Now that efi_loader subsystem provides interfaces that are equivalent
with bootefi command, we can replace command invocations with APIs.
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
To quote the author:
This little series corrects a problem I noticed with arm64 images,
where the kernel is not recognised if compression is used:
U-Boot> tftp image.fit
Using ethernet@7d580000 device
TFTP from server 192.168.4.7; our IP address is 192.168.4.147
Filename 'image.fit'.
Load address: 0x1000000
Loading: ################################################## 23 MiB
20.5 MiB/s
done
Bytes transferred = 24118272 (1700400 hex)
U-Boot> bootm
## Loading kernel from FIT Image at 01000000 ...
Using 'conf-768' configuration
Trying 'kernel' kernel subimage
Description: Linux
Type: Kernel Image (no loading done)
Compression: gzip compressed
Data Start: 0x01000120
Data Size: 13662338 Bytes = 13 MiB
Verifying Hash Integrity ... OK
Bad Linux ARM64 Image magic!
With this series:
U-Boot> tftp 20000000 image.fit
Using ethernet@7d580000 device
TFTP from server 192.168.4.7; our IP address is 192.168.4.147
Filename 'image.fit'.
Load address: 0x20000000
Loading: ################################################## 23.5 MiB
20.8 MiB/s
done
Bytes transferred = 24642560 (1780400 hex)
U-Boot> bootm 0x20000000
## Loading kernel from FIT Image at 20000000 ...
Using 'conf-768' configuration
Trying 'kernel' kernel subimage
Description: Linux
Type: Kernel Image (no loading done)
Compression: zstd compressed
Data Start: 0x20000120
Data Size: 14333475 Bytes = 13.7 MiB
Verifying Hash Integrity ... OK
Using kernel load address 80000
## Loading fdt from FIT Image at 20000000 ...
Using 'conf-768' configuration
Trying 'fdt-768' fdt subimage
Description: Raspberry Pi 4 Model B
Type: Flat Device Tree
Compression: zstd compressed
Data Start: 0x215f820c
Data Size: 9137 Bytes = 8.9 KiB
Architecture: AArch64
Verifying Hash Integrity ... OK
Uncompressing Flat Device Tree to 3aff3010
Booting using the fdt blob at 0x3aff3010
Working FDT set to 3aff3010
Uncompressing Kernel Image (no loading done) to 80000
Moving Image from 0x80000 to 0x200000, end=2b00000
Using Device Tree in place at 000000003aff3010, end 000000003afff4c4
Working FDT set to 3aff3010
Starting kernel ...
[ 0.000000] Booting Linux on physical CPU 0x0000000000 [0x410fd083]
The problem is that the arm64 magic is checked before the image is
decompressed. However this is only part of it. The kernel_noload image
type doesn't work with compression, since the kernel is not loaded. So
this series deals with that by using an lmb-allocated buffer for the
uncompressed kernel.
Another issue is that the arm64 handling is done too early, before the
image is loaded. This series moves it to after loading, so that
compression can be handled.
A patch is included to show the kernel load-address, so it is easy to
see what is going on.
One annoying feature of arm64 is that the image is often copied to
another address. It might be possible for U-Boot to figure that out
earlier and decompress it to the right place, but perhaps not.
With all of this it should be possible to boot a compressed kernel on
any of the 990 arm64 boards supported by Linux, although I have only
tested two.
It is not currently possible to execute the kernel in-place without
loading it. Use lmb to allocate memory for it.
Co-developed-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
If the image is compressed, then the existing check fails, since the
header is wrong.
Move the check later in the boot process, after the kernel is
decompressed. This allows use of bootm with compressed kernels, while
still permitting an uncompressed kernel to be used.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
The destination address for decompression (or copying) is useful
information. Show this to the user while booting, e.g.:
Uncompressing Kernel Image (no loading done) to 2080000
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
efi_get_var() allocates memory which has to be freed after the value of
the variable is consumed. Free the memory properly
Fixes: f2bfa0cb17 ("bootstd: Make efi_mgr bootmeth work for non-sandbox setups")
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Mark Kettenis <kettenis@openbsd.org>
This may as well happen before the general event is emitted, so move it.
This will allow us to use the livetree for the event part, but the
flattree for the earlier part.
