long file name entry,
each entry use 32 bytes, one entry can store up to 13 characters.
The maximum number of entry possible is 20. So, total size is
32*20=640bytes.
Signed-off-by: Genevieve Chan <ccheauya@altera.com> Signed-off-by: Tien Fong Chee <tfchee@altera.com>
Stefan Agner [Mon, 21 Nov 2016 18:58:52 +0000 (10:58 -0800)]
spl: dfu: move DFU Kconfig to SPL Kconfig
The DFU Kconfig menu entries should be part of the SPL
Kconfig file. Also avoid using the top level Makefile by
moving the config dependent build artifacts to the driver/
and driver/usb/gadget/ Makfiles.
With that, DFU can be built again in SPL if
CONFIG_SPL_DFU_SUPPORT is enabled.
Fixes: 6ad6102246d8 ("usb:gadget: Disallow DFU in SPL for now") Signed-off-by: Stefan Agner <stefan.agner@toradex.com> Reviewed-by: Simon Glass <sjg@chromium.org> Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Stefan Agner [Mon, 21 Nov 2016 18:58:51 +0000 (10:58 -0800)]
spl: add RAM boot device only if it is actually defined
Some devices (e.g. dra7xx) support loading to RAM using DFU without
having direct boot from RAM support. Make sure the linker list
does not contain BOOT_DEVICE_RAM if CONFIG_SPL_RAM_DEVICE is not
enabled.
Fixes: 98136b2f26fa ("spl: Convert spl_ram_load_image() to use linker list") Signed-off-by: Stefan Agner <stefan.agner@toradex.com> Acked-by: Lukasz Majewski <l.majewski@samsung.com>
This fixes the loop delay when using a hw watchdog.
In case a watchdog is used that accesses CPU registers,
the defined delay of 20us in a tight loop will cause a
huge delay in the actual timeout seen. This is caused
by the fact that udelay will inheritantly call WATCHDOG_RESET.
Together with the omap wdt implementation, the seen timeout increases up to
around 30s. This makes the loop very slow and causes long
delays when using the modem.
Instead, implement the 2 sec loop by using the timer interface to know
when to break out of the timeout loop. Watchdog kicking is taken care of
by getc().
Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
Tang Yuantian [Mon, 21 Nov 2016 02:24:20 +0000 (10:24 +0800)]
sata: fix sata command can not being executed bug
Commit d97dc8a0 separated the non-command code into its own file
which caused variable sata_curr_device can not be set to a correct
value.
Before commit d97dc8a0, variable sata_curr_device can be set
correctly in sata_initialize().
After commit d97dc8a0, sata_initialize() is moved out to its own file.
Accordingly, variable sata_curr_device is removed from sata_initialize()
too. This caused sata_curr_device never gets a chance to be set properly
which prevent other commands from being executed.
This patch sets variable sata_curr_device properly.
Fixes: d97dc8a0 (dm: sata: Separate the non-command code into its
own file)
Signed-off-by: Tang Yuantian <yuantian.tang@nxp.com> Reviewed-by: Simon Glass <sjg@chromium.org>
For block devices the sample fw_env.config does not mandate a erase block size
for block devices. A missing setting defaults to the full env size.
Depending on the environment location the alignment check now errors out for
perfectly legal settings.
Fix this by defaulting to the standard blocksize of 0x200 for environments
stored in a block device.
That keeps the fw_env.config files for block devices working even with that
new check.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
The Makefile in tools/ tries to find the "swig" utility by calling "which".
If nothing is found in the path, some versions of which will print an error
message:
$ make clean
which: no swig in (/usr/local/bin:/usr/bin:/bin)
This does not apply to all version of "which", though:
$ echo $0
bash
$ type which
which is aliased to `type -path'
$ which foo <== this version is OK
$ /usr/bin/which foo <== this one is chatty
/usr/bin/which: no foo in (/usr/local/bin:/usr/bin:/bin)
$ sh <== make uses /bin/sh
sh-4.3$ which foo <== no alias here
which: no foo in (/usr/local/bin:/usr/bin:/bin)
This error message is rather pointless in our case, since we just have
this very check to care for this. So add stderr redirection to suppress
the message.
Signed-off-by: Andre Przywara <andre.przywara@arm.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Andrew F. Davis [Fri, 18 Nov 2016 17:56:16 +0000 (11:56 -0600)]
ti_armv7_keystone2: env: Add NFS loading support for PMMC and MON
NFS loading support has been added to the default environment for
most boot components, as PMMC and MON loading were added later they
did not originally get the NFS commands added, add these now.
Signed-off-by: Andrew F. Davis <afd@ti.com> Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
Andrew F. Davis [Fri, 18 Nov 2016 17:56:15 +0000 (11:56 -0600)]
keystone2: Move target selection to Kconfig
The config option TARGET_K2x_EVM is set by the k2x defconfigs to pick
a board target, but the header configs also set K2x_EVM. This config
is redundant, remove it and use TARGET_K2x_EVM everywhere in its place.
