Currently when using OF_PLATDATA the binding between devices and drivers
is done trying to match the compatible string in the node with a driver
name. However, usually a single driver supports multiple compatible strings
which causes that only devices which its compatible string matches a
driver name get bound.
To overcome this issue, this patch adds the U_BOOT_DRIVER_ALIAS macro,
which generates no code at all, but allows an easy way to declare driver
name aliases. Thanks to this, dtoc could be improve to look for the driver
name based on its alias when it populates the U_BOOT_DEVICE entry.
Signed-off-by: Walter Lozano <walter.lozano@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When using OF_PLATDATA, the bind process between devices and drivers
is performed trying to match compatible string with driver names.
However driver names are not strictly defined, and also there are different
names used when declaring a driver with U_BOOT_DRIVER, the name of the
symbol used in the linker list and the used in the struct driver_info.
In order to make things a bit more clear, rename the drivers names. This
will also help for further OF_PLATDATA improvements, such as checking
for valid driver names.
Signed-off-by: Walter Lozano <walter.lozano@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Add a fix for sandbox of-platdata to avoid using an invalid ANSI colour:
Signed-off-by: Simon Glass <sjg@chromium.org>
This change allows more fine tuning of driver model based SPI support in
SPL and TPL. It is now possible to explicitly enable/disable the DM_SPI
support in SPL and TPL via Kconfig option.
Before this change it was necessary to use:
/* SPI Flash Configs */
#if defined(CONFIG_SPL_BUILD)
#undef CONFIG_DM_SPI
#undef CONFIG_DM_SPI_FLASH
#undef CONFIG_SPI_FLASH_MTD
#endif
in the ./include/configs/<board>.h, which is error prone and shall be
avoided when we strive to switch to Kconfig.
The goal of this patch:
Provide distinction for DM_SPI support in both U-Boot proper and SPL (TPL).
Valid use case is when U-Boot proper wants to use DM_SPI, but SPL must
still support non DM driver.
Another use case is the conversion of non DM/DTS SPI driver to support
DM/DTS. When such driver needs to work in both SPL and U-Boot proper, the
distinction is needed in Kconfig (also if SPL version of the driver
supports OF_PLATDATA).
In the end of the day one would have to support following use cases (in
single driver file - e.g. mxs_spi.c):
- U-Boot proper driver supporting DT/DTS
- U-Boot proper driver without DT/DTS support (deprecated)
- SPL driver without DT/DTS support
- SPL (and TPL) driver with DT/DTS (when the SoC has enough resources to
run full blown DT/DTS)
- SPL driver with DT/DTS and SPL_OF_PLATDATA (when one have constrained
environment with no fitImage and OF_LIBFDT support).
Some boards do require SPI support (with DM) in SPL (TPL) and some only
have DM_SPI{_FLASH} defined to allow compiling SPL.
This patch converts #ifdef CONFIG_DM_SPI* to #if CONFIG_IS_ENABLED(DM_SPI)
and provides corresponding defines in Kconfig.
Signed-off-by: Lukasz Majewski <lukma@denx.de>
Tested-by: Adam Ford <aford173@gmail.com> #da850-evm
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
[trini: Fixup a few platforms]
Signed-off-by: Tom Rini <trini@konsulko.com>
Versal:
- xspi bootmode fix
- Removing one clock from clk driver
- Align u-boot memory setting with OS by default
- Map TCM and OCM by default
ZynqMP:
- Minor DT improvements
- Reduce console buffer for mini configurations
- Add fix for AMS
- Add support for XDP platform
Zynq:
- Support for AES engine
- Enable bigger memory test by default
- Extend documentation for SD preparation
- Use different freq for Topic miami board
mmc:
- minor GD pointer removal
net:
- Support fixed-link cases by zynq gem
- Fix phy looking loop in axi enet driver
spi:
- Cleanup global macros for xilinx spi drivers
firmware:
- Add support for pmufw reloading
fpga:
- Improve error status reporting
common:
- Remove 4kB addition space for FDT allocation
-----BEGIN PGP SIGNATURE-----
iF0EABECAB0WIQQbPNTMvXmYlBPRwx7KSWXLKUoMIQUCXvScEQAKCRDKSWXLKUoM
IeHUAJ9Z1shAlbILuHZKEpqQySLHdUVgBQCff8Nf+wi1rByTrwflKt14MtIsdFY=
=2Yuu
-----END PGP SIGNATURE-----
Merge tag 'xilinx-for-v2020.10' of https://gitlab.denx.de/u-boot/custodians/u-boot-microblaze into next
Xilinx changes for v2020.10
Versal:
- xspi bootmode fix
- Removing one clock from clk driver
- Align u-boot memory setting with OS by default
- Map TCM and OCM by default
ZynqMP:
- Minor DT improvements
- Reduce console buffer for mini configurations
- Add fix for AMS
- Add support for XDP platform
Zynq:
- Support for AES engine
- Enable bigger memory test by default
- Extend documentation for SD preparation
- Use different freq for Topic miami board
mmc:
- minor GD pointer removal
net:
- Support fixed-link cases by zynq gem
- Fix phy looking loop in axi enet driver
spi:
- Cleanup global macros for xilinx spi drivers
firmware:
- Add support for pmufw reloading
fpga:
- Improve error status reporting
common:
- Remove 4kB addition space for FDT allocation
Remove below config options and convert them to macros. They have never
been configured to different values than default one. And also it makes
sense to reduce the config_whitelist.
