- Add and enable brcmnand driver on a number of relevant platforms.
Also add and enable LED drivers on more bcm platforms.
- Various ARMv8 fixes/improvements, including extending PSCI
functionality.
- fs_loader improvments
- Various FIT/SPL improvements
- PCI bugfixes
- Poplar platform ethernet support
- MediaTek MMC improvements
- Android boot improvements
According to Android image format [1], kernel image resides at 1 page
offset from the boot image address. Grab the magic number from there
and allow U-Boot to handle LZ4-compressed KNL binaries instead of
hardcoding compression type to IH_COMP_NONE. Other compression types,
if needed, can be added later.
Tested on H3ULCB-KF using the image detailed in [2].
[1] Excerpt from include/android_image.h
+-----------------+
| boot header | 1 page
+-----------------+
| kernel | n pages
+-----------------+
| ramdisk | m pages
+-----------------+
| second stage | o pages
+-----------------+
[2] => iminfo 4c000000
## Checking Image at 4c000000 ...
Android image found
kernel size: 85b9d1
kernel address: 48080000
ramdisk size: 54ddbc
ramdisk addrress: 4a180000
second size: 0
second address: 48000800
tags address: 48000100
page size: 800
os_version: 1200012a (ver: 0.9.0, level: 2018.10)
name:
cmdline: buildvariant=userdebug
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
If CONFIG_UNIT_TEST is enabled we should enable the individual tests by
default to ensure good test coverage.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
PCI_MSI_FLAGS_MASKBIT was missing from include file,
add it.
Signed-off-by: Ramon Fried <ramon.fried@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The fdt_setprop_u32() function does everything that we need, so we
really only use the function as a convenience wrapper, in which case it
can simply be a static inline function.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Testing has shown that the current DM implementation of a platform /
board specific translation offset, as its needed for the SPL on MVEBU
platforms is buggy. The translation offset is confingured too late,
after the driver bind functions are run. This may result in incorrect
address translations. With the current implementation its not possible
to configure the offset earlier, as the DM code has not run at all.
This patch now removed the set_/get_translation_offset() calls and
moves the translation offset into the GD variable translation_offset.
This variable will get used when CONFIG_TRANSLATION_OFFSET is enabled.
This option is enabled only for MVEBU on ARM32 platforms, where its
currenty needed and configured in the SPL.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Pierre Bourdon <delroth@gmail.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Simon Glass <sjg@chromium.org>
Cc: Heiko Schocher <hs@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Tested-by: Pierre Bourdon <delroth@gmail.com>
Tested-by: Baruch Siach <baruch@tkos.co.il>
Boards such as qemu-riscv, which receive their device tree at runtime,
for example from QEMU or firmware, are unable to add the appropriate
device tree properties to make devices available pre relocation.
Instead, they must rely on the DM_FLAG_PRE_RELOC flag to be set for the
required drivers.
Add the DM_FLAG_PRE_RELOC flag to the simple-bus driver to make devices
under it with drivers that have set the flag as well available pre
relocation for these boards.
Signed-off-by: Lukas Auer <lukas.auer@aisec.fraunhofer.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Enable this by default so that tracing can be inspected if enabled. This
cannot rely on the 'imply' in lib/Kconfig since this method of enabling
tracing relates on an environment variable (FTRACE) and does not use
Kconfig.
Signed-off-by: Simon Glass <sjg@chromium.org>
This converts the following to Kconfig:
CONFIG_TRACE_BUFFER_SIZE
CONFIG_TRACE_EARLY_SIZE
CONFIG_TRACE_EARLY
CONFIG_TRACE_EARLY_ADDR
Signed-off-by: Simon Glass <sjg@chromium.org>
At present many functions in this file return -1. Update them to return a
valid error code. Also tidy up the 'return' statements at the same time,
since these should have a blank line before them.
Signed-off-by: Simon Glass <sjg@chromium.org>
This function may be called from tracing code, since that code needs to
read the timer and this often requires calling do_div(), which calls
__div64_32(). If this function is instrumented it causes an infinite loop,
since emitting a trace record requests the time, which in turn emits a
trace record, etc.
Update the prototype to prevent instrumentation code being added.
Signed-off-by: Simon Glass <sjg@chromium.org>
These functions still use uint32_t and uint64_t but checkpatch now
requests that the kernel types be used instead. Update them as well as a
few resulting checkpatch errors.
Signed-off-by: Simon Glass <sjg@chromium.org>
This file should not include header files. They have already been included
by the time initcall.h is included. Also, document how to enable debugging
in this file.
Signed-off-by: Simon Glass <sjg@chromium.org>
It is possible for this to happen if something goes wrong very early in
the init sequence. Add a check for this.
Signed-off-by: Simon Glass <sjg@chromium.org>
These two buffers currently conflict if tracing is enabled. Move the
pre-console buffer and update the documentation.
Signed-off-by: Simon Glass <sjg@chromium.org>
Allocation routines were adjusted to ensure that the returned addresses
are a multiple of the page size, but the header code was not updated to
take account of this. These routines assume that the header size is the
same as the page size which is unlikely.