Signed-off-by: Andrew F. Davis <afd@ti.com> Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
Commit c68c62 ("i2c: mvtwsi: Make delay times frequency-dependent")
extensively used the ndelay function with a calculated parameter
which is dependant on the configured frequency of the I2C bus. If
standard speed is employed, the parameter is usually 10000 (10000ns
period length for 100kHz frequency).
But, since the arm architecture does not implement a proper version of
ndelay, the fallback default from include/linux/compat.h is used,
which defines every ndelay as udelay(1). This causes problems for
slower speeds on arm, since the delay time is now 9us too short for
the desired frequency, which leads to random failures of the I2C
interface.
To remedy this, we implement a proper, parameter-aware ndelay fallback
for architectures that don't implement a real ndelay function.
Reported-By: Jason Brown <Jason.brown@apcon.com>
To: Tom Rini <trini@konsulko.com>
To: Heiko Schocher <hs@denx.de> Signed-off-by: Mario Six <mario.six@gdsys.cc>
Marcel Ziswiler [Wed, 16 Nov 2016 16:49:24 +0000 (17:49 +0100)]
apalis/colibri_t30: move environment location
Now with the config block handling in place move the U-Boot environment
location before the config block at the end of 1st "boot sector" as
deployed during production using our downstream BSP.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com> Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
Marcel Ziswiler [Wed, 16 Nov 2016 16:49:22 +0000 (17:49 +0100)]
toradex: config block handling
Add Toradex factory configuration block handling. The config block is a
data structure which gets stored to flash during production testing. The
structure holds such information as board resp. hardware revision,
product ID and serial number which is used as the NIC part of the
Ethernet MAC address as well. The config block will be read upon boot by
the show_board_info() function, displayed as part of the board
information and passed to Linux via device tree or ATAGs.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com> Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
Drop CONFIG_CUSTOM_BOARDINFO as it is not Kconfig compliant and anyway
not really used anywhere plus the upcoming weak show_board_info()
approach seems much superior.
Add check for FDT_MAGIC, otherwise also legacy images will be loaded as
a FIT. With this check in place, the loader works correct both
with legacy and FIT images.
Signed-off-by: Tomas Melin <tomas.melin@vaisala.com> Acked-by: Lokesh Vutla <lokeshvutla@ti.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Philipp Tomsich [Wed, 16 Nov 2016 01:40:27 +0000 (01:40 +0000)]
sun8i_emac: Fix mdio read sequence
To send a parametrized command to the PHY over MDIO, we should write
the data first, the trigger the execution by the command register
write. Fix the access pattern in our MDIO write routine.
Apparently this doesn't really matter with the Realtek PHY on the
Pine64, but other PHYs (which require more setup) will choke on
the wrong order.
[Andre: add commit message]
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com> Signed-off-by: Andre Przywara <andre.przywara@arm.com> Acked-by: Jagan Teki <jagan@openedev.com>
Boris Brezillon [Tue, 8 Nov 2016 16:21:12 +0000 (17:21 +0100)]
mtd: nand: add support for the TC58NVG2S0H chip
Add the description of the Toshiba TC58NVG2S0H SLC nand to the nand_ids
table so we can use the NAND ECC infos and the ONFI timings.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com> Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Emmanuel Vadot [Sat, 5 Nov 2016 19:51:11 +0000 (20:51 +0100)]
sunxi: mmc: Set CONFIG_SYS_MMC_MAX_DEVICE
Set CONFIG_SYS_MMC_MAX_DEVICE to 4 for sunxi SoC.
This define is needed in the API code.
Signed-off-by: Emmanuel Vadot <manu@bidouilliste.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Yann E. MORIN [Mon, 31 Oct 2016 21:33:40 +0000 (22:33 +0100)]
arm: sunxi: do not force USB for arch-sunxi
Currently, USB is forced-enabled for the sunxi familly, and there is no
way to disable it.
However, USB takes a long time to initiliase, delaying the boot by up to
5 seconds (without any USB device attached!). This is a very long delay,
especially in cases where USB booting is not wanted at all, and where
the device is expected to boot relatively often (even in production).
Change the way the dependencies are handled, by only forcibly selecting
USB when CONFIG_DISTRO_DEFAULTS ("defaults suitable for booting general
purpose Linux distributions") is set. This option defaults to y for the
sunxi familly, so the current default behaviour is kept unchanged. Users
interested in boot time and/or size will be able to disable this to
further disable USB.
With USB disabled, the time spent in U-Boot before handing control to
the Linux kernel is about 1s now, down from ~5s (Nanopi Neo, sunxi H3).
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr> Cc: Ian Campbell <ijc@hellion.org.uk> Cc: Hans De Goede <hdegoede@redhat.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Alexander Graf [Fri, 18 Nov 2016 12:18:00 +0000 (13:18 +0100)]
travis: Add efi_loader grub2 test
We have all the building blocks now to run arbitrary efi applications
in travis. The most important one out there is grub2, so let's add
a simple test to verify that grub2 still comes up.
The deviation is tiny, but fails the overall build result. Since
the sleep test is not terribly important as gate keeper for travis
tests, let's just exclude it for this board.
Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Tom Rini <trini@konsulko.com>
Alexander Graf [Thu, 17 Nov 2016 21:40:10 +0000 (22:40 +0100)]
efi_loader: Allow to compile helloworld.efi w/o bundling it
Today we can compile a self-contained hello world efi test binary that
allows us to quickly verify whether the EFI loader framwork works.
We can use that binary outside of the self-contained test case though,
by providing it to a to-be-tested system via tftp.
This patch separates compilation of the helloworld.efi file from
including it in the u-boot binary for "bootefi hello". It also modifies
the efi_loader test case to enable travis to pick up the compiled file.
Because we're now no longer bloating the resulting u-boot binary, we
can enable compilation always, giving us good travis test coverage.
Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Tom Rini <trini@konsulko.com>
Alexander Graf [Thu, 17 Nov 2016 17:31:05 +0000 (18:31 +0100)]
tests: Add efi_loader hello world test
Now that we have working network tests and a hello world efi application
built inside our tree, we can automatically test that efi binary running
inside of U-Boot.
Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Tom Rini <trini@konsulko.com>
Alexander Graf [Thu, 17 Nov 2016 17:31:04 +0000 (18:31 +0100)]
travis: Add python path for environments
When running in travis-ci, we want to pass environment configuration to
the tests. These reside in a path available through PYTHONPATH, so let's
define that one to point to the unit test repo.
Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Tom Rini <trini@konsulko.com>
Alexander Graf [Thu, 17 Nov 2016 17:31:03 +0000 (18:31 +0100)]
Travis: Expose build dir as variable
Some travis QEMU tests can transfer files between the build directory
and the guest U-Boot instance. For that to work, both need to have access
to the same directory.
This patch puts the current build path into an environment variable, so
that the environment generating python scripts can extract it from there
and read the respective files.
Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Tom Rini <trini@konsulko.com>
Alexander Graf [Thu, 17 Nov 2016 17:31:02 +0000 (18:31 +0100)]
tests: net: Offset downloads to 4MB
The network test currently downloads files at 0MB offset of RAM start.
This works for most ARM systems, but x86 has weird memory layout constraints
on the first MB of RAM.
To not get caught into any of these, let's add a 4MB pad from start
of RAM to the default memory offset.
Signed-off-by: Alexander Graf <agraf@suse.de> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Tom Rini <trini@konsulko.com>
Simon Glass [Sun, 13 Nov 2016 21:24:54 +0000 (14:24 -0700)]
rockchip: Add support for veyron-minnie (ASUS Chromebook Flip)
This adds support for the Asus Chromebook Flip, an RK3288-based clamshell
device which can flip into 'tablet' mode. The device tree file comes from
Linux v4.8. The SDRAM parameters are for 4GB Samsung LPDDR3.
Simon Glass [Sun, 13 Nov 2016 21:22:16 +0000 (14:22 -0700)]
rockchip: Add support for veyron-mickey (Chromebit)
This adds support for the Asus Chromebit, and RK3288-based device designed
to plug directly into an HDMI monitor. The device tree file comes from
Linux v4.8.
Simon Glass [Sun, 13 Nov 2016 21:22:15 +0000 (14:22 -0700)]
rockchip: video: Avoid using u8 in the HDMI driver
It makes not sense using u8 to hold a value on a 32-bit or 64-bit machine.
It can only bloat the code by forcing the compiler to mask the value.
Change it to uint.
Simon Glass [Sun, 13 Nov 2016 21:22:14 +0000 (14:22 -0700)]
rockchip: veyron: Adjust ARM clock after relocation
Update board_init() to increase the ARM clock to the maximum speed on
veyron boards. This makes quite a large difference in performance. With
this change, speed goes from about 750 DMIPS to 2720 DMIPs.
Simon Glass [Sun, 13 Nov 2016 21:22:13 +0000 (14:22 -0700)]
rockchip: clk: Support setting ACLK
Add basic support for setting the ARM clock, since this allows us to run
at maximum speed in U-Boot. Currently only a single speed is supported
(1.8GHz).
Simon Glass [Sun, 13 Nov 2016 21:22:10 +0000 (14:22 -0700)]
rockchip: Rename jerry files to veyron
At present we have a single rk3288-based Chromebook: chromebook_jerry. But
all such Chromebooks can use the same binary with only device-tree
differences. The family name is 'veyron', so rename the files accordingly.
Also update the device-tree filename since this currently differs from
Linux.
Simon Glass [Sun, 13 Nov 2016 21:22:00 +0000 (14:22 -0700)]
stdio: Correct numbering logic in stdio_probe_device()
The current code assumes that the devices are ordered corresponding to
their alias value. But (for example) video1 can come before video0 in the
device tree.
Correct this, by always looking for device 0 first. After that we can fall
back to finding the first available device.
Keerthy [Wed, 26 Oct 2016 08:12:30 +0000 (13:42 +0530)]
power: regulator: Introduce regulator_set_value_force function
In case we want to force a particular value on a regulator
irrespective of the min/max constraints for testing purposes
one can call regulator_set_value_force function.
Signed-off-by: Keerthy <j-keerthy@ti.com> Reviewed-by: Simon Glass <sjg@chromium.org>