This function does not exist anymore. Drop it from the header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Writing tables is currently pretty opaque. Add a bit of debugging to the
process so we can see what tables are written and where they start/end in
memory.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Use the new binman memory-mapping function to access the VBT, to simplify
the code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Generation of this table can fail, so update the function to return an
error code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
The Intel Non-High-Definition-Audio Link Table (NHLT) table describes the
audio codecs and connections in a system. Various devices can contribute
information to produce the table.
Add functions to allow adding to the structure that is eventually written
to the ACPI tables. Also add the device-tree bindings.
Signed-off-by: Simon Glass <sjg@chromium.org>
In some cases an internal error may prevent this from working. Update the
function return value and report the error. At present the API for writing
tables does not easily support reporting errors, but once it is fully
updated to use a context pointer, this will be easier.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Call the new core function to inject ASL programmatically into the DSDT.
This is made up of fragments generated by devices that have the
inject_dsdt() method. The normal, compiled ASL file is added after this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Call the new core function to write the SSDT. This is made up of fragments
generated by devices that have the fill_ssdt() method.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Put this table before MCFG so that it matches the order that coreboot uses
when passing tables to Linux. This is a cosmetic change since the order of
the tables does not otherwise matter.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With DDR4, Intel SOCs take quite a long time to init their memory. During
this time, if the user is watching, it looks like SPL has hung. Add a
message in this case.
This works by adding a return code to fspm_update_config() that indicates
whether MRC data was found and a new property to the device tree.
Also add one more debug message while starting.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Tested-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
At present this enables a few arch-specific members of the global_data
struct which are otherwise not part of the struct. As a result we have to
use #ifdef in various places.
The cost of always having these in the struct is small. Adjust things so
that we can use compile-time code instead of #ifdefs.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
When the copy framebuffer is in use, we must also have the standard U-Boot
framebuffer available. Update the FSP driver to support this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Anatolij Gustschin <agust@denx.de>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
The APL FSP appears to leave the FPU in a bad state in that it has
registers in use. This causes an error when the next FPU operation is
performed.
Work around this by re-resetting the FPU after calling FSP-M. This allows
the freetype console to work correctly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
In the future if we have separate symbols for DM_SPI_FLASH and
SPL_DM_SPI_FLASH we will not always have function declarations available
for some DM calls. This in turn leads to build warnings but not
failures as the code isn't used and is discarded at link time.
Restructure things to not build code we won't use for TPL anyways.
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Rini <trini@konsulko.com>
Set this value in global_data so that it is reported correctly on x86
boards.
In fact, U-Boot allocates space for the frame buffer even though it is not
used. Then the FSP picks the address itself (e.g. 0xb0000000). So the
value set by U-Boot (high in memory with everything else that is
relocated), is not actually the correct value.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The change to avoid searching the device tree does not work on boards
wich don't have driver model set up this early, for example minnowmax.
Put back the old code (converted to livetree) as a fallback for these
devices. Also update the documentation.
This is tested on minnowmax, link, samus and coral.
Fixes: 87f1084a63 (x86: Adjust mrccache_get_region() to use livetree)
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com> (on Intel minnowmax)
The error code is often useful to figure out what is going on. Printing it
does not increase code size much, so print out the error and then hang.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The size is not actually used since it is present in the header. Drop this
parameter. Also tidy up error handling while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
We should not use typedefs in U-Boot. They cannot be used as forward
declarations which means that header files must include the full header to
access them.
Drop the typedef and rename the struct to remove the _s suffix which is
now not useful.
This requires quite a few header-file additions.
Signed-off-by: Simon Glass <sjg@chromium.org>
Move this header out of the common header. Network support is used in
quite a few places but it still does not warrant blanket inclusion.
Note that this net.h header itself has quite a lot in it. It could be
split into the driver-mode support, functions, structures, checksumming,
etc.
Signed-off-by: Simon Glass <sjg@chromium.org>
Virtually all callers of this function do the rounding on their own.
Some do it right, some don't. Instead of doing this in each caller,
do the rounding in efi_add_memory_map(). Change the size parameter
to bytes instead of pages and remove aligning and size calculation in
all callers.
