To quote the author:
So far, standard boot does not replicate all the of the functionality
of the distro_bootcmd scripts. In particular it lacks some bootdevs and
some of the bootmeths are incomplete.
Also there is currently no internal mechanism to enumerate buses in order
to discover bootdevs, e.g. with USB.
This series addresses these shortcomings:
- Adds the concept of a 'bootdev hunter' to enumerate buses, etc. in an
effort to find bootdevs of a certain priority
- Adds bootdevs for SCSI, IDE, NVMe, virtio, SPI flash
- Handles PXE and DHCP properly
- Supports reading the device tree with EFI and reading scripts from the
network
It also tidies up label processing, so it is possible to use:
bootflow scan mmc2
to scan just one MMC device (with BOOTSTD_FULL).
As before this implementation still relies on CONFIG_CMDLINE being
enabled, mostly for the network stack. Further work would be required to
disentangle that.
Quite a few tests are added but there are some gaps:
- SPI flash bootdev
- EFI FDT loading
Note that SATA works via SCSI (CONFIG_SCSI_AHCI) and does not use
driver model. Only pogo_v4 seems to be affected. Probably all thats is
needed is to call bootdev_setup_sibling_blk() in the Marvell SATA driver.
Also, while it would be possible to init MMC in a bootdev hunter, there is
no point since U-Boot always inits MMC on startup, if present.
With this series it should be possible to migrate boards to standard boot
by removing the inclusion of config_distro_bootcmd.h and instead adding
a suitable value for boot_targets to the environment, e.g.:
boot_targets=mmc1 mmc0 nvme scsi usb pxe dhcp spi
Thus it is possible to boot automatically without scripts and boards can
use a text-based environment instead of the config.h files.
To demonstrate this, rockpro64-rk3399 is migrated to standard boot in this
series. Full migration could probably be automated using a script, similar
in concept to moveconfig:
- obtain the board environment via 'make u-boot-initial-env'
- get the value of "boot_targets"
- drop config_distro_bootcmd.h from the config.h file
- rebuild again to get the environment without distro scripts
- write the environment (adding boot_targets) to board.env
- remove CONFIG_EXTRA_ENV_SETTINGS from the config.h file
Simon Glass [Tue, 17 Jan 2023 17:48:15 +0000 (10:48 -0700)]
bootstd: Allow scanning a single bootdev label
We want to support scanning a single label, like 'mmc' or 'usb0'. Add
this feature by plumbing the label through to the iterator, setting a
flag to indicate that only siblings of the initial device should be used.
This means that scanning a bootdev by its name is not supported anymore.
That feature doesn't seem very useful in practice, so it is no great loss.
Add a test for bootdev_find_by_any() while we are here.
Simon Glass [Tue, 17 Jan 2023 17:48:14 +0000 (10:48 -0700)]
bootstd: Switch bootdev scanning to use labels
At present we set up the bootdev order at the start, then scan the
bootdevs one by one.
However this approach cannot be used with hunters, since the bootdevs may
not exist until the hunter is used. Nor can we just run all the hunters at
the start, since that violate's U-Boot's 'lazy init' requirement. It also
increases boot time.
So we need to adjust the algorithm to scan by labels instead. As a first
step, drop the dev_order[] array in favour of a list of labels. Update the
name of bootdev_setup_iter_order() to better reflect what it does.
Update some related comments and log messages. Also disable a few tests
until a later commit where we can use them.
Simon Glass [Tue, 17 Jan 2023 17:48:12 +0000 (10:48 -0700)]
extension: Refactor to allow non-command usage
The current extension code is designed to be used from commands. We want
to add a boot driver which uses it. To help with this, split the code into
the command processing and a function which actually does the scan.
Really the extension code should be in common/ or use driver model, but
this is a start.
Simon Glass [Tue, 17 Jan 2023 17:48:08 +0000 (10:48 -0700)]
bootstd: Add a new pre-scan priority for bootdevs
We need extensions to be set up before we start trying to boot any of the
bootdevs. Add a new priority before all the others for tht sort of thing.
Also add a 'none' option, so that the first one is not 0.