At present os_realloc() does not work correctly due to this bug. The only
user is the hostfs 'ls' command, and only if the directory contains a
unusually long filename, which likely explains why this bug was not
caught earlier.
Fix this by doing the calculations using the obtained page size.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present if one of the initcalls fails on sandbox the address printing
is not help, e.g.:
initcall sequence 0000557678967c80 failed at call 00005576709dfe1f (err=-96)
This is because U-Boot gets relocated high into memory and the relocation
offset (gd->reloc_off) does not work correctly for sandbox.
Add support for finding the base address of the text region (at least on
Linux) and use that to set the relocation offset. This makes the output
better:
initcall sequence 0000560775957c80 failed at call 0000000000048134 (err=-96)
Then you use can use grep to see which init call failed, e.g.:
$ grep 0000000000048134 u-boot.map
stdio_add_devices
Of course another option is to run it with a debugger such as gdb:
$ gdb u-boot
...
(gdb) br initcall.h:41
Breakpoint 1 at 0x4db9d: initcall.h:41. (2 locations)
Note that two locations are reported, since this function is used in both
board_init_f() and board_init_r().
(gdb) r
Starting program: /tmp/b/sandbox/u-boot
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
U-Boot 2018.09-00264-ge0c2ba9814-dirty (Sep 22 2018 - 12:21:46 -0600)
DRAM: 128 MiB
MMC:
Breakpoint 1, initcall_run_list (init_sequence=0x5555559619e0 <init_sequence_f>)
at /scratch/sglass/cosarm/src/third_party/u-boot/files/include/initcall.h:41
41 printf("initcall sequence %p failed at call %p (err=%d)\n",
(gdb) print *init_fnc_ptr
$1 = (const init_fnc_t) 0x55555559c114 <stdio_add_devices>
(gdb)
Signed-off-by: Simon Glass <sjg@chromium.org>
When using boot scripts it can become quite hard to understand
which commands are actually executed during bootup (e.g. where
is a kernel image loaded from or which DTB is in use).
Shell scripts suffer from a similar problem and many shells address
this problem with a command execution tracer (e.g. BASH has xtrace,
which can be enabled by "set -x").
This patch introduces a command tracer for U-Boot, which prints
every command with its arguments before it is executed.
Signed-off-by: Christoph Muellner <christoph.muellner@theobroma-systems.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Our platform doesn't store the DTB into the Android image second area,
but rather copies the DTB to RAM from a dedicated dtb.img partition [0],
prior to booting the Android image by calling bootm.
Similar to [1], we find it useful to just call 'bootm' and have the
right DTB being passed to OS (assuming its address has been previously
stored in 'fdtaddr' by calling `fdt addr <dtb-addr>`).
Booting Android with DTB from 'fdtaddr' will only occur if:
- No DTB is embedded in the second area of Android image
- 'fdtaddr' points to a valid DTB in RAM
[0] https://source.android.com/devices/architecture/dto/partitions
[1] https://patchwork.ozlabs.org/patch/1046652/
("Support boot Android image without address on bootm command")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Prepare for booting Android images which lack any DTB in the second
area by using 'fdtaddr' environment variable as source/address of FDT.
No functional/behavioral change expected in this patch.
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Increase the readability of boot_get_fdt().
No change in behavior is expected.
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The 'no_fdt' goto label was introduced by v2015.01 commit [0] and it
had two review stages [1-2]. The *documented* purpose behind commit [0]
is (excerpt from commit description):
> allows both FDT and non-FDT kernels to boot by making the
> third parameter to the bootm/bootz optional
While [1] and [2] share the same goal, they have very different
implementations:
- [1] was based on a very simple 'argc' check at function error out
with returning success to the caller if the third parameter was NOT
passed to bootm/bootz command. This approach had the downside of
returning success to the caller even in case of legitimate internal
errors, which should halt booting.
- [2] added the "no_fdt" label and several "goto no_fdt" statements.
This allowed to report the legitimate internal errors to the caller.
IOW the major difference between [1] and [2] is:
- [1] boot w/o FDT if FDT address is not passed to boot{m,z,*}
- [2] give *freedom* to the developer to boot w/o FDT from any
(more or less) arbitrary point in the function flow (and here
comes the peculiar aspect, which looks to be a leftover from [1])
with the precondition that the 3rd argument (FDT address) is NOT
provided to boot{m,z,*}. In practice, this means that only a subset
of "goto no_fdt" end up booting w/o FDT while the other subset is
returning an error to the caller.
This patch removes the peculiar behavior described above, such that
"goto no_fdt" performs really what it tells to the developer.
The motivation of this patch is to decrease the unneeded complexity
and increase the readability of boot_get_fdt().
[0] 48aead71c1 ("fdt: Allow non-FDT kernels to boot when CONFIG_OF_LIBFDT is defined")
[1] https://patchwork.ozlabs.org/patch/412923/
("[U-Boot,v1] fdt: Allow non-FDT kernels to boot when CONFIG_OF_LIBFDT is defined")
[2] https://patchwork.ozlabs.org/patch/415635/
("[U-Boot,v2] fdt: Allow non-FDT kernels to boot when CONFIG_OF_LIBFDT is defined")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Paranoid programming [1] lies at the foundation of proper software
development, but the repetitive zeroing-out of output arguments in the
context of the same function rather clutters the code and inhibits
further refactoring/optimization than is doing any good.