There is no more need to make the original efi_add_memory_map() (which
takes pages as size) available outside the module. Thus rename it to
efi_add_memory_map_pg() and make it static to prevent further misuse
outside the module.
Signed-off-by: Michael Walle <michael@walle.cc>
Add missing comma in sunxi_display.c.
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
At present SPL only works on bare-metal builds. With a few tweaks it can
be used for coreboot also.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
At present this function copies U-Boot from the last 1MB of ROM. This is
not the right way to do it. Instead, the binman symbol should provide the
location.
But in any case the code should live in the caller,
spl_board_load_image(), so that the 64-bit jump function can be used
elsewhere. Move it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
When U-Boot is not the first-stage bootloader the interrupt and cache init
must be skipped, as well as init for various peripherals. Update the code
to add checks for this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
It is useful to be able to boot the same x86 image on a device with or
without a first-stage bootloader. For example, with chromebook_coral, it
is helpful for testing to be able to boot the same U-Boot (complete with
FSP) on bare metal and from coreboot. It allows checking of things like
CPU speed, comparing registers, ACPI tables and the like.
When U-Boot is not the first-stage bootloader much of this code is not
needed and can break booting. Add checks for this to the FSP code.
Rather than checking for the amount of available SDRAM, just use 1GB in
this situation, which should be safe. Using 2GB may run into a memory
hole on some SoCs.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
We always write three basic tables to ACPI at the start. Move this into
its own function, along with acpi_fill_header(), so we can write a test
for this code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Move this code to a generic location so that we can test it with sandbox.
This requires adding a few new fields to acpi_ctx, so drop the local
variables used in the original code.
Also use mapmem to avoid pointer-to-address casts which don't work on
sandbox.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
We don't actually support tables without an XSDT so we can drop this dead
code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Call the new core function to permit devices to write their own ACPI
tables. These tables will appear after all other tables.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
The current code uses an address but a pointer would result in fewer
casts. Also it repeats the alignment code in a lot of places so this would
be better done in a helper function.
Update write_acpi_tables() to make use of the new acpi_ctx structure,
adding a few helpers to clean things up.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
A device may want to write out ACPI tables to describe itself to Linux.
Add a method to permit this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
This file is potentially useful to other architectures saddled with ACPI
so move most of its contents to a common location.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
This header relates to ACPI and we are about to add some more ACPI
headers. Move this one into a new directory so they are together.
The header inclusion in pci_rom.c is not specific to x86 anymore, so drop
the #ifdef CONFIG_X86.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Wolfgang Wallner <wolfgang.wallner@br-automation.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
SPCR has no clue if the UART base clock speed is different to
the default one. However, the SPCR 1.04 defines baud rate 0 as
a preconfigured state of UART and OS is supposed not to touch
the configuration of the serial device.
Linux kernel supports that starting from v5.0, see commit
b413b1abeb21 ("ACPI: SPCR: Consider baud rate 0 as preconfigured state")
for the details.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
There is no need to have an assignment to NULL for XSDT pointer.
Therefore, no need to assign it when rsdt_address is not set.
Because of above changes we may decrease indentation level as well.
While here, drop unnecessary parentheses.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This function doesn't use uclass_find_first_device() correctly. Add a
check that the device is found so we don't try to read properties from a
NULL device.
The fixes booting on minnoxmax.
Fixes: 87f1084a63 ("x86: Adjust mrccache_get_region() to use livetree")
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present dm/device.h includes the linux-compatible features. This
requires including linux/compat.h which in turn includes a lot of headers.
One of these is malloc.h which we thus end up including in every file in
U-Boot. Apart from the inefficiency of this, it is problematic for sandbox
which needs to use the system malloc() in some files.
Move the compatibility features into a separate header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present panic() is in the vsprintf.h header file. That does not seem
like an obvious choice for hang(), even though it relates to panic(). So
let's put hang() in its own header.
Signed-off-by: Simon Glass <sjg@chromium.org>
[trini: Migrate a few more files]
Signed-off-by: Tom Rini <trini@konsulko.com>
We don't generally have enough space to run this, so don't build it into
TPL. This helps reduce the size of TPL.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The FSP-S may do this but at least for coral it does not. Set this up so
that graphics is not deathly slow.