CONFIG_SYS_ZYNQ_SPI_WAIT
CONFIG_SYS_ZYNQ_QSPI_WAIT
CONFIG_XILINX_SPI_IDLE_VAL
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Modify the Freescale ESPI driver to support the driver model.
Also resolved the following problems:
===================== WARNING ======================
This board does not use CONFIG_DM_SPI. Please update
the board before v2019.04 for no dm conversion
and v2019.07 for partially dm converted drivers.
Failure to update can lead to driver/board removal
See doc/driver-model/MIGRATION.txt for more info.
====================================================
===================== WARNING ======================
This board does not use CONFIG_DM_SPI_FLASH. Please update
the board to use CONFIG_SPI_FLASH before the v2019.07 release.
Failure to update by the deadline may result in board removal.
See doc/driver-model/MIGRATION.txt for more info.
====================================================
Signed-off-by: Chuanhua Han <chuanhua.han@nxp.com>
Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Now all boards which are using davinci SPI driver
have moved to SPL_DM so drop the unneeded non-dm code.
Cc: Adam Ford <aford173@gmail.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Tested-by: Adam Ford <aford173@gmail.com> #da850-evm
atmel_spi.h has register offsets, and atmel_spi_slave
structure, move it into .c file for better readability
and drop atmel_spi.h
Cc: Wenyou Yang <wenyou.yang@atmel.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
atmel spi driver now accessible only when DM_SPI enabled.
So, remove nondm code.
Cc: Wenyou Yang <wenyou.yang@atmel.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
- Deadline for DM migration already passed by months.
- Sent couple of zap patches and
- No response on dm conversation
hence removed the driver.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Order the macros, private structures and simple functions
in a proper way to have more code readability.
No functionality changed.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Deadline for DM migration already passed by months
and no response on full dm conversation hence removed
the nondm code.
Note: Look like there is no user for nondm code
for this driver.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
At present this logic does not work on link and samus, since their SPI
controller is not a PCI device, but a child of the PCH.
Unfortunately, fixing this involves a lot of extra logic. Still, this was
requested in the review of the fix-up patch, so here it is.
Fixes: 92842147c3 ("spi: ich: Add support for get_mmap() method")
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)
sh_spi driver is deprecated, no active updates and
no board user, hence dropped the same.
Cc: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
CF_SPI kconfig option defined twice with DM_SPI
and non DM_SPI.
Drop the non DM_SPI side kconfig definition.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
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>
Have this symbol follow the pattern of all other such symbols.
This patch also removes a TODO from the code.
Reviewed-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Trevor Woerner <twoerner@gmail.com>
Board gets reset when performing burst read/write operations. On the
other hand, no such behaviour is observed on small size operations.
In Linux, readl_poll_timeout API already adds delay of 1us which is
further skipped in U-boot. Hence, use new "readl_poll_sleep_timeout" API
which adds delay alongwith timeout functionality.
Signed-off-by: Kuldeep Singh <kuldeep.singh@nxp.com>
lpc32xx_ssp driver is deprecated, no active updates
and no board user, hence dropped the same.
Cc: Vladimir Zapolskiy <vz@mleia.com>
Cc: Albert ARIBAUD <albert.aribaud@3adev.fr>
Cc: Tom Rini <trini@konsulko.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
The guessed reason is that the existing logic of filling
tx fifo with data, rx fifo with NULL for tx transfer and
filling rx fifo with data, tx fifo with NULL for rx transfer
is not clear enough to support the Quad Page Program.
