oweals/u-boot.git
7 years agoMerge git://git.denx.de/u-boot-mips
Tom Rini [Tue, 28 Nov 2017 21:54:30 +0000 (16:54 -0500)]
Merge git://git.denx.de/u-boot-mips

7 years agoMerge git://git.denx.de/u-boot-uniphier
Tom Rini [Tue, 28 Nov 2017 21:54:09 +0000 (16:54 -0500)]
Merge git://git.denx.de/u-boot-uniphier

7 years agoboston: Add u-boot.mcs make target
Paul Burton [Tue, 21 Nov 2017 22:31:07 +0000 (14:31 -0800)]
boston: Add u-boot.mcs make target

U-Boot is generally flashed to a MIPS Boston development board by means
of a .mcs file which Xilinx Vivado software can write to the flash
present on the board. As such we'd generally want to produce an mcs file
when building U-Boot to target the Boston board. Introduce a make target
for u-boot.mcs which generates it using the srec_cat tool available from
the SRecord project, and build it by default when srec_cat is present.

Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: u-boot@lists.denx.de
7 years agoboston: Set CONFIG_SYS_LOAD_ADDR to 0x88000000
Paul Burton [Tue, 21 Nov 2017 20:35:31 +0000 (12:35 -0800)]
boston: Set CONFIG_SYS_LOAD_ADDR to 0x88000000

Generally we load Linux kernels on Boston boards in the form of FIT
images containing a compressed kernel binary. Linux is linked at
0x80100000 and so we need to decompress the kernel binary to that
address, however this is our default load address which means that
unless explicitly avoided we hit a decompression error as the
uncompressed kernel binary overwrites its compressed version from the
FIT image.

Avoid this by adjusting CONFIG_SYS_LOAD_ADDR to 0x88000000 (or
0xffffffff88000000 for MIPS64 builds) which avoids the address overlap
between compressed & uncompressed kernel binaries.

Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: u-boot@lists.denx.de
7 years agoMIPS: Break out of cache loops for unimplemented caches
Paul Burton [Tue, 21 Nov 2017 19:18:39 +0000 (11:18 -0800)]
MIPS: Break out of cache loops for unimplemented caches

If we run on a CPU which doesn't implement a particular cache then we
would previously get stuck in an infinite loop, executing a cache op on
the first "line" of the missing cache & then incrementing the address by
0. This was being avoided for the L2 caches, but not for the L1s. Fix
this by generalising the check for a zero line size & avoiding the cache
op loop when this is the case.

Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: u-boot@lists.denx.de
7 years agoMIPS: Clear instruction hazards in flush_cache()
Paul Burton [Tue, 21 Nov 2017 19:18:38 +0000 (11:18 -0800)]
MIPS: Clear instruction hazards in flush_cache()

When writing code, for example during relocation, we ensure that the
icache has a coherent view of the new instructions with a call to
flush_cache(). This handles the bulk of the work to ensure the new
instructions will execute as expected, however it does not ensure that
the CPU pipeline doesn't already contain instructions taken from a stale
view of the affected memory. This could theoretically be a problem for
relocation, but in practice typically isn't because we sync caches for
enough code after the entry point of the newly written code that by the
time the CPU pipeline might possibly fetch any of it we'll have long ago
written it back & invalidated any stale icache entries. This is however
a problem for shorter regions of code.

In preparation for later patches which write shorter segments of code,
ensure any instruction hazards are cleared by flush_cache() by
introducing & using a new instruction_hazard_barrier() function which
makes use of the jr.hb instruction to clear the hazard.

Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: u-boot@lists.denx.de
7 years agoMIPS: Ensure cache ops complete in cache maintenance functions
Paul Burton [Tue, 21 Nov 2017 19:18:37 +0000 (11:18 -0800)]
MIPS: Ensure cache ops complete in cache maintenance functions

A typical use of cache maintenance functions is to force writeback of
data which a device is about to read using DMA - for example a
descriptor or command structure. Such users of cache maintenance
functions require that operations on the cache have completed before
they proceed to instruct a device to read memory. This requires that we
place a completion barrier (ie. sync instruction) between the cache ops
and whatever write informs the device to perform DMA.

Whilst strictly speaking this isn't all users of the cache maintenance
functions & we could instead place the barriers in the drivers that
require them, it would be much more invasive to do so than to just have
the barrier be the default by placing it in the cache functions
themselves. The cost is low enough that it shouldn't matter to us in any
rare cases that we use the cache functions when not performing DMA.

Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: u-boot@lists.denx.de
7 years agoUpdate Paul Burton's email address
Paul Burton [Mon, 30 Oct 2017 23:58:21 +0000 (16:58 -0700)]
Update Paul Burton's email address

MIPS is no longer a part of Imagination Technologies, and as such my
@imgtec.com email address will soon cease to function. This patch
updates occurrances of it with my new @mips.com email address, and adds
an entry in .mailmap such that git (& tools such as get_maintainer.pl
when examining history) will use the new address.

Signed-off-by: Paul Burton <paul.burton@mips.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: u-boot@lists.denx.de
7 years agoMIPS: Drop unused PTR_COUNT_SHIFT from u-boot.lds
Paul Burton [Fri, 15 Sep 2017 18:35:54 +0000 (11:35 -0700)]
MIPS: Drop unused PTR_COUNT_SHIFT from u-boot.lds

The u-boot.lds linker script for MIPS defines a PTR_COUNT_SHIFT macro to
2 or 3 for 32 bit or 64 bit builds respectively. This macro is never
actually used though, so remove the dead code.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
7 years agoboston: Remove unused label in lowlevel_display
Paul Burton [Fri, 15 Sep 2017 18:34:31 +0000 (11:34 -0700)]
boston: Remove unused label in lowlevel_display

The lowlevel_display() function includes a "1:" label which is never
used. Remove it.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
7 years agoboston: Drop unused return value
Paul Burton [Fri, 15 Sep 2017 18:33:53 +0000 (11:33 -0700)]
boston: Drop unused return value

The boston lowlevel_init() function zeroes the return register v0,
despite the function not being expected to return a value & that value
never being used.