In boot_get_fdt(), we already perform zero/NULL-initialization of
*of_flat_tree and *of_size at the beginning of the function, so doing
the same at function error-out is redundant/superfluous.
Moreover, keeping the code unchanged might encourage the developers to
update *of_flat_tree and *of_size during some interim computations,
which is against the current design of boot_get_fdt(). Currently,
writing useful data into these arguments happens just before
successfully returning from boot_get_fdt() and it should better stay so.
[1] https://blog.regehr.org/archives/1106
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This fixes relaction isses with the PSCI_TABLE entries in
the psci_32_table and psci_64_table.
When using 32-bit adress pointers relocation was not being applied to
the tables, causing PSCI handlers to point to the un-relocated code
area. By using 64-bit data relocation is properly applied. The
handlers are thus in the "secure data" area, which is protected by
/memreserve/ in the FDT.
Signed-off-by: Lars Povlsen <lars.povlsen@microchip.com>
The cache flush of the kernel load area needs to be aligned outward to
the DMA cache alignment. The operations are simpler if we think of this
as aligning the start down, ALIGN_DOWN(load, ARCH_DMA_MINALIGN), and
aligning the end up, ALIGN(load_end, ARCH_DMA_MINALIGN), and then find
the length of the flushed region by subtracting the former from the
latter.
Cc: Tom Rini <trini@konsulko.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Trent Piepho <tpiepho@impinj.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When using memalign() in a scenario where U-Boot is configured for full
malloc support with simple malloc not explicitly enabled and before the
full malloc support is initialized, a memory block is being allocated
and returned without the alignment parameter getting honored.
Fix this issue by replacing the existing memalign pre-full malloc init
logic with a call to memalign_simple() this way ensuring proper alignment
of the returned memory block.
Fixes: ee038c58d5 ("malloc: Use malloc simple before malloc is fully initialized in memalign()")
Signed-off-by: Andreas Dannenberg <dannenberg@ti.com>
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
This patch adds a short message to the SPL NAND loader, which displays
the source and destinations addresses including the size of the
loaded image, like this:
U-Boot SPL 2019.04-rc3-00113-g486efd8aaf (Mar 15 2019 - 14:18:02 +0100)
Trying to boot from NAND
Loading U-Boot from 0x00040000 (size 0x000a0000) to 0x22900000
I find this message quite helpful - hopefully others do so as well.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Heiko Schocher <hs@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add support for MediaTek MT8516 SoC. This include the file
that will initialize the SoC after boot and its device tree.
Signed-off-by: Fabien Parent <fparent@baylibre.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add the implementation for the CLK_GATE_SETCLR_INV and
CLK_GATE_NO_SETCLR flags.
Signed-off-by: Fabien Parent <fparent@baylibre.com>
Acked-by: Ryder Lee <ryder.lee@mediatek.com>
We either need to use IS_ENABLED(CONFIG_FOO) or CONFIG_IS_ENABLED(FOO).
IS_ENABLE(FOO) will always return false.
This commit fixes the comparison by using the CONFIG_IS_ENABLED(FOO)
syntax.
Signed-off-by: Fabien Parent <fparent@baylibre.com>
Some MediaTek SoC need an additional clock "source_cg". Enable
this new clock. We reuse the same clock name as in the kernel.
Signed-off-by: Fabien Parent <fparent@baylibre.com>
Acked-by: Ryder Lee <ryder.lee@mediatek.com>
The 'phy' reset of gmac device in kernel device tree is not generic
enough for u-boot to use, so we need to overwrite the 'resets' property
as needed. With this device tree fixup and poplar_defconfig changes,
Ethernet starts working on Poplar board.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
It adds the driver for HIGMACV300 Ethernet controller found on HiSilicon
SoCs like Hi3798CV200. It's based on a downstream U-Boot driver, but
quite a lot of code gets rewritten and cleaned up to adopt driver model
and PHY API.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
It adds a Driver Model compatible reset driver for HiSlicon platform.
The driver implements a custom .of_xlate function, and uses .data field
as reset register offset and .id field as bit shift.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Some reset controllers support different polarities for reset operation,
so let's add a polarity field into struct reset_ctl.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
If a platform defines CONFIG_NR_DRAM_BANKS, each DRAM bank will be added
as a PCI region. The number of MAX_PCI_REGIONS therefore needs to scale
with the number of DRAM banks, otherwise we will end up with too little
space in the hose->regions array to store all system memory regions.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Make sure that we don't overflow the hose->regions array, otherwise we
would end up overwriting the hose->region_count field and cause mayhem
to ensue. Also print an error message when we'd be overflowing because
it indicates that there aren't enough regions available and the number
needs to be increased.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
With the migration to DM in SPL and the DT support, the
old legacy code is no longer neaded, so this patch removes it
Signed-off-by: Adam Ford <aford173@gmail.com>
Allow the led bcm6858 driver to be used on bcm63158.
They have the same led controller.
Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