SiFive SPI controllers have specific sets of watermark
registers and SPI I/O directions bits in order to program
SPI controllers clear enough to support all sets of operating
modes.
Here is the exact programing sequence that would follow on this
patch and tested via SPI-NOR and MMC_SPI.
- set the frame format proto, endian
- set the frame format dir, set it for tx and clear it for rx
- TX transfer:
fill tx fifo with data.
wait for TX watermark bit to clear.
- RX transfer:
fill tx fifo with 0xff.
write nbytes to rx watermark register
wait for rx watermark bit to clear.
read the rx fifo data.
So, this patch adopts this program sequence and fixes the existing
I/O direction bit.
Cc: Vignesh R <vigneshr@ti.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Tested-by: Sagar Kadam <sagar.kadam@sifive.com>
SiFive SPI controller has a proto bit field in frame format
register which would be used to configure the SPI I/O protocol
lines used on specific transfer.
Right now the driver is configuring this proto using slave->mode,
for all types of transctions. This makes the driver unable to
function since the proto needs to configure dynamically for
each and every transaction separately at runtime.
Now, the controller driver supports per transfer via spi-mem
exec_opo, so add the fmt_proto flag and fill the per transfer
buswidth so that the controller configures the proto bit at
runtime.
This patch fixes the SPI controller works with SPI NOR flash
on quad read with page program.
Cc: Vignesh R <vigneshr@ti.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Tested-by: Sagar Kadam <sagar.kadam@sifive.com>
SiFive SPI controller is responsible to handle the
slave devices like mmc spi and spi nor flash.
The controller is designed such a way that it would
handle the slave transactions based on the I/O protocol
numbers, example if spi nor slave send quad write opcode
it has to send alone with I/O protocol number of 4 and
if it try to send data it has to send I/O protocol number
along with 4 line data.
But the current spi-xfer code from spi-mem is combining
the opcode and address in a single transaction, so the
SPI controller will be unable to identify the I/O protocol
number of opcode vs address.
So, add the spi-mem exec_op with spi-xfer of opcode, address
and data as a separate transaction. This doesn't remove
the .xfer of dm_spi_ops since mmc spi will make use of it.
Note: This code might have moved to the spi-mem core area
once we have done the dedicated tests on other controllers
and have real reason to move.
Cc: Vignesh R <vigneshr@ti.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Tested-by: Sagar Kadam <sagar.kadam@sifive.com>
Usually variable name slave is used for spi_slave structure
and slave_plat for the dm_spi_slave_platdata.
Let's follow this meaningful notation by replacing slave
with slave_plat for dm_spi_slave_platdata structure.
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Commit: 0ebb261a0b2d ("spi: spi-mem: Add SPI_MEM_NO_DATA to the spi_mem_data_dir enum")
in linux.
When defining spi_mem_op templates we don't necessarily know the size
that will be passed when the template is actually used, and basing the
supports_op() check on op->data.nbytes to know whether there will be
data transferred for a specific operation is not possible.
Add SPI_MEM_NO_DATA to the spi_mem_data_dir enum so that we can base
our checks on op->data.dir instead of op->data.nbytes.
This also fixes a bug identified with the atmel-quaspi driver.
The spi-nor core, when erasing sectors, fills the spi_mem_op template
using SPI_MEM_OP_NO_DATA, which initializes all the data members with
value zero. This is wrong because data.dir is treated as SPI_MEM_DATA_IN,
which translates in our driver to read accesses for erases (RICR), while
the controller expects write accesses (WICR).
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
"assigned-clock-parents" and "assigned-clock-rates" DT properties take
effect only after ofdata_to_platdata() when clk_set_defaults() is called
in device_probe(). Therefore clk get rate() would return a wrong value
in ofdata_to_platdata() when compared with probe. Hence it needs to be
moved to probe.
Tested on u-boot-ti/next.
Signed-off-by: Pratyush Yadav <p.yadav@ti.com>
Acked-by: Vignesh Raghavendra <vigneshr@ti.com>
Acked-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
To support the SPI MEM API, instead of modifying the existing U-Boot
driver, this patch adds a port of the existing Linux driver.
This also has the advantage that porting changes and fixes from Linux will
be easier.