Remove the redundant assignment to v0.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
7 years agoARM: uniphier: remove unused NAND CONFIG options
Masahiro Yamada [Mon, 27 Nov 2017 05:13:49 +0000 (14:13 +0900)]
ARM: uniphier: remove unused NAND CONFIG options

The Denali NAND driver does not use these options any more.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agoARM: dts: uniphier: Sync with Linux 4.15-rc1
Masahiro Yamada [Fri, 24 Nov 2017 15:25:35 +0000 (00:25 +0900)]
ARM: dts: uniphier: Sync with Linux 4.15-rc1

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agogpio: uniphier: import dt-binginds header from Linux
Masahiro Yamada [Fri, 24 Nov 2017 15:25:34 +0000 (00:25 +0900)]
gpio: uniphier: import dt-binginds header from Linux

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agoARM: uniphier: remove XIRQ pin settings
Masahiro Yamada [Fri, 24 Nov 2017 15:25:33 +0000 (00:25 +0900)]
ARM: uniphier: remove XIRQ pin settings

The XIRQ pins are now set up on the Linux side by the GPIO hogging.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agoARM: uniphier: remove IRQ settings
Masahiro Yamada [Fri, 24 Nov 2017 15:25:32 +0000 (00:25 +0900)]
ARM: uniphier: remove IRQ settings

This work-around has been here in U-Boot because the AIDET and GPIO
drivers were missing in the upstream Linux.  Both are now available
in Linus' tree:
  - drivers/irqchip/irq-uniphier-aidet.c
  - drivers/gpio/gpio-uniphier.c

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agoARM: uniphier: set CONFIG_LOGLEVEL to 6
Masahiro Yamada [Fri, 24 Nov 2017 15:25:31 +0000 (00:25 +0900)]
ARM: uniphier: set CONFIG_LOGLEVEL to 6

Print out KERN_NOTICE or higher level log messages.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: denali: sync with Linux 4.15-rc1
Masahiro Yamada [Tue, 21 Nov 2017 17:38:32 +0000 (02:38 +0900)]
mtd: nand: denali: sync with Linux 4.15-rc1

I largely reworked the Denali NAND controller driver in Linux.
This commit imports the improvements from Linux.  The code is
almost synced with Linux 4.15-rc1.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: introduce NAND_ROW_ADDR_3 flag
Masahiro Yamada [Tue, 21 Nov 2017 17:38:31 +0000 (02:38 +0900)]
mtd: nand: introduce NAND_ROW_ADDR_3 flag

Several drivers check ->chipsize to see if the third row address cycle
is needed.  Instead of embedding magic sizes such as 32MB, 128MB in
drivers, introduce a new flag NAND_ROW_ADDR_3 for clean-up.  Since
nand_scan_ident() knows well about the device, it can handle this
properly.  The flag is set if the row address bit width is greater
than 16.

Delete comments such as "One more address cycle for ..." because
intention is now clear enough from the code.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Wenyou Yang <wenyou.yang@microchip.com>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 14157f861437ebe2d624b0a845b91bbdf8ca9a2d]

7 years agomtd: nand: add a shorthand to generate nand_ecc_caps structure
Masahiro Yamada [Tue, 21 Nov 2017 17:38:30 +0000 (02:38 +0900)]
mtd: nand: add a shorthand to generate nand_ecc_caps structure

struct nand_ecc_caps was designed as flexible as possible to support
multiple stepsizes (like sunxi_nand.c).

So, we need to write multiple arrays even for the simplest case.
I guess many controllers support a single stepsize, so here is a
shorthand macro for the case.

It allows to describe like ...

NAND_ECC_CAPS_SINGLE(denali_pci_ecc_caps, denali_calc_ecc_bytes, 512, 8, 15);

... instead of

static const int denali_pci_ecc_strengths[] = {8, 15};
static const struct nand_ecc_step_info denali_pci_ecc_stepinfo = {
        .stepsize = 512,
        .strengths = denali_pci_ecc_strengths,
        .nstrengths = ARRAY_SIZE(denali_pci_ecc_strengths),
};
static const struct nand_ecc_caps denali_pci_ecc_caps = {
        .stepinfos = &denali_pci_ecc_stepinfo,
        .nstepinfos = 1,
        .calc_ecc_bytes = denali_calc_ecc_bytes,
};

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: a03c60178c181767ecfb26fb311a88742d228118]

7 years agomtd: nand: add generic helpers to check, match, maximize ECC settings
Masahiro Yamada [Tue, 21 Nov 2017 17:38:29 +0000 (02:38 +0900)]
mtd: nand: add generic helpers to check, match, maximize ECC settings

Driver are responsible for setting up ECC parameters correctly.
Those include:
  - Check if ECC parameters specified (usually by DT) are valid
  - Meet the chip's ECC requirement
  - Maximize ECC strength if NAND_ECC_MAXIMIZE flag is set

The logic can be generalized by factoring out common code.

This commit adds 3 helpers to the NAND framework:
nand_check_ecc_caps - Check if preset step_size and strength are valid
nand_match_ecc_req - Match the chip's requirement
nand_maximize_ecc - Maximize the ECC strength

To use the helpers above, a driver needs to provide:
  - Data array of supported ECC step size and strength
  - A hook that calculates ECC bytes from the combination of
    step_size and strength.

By using those helpers, code duplication among drivers will be
reduced.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 2c8f8afa7f92acb07641bf95b940d384ed1d0294]

7 years agomtd: nand: Pass the CS line to ->setup_data_interface()
Boris Brezillon [Tue, 21 Nov 2017 17:38:28 +0000 (02:38 +0900)]
mtd: nand: Pass the CS line to ->setup_data_interface()

Some NAND controllers can assign different NAND timings to different
CS lines. Pass the CS line information to ->setup_data_interface() so
that the NAND controller driver knows which CS line is concerned by
the setup_data_interface() request.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 104e442a67cfba4d0cc982384761befb917fb6a1]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: allow drivers to request minimum alignment for passed buffer
Masahiro Yamada [Tue, 21 Nov 2017 17:38:27 +0000 (02:38 +0900)]
mtd: nand: allow drivers to request minimum alignment for passed buffer

In some cases, nand_do_{read,write}_ops is passed with unaligned
ops->datbuf.  Drivers using DMA will be unhappy about unaligned
buffer.

