Pali Rohár [Sun, 7 Aug 2022 19:04:22 +0000 (21:04 +0200)]
distroboot: ubifs: Add support for specifying UBI header offset
Some UBI partitions may use non-standard UBI header offset. For attaching
these UBI partitions it is required to pass second argument with offset to
"ubi part" command.
Therefore extend distroboot to allow specifying additional optional 6th
argument with UBI header offset. This offset is set in new distroboot
variable ${bootubioff} which may be used by distroboot script to e.g.
properly pass this value to linux kernel command line for proper mounting
of rootfs by kernel. This variable is set to empty string (cleared) when
UBI header offset is not specified into distroboot BOOT_TARGET_DEVICES
macro.
Usage of helper macro BOOTENV_DEV_UBIFS_BOOTUBIOFF in this change is there
as a type check. It ensures that in BOOT_TARGET_DEVICES macro was specified
UBIFS func with either 5 or 6 arguments. If not then cpp throws compile
error.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Tom Rini <trini@konsulko.com>
Hector Martin [Sun, 31 Jul 2022 06:31:31 +0000 (15:31 +0900)]
nvme: Do a clean NVMe shutdown
The brute-force controller disable method can end up racing controller
initialization and causing a crash when we shut down Apple ANS2 NVMe
controllers. Do a proper controlled shutdown, which does block until
things are quiesced properly. This is nicer in general for all
controllers.
Signed-off-by: Hector Martin <marcan@marcan.st> Tested-by: Mark Kettenis <kettenis@openbsd.org> (firefly-rk3399)
PCIe config space has address range 0-4095. So do not allow reading from
addresses outside of this range. Lot of U-Boot drivers do not expect that
passed value is not in this range. PCI DM read function is extended to
fill read value to all ones or zeros when it fails as U-Boot callers
ignores return value.
Calling U-Boot command 'pci display.b 0.0.0 0 0x2000' now stops printing
config space at the end (before 0x1000 address).
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
- kirkwood: pogo_v4 & nsa310s: Add distro boot (Tony)
- kirkwood: add DM timer support and use it on lsxl boards (Michael)
- kirkwood: convert the Buffalo Linkstation LS-CHLv2 and XHL boards
to DM (Michael)
- mvebu: turris_mox/omnia: misc improments (Pali)
- mvebu: mbus: Fix mbus driver to work also after U-Boot relocation (Pali)
Michael Walle [Wed, 17 Aug 2022 19:38:08 +0000 (21:38 +0200)]
board: lsxl: disable eth0
The board has only one network interface. The linux kernel will
gracefully skip a the ethernet interface if no connected PHY could be
probed. u-boot on the other hand will throw an error message. The kernel
device tree is about to be fixed. For now, just disable the ethernet
interface in our -u-boot.dtsi.
Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Stefan Roese <sr@denx.de>
Michael Walle [Wed, 17 Aug 2022 19:38:06 +0000 (21:38 +0200)]
board: lsxl: convert to DM_SERIAL
DM_SERIAL needs early malloc. The on-chip RAM is pretty tight, it's only
2kiB, with DM_SERIAL enabled, this doesn't work anymore. Fortunately for
us, we don't need the on-chip RAM because the DRAM is already
initialized before u-boot starts. Just put the early malloc area there
and use the default early malloc size.
Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Stefan Roese <sr@denx.de>
Michael Walle [Wed, 17 Aug 2022 19:38:04 +0000 (21:38 +0200)]
board: lsxl: convert to DM_GPIO
Use the new mvebu GPIO driver and convert all the function calls to the
former kirkwood GPIO driver. This means that we are now using the LED
uclass and the regulator uclass. Unfortunately, the GPIO LED doesn't
offer a blinking method. Thus we are now stuck with solid on and off
states, which makes debugging a bit harder. Also, there is no GPIO fan
driver for now.
Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Stefan Roese <sr@denx.de>
Michael Walle [Wed, 17 Aug 2022 19:38:03 +0000 (21:38 +0200)]
board: lsxl: make last resort recovery more reliable
If something is wrong with the environment, we cannot rely on a proper
u-boot operation anymore. In fact, it is possible, that we never reach
misc_init_r() with a broken environment.
Also don't enable the netconsole by environment settings. This way the
user don't have to reconfigure the environment. Instead the network
console is only enabled when the push button is pressed during boot.
Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 17 Aug 2022 19:37:49 +0000 (21:37 +0200)]
arm: kirkwood: 88f6281: Detect CONFIG_SYS_TCLK from SAR register
Bit 21 in SAR register specifies if TCLK is running at 166 MHz or 200 MHz.