Porting of driver left most of the functions unchanged while few of the
changes are:
-Remove lock(mutexes) and irq handler as u-boot is a single core execution.
-Remove invalid masterid as it was required specially for multicore
execution in LS2088ARDB which is not the case in u-boot.
-Remove clock support as changing spi speed is not supported in uboot and
nor in linux.
Currently tested on LS1088ARDB, LS1012ARDB, LS1046ARDB, LS1046AFRWY,
LS1043AQDS, LS1021ATWR, LS2088ARDB, I.MX6ULL EVK.
Signed-off-by: Frieder Schrempf <frieder.schrempf@kontron.de>
Signed-off-by: Ashish Kumar <Ashish.Kumar@nxp.com>
Signed-off-by: Kuldeep Singh <kuldeep.singh@nxp.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Acked-by: Vignesh Raghavendra <vigneshr@ti.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
The build currently fails with
drivers/spi/mpc8xxx_spi.c:64:3: warning: implicit declaration of function ‘dev_err’ [-Wimplicit-function-declaration]
...
drivers/spi/built-in.o: In function `mpc8xxx_spi_set_speed':
drivers/spi/mpc8xxx_spi.c:227: undefined reference to `dev_err'
Fixes: 4856cc7a97 (mpc8xxx_spi: implement real ->set_speed)
Fixes: 1a7b462dee (mpc8xxx_spi: put max_cs to use)
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
This Tegra QSPI driver hadn't been brought up to date with how
DM drivers are fetching data from the FDT now, and was pulling
in bogus data for base, max freq, etc. Fixed ofdata_to_platdata
to work the same way it does in the tegra114 SPI driver, using
dev_read_ functions.
Signed-off-by: Tom Warren <twarren@nvidia.com>
When claim_bus was setting the clock, it reset the QSPI controller,
which wipes out any tap delays set by previous bootloaders (nvtboot,
CBoot for example on Nano). Instead of doing that in claim_bus, which
gets called a lot, moved clock setting to probe(), and set tap delays
there, too. Also updated clock to 80MHz to match CBoot. Now QSPI env
save works reliably again.
Signed-off-by: Tom Warren <twarren@nvidia.com>
claim_bus() is passed a udevice *dev, which is the bus device's parent.
In this driver, claim_bus assumed it was the bus, which caused the
'priv' info pointer to be wrong, and periph_id was incorrect. This in
turn caused the periph clock call to assign the wrong clock (PLLM
instead of PLLP0), which caused a kernel warning. I only saw the 'bad'
periph_id when enabling DEBUG due to an assert. Not sure how QSPI was
working w/this errant clock, but it was moot as QSPI wasn't active
unless you probed it, and that wasn't happening until I posted a patch
to enable env save to QSPI for Nano (coming soon).
Signed-off-by: Tom Warren <twarren@nvidia.com>
This feature should not be enabled in release but can be useful for
developers who need to monitor register accesses at some specific places.
Helped me identify a bug in u-boot, by comparing the register accesses
from the u-boot driver with the ones from its linux variant.
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
[jagan: use 16 bit array with tmp variable]
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
The sama5d2 QSPI controller memory space is limited to 128MB:
0x9000_00000-0x9800_00000/0XD000_0000--0XD800_0000.
There are nor flashes that are bigger in size than the memory size
supported by the controller: Micron MT25QL02G (256 MB).
Check if the address exceeds the MMIO window size. An improvement
would be to add support for regular SPI mode and fall back to it
when the flash memories overrun the controller's memory space.
Fixes: 24c8ff4684 ("spi: Add Atmel QuadSPI driver")
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Not all boards have the same CSB frequency, nor do every SPI slave
necessarily support running at 16.7 MHz. So implement ->set_speed;
that also allows using a smaller PM (i.e., 0) for slaves that do
support a higher speed.
Based on work by Klaus H. Sørensen.
Cc: Klaus H. Sorensen <khso@prevas.dk>
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
There are a few problems with the current driver.
First, it unconditionally reads from dout/writes to din whether or not
those pointers are NULL. So for example a simple "sf probe" ends up
writing four bytes at address 0:
=> md.l 0x0 8
00000000: 45454545 45454545 05050505 05050505 EEEEEEEE........
00000010: 00000000 00000000 07070707 07070707 ................