The new struct member, buf_align, represents the minimum alignment
the driver require for the buffer.  If the buffer passed from the
upper MTD layer does not have enough alignment, nand_do_*_ops will
use bufpoi.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 477544c62a84d3bacd9f90ba75ffc16c04d78071]

7 years agomtd: nand: Wait for PAGEPROG to finish in drivers setting NAND_ECC_CUSTOM_PAGE_ACCESS
Boris Brezillon [Tue, 21 Nov 2017 17:38:26 +0000 (02:38 +0900)]
mtd: nand: Wait for PAGEPROG to finish in drivers setting NAND_ECC_CUSTOM_PAGE_ACCESS

Drivers setting NAND_ECC_CUSTOM_PAGE_ACCESS are supposed to handle the
full read/write page sequence, and waiting for a page to actually be
programmed is part of this write-page sequence.
This is also what is done in ->write_oob_xxx() hooks, so let's do that in
->write_page_xxx() as well to make it consistent.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 41145649f4acb30249b636b945053db50c9331c5]
[masahiro:
 There is no driver setting NAND_ECC_CUSTOM_PAGE_ACCESS in U-Boot.
 No driver is affected by this change.]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Drop the ->errstat() hook
Boris Brezillon [Tue, 21 Nov 2017 17:38:25 +0000 (02:38 +0900)]
mtd: nand: Drop the ->errstat() hook

The ->errstat() hook is no longer implemented NAND controller drivers.
Get rid of it before someone starts abusing it.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 7d135bcced20be2b50128432c5426a7278ec4f6d]
[masahiro: modify davinci_nand.c for U-Boot]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Drop unused cached programming support
Boris Brezillon [Tue, 21 Nov 2017 17:38:24 +0000 (02:38 +0900)]
mtd: nand: Drop unused cached programming support

Cached programming is always skipped, so drop the associated code until
we decide to really support it.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 0b4773fd1649e0d418275557723a7ef54f769dc9]
[masahiro: modify davinci_nand.c for U-Boot]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: add mtd_ooblayout_xxx() helper functions
Boris Brezillon [Tue, 21 Nov 2017 17:38:23 +0000 (02:38 +0900)]
mtd: add mtd_ooblayout_xxx() helper functions

In order to make the ecclayout definition completely dynamic we need to
rework the way the OOB layout are defined and iterated.

Create a few mtd_ooblayout_xxx() helpers to ease OOB bytes manipulation
and hide ecclayout internals to their users.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 75eb2cec251fda33c9bb716ecc372819abb9278a]
[masahiro:
 cherry-pick more code from adbbc3bc827eb1f43a932d783f09ba55c8ec8379]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Support controllers with custom page
Marc Gonzalez [Tue, 21 Nov 2017 17:38:22 +0000 (02:38 +0900)]
mtd: nand: Support controllers with custom page

If your controller already sends the required NAND commands when
reading or writing a page, then the framework is not supposed to
send READ0 and SEQIN/PAGEPROG respectively.

Signed-off-by: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 3371d663bb4579f1b2003a92162edd6d90edd089]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Add a few more timings to nand_sdr_timings
Boris Brezillon [Tue, 21 Nov 2017 17:38:21 +0000 (02:38 +0900)]
mtd: nand: Add a few more timings to nand_sdr_timings

Add the tR_max, tBERS_max, tPROG_max and tCCS_min timings to the
nand_sdr_timings struct.
Assign default/safe values for the statically defined timings, and
extract them from the ONFI parameter table if the NAND is ONFI
compliant.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Tested-by: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
[Linux commit: 204e7ecd47e26cc12d9e8e8a7e7a2eeb9573f0ba
 Fixup commit: 6d29231000bbe0fb9e4893a9c68151ffdd3b5469]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Fix data interface configuration logic
Boris Brezillon [Tue, 21 Nov 2017 17:38:20 +0000 (02:38 +0900)]
mtd: nand: Fix data interface configuration logic

When changing from one data interface setting to another, one has to
ensure a specific sequence which is described in the ONFI spec.

One of these constraints is that the CE line has go high after a reset
before a command can be sent with the new data interface setting, which
is not guaranteed by the current implementation.

Rework the nand_reset() function and all the call sites to make sure the
CE line is asserted and released when required.

Also make sure to actually apply the new data interface setting on the
first die.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Fixes: d8e725dd8311 ("mtd: nand: automate NAND timings selection")
Reviewed-by: Sascha Hauer <s.hauer@pengutronix.de>
Tested-by: Marc Gonzalez <marc_gonzalez@sigmadesigns.com>
[Linux commit: 73f907fd5fa56b0066d199bdd7126bbd04f6cd7b]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: automate NAND timings selection
Boris Brezillon [Tue, 21 Nov 2017 17:38:19 +0000 (02:38 +0900)]
mtd: nand: automate NAND timings selection

The NAND framework provides several helpers to query timing modes supported
by a NAND chip, but this implies that all NAND controller drivers have
to implement the same timings selection dance. Also currently NAND
devices can be resetted at arbitrary places which also resets the timing
for ONFI chips to timing mode 0.

Provide a common logic to select the best timings based on ONFI or
->onfi_timing_mode_default information. Hook this into nand_reset()
to make sure the new timing is applied each time during a reset.

NAND controller willing to support timings adjustment should just
implement the ->setup_data_interface() method.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
[Linux commit: d8e725dd831186a3595036b2b1df9f68cbc6efa3]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Expose data interface for ONFI mode 0
Sascha Hauer [Tue, 21 Nov 2017 17:38:18 +0000 (02:38 +0900)]
mtd: nand: Expose data interface for ONFI mode 0

The nand layer will need ONFI mode 0 to use it as timing mode
before and right after reset.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 6e1f9708dbf3c50a8da93c1952a01a7a2acb5e66]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: convert ONFI mode into data interface
Sascha Hauer [Tue, 21 Nov 2017 17:38:17 +0000 (02:38 +0900)]
mtd: nand: convert ONFI mode into data interface

struct nand_data_interface is the designated type to pass to
the NAND drivers to configure the timing. To simplify further
patches convert the onfi_sdr_timings array from type struct
nand_sdr_timings nand_data_interface.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: b1dd3ca203fccd111926c3f6ac59bf903ec62b05]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Introduce nand_data_interface
Sascha Hauer [Tue, 21 Nov 2017 17:38:16 +0000 (02:38 +0900)]
mtd: nand: Introduce nand_data_interface

Currently we have no data structure to fully describe a NAND timing.
We only have struct nand_sdr_timings for NAND timings in SDR mode,
but nothing for DDR mode and also no container to store both types
of timing.
This patch adds struct nand_data_interface which stores the timing
type and a union of different timings. This can be used to pass to
drivers in order to configure the timing.
Add kerneldoc for struct nand_sdr_timings while touching it anyway.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: eee64b700e26b9bcc6fce024681c31f5e12271fc]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Create a NAND reset function
Sascha Hauer [Tue, 21 Nov 2017 17:38:15 +0000 (02:38 +0900)]
mtd: nand: Create a NAND reset function

When NAND devices are resetted some initialization may have to be done,
like for example they have to be configured for the timing mode that
shall be used. To get a common place where this initialization can be
implemented create a nand_reset() function. This currently only issues
a NAND_CMD_RESET to the NAND device. The places issuing this command
manually are replaced with a call to nand_reset().