While we are here, comment enum bootdev_prio_t fully and expand the test
for the 'bootdev hunt' command.
Simon Glass [Tue, 17 Jan 2023 17:48:05 +0000 (10:48 -0700)]
bootstd: Treat DHCP and PXE as bootdev labels
These are associated with the ethernet boot device but do not match its
uclass name, so handle them as special cases.
Provide a way to pass flags through with the bootdev so that we know
how to process it. The flags are checked by the bootmeths, to ensure that
only the selected bootmeth is used.
While these both use the network device, they work quite differently. It
is common to run only one of these, or to run PXE before DHCP. Provide
bootflow flags to control which methods are used. Check these in the two
bootmeths so that only the chosen one is used.
Simon Glass [Tue, 17 Jan 2023 17:48:04 +0000 (10:48 -0700)]
bootstd: Support reading a script from network or SPI flash
At present this bootmeth only supports a block device and the sandbox
host filesystem. Add support for obtaining the script from a network
device. Also implement the set_bootflow() method so that it is easy
for other bootdevs (such as enabling SPI flash to support scripts).
Simon Glass [Tue, 17 Jan 2023 17:48:02 +0000 (10:48 -0700)]
sandbox: Allow SPI flash bootdevs to be disabled for tests
Most tests don't want these and they can create a lot of noise. Add a way
to disable them. Use that in tests, with a flag provided to enable them
for tests that need this feature.
Simon Glass [Tue, 17 Jan 2023 17:48:00 +0000 (10:48 -0700)]
bootstd: Move label parsing into its own function
This is complicated enough to merit its own function, particularly as we
are about to add to it. Create a new label_to_uclass() function to decode
a label.
Also update the code to ignore an empty label or one consisting of just a
number.
Simon Glass [Tue, 17 Jan 2023 17:47:57 +0000 (10:47 -0700)]
bootstd: Support reading the device tree with EFI
With EFI booting the device tree is required but is not actually specified
in any way. The normal method is to use a fdtfile environment variable to
get the filename, then look for that file on the media.
Simon Glass [Tue, 17 Jan 2023 17:47:56 +0000 (10:47 -0700)]
bootstd: Include the device tree in the bootflow
Some bootmeths provide a way to load a device tree as well as the base
OS image. Add a way to store this in the bootflow. Update the
'bootflow info' command to show this information.
Note that the device tree is not allocated, but instead is stored at
an address provided by an environment variable. This may need to be
adjusted at some point, but for now it works well and fits in with the
existing distro-boot scripts.
Simon Glass [Tue, 17 Jan 2023 17:47:53 +0000 (10:47 -0700)]
ata: Don't try to use non-existent ports
The controller indicates the number of ports but also has a port map
which specifies which ports are actually valid. Make use of this to
avoid trying to send commands to an invalid port.
Simon Glass [Tue, 17 Jan 2023 17:47:51 +0000 (10:47 -0700)]
virtio: Add a block device
The test code for virtio is fairly simplistic and does not actually create
a block device. Add a way to specify the device type in the device tree.
Add a block device so that we can do more testing.
Simon Glass [Tue, 17 Jan 2023 17:47:38 +0000 (10:47 -0700)]
bootstd: Add an MMC hunter
Add a hunter for MMC. This doesn't do anything at present, since MMC is
currently set up when U-Boot starts. If MMC moves to lazy init then we can
add a hunter function.
Simon Glass [Tue, 17 Jan 2023 17:47:34 +0000 (10:47 -0700)]
bootstd: Support running bootdev hunters
Add a way to run a bootdev hunter to find bootdevs of a certain type. Add
this to the 'bootdev hunt' command. Test for this are added in a later
patch, since a useful test needs some hunters to work with.
Simon Glass [Tue, 17 Jan 2023 17:47:33 +0000 (10:47 -0700)]
bootstd: Add the concept of a bootdev hunter
Some bootdevs must be enumerated before they appear. For example, USB
bootdevs are not visible until USB is enumerated.
With standard boot this needs to happen automatically, since we only
want to enumerate a bus if it is needed.