It isn't clear whether the FSP is expected to set up MTRR. It is not
mentioned in the APL FSP document.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Add support for some important configuration options and FSP memory init.
The memory init uses swizzle tables from the device tree.
Support for the FSP_S binary is also included.
Bootstage timing is used for both FSP_M and FSP_S and memory-mapped SPI
reads.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This generic FSP file should include the generic FSP support header, not
the FSP1 version. Fix it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Both versions of FSP can use the same graphics support, so move it into
the common directory.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present the interrupt table is included in all phases of U-Boot. Allow
it to be omitted, e.g. in TPL, to reduce size.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Set up MTRRs for the FSP SDRAM regions to improve performance.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present with fsp a single DRAM bank is added which extends to the
whole size of memory. However there is typically only 2GB of memory
available below the 4GB boundary, and this is what is used by U-Boot while
running in 32-bit mode.
Scan the tables to set the banks correct. The first bank is set to memory
below 4GB, and the rest of memory is put into subsequent banks.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This function needs to be different for FSP2, so move the existing
function into the fsp1 directory. Since it is only called from one file,
drop it from the header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This function is only used within the implementation so make it static.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With Apollo Lake we need to support a normal cache, which almost never
changes and a much smaller 'variable' cache which changes every time.
Update the code to add a cache type, use an array for the caches and use a
for loop to iterate over the caches.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This function is a bit confusing at present due to the error handling.
Update it to remove the goto, returning errors as they happen.
While we are here, use hex for the data size since this is the norm in
U-Boot.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present we reuse the mrc_output char * to also point to the cache
record after it has been set up. This is confusing and doesn't save much
data space.
Add a new mrc_cache member instead.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
It is now possible to obtain the memory map for a SPI controllers instead
of having it hard-coded in the device tree. Update the code to support
this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Change the algorithm to first find the flash device then read the
properties using the livetree API. With this change the device is not
probed so this needs to be done in mrccache_save().
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This should take account of the end of the new cache record since a record
cannot extend beyond the end of the flash region. This problem was not
seen before due to the alignment of the relatively small amount of MRC
data.
But with Apollo Lake the MRC data is about 45KB, even if most of it is
zeroes.
Fix this bug and update the parameter name to be less confusing.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Move this function into init.h which seems to be designed for this sort
of thing. Also update the header to declare struct global_data so that it
can be included without global_data.h being needed.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Move these two functions into the irq_funcs.h header file. Also move
interrupt_handler_t as this is used by the irq_install_handler() function.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
These functions do not use driver model but are fairly widely used in
U-Boot. But it is not clear that they will use driver model anytime soon,
so we don't want to label them as 'legacy'.
Move them to a new irq_func.h header file. Avoid the name 'irq.h' since it
is widely used in U-Boot already.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
These functions belong in cpu_func.h. Another option would be cache.h
but that code uses driver model and we have not moved these cache
functions to use driver model. Since they are CPU-related it seems
reasonable to put them here.
Move them over.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
We already a message indicating that U-Boot is about to jump to SPL, so
make this one a debug() to reduce code size.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
We don't need to do this and it is done (in more detail) in U-Boot proper.
Drop this to save code space.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present we call spl_init() before identifying the CPU. This is not a
good idea - e.g. if bootstage is enabled then it will try to set up the
timer which works better if the CPU is identified.
Put explicit code at each entry pointer to identify the CPU.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
In TPL we try to minimise code size so do not include the PCI subsystem.
We can use fixed BARs and drivers can directly program the devices that
they need.
However we do need to bind the devices on the PCI bus and without PCI this
does not ordinarily happen. As a work-around, define a fake PCI bus which
does this binding, but no other PCI operations. This is a convenient way
to ensure that we can use the same device tree for TPL, SPL and U-Boot
proper:
TPL - CONFIG_TPL_PCI is not set (no auto-config, fake PCI bus)
SPL - CONFIG_SPL_PCI is set (no auto-config but with real PCI bus)
U-Boot - CONFIG_PCI is set (full auto-config after relocation)
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The x86 power unit handles power management. Support initing this device
which is modelled as a new type of system controller since there are no
operations needed.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Move the code to determine the size of a cache record into a function so
we can use it elsewhere in this file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present when these fail to boot there is no message, just a hang. Add a
panic so it is obvious that something when wrong.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With FSP2 we use MTRRs in U-Boot proper even though the 32-bit init
happens in TPL. Enable this, using a variable to try to make the
conditions more palatable.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present these uclasses assumes that they are used with a device tree.
