Peter Hoyes [Mon, 12 Jul 2021 14:04:21 +0000 (15:04 +0100)]
armv8: Initialize CNTFRQ if at highest exception level
CNTFRQ_EL0 is only writable from the highest supported exception
level on the platform. For Armv8-A, this is typically EL3, but
technically EL2 and EL3 are optional so it may need to be
initialized at EL2 or EL1. For Armv8-R, the highest exception
level is always EL2.
This patch moves the initialization outside of the switch_el
block and uses a new macro branch_if_not_highest_el which
dynamically detects whether it is at the highest supported
exception level.
Linux's docs state that CNTFRQ_EL0 should be initialized by the
bootloader. If not set, the the U-Boot prompt countdown hangs.
Stephan Gerhold [Wed, 14 Jul 2021 08:56:26 +0000 (10:56 +0200)]
serial: serial_msm: Delay initialization to let pins stabilize
For some reason, the DragonBoard 410c aborts autoboot immediately if
U-Boot is started without LK. It looks like it picks up a single broken
character via serial and therefore believes a key was pressed to abort
autoboot.
After some debugging, it seems like adding some delay after pinctrl
setup but before UART initialization fixes the issue. It's also worth
mentioning that unlike when booting from LK, the pinctrl setup is
actually necessary when booting U-Boot without LK since UART is broken
if the pinctrl line is removed.
I suspect that reconfiguring the pins might take some time to stabilize
and if the UART controller is enabled too quickly it will pick up some
random noise. Adding a few milliseconds of delay fixes the issue and
shouldn't have any other negative side effects.
3ms seems to be the minimum delay required in my tests, use 5ms instead
just to be sure.
Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
Stephan Gerhold [Wed, 14 Jul 2021 08:56:25 +0000 (10:56 +0200)]
board: dragonboard410c: Fix fastboot
At the moment pressing the volume down key does not actually launch
fastboot. This is because setting "bootdelay" to "-1" actually
disables autoboot and drops to the U-Boot console. It does not execute
the "bootcmd".
The correct value for "bootdelay" here would be "-2", which disables
the delay and key checking and would immediately execute the "bootcmd".
However, even better in this case is using "preboot" to trigger Fastboot.
The advantage is that running "fastboot continue" will actually continue
the autoboot process instead of ending up in the U-Boot shell.
Also make sure to unset "preboot" again immediately in case the user
saves the environment after triggering fastboot.
Cc: Ramon Fried <rfried.dev@gmail.com> Fixes: aa043ee91a47 ("db410c: automatically launch fastboot") Signed-off-by: Stephan Gerhold <stephan@gerhold.net> Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
Stephan Gerhold [Wed, 14 Jul 2021 08:56:24 +0000 (10:56 +0200)]
board: dragonboard410c: Fix PHYS_SDRAM_1_SIZE
The DragonBoard 410c has proprietary firmware from Qualcomm that
reserves 8 MiB of memory for tz/smem/hyp/rmtfs/rfsa from 0x86000000
to 0x86800000. I'm not aware of any ATF (ARM Trusted Firmware) port
for DB410c that would reserve 30 MiB of memory at the end of RAM.
I suspect the comment might have been copied from hikey.h which has
a very similar comment (and which actually does have an ATF port).
Reducing the memory size just prevents U-Boot from using the end of
the RAM, not the reserved region inbetween. Therefore we might as well
display the correct DRAM size (1 GiB) instead of strange 986 MiB.
Fixes: 626f048bbc14 ("board: Add Qualcomm Dragonboard 410C support") Signed-off-by: Stephan Gerhold <stephan@gerhold.net> Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
Stephan Gerhold [Wed, 7 Jul 2021 09:06:02 +0000 (11:06 +0200)]
board: dragonboard410c: Load U-Boot directly without LK
At the moment the U-Boot port for the DragonBoard 410c is designed
to be loaded as an Android boot image after Qualcomm's Little Kernel (LK)
bootloader. This is simple to set up but LK is redundant in this case,
since everything done by LK can be also done directly by U-Boot.