Add a way to define bootdev 'hunters' which can be used to hunt for
bootdevs of a given type. Track which ones have been used and add a
command to list them.
Include a clang work-around which seems to be needed.
Simon Glass [Tue, 17 Jan 2023 17:47:30 +0000 (10:47 -0700)]
lib: Support printing an error string
It is often useful to show an error code to give the user a clue as to
what went wrong. When error strings are compiled into U-Boot it is
possible to show a message as well.
But at present it is not very convenient, since code must check if the
error strings are present, then obtain the error string and use it in
a printf() string.
Add a %dE option which shows an error code along with an error string,
if available. This makes it easy to show one or both.
Simon Glass [Tue, 17 Jan 2023 17:47:29 +0000 (10:47 -0700)]
sandbox: Enable the Ethernet bootdev
This bootdev is disabled at present since it messes with the tests. Now
that there is a way to disable networking at runtime, enable the driver.
This allows running tests with it if necessary.
Simon Glass [Tue, 17 Jan 2023 17:47:28 +0000 (10:47 -0700)]
sandbox: Allow ethernet bootdevs to be disabled for tests
Most tests don't want these and can create a lot of noise. Add a way to
disable them. Use that in tests, with a flag provided to enable them for
tests that need this feature.
Simon Glass [Tue, 17 Jan 2023 17:47:26 +0000 (10:47 -0700)]
bootstd: Add a default method to get bootflows
The code in these functions turns out to often be the same. Add a default
get_bootflow() function and allow the drivers to select it by setting
the method to NULL.
Simon Glass [Tue, 17 Jan 2023 17:47:25 +0000 (10:47 -0700)]
dm: mmc: Use bootdev_setup_sibling_blk()
At present MMC uses the bootdev_setup_for_dev() function to set up the
bootdev. This is because MMC only has one block-device child, so does not
need to worry about naming of the bootdev.
However this inconsistency with other bootdevs that use block devices is a
bit annoying. The only real reason for it is to have a name like
'mmc0.bootdev' instead of 'mmc0.blk.bootdev'.
Update bootdev_setup_sibling_blk() to drop '.blk' from the name where it
appears, thus removing the only reason to use the bootdev_setup_for_dev().
Switch MMC over to the subling function.
Simon Glass [Tue, 17 Jan 2023 17:47:23 +0000 (10:47 -0700)]
dm: test: Correct ordering of DM setup
We must call dm_scan_other() after devices from the device tree have been
created, since that function behaves differently if there is no bootstd
device.
Adjust the logic to achieve this.
Also fix the bootflow_system() test which was relying on this broken
behaviour.
Simon Glass [Tue, 17 Jan 2023 17:47:22 +0000 (10:47 -0700)]
dm: part: Update test to use mmc2
At present this test sets up a partition table on mmc1. But this is used
by the bootstd tests, so it is not possible to run those after this test
has run, without restarting the Python test harness.
This is inconvenient when running tests repeatedly with 'ut dm'. Move the
test to use mmc2, which is not used by anything.
Simon Glass [Tue, 17 Jan 2023 17:47:19 +0000 (10:47 -0700)]
sandbox: mmc: Start off with a zeroed file
When running multiple tests the mmc emulator calls malloc() to obtain the
memory for its disk image. Since the memory is not cleared, it is possible
that it happens to contain a partition table.
The dm_test_part() test (for one) relies on mmc0 being empty on startup.
Zero the memory to ensure that it is.
Simon Glass [Tue, 17 Jan 2023 17:47:16 +0000 (10:47 -0700)]
bootstd: Simplify locating existing bootdevs
There is no point in trying to match the alias order for bootdevs, since
build_order() either sorts them by priority, uses the boot_targets
environment variable or the bootdev-order property.
Just use the iterator instead, to simplify the code.
Simon Glass [Tue, 17 Jan 2023 17:47:15 +0000 (10:47 -0700)]
bootstd: Remove special-case code for boot_targets
Rather than implement this as its own case in build_order(), process the
boot_targets environment variable in the bootstd_get_bootdev_order()
function. This allows build_order() to be simplified.