Update them to support of-platdata as well.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
At present this hedaer is only available on x86. To allow sandbox to use
it for testing, move it to a common location.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The useable RAM is calculated when the RAM is inited. Save this value so
that it can be easily used in U-Boot proper.
Also save a pointer to the hob list so that it is accessible (before
relocation only) in U-Boot proper. This avoids having to scan it in SPL,
for everything U-Boot proper might need later.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: guard handoff_arch_save() with IS_ENABLED(CONFIG_USE_HOB)]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
When TPL is running, broadwell needs to do different init from SPL. There
is no need for this code to be in the generic x86 SPL file, so move it to
arch_cpu_init().
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This image loader works on systems where the flash is directly mapped to
the last part of the 32-bit address space. On recent Intel systems (such
as apollolake) this is not the case.
Reduce the priority of this loader so that another one can override it.
While we are here, rename the loader to BOOT_DEVICE_SPI_MMAP since
BOOT_DEVICE_BOARD is not very descriptive.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Use the standard hang() function when booting fails since this implements
the defined U-Boot behaviour for this situation.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
With FSP2 the non-volatile storage used by the FSP to init memory can be
split into a fixed piece (determined at compile time) and a variable piece
(determined at run time). Add support for reading the latter.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Some of this file can be shared between FSP1 and FSP2. Move it into a
shared file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Some of this file can be shared between FSP1 and FSP2. Move it into a
shared file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Most of the DRAM functionality can be shared between FSP1 and FSP2. Move
it into a shared file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: rebase the patch against u-boot-x86/next to get it applied cleanly]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
The comments in the FSP code use a different style from the rest of the
x86 code. I am not sure it this is intentional.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: fix 2 comment style issues]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Update a few #ifdefs to if() to improve build coverage.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: recover the codes that got wrongly deleted in dram_init()]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Given these exported function an fsp_ prefix since they are declared in an
fsp.h header.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Since there is now a new version of the FSP and it is incompatible with
the existing version, move the code into an fsp1 directory. This will
allow us to put FSP v2 code into an fsp2 directory.
Add a Kconfig which defines which version is in use.
Some of the code in this new fsp1/ directory is generic across both FSPv1
and FSPv2. Future patches will address this.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
This is reincarnation of the U-Boot
commit 3469bf4274
Author: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Date: Wed Jan 10 19:40:15 2018 +0200
x86: zImage: Propagate acpi_rsdp_addr to kernel via boot parameters
after upstream got eventually the Linux kernel
commit e6e094e053af75cbc164e950814d3d084fb1e698
Author: Juergen Gross <jgross@suse.com>
Date: Tue Nov 20 08:25:29 2018 +0100
x86/acpi, x86/boot: Take RSDP address from boot params if available
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Use efi_add_conventional_memory_map() to configure EFI conventional memory
properly with ram_top value. This will give 32-bit mode U-Boot proper
conventional memory regions even if e820 has an entry which is greater than
32-bit address space.
Signed-off-by: Aiden Park <aiden.park@intel.com>
Tested-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
[bmeng: fixed some typos in the commit message]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Every x86 platform provides board_get_usable_ram_top(), hence there
is no need to provide a weak version board_get_usable_ram_top(), not
to mention there is another weak version board_get_usable_ram_top()
in common/board_f.c.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Aiden Park <aiden.park@intel.com>
The setting up MTRRs have already been done in previous
Slim Bootloader stages.
Signed-off-by: Aiden Park <aiden.park@intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
FSP (CONFIG_HAVE_FSP) and Slim Bootloader (CONFIG_SYS_SLIMBOOTLOADER)
consume HOB (CONFIG_USE_HOB) data from the each HOB list pointer.
Add a common HOB library in lib/hob.c and include/asm/hob.h.
Signed-off-by: Aiden Park <aiden.park@intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Use existing EFI_GUID and efi_guid_t instead of struct efi_guid.