Signed-off-by: Simon Glass <sjg@chromium.org>
This code dates from around 2008:
56844a22b7 powerpc: Fix bootm to boot up again with a Ramdisk
Since then we have added FDT relocation which provides enough space
for expansion. We have also added all sorts of fixups earlier in
image_setup_libfdt() which require more space, with ramdisk being the
least of them.
Therefore this extra hack for ramdisk seems unnecessary. Drop it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
To quote the author:
It would be useful to be able to boot an OS when CONFIG_CMDLINE is
disabled. This could allow reduced code size.
Standard boot provides a way to handle programmatic boot, without
scripts, so such a feature is possible. The main impediment is the
inability to use the booting features of U-Boot without a command line.
So the solution is to avoid passing command arguments and the like to
code in boot/
A similar process has taken place with filesystems, for example, where
we have (somewhat) separate Kconfig options for the filesystem commands
and the filesystems themselves.
This series starts the process of refactoring the bootm logic so that
it can be called from standard boot without using the command line.
Mostly it removes the use of argc, argv and cmdtbl from the internal
logic.
Some limited tidy-up is included, but this is kept to smaller patches,
rather than trying to remove all #ifdefs etc. Some function comments
are added, however.
A simple programmatic boot is provided as a starting point.
This work will likely take many series, so this is just the start.
Size growth with this series for firefly-rk3288 (Thumb2) is:
arm: (for 1/1 boards) all +23.0 rodata -49.0 text +72.0
This should be removed by:
https://source.denx.de/u-boot/custodians/u-boot-dm/-/issues/11
but it is not included in this series as it is already large enough.
No functional change is intended in this series.
Changes in v3:
- Add a panic if programmatic boot fails
- Drop RFC tag
Changes in v2:
- Add new patch to adjust position of unmap_sysmem() in boot_get_kernel()
- Add new patch to obtain command arguments
- Fix 'boot_find_os' typo
- Pass in the command name
- Use the command table to provide the command name, instead of "bootm"
Add some functions which provide an argument to a command, or NULL if
the argument does not exist.
Use the same numbering as argv[] since it seems less confusing than the
previous idea.
Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: Tom Rini <trini@konsulko.com>
At present bootstd requires CONFIG_CMDLINE to operate. Add a new
'programmatic' boot which can be used when no command line is available.
For now it does almost nothing, since most bootmeths require the
command line.
Signed-off-by: Simon Glass <sjg@chromium.org>
Rather than passing the full list of command arguments, pass only those
which are needed.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Move this code into a function to reduce code size and make it easier
to understand. Drop the unnecessary 0x to help a little with code size.
Use this in bootm_find_images()
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Rather than passing it all the command-line args, pass in the pieces
that it needs. These are the image address, the ramdisk address/name
and the FDT address/name.
Ultimately this will allow usage of this function without being called
from the command line.
Move the function comment to the header file and tidy it a little.
Signed-off-by: Simon Glass <sjg@chromium.org>
The normal bootm flow calls bootm_find_other() can call the
BOOTM_STATE_FINDOTHER state as part of its processing. Fix the
condition there so that this hack can be removed.
Also drop the confusing check for the OS type, since do_bootm_tee() is
only called if the condition is met - see bootm_os_get_boot_func()
Signed-off-by: Simon Glass <sjg@chromium.org>
The normal bootm flow calls bootm_find_other() can call the
BOOTM_STATE_FINDOTHER state as part of its processing. Fix the
condition there so that this hack can be removed.
Signed-off-by: Simon Glass <sjg@chromium.org>
The Android mechanism uses the loadaddr envrionment-variable to get the
load address, if none is provided. This is equivalent to
image_load_addr so use that instead, converting it to a string as
needed.
This change will permit passing img_addr to this function, in a future
change.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function only uses two arguments. The 'arch' always has a constant
value, so drop it. This simplifies the function call.
Tidy up the function comment while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function only uses two arguments. The 'arch' always has a constant
value, so drop it. This simplifies the function call.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function only uses one argument from bootm (argv[2]) so pass it in
directly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Rather than having boot_get_fdt() calculate this, move the calculation
into the caller. This removes the access to argv[0] in this function,
so we can later refactor it to just accept argv[2] instead of the whole
argv[].