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 2f94abfe35b210e7711af9202a3dcfc9e779219a]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: remove unnecessary 'extern' from function declarations
Sascha Hauer [Tue, 21 Nov 2017 17:38:14 +0000 (02:38 +0900)]
mtd: nand: remove unnecessary 'extern' from function declarations

'extern' is not necessary for function declarations. To prevent
people from adding the keyword to new declarations remove the
existing ones.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
[Linux commit: 79022591839f110f465cac0223e117b91d47d5db]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: Add an option to maximize the ECC strength
Boris Brezillon [Tue, 21 Nov 2017 17:38:13 +0000 (02:38 +0900)]
mtd: nand: Add an option to maximize the ECC strength

The generic NAND DT bindings allows one to tweak the ECC strength and
step size to their need. It can be used to lower the ECC strength to
match a bootloader/firmware config, but might also be used to get a better
reliability.

In the latter case, the user might want to use the maximum ECC strength
without having to explicitly calculate the exact value (this value not
only depends on the OOB size, but also on the NAND controller, and can
be tricky to extract).

Add a generic 'nand-ecc-maximize' DT property and the associated
NAND_ECC_MAXIMIZE flag, to let ECC controller drivers select the best
ECC strength and step-size on their own.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Rob Herring <robh@kernel.org>
[Linux commit: ba78ee00e1ff84de9b3ad33edbd3ec599099ee82]
[masahiro: of_property_read_bool -> fdt_getprop for U-Boot]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agomtd: nand: add onfi_* stubs in case ONFI_DETECTION is disabled
Masahiro Yamada [Tue, 21 Nov 2017 17:38:12 +0000 (02:38 +0900)]
mtd: nand: add onfi_* stubs in case ONFI_DETECTION is disabled

Add stubs to the header in case CONFIG_SYS_NAND_ONFI_DETECTION is
disabled.  This is much easier than adding around #ifdef to the
caller side.

Also, I removed the #ifdef around onfi_params.  In Linux, onfi_params
and jedec_params are unified as union.  It will be the right thing
to do.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agobitops: collect BIT macros to include/linux/bitops.h
Masahiro Yamada [Tue, 21 Nov 2017 17:38:11 +0000 (02:38 +0900)]
bitops: collect BIT macros to include/linux/bitops.h

Same macros are defined in various places.  Collect them into
include/linux/bitops.h like Linux.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agoARM: openrd: set CONFIG_LOGLEVEL to 2
Masahiro Yamada [Tue, 28 Nov 2017 12:23:20 +0000 (21:23 +0900)]
ARM: openrd: set CONFIG_LOGLEVEL to 2

These boards are on the boundary of "u-boot-nodtb.bin exceeds file
size limit" error.

Reduce the log-level to save memory footprint.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
7 years agoMerge git://git.denx.de/u-boot-socfpga
Tom Rini [Tue, 28 Nov 2017 14:30:36 +0000 (09:30 -0500)]
Merge git://git.denx.de/u-boot-socfpga

7 years agoMerge git://git.denx.de/u-boot-usb
Tom Rini [Tue, 28 Nov 2017 14:30:27 +0000 (09:30 -0500)]
Merge git://git.denx.de/u-boot-usb

7 years agotest: Correct operation when tests pass
Simon Glass [Mon, 27 Nov 2017 03:25:55 +0000 (20:25 -0700)]
test: Correct operation when tests pass

When tests pass an error message is printed because of a variable that is
not initialised. Fix this.

Signed-off-by: Simon Glass <sjg@chromium.org>
7 years agospl: TI: Do not default to SPL_FIT_IMAGE_TINY being enabled
Tom Rini [Tue, 28 Nov 2017 01:07:47 +0000 (20:07 -0500)]
spl: TI: Do not default to SPL_FIT_IMAGE_TINY being enabled

This option prevents booting on am335x_evm at least along with most
likely other platforms.

Fixes: 337bbb629777 ("spl: fit: add SPL_FIT_IMAGE_TINY config to reduce code-size")
Signed-off-by: Tom Rini <trini@konsulko.com>
7 years agousb: r8a66597: Add support for RZ/A series
Chris Brandt [Mon, 27 Nov 2017 19:04:10 +0000 (14:04 -0500)]
usb: r8a66597: Add support for RZ/A series

While the USB HW in the RZ/A is basically the same, there are some
differences from the original versions that were in the SH SoCs.

Signed-off-by: Chris Brandt <chris.brandt@renesas.com>
7 years agoMerge git://www.denx.de/git/u-boot-imx
Tom Rini [Mon, 27 Nov 2017 12:22:11 +0000 (07:22 -0500)]
Merge git://www.denx.de/git/u-boot-imx

Signed-off-by: Tom Rini <trini@konsulko.com>
7 years agoconfigs: icore-rqs: Enable falcon mode
Jagan Teki [Mon, 20 Nov 2017 18:32:18 +0000 (00:02 +0530)]
configs: icore-rqs: Enable falcon mode

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoengicam: imx6q: Return mmc dev 0 for icore
Jagan Teki [Mon, 20 Nov 2017 18:32:17 +0000 (00:02 +0530)]
engicam: imx6q: Return mmc dev 0 for icore

icorem6 has sd on usdhci1 which is devno 0 so return proper
devno from board_mmc_get_env_dev for icorem6 and icorem_6rqs

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoi.MX6: engicam: Add imx6q/imx6ul boards for existing boards
Jagan Teki [Mon, 20 Nov 2017 18:32:16 +0000 (00:02 +0530)]
i.MX6: engicam: Add imx6q/imx6ul boards for existing boards

Add new board names for existing board support
imx6q - icore and icore_rqs boards
imx6ul - geam6ul and isiot boards

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoboard: icore-rqs: Fix mmc get env device
Jagan Teki [Mon, 20 Nov 2017 18:32:15 +0000 (00:02 +0530)]
board: icore-rqs: Fix mmc get env device

As per USDHC boot eFUSE descriptions:
USDHC3 => devno 2
USDHC4 => devno 3

Linux will detect mmc0, mmc1, mmc2 based on the status
"okay" on usdhc so imx6qdl-icore-rqs.dtsi has enabled
usdhc1, usdhc3 and usdhc4.But U-Boot can detect based
on the aliases so add mmc1, mmc2 for usdhc3 and usdhc4
respectively and return the board_mmc_get_env_dev
by subtracting -1