- Merge in the final batch of CONFIG to Kconfig/CFG migration work. This
includes a fix for a number of ns16550 or similar UARTs due to a
migration bug. We also pull in a revert for enabling CONFIG_VIDEO on
tools-only_defconfig.
Tom Rini [Tue, 10 Jan 2023 16:19:46 +0000 (11:19 -0500)]
CI: Make check for new defined CONFIG symbols even more robust
Now that all remaining in-tree cases where we define or undef a CONFIG
symbol have been migrated to Kconfig or renamed to CFG we can make the
CI check more robust. We will exclude the doc, tools and arch/arm/dts
directories from this check as they are special cases. Further, we can
exclude the scripts/kconfig/lkc.h and include/linux/kconfig.h files as
the CONFIG values they define are special tooling cases and not real
symbols.
In the case of docs, the only places that currently fail this test are
old documentation that should be rewritten so that we can remove this
special case.
Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Tom Rini [Tue, 10 Jan 2023 16:19:45 +0000 (11:19 -0500)]
global: Finish CONFIG -> CFG migration
At this point, the remaining places where we have a symbol that is
defined as CONFIG_... are in fairly odd locations. While as much dead
code has been removed as possible, some of these locations are simply
less obvious at first. In other cases, this code is used, but was
defined in such a way as to have been missed by earlier checks. Perform
a rename of all such remaining symbols to be CFG_... rather than
CONFIG_...
Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Tom Rini [Tue, 10 Jan 2023 16:19:44 +0000 (11:19 -0500)]
watchdog: Clean up defaults for imx_watchdog / ulp_wdog
In imx_watchdog, clean up the comment to just note the range now, as we
do not need to set the default here as Kconfig does this for us. For
ulp_wdog, set the default value via Kconfig instead.
Cc: Stefan Roese <sr@denx.de> Cc: Stefano Babic <sbabic@denx.de> Cc: Peng Fan <peng.fan@nxp.com> Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Stefan Roese <sr@denx.de>
Tom Rini [Tue, 10 Jan 2023 16:19:42 +0000 (11:19 -0500)]
nxp: Finish migration of SYS_FSL_SRDS_[12] to Kconfig
As this is used on both ARM and PowerPC platforms, this needs to be
listed in arch/Kconfig.nxp and match how they're currently used by
select'ing them under the required PowerPC ARCH_xxx options.
Tom Rini [Tue, 10 Jan 2023 16:19:41 +0000 (11:19 -0500)]
nxp: Finish migration of SYS_FSL_IFC_BANK_COUNT to Kconfig
As this is used on both ARM and PowerPC platforms, this needs to be
asked in arch/Kconfig.nxp. Set the PowerPC defaults based on
arch/powerpc/include/asm/config_mpc85xx.h and remove the default set in
drivers/mtd/nand/raw/fsl_ifc_nand.c
Tom Rini [Tue, 10 Jan 2023 16:19:36 +0000 (11:19 -0500)]
common/update: Finish Kconfig migration
We can enforce the dependencies of this module via Kconfig now, so do so
rather than with #error statements. Further, we can ensure that all
required values are set to their defaults in Kconfig, and in fact
already do so, so remove the tests here. The exception is
CONFIG_UPDATE_LOAD_ADDR which needed to be migrated to Kconfig in the
first place.
Tom Rini [Tue, 10 Jan 2023 16:19:35 +0000 (11:19 -0500)]
spl: sata: Rework the loading case it not use IS_ENABLED(...)
In this case, using IS_ENABLED(...) to attempt to load the image results
in harder to read and less useful code, along with having to define a
CONFIG value that would be unused. To maintain the current albeit
slightly odd behavior, maintain that if we have both SPL_FS_FAT and
SPL_SATA_RAW_U_BOOT_USE_SECTOR enabled, we use SPL_FS_FAT for the load.