This information is undocumented in public Marvell Kirkwood Functional
Specifications [2], but is available in Linux v3.15 kirkwood code [1].
Commit 8ac303d49f89 ("arm: kirkwood: Do not overwrite CONFIG_SYS_TCLK")
broke support for Marvell 88F6281 SoCs because it was expected that all
those SoCs have TCLK running at 200 MHz as specified in Marvell 88F6281
Hardware Specifications [3].
Fix broken support for 88F6281 by detecting CONFIG_SYS_TCLK from SAR
register, like it was doing Linux v3.15.
Michael Walle [Wed, 17 Aug 2022 19:37:48 +0000 (21:37 +0200)]
time: move the CONFIG_SYS_TIMER_RATE handling to the compiler
CONFIG_SYS_TIMER_RATE might be a dynamic value, i.e. a function call
instead of a static value, thus it has to be evaluated at runtime. If it
is a static value, the compiler should be able to optimize the unused
branches out.
This will be needed for kirkwoods dynamic CONFIG_SYS_TCLK setting.
Cc: Pali Rohár <pali@kernel.org> Signed-off-by: Michael Walle <michael@walle.cc> Reviewed-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
- Add distro boot to board include file and deconfig file
- Miscellaneous changes:
- Remove Gerald from maintainer list (email bounced)
- Add CONFIG_SUPPORT_PASSING_ATAGS and friends to support legacy
kernel method of booting (e.g. OpenWrt) with appended DTB.
- Add CONFIG_UBIFS_SILENCE_MSG to reduce binary size.
Note that this patch is depended on the following patch:
https://patchwork.ozlabs.org/project/uboot/patch/20220807192709.21717-1-pali@kernel.org/
Pali Rohár [Wed, 10 Aug 2022 12:46:09 +0000 (14:46 +0200)]
arm: mvebu: mbus: Fix mbus driver to work also after U-Boot relocation
mbus driver is initialized from arch_cpu_init() callback which is called
before relocation. This driver stores lot of functions and structure
pointers into global variables, so it is data position dependent.
Therefore after relocations all pointers are invalid and driver does not
work anymore as all pointers referes to the old memory, which overlaps with
CONFIG_SYS_LOAD_ADDR and ${loadaddr}.
For example U-Boot fuse command crashes if loadaddr memory is cleared or
rewritten by some image loaded by U-Boot load command.
mw.w ${loadaddr} 0x0 10000
fuse read 0 1 2
Fix this issue by removing of all mbus global variables in which are stored
pointers to structures or functions which changes during relocation. And
replace it by direct function calls (not via pointers). With this change
fuse command finally works.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Chris Packham <chris.packham@alliedtelesis.co.nz> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 10 Aug 2022 10:54:11 +0000 (12:54 +0200)]
arm: mvebu: turris_mox: Set "sfp" label in eth1 DT node when only Mox SFP is detected
When Mox SFP module is connected after Topaz or Peridot module then port DT
node already contains "sfp" label. But Mox SFP module can be connected also
without Topaz or Peridot module in which case it is connected directly into
he eth1 DT node, which is without any label. So add "sfp" label into eth1
DT node in this case.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Marek Behún <kabel@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Wed, 10 Aug 2022 09:00:25 +0000 (11:00 +0200)]
arm: mvebu: turris_omnia: Show MCU version
There are already more MCU firmware versions for Turris Omnia in
production, so display git commit (version) of the MCU firmware during
U-Boot startup. It will help to identify what version of MCU firmware is
Turris Omnia using.
MCU firmware for Turris Omnia is open source and available at website:
https://gitlab.nic.cz/turris/hw/omnia_hw_ctrl
It can be updated from running system via i2c bus with this tool:
https://gitlab.nic.cz/turris/omnia-mcutool
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
- Add distro boot to board include file and deconfig file
- Miscellaneous changes:
- Add CONFIG_SUPPORT_PASSING_ATAGS and friends to support legacy
kernel method of booting (e.g. OpenWrt) with appended DTB.
- Add CONFIG_LTO and CONFIG_UBIFS_SILENCE_MSG, and disable some
unused configs to reduce binary size.
Note that this patch is depended on the following patch:
https://patchwork.ozlabs.org/project/uboot/patch/20220807192709.21717-1-pali@kernel.org/
Pali Rohár [Mon, 8 Aug 2022 17:13:43 +0000 (19:13 +0200)]
arm: mvebu: Define env_sf_get_env_addr() for all Armada boards in SPL
SPI0 CS0 Flash is mapped to address range 0xD4000000 - 0xD7FFFFFF by BootROM.