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoi.MX6UL: icore: Add SPL_OF_CONTROL support
Jagan Teki [Mon, 20 Nov 2017 18:32:14 +0000 (00:02 +0530)]
i.MX6UL: icore: Add SPL_OF_CONTROL support

Add OF_CONTROL support for SPL code.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agopinctrl: imx6ul: Fix pinctrl data overlapped with DT area
Jagan Teki [Mon, 20 Nov 2017 18:32:13 +0000 (00:02 +0530)]
pinctrl: imx6ul: Fix pinctrl data overlapped with DT area

before relocation pinctrl data BSS is overlapping DT area,
when .data is using uninitialized global variable,
imx6_pinctrl_soc_info. So assign them flags ZERO_OFFSET_VALID
to prevent BSS overlap

Suggested-by: Lokesh Vutla <lokeshvutla@ti.com>
Reported-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoi.MX6Q: icore: Add SPL_OF_CONTROL support
Jagan Teki [Mon, 20 Nov 2017 18:32:12 +0000 (00:02 +0530)]
i.MX6Q: icore: Add SPL_OF_CONTROL support

Add OF_CONTROL support for SPL code.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoi.MX6Q: icorem6: Move spl load fit to common/spl
Jagan Teki [Mon, 20 Nov 2017 18:32:11 +0000 (00:02 +0530)]
i.MX6Q: icorem6: Move spl load fit to common/spl

Move spl load fit code into common/spl

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agoi.MX6: engicam: Move set_fdt_file to common
Jagan Teki [Mon, 20 Nov 2017 18:32:10 +0000 (00:02 +0530)]
i.MX6: engicam: Move set_fdt_file to common

setenv_fdt_file to common code and set dtb based on
CONFIG_DEFAULT_DEVICE_TREE and cpu_type.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
7 years agomx6sxsabresd: Use PARTUUID to specify the rootfs location
Fabio Estevam [Wed, 22 Nov 2017 13:15:28 +0000 (11:15 -0200)]
mx6sxsabresd: Use PARTUUID to specify the rootfs location

mx6sxsabresd can run different kernel versions, such as NXP 4.1 or mainline.

Currently the rootfs location is passed via mmcblk number and the
problem with this approach is that the mmcblk number for the SD
card changes depending on the kernel version.

In order to avoid such issue, use the UUID method to specify the
rootfs location.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Reviewed-by: Stefano Babic <sbabic@denx.de>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
7 years agoimx: Remove boolean parameter from wdog powerdown function
Fabio Estevam [Thu, 23 Nov 2017 12:55:33 +0000 (10:55 -0200)]
imx: Remove boolean parameter from wdog powerdown function

imx_set_wdog_powerdown() is always used to disable the power down
enable bit, so remove the boolean parameter of the function.

It is also a bit strange to write a boolean value into registers,
so this new version makes explicit that we are writing 0.

While at it, rename it to imx_wdog_disable_powerdown().

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
7 years agoimx: Also clear powerdown enable bit for WDOG3 on i.MX6ULL
Fabio Estevam [Thu, 23 Nov 2017 11:18:54 +0000 (09:18 -0200)]
imx: Also clear powerdown enable bit for WDOG3 on i.MX6ULL

Clear powerdown enable bit for WDOG3 on i.MX6ULL to avoid unwanted
kernel reboots.

Suggested-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Stefano Babic <sbabic@denx.de>
7 years agomx6ull: Fix WDOG3 base address
Fabio Estevam [Thu, 23 Nov 2017 11:18:53 +0000 (09:18 -0200)]
mx6ull: Fix WDOG3 base address

i.MX6ULL has the same WDOG3 base address as i.MX6UL, so take this
into account in the base address definition.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Stefano Babic <sbabic@denx.de>
7 years agoMerge git://git.denx.de/u-boot-rockchip
Tom Rini [Sun, 26 Nov 2017 23:10:21 +0000 (18:10 -0500)]
Merge git://git.denx.de/u-boot-rockchip

7 years agorockchip: defconfig: puma-rk3399: bypass ADC-based boot_mode check
Philipp Tomsich [Fri, 24 Nov 2017 13:44:59 +0000 (14:44 +0100)]
rockchip: defconfig: puma-rk3399: bypass ADC-based boot_mode check

The boot (and fallback/emergency boot) concept for the RK3399-Q7
differs from Rockchip's reference platforms.

On the RK3399-Q7, some of this functionality is present in the
bootloader itself (and configurable); some is backed in hardware by
the Qseven BIOS_DISABLE signal to invoke the final stages of fallbacks
(i.e. either an external boot bypassing on-module memories or falling
back to the BROM for USB recovery).

In summary: the ADC-based boot_mode check does not apply for the
RK3399-Q7 and we therefore disable it (in this commit) by setting
CONFIG_BOOT_MODE_REG to 0.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
7 years agorockchip: fix turning off boot-mode via Kconfig
Philipp Tomsich [Fri, 24 Nov 2017 13:44:58 +0000 (14:44 +0100)]
rockchip: fix turning off boot-mode via Kconfig

The ROCKCHIP_BOOT_MODE_REG option defaults to a hex value, so 0 will
show as 0x0 if a default is provided and changed via Kconfig.
However, it still will show as 0, if no default is given.

Consequently, the "is set to something other than 0" test in a
Makefile is cumbersome.  Instead this check can easily be performed in
the C-code.

This removes the ifeq-check from mach-rockchip/Makefile, adds a
matching #if-check to boot_mode.c and fixes resulting link issues (if
boot_mode.o was not included due to the Makefile check) by defining a
stub function (in case the functionality is not built in) for
setup_boot_mode in boot_mode.c.

Fixes: e306779 (rockchip: make boot_mode related codes reused across all platforms)
Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
7 years agofdt: Add compatible strings for Arria 10
Tien Fong Chee [Mon, 25 Sep 2017 08:40:03 +0000 (16:40 +0800)]
fdt: Add compatible strings for Arria 10

Add compatible strings for Intel Arria 10 SoCFPGA device.

Signed-off-by: Tien Fong Chee <tien.fong.chee@intel.com>
7 years agoarm: socfpga: Enhance FPGA program write rbf data with size >= 4 bytes
Tien Fong Chee [Mon, 25 Sep 2017 08:40:01 +0000 (16:40 +0800)]
arm: socfpga: Enhance FPGA program write rbf data with size >= 4 bytes

Existing FPGA program write is always assume RBF data >= 32 bytes, so
any rbf data less than 32 bytes writing to FPGA would be failed.
This patch enhances the FPGA program write to support rbf data with
size >= 4 bytes.