=> sf probe 0
mpc8xxx_spi_xfer: slave spi@7000:0 dout 0FB53618 din 00000000 bitlen 8
mpc8xxx_spi_xfer: slave spi@7000:0 dout 00000000 din 0FB536B8 bitlen 48
SF: Detected s25sl032p with page size 256 Bytes, erase size 64 KiB, total 4 MiB
=> md.l 0x0 8
00000000: ff000000 45454545 05050505 05050505 ....EEEE........
00000010: 00000000 00000000 07070707 07070707 ................
(here I've change the first debug statement to a printf, and made it
print the din/dout pointers rather than the uints they point at).
Second, as we can also see above, it always writes a full 32 bits,
even if a smaller amount was requested. So for example
=> mw.l $loadaddr 0xaabbccdd 8
=> md.l $loadaddr 8
02000000: aabbccdd aabbccdd aabbccdd aabbccdd ................
02000010: aabbccdd aabbccdd aabbccdd aabbccdd ................
=> sf read $loadaddr 0x400 6
device 0 offset 0x400, size 0x6
mpc8xxx_spi_xfer: slave spi@7000:0 dout 0FB536E8 din 00000000 bitlen 40
mpc8xxx_spi_xfer: slave spi@7000:0 dout 00000000 din 02000000 bitlen 48
SF: 6 bytes @ 0x400 Read: OK
=> sf read 0x02000010 0x400 8
device 0 offset 0x400, size 0x8
mpc8xxx_spi_xfer: slave spi@7000:0 dout 0FB53848 din 00000000 bitlen 40
mpc8xxx_spi_xfer: slave spi@7000:0 dout 00000000 din 02000010 bitlen 64
SF: 8 bytes @ 0x400 Read: OK
=> md.l $loadaddr 8
02000000: 45454545 45450000 aabbccdd aabbccdd EEEEEE..........
02000010: 45454545 45454545 aabbccdd aabbccdd EEEEEEEE........
Finally, when the bitlen is 24 mod 32 (e.g. requesting to read 3 or 7
bytes), the last three bytes and up being the wrong ones, since the
driver does a full 32 bit read and then shifts the wrong byte out:
=> mw.l $loadaddr 0xaabbccdd 4
=> md.l $loadaddr 4
02000000: aabbccdd aabbccdd aabbccdd aabbccdd ................
=> sf read $loadaddr 0x444 10
device 0 offset 0x444, size 0x10
mpc8xxx_spi_xfer: slave spi@7000:0 dout 0FB536E8 din 00000000 bitlen 40
mpc8xxx_spi_xfer: slave spi@7000:0 dout 00000000 din 02000000 bitlen 128
SF: 16 bytes @ 0x444 Read: OK
=> md.l $loadaddr 4
02000000: 552d426f 6f742032 3031392e 30342d30 U-Boot 2019.04-0
=> mw.l $loadaddr 0xaabbccdd 4
=> sf read $loadaddr 0x444 0xb
device 0 offset 0x444, size 0xb
mpc8xxx_spi_xfer: slave spi@7000:0 dout 0FB536E8 din 00000000 bitlen 40
mpc8xxx_spi_xfer: slave spi@7000:0 dout 00000000 din 02000000 bitlen 88
SF: 11 bytes @ 0x444 Read: OK
=> md.l $loadaddr 4
02000000: 552d426f 6f742032 31392e00 aabbccdd U-Boot 219......
Fix all of that by always using a character size of 8, and reject
transfers that are not a whole number of bytes. While it ends being
more work for the CPU, we're mostly bounded by the speed of the SPI
bus, and we avoid writing to the mode register in every loop.
Based on work by Klaus H. Sørensen.
Cc: Klaus H. Sorensen <khso@prevas.dk>
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Currently, max_cs is write-only; it's just set in
mpc8xxx_spi_ofdata_to_platdata and not otherwise used.
My mpc8309 was always resetting during an "sf probe 0". It turns out
dm_gpio_set_dir_flags() was being called with garbage, since nothing
had initialized priv->gpios[0] - our device tree used "cs-gpios"
rather than "gpios", so gpio_request_list_by_name() had returned 0.
That would have been a lot easier to figure out if the chip select
index was sanity checked, so rename max_cs to cs_count, and reject a
xfer with a too large cs index.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Most files don't need this header and it pulls in quite of lots of stuff,
malloc() in particular. Drop it.
Signed-off-by: Simon Glass <sjg@chromium.org>
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>