Proper U-Boot removes this direct mapping. So it is available only in SPL.
This applies for all 32-bit Armada BootROMs. SPL mvebu code is used only on
32-bit Armada SoCs. So move env_sf_get_env_addr() function from Turris
Omnia board to common SPL mvebu code and add proper checks for SPI0 CS0.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Tom Rini [Wed, 10 Aug 2022 01:08:54 +0000 (21:08 -0400)]
CI: Move to Ubuntu 2022.04 "Jammy" for CI base
- We now have a new enough sbsigntools in the distro, stop building.
- Use the 20220801 tag for Jammy.
- Move to pygit2 1.9.2 (current version) as the old one doesn't build on
"Jammy".
- Add the working directory to the list of safe directories for git.
- Move to pytest 6.2.5 to address other issues.
- This move exposed a number of minor issues in the existing scripts we
used within CI to perform the jobs themselves. The most notable changes
here involve using 'set +e / set -e' to enforce when we should or should
not make non-zero buildman status be a fatal error.
Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Holger Brunck [Fri, 19 Aug 2022 14:55:06 +0000 (16:55 +0200)]
powerpc: remove support for kmtergr1 and MPC8309
The kmtegr1 board is out of maintenance and can be removed. As it is the
only board in the tree using MPC8309 the support for this CPU is dropped
completely.
Enforce requiring DM_ETH to be enabled for ethernet drivers, as the
migration deadline has well passed. To facilitate this, we remove some
non-migrated platforms and disable networking on a few others. Finally
we remove some of the now-useless non-DM_ETH code in some platforms as a
prerequisite for DM_ETH being set.
Tom Rini [Tue, 2 Aug 2022 11:33:47 +0000 (07:33 -0400)]
net: Make DM_ETH be selected by NETDEVICE
The deadline for DM_ETH migration passed 2 years ago. Now that
platforms which cannot be migrated have been either removed or had
drivers disabled, and platforms that needed minor help to migrate have
been forcefully migrated, we can complete the migration.
This entails select'ing DM_ETH under NETDEVICES, and then removing now
extraneous depends on lines. In a few places, we can now either remove
options or just simplify later dependencies.
Cc: Ramon Fried <rfried.dev@gmail.com> Cc: Simon Glass <sjg@chromium.org> Signed-off-by: Tom Rini <trini@konsulko.com> Acked-by: Ramon Fried <rfried.dev@gmail.com>
Add gzip bintool to binman to support on-the-fly compression of Linux
kernel images and FPGA bitstreams. The SPL basic fitImage implementation
supports only gzip decompression.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Rename the module and support this, since gzip.py is a system module: Signed-off-by: Simon Glass <sjg@chromium.org>
Add a bintools base class for packers which compression / decompression
entry contents.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Dropped dead/untested code in version(): Signed-off-by: Simon Glass <sjg@chromium.org>
Add common test functions to test all supported compressions.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Simon Glass <sjg@chromium.org>
binman: Select compression bintools in cbfs_util class
Select the lz4 and lzma_alone bintools in cbfs_util class to centralize
the supported compression algorithm evaluation inside the class and over
multiple classes.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
binman: Move compression bintools creation into test setup
Move compression bintools (packer) creation into test setup to reuse
bintool objects between tests.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Put comp_util import back in, since it is still needed here: Signed-off-by: Simon Glass <sjg@chromium.org>
binman: Remove obsolete compressed data header handling
Remove the obsolete compressed data header handling from the utilities
to compress and decompress data. The header is uncommon, not supported
by U-Boot and incompatible with external compressed artifacts.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Disable the compressed data header of the utilities to compress and
decompress data. The header is uncommon, not supported by U-Boot and
incompatible with external compressed artifacts.
The header was introduced as part of commit eb0f4a4cb402 ("binman:
Support replacing data in a cbfs") to allow device tree entries to be
larger than the compressed contents.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Add an optional length header attribute to the device tree blob entry
class based on the compressed data header from the utilities to compress
and decompress data.
If needed the header could be enabled with the following
attribute beside the compress attribute:
prepend = "length";
The header was introduced as part of commit eb0f4a4cb402 ("binman:
Support replacing data in a cbfs") to allow device tree entries to be
larger than the compressed contents. Regarding the commit "this is
necessary to cope with a compressed device tree being updated in such a
way that it shrinks after the entry size is already set (an obscure
case)". This case need to be fixed without influence any compressed data
by itself.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Collect and thereby initialize bintools before any usage but after
generation of entries. This is needed to handle bintools for compress
and decompress like other bintools.