This is pre-work before making a common HOB library.
- Change 'struct efi_guid' to efi_guit_t
- Remove 'struct efi_guid'
- Define GUIDs with EFI_GUID() macro
- Use guidcmp() instead of compare_guid()
- Remove compare_guid()
Signed-off-by: Aiden Park <aiden.park@intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested on MinnowMax
Tested-by: Bin Meng <bmeng.cn@gmail.com>
This slimbootloader CPU type is to enable U-Boot as a payload which
runs on top of Slim Bootloader (https://github.com/slimbootloader).
The Slim Bootloader is designed with multi-stage architecture for
the execution from reset vector to OS booting, and supports QEMU,
Apollolake, Whiskeylake and Coffeelake platforms consuming Intel
FSP (https://github.com/IntelFsp) for silicon initialization
including CAR and memory initialization.
The Slim Bootloader generates new HOB (Hand Off Block) which are
serial port info, memory map info, performance data info and so on,
and passes it to a Payload. U-Boot as a payload will use these HOB
information for basic initialization such as serial console.
As an initial commit,
- Add CONFIG_SYS_SLIMBOOTLOADER to enable slimbootloader CPU type
- Add new arch/x86/cpu/slimbootloader directory with minimum codes
- Get hob_list pointer from Slim Bootloader
Signed-off-by: Aiden Park <aiden.park@intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Here is a stub function that generates an empty CSRT. If the target platform
provides acpi_fill_csrt() function, it will be used to populate the table.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This breaks Intel Edison to work. It gets laggish and unable to boot kernel.
Reverts commit 665cb18ea6 for now
till better solution will be proposed.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Bin Meng <bmeng.cn@gmail.com>
Add the required CPU code so that TPL builds correctly. Also update the
SPL code to deal with being booted from TPL.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This report is normally generated by U-Boot proper. Correct the condition
here so that it respects the Kconfig options for bootstage.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The MTRRs are normally set up in U-Boot proper, so avoid setting them up
in SPL as well.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
When the MRC cache fails to save it is useful to have some debugging info
to indicate what when wrong. Add some more debug() calls.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
When SPL is used to set up the memory controller we want to save the MRC
data in SPL to avoid needing to pass it up to U-Boot proper to save. Add a
function to handle that.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
We use binman to build the x86 image now. Update a comment which still
refers to ifdtool.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The i8254 timer control IO port (0x43) should be setup correctly
by using PIT counter 2 to generate beeps, however in U-Boot other
codes like TSC driver utilizes PIT for TSC frequency calibration
and configures the counter 2 to a different mode that does not
beep. Fix this by always ensuring the PIT counter 2 is correctly
initialized so that the i8254 beeper driver works as expected.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
We may not do an assumption that current console device is always a first
of UCLASS_SERIAL one.
For example, on properly described Intel Edison board the console UART
is a third one.
Use current serial device as described in global data.
Fixes: a61cbad78e ("dm: serial: Adjust serial_getinfo() to use proper API")
Cc: Simon Glass <sjg@chromium.org>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Adjust the code to allow beeping at different frequencies, using a
calculated value for timer 2.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
arch/x86/lib/string.c contains assembler implementations of memcpy(),
memmove(), and memset() written for i386. Don't use it on x86_64.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
All driver-model functions should have a device as the first parameter.
Update this function accordingly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
All driver-model functions should have a device as the first parameter.
Update this function accordingly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
mask_irq(), unmask_irq() and specific_eoi() are provided by the
i8259 PIC driver and should be wrapped with CONFIG_I8259_PIC.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Hannes Schmelzer <oe5hpm@oevsv.at>
Currently we support only relocations of type ELF64_R_TYPE or ELF32_R_TYPE.
We should be warned if other relocation types appear in the relocation
sections.
This type of message has helped to identify code overwriting a relocation
section before relocation and incorrect parsing of relocation tables.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Once we get a zero pointer from load_zimage(...) we must bunch out
instead of continue boot.