Signed-off-by: Tien Fong Chee <tien.fong.chee@intel.com>
7 years agoARM: socfpga: add bindings doc for arria10 fpga manager
Tien Fong Chee [Mon, 25 Sep 2017 08:39:57 +0000 (16:39 +0800)]
ARM: socfpga: add bindings doc for arria10 fpga manager

This DT binding doc is porting from Linux DT binding doc.
commit 1adcbea4201a6852362aa5ece573f1f169b28113

Add a device tree bindings document for the SoCFPGA Arria10
FPGA Manager driver.

Signed-off-by: Alan Tull <atull@opensource.altera.com>
Acked-by: Rob Herring <robh@kernel.org>
Acked-By: Moritz Fischer <moritz.fischer@ettus.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Tien Fong Chee <tien.fong.chee@intel.com>
7 years agoMAINTAINERS: Update maintainer for USB xHCI
Bin Meng [Mon, 6 Nov 2017 15:39:01 +0000 (07:39 -0800)]
MAINTAINERS: Update maintainer for USB xHCI

As discussed with Marek, this adds myself as the maintainer to
offload the USB xHCI part.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
7 years agoTrigger watchdog before calling usb_gadget_handle_interrupts
Vincent Prince [Mon, 23 Oct 2017 09:16:35 +0000 (11:16 +0200)]
Trigger watchdog before calling usb_gadget_handle_interrupts

This prevents board resets when calling sdp command on boards which have a watchdog.

Signed-off-by: Vincent Prince <vincent.prince.fr@gmail.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Stefan Agner <stefan.agner@toradex.com>
7 years agousb: gadget: storage: Increase FSG_BUFLEN
Marek Vasut [Thu, 19 Oct 2017 19:45:53 +0000 (21:45 +0200)]
usb: gadget: storage: Increase FSG_BUFLEN

Increase the buffer length to be just above maximum permissible value
of 128 kiB . This increases the performance of the UMS and alike by a
factor of 2 - 2.5 as the buffers are less fragmented.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Lukasz Majewski <lukma@denx.de>
7 years agousb: hub: identify the hub-device to usb_hub_reset_devices
Philipp Tomsich [Wed, 22 Nov 2017 16:15:17 +0000 (17:15 +0100)]
usb: hub: identify the hub-device to usb_hub_reset_devices

When usb_hub_reset_devices is called, it should be passed both an
indicator which hub it should operate on and what port number (local
to that hub) should be reset.

Previously, the usb_hub.c code did not include such context and
always started resets from port number 1, performing multiple
reset-requests for the same devices:

       /*
        * Reset any devices that may be in a bad state when applying
        * the power.  This is a __weak function.  Resetting of the devices
        * should occur in the board file of the device.
        */
       for (i = 0; i < dev->maxchild; i++)
              usb_hub_reset_devices(i + 1);

This adds an additional 'hub' parameter to usb_hub_reset_devices
that provides the context to fully qualify the port-number in.

Existing implementations are changed to accept and ignore the new
parameter.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Tested-by: Jakob Unterwurzacher <jakob.unterwurzacher@theobroma-systems.com>
7 years agousb: ehci: do not invalidate a NULL buffer
Dirk Behme [Fri, 17 Nov 2017 14:28:36 +0000 (15:28 +0100)]
usb: ehci: do not invalidate a NULL buffer

Its a valid use case to call ehci_submit_async() with a NULL buffer
with length 0. E.g. from usb_set_configuration().

As invalidate_dcache_range() isn't able to judge if the address
NULL is valid or not (depending on the SoC hardware configuration it
might be valid) do the check in ehci_submit_async() as here we know
that we don't have to invalidate such a buffer.

Signed-off-by: Dirk Behme <dirk.behme@de.bosch.com>
7 years agodm: usb: ehci: avoid possible NULL dereference
Heinrich Schuchardt [Mon, 20 Nov 2017 18:33:39 +0000 (19:33 +0100)]
dm: usb: ehci: avoid possible NULL dereference

Currently we check in ehci_shutdown() if ctrl is NULL after
dereferencing it.

Before this we have already dereferenced ctrl, ctrl->hccr,
and ctrl->hcor in ehci_get_portsc_register(), ehci_submit_root(),
and hci_common_init().

A better approach is to already check ctrl, ctrl->hccr, and ctrl->hcor
during the initialization in ehci_register() and usb_lowlevel_init()
and signal an error here via the return code.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
7 years agousb: xhci: Add Renesas R-Car xHCI driver
Marek Vasut [Sun, 15 Oct 2017 13:01:29 +0000 (15:01 +0200)]
usb: xhci: Add Renesas R-Car xHCI driver

Add firmware V3, firmware loader and XHCI glue for the Renesas R-Car
Gen3 SoCs XHCI controller. Thus far only the R-Car Gen3 R8A7795 ES2.0+
and R8A7796 are supported.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
Cc: Bin Meng <bmeng.cn@gmail.com>
7 years agousb: ehci: Fix accessors for big-endian platforms and descriptors
Alexey Brodkin [Fri, 17 Nov 2017 13:26:30 +0000 (16:26 +0300)]
usb: ehci: Fix accessors for big-endian platforms and descriptors

Commit 9000eddbae0d ("drivers/usb/ehci: Use platform-specific accessors")
broke USB 2.0 on big-endian platforms because for them writel/readl()
does automatic conversion of BE data to LE.

Proper implementation requires to use "raw" variant of these accessors
which read/write data without messing with endianess.

While at it replace cpu_to_be32() to be32_to_cpu() in readl() to
keep sane semantics.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Marek Vasut <marex@denx.de>
Reported-by: Vladimir Boroda <boroda@yahoo.com>
7 years agocmd: usb: ignore blk, emulation devices in usb tree/info display
Suneel Garapati [Tue, 24 Oct 2017 00:28:40 +0000 (17:28 -0700)]
cmd: usb: ignore blk, emulation devices in usb tree/info display

Usb tree/info commands iterate over all usb uclass devices recursively.
Blk uclass devices based on struct blk_desc are created for mass storage
device, treating them as usb uclass devices based on struct usb_device
and referencing usb config interface descriptors cause crash.
To fix, ignore blk and usb_emul uclass devices in usb_show_info
and usb_tree_graph. Also avoid addition of preamble for blk uclass
child devices, otherwise tree dump gets messed up.

Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agorockchip: pinctrl: rk3399: add support for I2C8
Philipp Tomsich [Wed, 22 Nov 2017 18:47:37 +0000 (19:47 +0100)]
rockchip: pinctrl: rk3399: add support for I2C8

The RK3399 has a total of 9 I2C controllers.  To support these, the
enum in periph.h is extended and the mapping from the IRQ numbers to
the peripheral-ids is extended to ensure that pinctrl requests are
passed through to the function configuring the I2C pins.

For I2C8, the pinctrl is implemented and tested (on a RK3399-Q7) using
communication with the FAN53555 connected on I2C8.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Tested-by: Klaus Goger <klaus.goger@theobroma-systems.com>
7 years agorockchip: clk: rk3399: change extract_bits to bitfield_extract
Philipp Tomsich [Wed, 22 Nov 2017 18:45:04 +0000 (19:45 +0100)]
rockchip: clk: rk3399: change extract_bits to bitfield_extract

The RK3399 clk driver still has a left-over use of extract_bits, which
can be replaced by using bitfield_extract from include/bitfield.h.
This rewrites the invocation to use the shared function.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
7 years agospl: fit: add SPL_FIT_IMAGE_TINY config to reduce code-size
Philipp Tomsich [Fri, 24 Nov 2017 12:26:03 +0000 (13:26 +0100)]
spl: fit: add SPL_FIT_IMAGE_TINY config to reduce code-size

A minor code-size increase from the changes for tracking the os-type
of FIT images and from infrastructure for recording the loadables into
the the loaded FDT, broke the builds for sun50i and some OMAP2+ devices.

This change adds a new config option (enabled by default for
MACH_SUN50I, MACH_SUN50I_H5 and ARCH_OMAP2PLUS) that does skips these
processing steps (bringing code size down to below the limit
again). The os-type is not evaluated, but assumed to be IH_OS_UBOOT
(i.e. taking the code-paths intended for backward-compatibility).

Note that enabling this config option precludes any useful downstream
processing, such as utilising a special calling convention for ATF or
OPTEE, based on the os-type of the loadables.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
7 years agorockchip: defconfig: lion-rk3368: sync up with SPL changes for ATF
Philipp Tomsich [Wed, 13 Sep 2017 19:29:43 +0000 (21:29 +0200)]
rockchip: defconfig: lion-rk3368: sync up with SPL changes for ATF

This tracks the SPL changes for ATF for the RK3368-uQ7:
 * renames ATF_SUPPORT to ATF
 * drops CONFIG_SPL_ATF_TEXT_BASE (now dynamically retrieved from
   the .itb file)

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agorockchip: defconfig: puma-rk3399: sync up with SPL changes for ATF
Philipp Tomsich [Wed, 13 Sep 2017 19:29:42 +0000 (21:29 +0200)]
rockchip: defconfig: puma-rk3399: sync up with SPL changes for ATF

This defconfig update makes use of the new features:
 * CONFIG_ROCKCHIP_SPL_RESERVE_IRAM is now set to 0, as there is no
   overlap between the M0 firmware and the ATF (we load this to DRAM
   and relocate it to its final location within the ATF)
 * tracks the ATF_SUPPORT -> ATF renaming

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agorockchip: board: lion-rk3368: update .its file
Philipp Tomsich [Wed, 13 Sep 2017 19:29:41 +0000 (21:29 +0200)]
rockchip: board: lion-rk3368: update .its file

For the RK3368-uQ7, we can now update the .its file to mark the
Trusted Firmware as out 'firmware' bootable and annotate both ATF and
U-Boot with an OS-type.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agorockchip: board: puma-rk3399: update .its file to use new features
Philipp Tomsich [Wed, 13 Sep 2017 19:29:40 +0000 (21:29 +0200)]
rockchip: board: puma-rk3399: update .its file to use new features

This commit updates the .its file for the RK3399-Q7 to use the new
features and demonstrates how to use those:
 * it marks the ATF as the 'firmware'
 * it tracks the OS-type for U-Boot and ATF
 * it loads the PMU (M0) firmware to DRAM and records the location
   to /fit-images (where our ATF reads it from)

With the handoff of the next-stage FDT to ATF in place, we can now use
this to pass information about the load addresses and names of each
loadables to ATF: now we can load the M0 firmware into DRAM and avoid
overwriting parts of the SPL stage.  This is achieved by changing our
.its-file to use an available area of DRAM as the load-address.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agorockchip: defconfig: firefly-rk3399: sync up with SPL changes for ATF
Philipp Tomsich [Wed, 13 Sep 2017 19:29:39 +0000 (21:29 +0200)]
rockchip: defconfig: firefly-rk3399: sync up with SPL changes for ATF

This tracks the SPL changes for ATF for the Firefly:
 * renames ATF_SUPPORT to ATF
 * drops CONFIG_SPL_ATF_TEXT_BASE

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: atf: drop the SPL_ATF_TEXT_BASE configuration item
Philipp Tomsich [Wed, 13 Sep 2017 19:29:37 +0000 (21:29 +0200)]
spl: atf: drop the SPL_ATF_TEXT_BASE configuration item

The SPL_ATF_TEXT_BASE configuration item has become obsolete.
Remove it from Kconfig.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: rename config item SPL_ATF_SUPPORT to SPL_ATF
Philipp Tomsich [Wed, 13 Sep 2017 19:29:36 +0000 (21:29 +0200)]
spl: rename config item SPL_ATF_SUPPORT to SPL_ATF

Having CONFIG_SPL_ATF seems more natural.
Rename it, while it it is easy and there's few boards that use it
(only RK3399 and RK3368 boards).

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: atf: introduce spl_invoke_atf and make bl31_entry private
Philipp Tomsich [Wed, 13 Sep 2017 19:29:35 +0000 (21:29 +0200)]
spl: atf: introduce spl_invoke_atf and make bl31_entry private

This adds a new interface spl_invoke_atf() that takes a spl_image_info
argument and then derives the necessary parameters for the ATF entry.
Based on the additional information recorded (into /fit-images) from
the FIT loadables, we can now easily locate the next boot stage.

We now pass a pointer to a FDT as the platform-specific parameter
pointer to ATF (so we don't run into the future headache of every
board/platform defining their own proprietary tag-structure), as
FDT access is already available in ATF.