Tom Rini [Tue, 10 Jan 2023 16:19:34 +0000 (11:19 -0500)]
autoboot: Rework CONFIG_AUTOBOOT_STOP_STR_* usage
In order to not define a CONFIG value when the
CONFIG_AUTOBOOT_STOP_STR_* functionality is not enabled, rework the
assignment of empty and unused (as the code will be discarded under if
0, in the end) values to be AUTOBOOT_STOP_STR_* instead of
CONFIG_AUTOBOOT_STOP_STR_*.
Tom Rini [Tue, 10 Jan 2023 16:19:33 +0000 (11:19 -0500)]
arm: lib1funcs.S: Update compatibility with Linux comment slightly
At this point, the Linux code for "lib1funcs" has changed rather
dramatically. While a resync would be beneficial, it's outside the scope
of what we need here. Simply remove the define for CONFIG_AEABI and
tests for it.
Tom Rini [Tue, 17 Jan 2023 22:10:50 +0000 (17:10 -0500)]
dm: ns16550: Restore how we define UART_REG
Prior to commit 9591b63531fa ("Convert CONFIG_SYS_NS16550_MEM32 et al to
Kconfig") we had defined CONFIG_SYS_NS16550_REG_SIZE to -1 with
DM_SERIAL such that we would then have a size 0 character array. This
resulted in functionally no padding. The confusion on my part came from
dealing with the constraints around platforms that do not use DM_SERIAL
in SPL/TPL. After Andre Przywara reported that sunxi was broken, I've
re-read the code and comments again and thought on this harder. What we
want I believe is what this patch does now.
If DM_SERIAL is defined for this stage, regardless of
CONFIG_SYS_NS16550_REG_SIZE then we will dynamically handle reg shifts
and 'struct ns16550' needs no padding (which is functionally what
unsigned char foo[0] provides). This is the same case as NS16550_DYNAMIC
and DEBUG_UART. Expand the existing comment here slightly.
Otherwise, we will have CONFIG_SYS_NS16550_REG_SIZE set to a non-zero
value, and handle padding within the struct.
Cc: Simon Glass <sjg@chromium.org> Cc: Sergei Antonov <saproj@gmail.com> Cc: Quentin Schulz <quentin.schulz@theobroma-systems.com> Cc: Andre Przywara <andre.przywara@arm.com> Fixes: 9591b63531fa ("Convert CONFIG_SYS_NS16550_MEM32 et al to Kconfig") Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Andre Przywara <andre.przywara@arm.com> Tested-by: Andre Przywara <andre.przywara@arm.com> Tested-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Tom Rini [Tue, 10 Jan 2023 16:19:31 +0000 (11:19 -0500)]
arm: Rework usage of CONFIG_ARMV[78]_SECURE_BASE in linker scripts
In order to avoid defining CONFIG_ARMV[78_]SECURE_BASE as empty in the
linker scripts, if not already defined, add and use
__ARMV[78_]SECURE_BASE for when the base is not defined and we want the
linker scripts to continue.
Tom Rini [Tue, 10 Jan 2023 16:19:30 +0000 (11:19 -0500)]
i2c: fsl_i2c: Rework usage of CONFIG_SYS_IMMR
This driver is used on both m68k, where CONFIG_SYS_IMMR is not used, and
PowerPC an ARM where it is. Abstract this to a new value rather than
re-defining a CONFIG symbol on m68k.
Ilias Apalodimas [Fri, 20 Jan 2023 11:58:48 +0000 (13:58 +0200)]
efi_loader: fix CapsuleMax variable reporting
Currently the code that adds the CapsuleMax variable is under a
Kconfig named 'EFI_HAVE_CAPSULE_UPDATE. Git history only shows a
single occurrence of that. The IS_ENABLED should be checking for
EFI_HAVE_CAPSULE_SUPPORT
Reported-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Jan Kiszka [Wed, 18 Jan 2023 21:24:59 +0000 (22:24 +0100)]
efi_loader: Avoid overwriting previous outputs on console screen clearing
Before clearing the screen, ensure that no previous output of firmware
or UEFI programs will be overwritten on serial devices or other
streaming consoles. This helps generating complete boot logs.
Tested regarding multi-output against qemu-x86_defconfig. Still, there
were remaining concerns about side effects, so this is provided as an
opt-in feature.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Reviewed-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>