Signed-off-by: Stefan Herbrechtsmeier <stefan.herbrechtsmeier@weidmueller.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Simon Glass [Sat, 13 Aug 2022 17:40:50 +0000 (11:40 -0600)]
binman: Allow collection to use entries from other sections
At present the collections etype only works with entries in the same
section. This can be limiting, since in some cases the data may be inside
a subsection, e.g. if there are alignment constraints.
Add a function to find the entries in an etype and have it search
recursively. Make use of this for mkimage also.
Simon Glass [Sat, 13 Aug 2022 17:40:46 +0000 (11:40 -0600)]
binman: Avoid use of expected failure
The testReplaceSectionSimple() test is the only one which expects failure.
It looks odd in the output and takes time to glance at it to see that all
is in fact well. Also it does not check that the right exception is
generated.
Use the more common (in binman) approach of checking for an exception.
Simon Glass [Sat, 13 Aug 2022 17:40:45 +0000 (11:40 -0600)]
binman: Adjust mkimage etype node reading
Since this is implemented as a section, it should really be split into
several functions, one to read the node and one to read the entries. Do
this so that it matches how Entry_section works.
Simon Glass [Sat, 13 Aug 2022 17:40:44 +0000 (11:40 -0600)]
binman: Add a way to check for missing properties
Some new entries are likely to have required properties. Support this in a
standard way, with a list of required properties which can be set up by
base classes. Check for missing properties when the entry is read.
Simon Glass [Sat, 13 Aug 2022 17:40:42 +0000 (11:40 -0600)]
patman: Don't buffer test output with a single test
When a single test is run we don't need to buffer the test output. This
has the unfortunate side effect of suppressing test output, in particular
the binman output directory normally printed with the -X option. This is
a huge problem since it blocks debugging of tests.
We don't actually know how many tests will be run when we set up the
suite, so as a work-around, assume that test_name being specified
indicates that there is likely only one.
Michal Simek [Fri, 12 Aug 2022 08:54:51 +0000 (10:54 +0200)]
bootstage: Show func name for bootstage_mark/error
bootstage_mark() and bootstate_error() are not recording any name and in
report it is showing as id=<value>. That's not useful and it is better to
show function name which calls it.
That's why use macros with passing __func__ as recorded name for bootstage.
Simon Glass [Sun, 7 Aug 2022 15:46:46 +0000 (09:46 -0600)]
binman: Put fake files in a subdirectory
At present fake files from a previous build appear to be real files for
a subsequent build, since they sit in the output directory.
This can cause problems, since binman may need to parse the file, e.g.
with the Intel description.bin files.
Fix this by putting them in a 'binman-fake' subdirectory. Keep a track
of the fake filename so we only create it once. Subsequent builds will
still see that the file is missing and mark it as fake.
Pali Rohár [Sun, 7 Aug 2022 19:06:04 +0000 (21:06 +0200)]
hwconfig: Allow to use restricted env
During early boot phase GD_FLG_ENV_READY is not set but env_get() may work
when env is ready in restricted mode. Do not fail with error message
"WARNING: Calling __hwconfig without a buffer and before environment is ready"
when env is already working by checking for ENV_VALID flag.
The Yocto project builds their aarch64 cross-compiler with the
configure knob --enable-standard-branch-protection, which means that
their gcc behaves as if -mbranch-protection=standard is passed; the
default (lacking that configure knob) is -mbranch-protection=none.
This means that when building U-Boot using the Yocto toolchain, most
functions end up containing paciasp/autiasp/bti instructions. However,
since U-Boot is not an ordinary userspace application, there's no OS
kernel which has set up the required authentication keys, so these
instructions do nothing at all (even on arm64 hardware that does have
the pointer authentication capability). They do however make the image
larger.
It is theoretically possible for U-Boot to make use of the pointer
authentication protection - cf. the linux kernel's
CONFIG_ARM64_PTR_AUTH_KERNEL - but it is far from trivial, and it's
hard to see just what threat model it would protect against in a
bootloader context. Regardless, we certainly have none of the required
infrastructure now, so explicitly pass -mbranch-protection=none to
ensure those useless instructions do not get emitted.
For a toolchain not configured with
--enable-standard-branch-protection, this changes nothing. For the
Yocto toolchain, this reduces the size of both SPL and U-Boot proper
by about 3% for my imx8mp target.
If you don't have a Yocto toolchain, the effect can easily be
reproduced by applying this patch and changing =none to =standard.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk> Reviewed-by: Simon Glass <sjg@chromium.org>
projects / u-boot.git / log