With the necessary infrastructure in place, we can now update the
support for the ARM Trusted Firmware to dispatch into the
spl_invoke_atf function only if a IH_OS_ARM_TRUSTED_FIRMWARE image is
loaded.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: fit: implement recording of loadables into /fit-images
Philipp Tomsich [Wed, 13 Sep 2017 19:29:34 +0000 (21:29 +0200)]
spl: fit: implement recording of loadables into /fit-images

If a FDT was loaded (e.g. to append it to U-Boot image), we store it's
address and record information for all loadables into this FDT.  This
allows us to easily keep track of images for multiple privilege levels
(e.g. with ATF) or of firmware images preloaded into temporary
locations (e.g. PMU firmware that may overlap the SPL stage).

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: fit: implement fdt_record_loadable
Philipp Tomsich [Wed, 13 Sep 2017 19:29:33 +0000 (21:29 +0200)]
spl: fit: implement fdt_record_loadable

During the loading of more complex FIT images (e.g. when the invoked
next stage needs to find additional firmware for a power-management
core... or if there are multiple images for different privilege levels
started in parallel), it is helpful to create a record of what images
are loaded where: if a FDT is loaded for one of the next stages, it
can be used to convey the status and location of loadables.

This adds a fdt_record_loadable() function that can be invoked to
record the status of each loadable below the /fit-images path.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: fit: simplify logic for FDT loading for non-OS boots
Philipp Tomsich [Wed, 13 Sep 2017 19:29:32 +0000 (21:29 +0200)]
spl: fit: simplify logic for FDT loading for non-OS boots

To better support bootin through an ATF or OPTEE, we need to
streamline some of the logic for when the FDT is appended to an image:
depending on the image type, we'd like to append the FDT not at all
(the case for the OS boot), to the 'firmware' image (if it is a
U-Boot) or to one of the loadables (if the 'firmware' is an ATF, an
OPTEE, or some other image-type and U-Boot is listed in the
loadabled).

To achieve this goal, we drop the os_boot flag and track the type of
image loaded.  If it is of type IH_OS_U_BOOT, we append the FDT.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Acked-by: York Sun <york.sun@nxp.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: change load_addr and entry_point to uintptr_t
Philipp Tomsich [Wed, 13 Sep 2017 19:29:31 +0000 (21:29 +0200)]
spl: change load_addr and entry_point to uintptr_t

Mainly a stylistic change: convert the load_addr and entry_point
fields of struct spl_image_info to uintptr_t (from ulong).

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agospl: add a fdt_addr field to spl_image_info
Philipp Tomsich [Wed, 13 Sep 2017 19:29:30 +0000 (21:29 +0200)]
spl: add a fdt_addr field to spl_image_info

When loading a full U-Boot with detached device-tree using the SPL FIT
backend, we should store the address of the FDT loaded as part of the
SPL image info: this allows us to fixup the FDT with additional info
we may want to propagate onward.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agoimage: add IH_OS_ARM_TRUSTED_FIRMWARE for ARM Trusted Firmware
Philipp Tomsich [Wed, 13 Sep 2017 19:29:29 +0000 (21:29 +0200)]
image: add IH_OS_ARM_TRUSTED_FIRMWARE for ARM Trusted Firmware

To boot on ARMv8 systems with ARM Trusted Firmware, we need to
assemble an ATF-specific parameter structure and also provide the
address of the images started by ATF (e.g. BL3-3, which may be the
full U-Boot).

To allow us to identify an ARM Trusted Firmware contained in a FIT
image, this adds the necessary definitions.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
7 years agoMerge git://git.denx.de/u-boot-arc
Tom Rini [Fri, 24 Nov 2017 16:45:15 +0000 (11:45 -0500)]
Merge git://git.denx.de/u-boot-arc

7 years agoarc: cache: Add required NOPs after invalidation of instruction cache
Alexey Brodkin [Fri, 17 Nov 2017 13:02:17 +0000 (16:02 +0300)]
arc: cache: Add required NOPs after invalidation of instruction cache

As per ARC HS databook (see chapter 5.3.3.2) it is required to add
3 NOPs after each write to IC_IVIC which we do from now on.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Eugeniy Paltsev <paltsev@synopsys.com>
7 years agoarc: bootm: Move slave cores kick-starting under !fake
Alexey Brodkin [Fri, 17 Nov 2017 12:53:20 +0000 (15:53 +0300)]
arc: bootm: Move slave cores kick-starting under !fake

Currently slave cores will be kick-started even if we want
to dry run bootm which is not what we really want.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Eugeniy Paltsev <paltsev@synopsys.com>
7 years agoARC: HSDK: introduce CREG GPIO driver
Eugeniy Paltsev [Mon, 16 Oct 2017 13:21:32 +0000 (16:21 +0300)]
ARC: HSDK: introduce CREG GPIO driver

The HSDK can manage some pins via CREG registers block.

Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
7 years agoMerge git://git.denx.de/u-boot-dm
Tom Rini [Fri, 24 Nov 2017 03:48:35 +0000 (22:48 -0500)]
Merge git://git.denx.de/u-boot-dm

7 years agobinman: Return non-zero exit code on test failure
Simon Glass [Mon, 13 Nov 2017 04:52:29 +0000 (21:52 -0700)]
binman: Return non-zero exit code on test failure

Return exit code 1 when test fail so that callers can detect this.

Signed-off-by: Simon Glass <sjg@chromium.org>
7 years agobinman: Add add test for using an Intel MRC binary
Simon Glass [Mon, 13 Nov 2017 04:52:28 +0000 (21:52 -0700)]
binman: Add add test for using an Intel MRC binary

MRC (Memory Reference Code) is a binary blob used to set up the SDRAM
controller on some Intel boards. Add a test for this feature.

With this test coverage on binman is back up to 100%.

Signed-off-by: Simon Glass <sjg@chromium.org>
7 years agobinman: Add add test for SPL with a microcode pointer
Simon Glass [Mon, 13 Nov 2017 04:52:27 +0000 (21:52 -0700)]
binman: Add add test for SPL with a microcode pointer

Add a test for this feature. It allows SPL to hold a pointer to the
microcode block. This is used for 64-bit U-Boot on x86.

Signed-off-by: Simon Glass <sjg@chromium.org>
7 years agobinman: Add a test for x86-start16-spl
Simon Glass [Mon, 13 Nov 2017 04:52:26 +0000 (21:52 -0700)]
binman: Add a test for x86-start16-spl

This allows us to put the 16-bit x86 start-up code in SPL. Add a test for
it.

Signed-off-by: Simon Glass <sjg@chromium.org>