Signed-off-by: Hannes Schmelzer <hannes.schmelzer@br-automation.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Use CONFIG_IS_ENABLED(EFI_LOADER) to avoid explicitly checking CONFIG_SPL
too. This simplifies the conditional.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
These comments were copied from the Linux kernel driver in
drivers/platform/x86/intel_scu_ipc.c
Signed-off-by: Georgii Staroselskii <georgii.staroselskii@emlid.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This interface will be used to configure properly some pins on
Merrifield that are shared with SCU.
scu_ipc_raw_command() writes SPTR and DPTR registers before sending
a command to SCU.
This code has been ported from Linux work done by Andy Shevchenko.
Signed-off-by: Georgii Staroselskii <georgii.staroselskii@emlid.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
In int-ll64.h, we always use the following typedefs:
typedef unsigned int u32;
typedef unsigned long uintptr_t;
typedef unsigned long long u64;
This does not need to match to the compiler's <inttypes.h>.
Do not include it.
The use of PRI* makes the code super-ugly. You can simply use
"l" for printing uintptr_t, "ll" for u64, and no modifier for u32.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
As of today, the proposal of adding "acpi_rsdp_addr" to the kernel
boot protocol does not make its way to the kernel mainline. This
creates some confusion if we leave it in the U-Boot code base.
Remove it for now until we have a clear picture with kernel upstream.
Note this eventually does a partial revert to commit 3469bf4274
("x86: zImage: Propagate acpi_rsdp_addr to kernel via boot parameters")
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
At present Linux kernel loaded from U-Boot as an EFI payload does
not boot. This fills in kernel's boot params structure with the
required critical EFI information like system table address and
memory map stuff so that kernel can obtain essential data like
runtime services and ACPI table to boot.
With this patch, now U-Boot as an EFI payload becomes much more
practical: it is another option of kernel bootloader, ie, can be
a replacement for grub.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When we build with -fdata-sections we may end up with bss subsections. Our
linker script explicitly lists only a single consecutive bss section though.
Adapt the statement to also include subsections.
This fixes booting efi-x86_app_defconfig.
Signed-off-by: Alexander Graf <agraf@suse.de>
We need to know about x86 relocation definitions even in cases where
we don't officially build against the x86 target, such as with sandbox.
So let's move the x86 definitions into the common elf header, where all
other architectures already have them.
Signed-off-by: Alexander Graf <agraf@suse.de>
write_acpi_tables() currently touches ACPI hardware to switch to
ACPI mode at the end. Move such operation out of this function,
so that it only does what the function name tells us.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
acpi_find_fadt(), acpi_find_wakeup_vector() and enter_acpi_mode()
are something unrelated to ACPI tables generation. Move these to
a separate library.
This also fixes several style issues reported by checkpatch in the
original codes.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
In preparation for the reset driver conversion, eliminate the
reset_cpu() call in the FSP init path as it's too early for the
reset driver to work.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
On x86 traditional E820 table is used to pass the memory information
to kernel. With EFI loader we can build the EFI memory map from it.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Built without a ROM image with FSP (u-boot.rom), the U-Boot loader applies
the microcode update data block encoded in Device Tree to the bootstrap
processor but not passed to the other CPUs when multiprocessing is enabled.
If the bootstrap processor successfully performs a microcode update
from Device Tree, use the same data block for the other processors.
Signed-off-by: Ivan Gorinov <ivan.gorinov@intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: fixed build errors on edison and qemu-x86]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
At present the EFI application and payload support codes in the x86
directory is distributed in a hybrid way. For example, the Kconfig
options for both app and payload are in arch/x86/lib/efi/Kconfig,
but the source codes in the same directory get built only for
CONFIG_EFI_STUB.
This refactors the codes by consolidating all the EFI support codes
into arch/x86/cpu/efi, just like other x86 targets.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
UEFI specifies the calling convention used in Microsoft compilers;
first arguments of a function are passed in (%rcx, %rdx, %r8, %r9).
All other compilers use System V ABI by default, passing first integer
arguments of a function in (%rdi, %rsi, %rdx, %rcx, %r8, %r9).
These ABI also specify different sets of registers that must be preserved
across function calls (callee-saved).
GCC allows using the Microsoft calling convention by adding the ms_abi
attribute to a function declaration.
Current EFI implementation in U-Boot specifies EFIAPI for efi_main()
in the test apps but uses default calling convention in lib/efi.