Move the function comment to the header file and fix the u8 argument,
while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
The OF_LIBFDT #ifdef makes it harder to use a local variable for that
code block. Convert it to if() instead.
Signed-off-by: Simon Glass <sjg@chromium.org>
This function normally only uses one argument so pass it in directly.
Move comments to the header file so could one day include these
functions in API docs. Fix up the u8 argument while here, since it
avoids the compiler having to mask the value on some machines.
The Android case here is bit strange, since it can use argv[0], so deal
with that in the caller.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
The same error message is printed in two places. Move it out to the
caller so we can avoid passing in the command name. Leave the bootstage
handling where it is.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This is a misnomer since we don't necessarily know that the image is a
FIT. Use the existing BOOTSTAGE_ID_CHECK_IMAGETYPE instead.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Rather than looking for a zero-sized image, use the error code returned
to determine if things are OK.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
These unmaps should happen regardless of the return value. Move them
before the 'return' statement.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function obtains lots of error codes and then throws them away.
Update it to return the error, moving the image pointer to an
argument.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function only uses one argument and just needs to know the name of
the command which called it. Adjust the function to use only what it
needs. This will make it easier to call from a non-command context.
Tidy up the function comment while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
This function does not modify its first argument, so mark it const. Also
move the comments to the header file and expand them to provide more
useful information.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Move this code and image_get_kernel() higher in the file to avoid the
need for a forward declaration.
No attempt is made to remove #ifdefs or adjust the code in any other
way.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Move the argument decoding to the caller, to avoid needing to pass the
command-line arguments.
Add a function comment while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
bootmeth_efi doesn't allocate any buffer to load efi in any case.
enable static buffer flag for all cases.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Shantur Rathore <i@shantur.com>
While booting with efi, if fdt isn't available externally,
just use the built-in one.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Shantur Rathore <i@shantur.com>
We need to set boot->fname before calling efi_set_bootdev
otherwise this crashes as bflow->fname is null.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Shantur Rathore <i@shantur.com>
bootmeth_efi sets up bootp_arch which is read later in bootp.c
Currently bootp_arch is being set as integer string and being
read in bootp.c as hex, this sends incorrect arch value to dhcp server
which in return sends wrong file for network boot.
For ARM64 UEFI Arch value is 0xb (11), here we set environment as 11
and later is read as 0x11 and 17 is sent to dhcp server.
Setting it as hex string fixes the problem.
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Shantur Rathore <i@shantur.com>
EFI applications can be very large and thus used to cause boot failures
when malloc() space was exhausted.
A recent changed fixed this by using the kernel_addr_r environment var
as the address of the buffer. However, it still frees the buffer when
the bootflow is discarded.
Fix this by introducing a flag to indicate whether the buffer was
allocated, or not.
Note that kernel_addr_r is not the last word here. It might be better
to use lmb to place images. But there is a lot of refactoring to do
before we can remove the environment variables. The distro scripts rely
on them so it is safe for bootstd to do so too.
Fixes: 6a8c2f9781 bootstd: Avoid allocating memory for the EFI file
Signed-off-by: Simon Glass <sjg@chromium.org>
Reported by: Simon Glass <sjg@chromium.org>
Reported by: Shantur Rathore <i@shantur.com>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Tested-by: Shantur Rathore <i@shantur.com>
During bootstd scanning for bootdevs, if bootdev_hunt_drv() encounters
a device not found error (e.g. ENOENT), let it return a successful status
so that bootstd will continue scanning the next devices, not stopping
prematurely.
Background:
During scanning for bootflows, it's possible for bootstd to encounter a
faulty device controller. Also when the same u-boot is used for another
variant of the same board, some device controller such as SATA might
not exist.
I've found this issue while converting the Marvell Sheevaplug board to
use bootstd. This board has 2 variants, the original Sheevaplug has MMC and
USB only, but the later variant comes with USB, MMC, and eSATA ports. We
have been using the same u-boot (starting with CONFIG_IDE and later with DM
CONFIG_SATA) for both variants. This worked well with the old
envs-scripting booting scheme.
Signed-off-by: Tony Dinh <mibodhi@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
In order for a predefined "preboot" or "bootcmd" to be executed by the
running system we must have a command line. Add CMDLINE as a
dependency.
Signed-off-by: Tom Rini <trini@konsulko.com>