Save efi_main() arguments in the startup code on x86_64;
use EFI calling convention for _relocate() on x86_64;
consistently use EFI calling convention for efi_main() everywhere.
Signed-off-by: Ivan Gorinov <ivan.gorinov@intel.com>
Reviewed-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Attempting to use a toolchain that is preconfigured to generate code
for the 32-bit architecture (i386), for example, the i386-linux-gcc
toolchain on kernel.org, to compile the 64-bit EFI payload does not
build. This updates the makefile fragments to ensure '-m64' is passed
to toolchain when building the 64-bit EFI payload stub codes.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The pinctrl_ich6 driver is currently unconditionally built for all
x86 boards. Let's use a Kconfig option to control the build.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This undocumented function relies on arch-specific code to declare a nop
weak version. Add the weak function in common code instead to avoid having
to duplicate the same function in each arch.
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
The efi selftest and the hello application require CRT0 and RELOC to be
built.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
We have a large number of places where while we historically referenced
gd in the code we no longer do, as well as cases where the code added
that line "just in case" during development and never dropped it.
Signed-off-by: Tom Rini <trini@konsulko.com>
If CONFIG_FRAMEBUFFER_SET_VESA_MODE is not set, don't switch
graphics card to VESA mode. This applies to both native mode
and emulator mode of running the VGA BIOS.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This changes 'struct e820entry' to 'struct e820_entry' to conform
with the coding style.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
This fixes the following checkpatch warning:
warning: Prefer 'unsigned int' to bare use of 'unsigned'
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
On x86 platforms, U-Boot does not pass Device Tree data to the kernel.
This prevents the kernel from using FDT loaded by U-Boot.
Read the working FDT address from the "fdtaddr" environment variable
and add a copy of the FDT data to the kernel setup_data list.
Signed-off-by: Ivan Gorinov <ivan.gorinov@intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: add #include <linux/libfdt.h> to zimage.c to fix build error]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Merge init_helpers.h in the new file init.h
with only prototypes for init_cache_f_r
used in common/board_f.c
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Only ARM and in some configs MIPS really implement arch_fixup_fdt().
Others just use the same boilerplate which is not good by itself,
but what's worse if we try to build with disabled CONFIG_CMD_BOOTM
and enabled CONFIG_OF_LIBFDT we'll hit an unknown symbol which was
apparently implemented in arch/xxx/lib/bootm.c.
Now with weak arch_fixup_fdt() right in image-fdt.c where it is
used we get both items highlighted above fixed.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: York Sun <york.sun@nxp.com>
Cc: Stefan Roese <sr@denx.de>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
At present the acpi_rsdp_addr variable is directly referenced in
setup_zimage(). This changes to use an API for better encapsulation
and extension.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
New field acpi_rsdp_addr, which has been introduced in boot protocol
v2.14 [1], in boot parameters tells kernel the exact address of RDSP
ACPI table. Knowing it increases robustness of the kernel by avoiding
in some cases traversal through a part of physical memory.
It will slightly reduce boot time by the same reason.
[1] See Linux kernel commit
2f74cbf ("x86/boot: Add the ACPI RSDP address to struct setup_header::acpi_rdsp_addr")
@ https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/commit/?id=2f74cbf
for the details.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: updated the kernel commit git URL and fixed one style issue]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
The commit
20bfac0599 ("x86: zImage: add Intel MID platforms support")
introduced an assignment of subarch field in boot parameters, though
missed the right place of doing that. It doesn't matter if we have or
not a kernel command line supplied, we just set that field. Although
guard it by protocol version which supports it.
Fixes: 20bfac0599 ("x86: zImage: add Intel MID platforms support")
Cc: Vincent Tinelli <vincent.tinelli@intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The commit
eece493a7a ("cmd: qfw: bring ACPI generation code into qfw core")
moves ACPI related code to another file and missed an update of
references in acpi_table.c.
Do it now.
Fixes: eece493a7a ("cmd: qfw: bring ACPI generation code into qfw core")
Cc: Miao Yan <yanmiaobest@gmail.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
We only need to compile and link these files when building for full
U-Boot. Move them to under cmd/x86/ to make sure they aren't linked in
and undiscarded due to u_boot_list_2_cmd_* being included).
Cc: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