Dropping LK entirely has at least the following advantages:
- Easier installation/board code (no need for Android boot images)
- (Slightly) faster boot
- Boot directly in 64-bit without a round trip to 32-bit for LK
So far this was not possible yet because of unsolved problems:
1. Signing tool: The firmware expects a "signed" ELF image with extra
(Qualcomm-specific) ELF headers, usually used for secure boot.
The DragonBoard 410c does not have secure boot by default but the
extra ELF headers are still required.
2. PSCI bug: There seems to be a bug in the PSCI implementation
(part of the TrustZone/tz firmware) that causes all other CPU cores
to be started in 32-bit mode if LK is missing in the boot chain.
This causes Linux to hang early during boot.
There is a solution for both problems now:
1. qtestsign (https://github.com/msm8916-mainline/qtestsign)
can be used as a "signing" tool for U-Boot and other firmware.
2. A workaround for the "PSCI bug" is to execute the TZ syscall when
entering U-Boot. That way PSCI is made aware of the 64-bit switch
and starts all other CPU cores in 64-bit mode as well.
Simplify the dragonboard410c board by removing all the extra code that
is only used to build an Android boot image that can be loaded by LK.
This allows dropping the custom linker script, special image magic,
as well as most of the special build/installation instructions.
CONFIG_REMAKE_ELF is used to build a new ELF image that has both U-Boot
and the appended DTB combined. The resulting u-boot.elf can then be
passed to the "signing" tool (e.g. qtestsign).
The PSCI workaround is placed in the "boot0" hook that is enabled
with CONFIG_ENABLE_ARM_SOC_BOOT0_HOOK. The extra check for EL1 allows
compatibility with custom firmware that enters U-Boot in EL2 or EL3,
e.g. qhypstub (https://github.com/msm8916-mainline/qhypstub).
As a first step these changes apply only to DragonBoard410c.
Similar changes could likely also work for the DragonBoard 820c.
Note that removing LK wouldn't be possible that easily without a lot of
work already done three years ago by Ramon Fried. A lot of missing
initialization, pinctrl etc was already added back then even though
it was not strictly needed yet.
Cc: Ramon Fried <rfried.dev@gmail.com> Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
configs: synquacer: Ignore OsIndications on DeveloperBox
Since we can not set OsIndications from Runtime Services
SetVariables at this moment, it is better to ignore the
OsIndications if there is any capsule file in the
correct place.
configs: synquacer: Use RAW capsule image instead of FIT
Since the recent commit;
commit b891ff18f899 ("efi_loader: Force a single FMP instance per hardware store")
forces a single FMP instances for a storage, we can not
enable both RAW and FIT capsule image support at once.
Since RAW capsule image support is simpler than FIT,
enable RAW capsule image instead of FIT by default.
configs: synquacer: Drop Ext2/4 support by default
Since the U-Boot for the SynQuacer DeveloperBox is designed for
compatible with EDK2 boot, we don't need to support Ext2/4 fs
support by default. Drop it.
board: synquacer: Initialize SCBM SMMU at board_init()
Since the SCBM SMMU is not only connected to the NETSEC
but also shared with the F_SDH30 (eMMC controller), that
should be initialized at board level instead of NETSEC.
Move the SMMU initialization code into board support
and call it from board_init().
Without this fix, if the NETSEC is disabled, the Linux
eMMC ADMA cause an error because SMMU is not initialized.
In case crypt-based hashing is enabled this will be the default mechanism
that is used. If a user wants to have support for both, the environment
variable `bootstopusesha256` can be set to `true` to allow plain SHA256
based hashing of the password.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de> Reviewed-by: Simon Glass <sjg@chromium.org>
The key-sequence based unlock mechanisms are sensitive to junk symbols
that could have been sent to stdin and are still waiting to be retrieved.
Enabling this option will read all symbols off stdin before displaying the
autoboot prompt (and starting to read the password from stdin).
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de> Reviewed-by: Simon Glass <sjg@chromium.org>
common: allow disabling of timeout for password entry
In case a user has to enter a complicated password it is sometimes
desireable to give the user more time than the default timeout.
Enabling this feature will disable the timeout entirely in case the user
presses the <Enter> key before entering any other character.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de> Reviewed-by: Simon Glass <sjg@chromium.org>
Add the basic functionality required to support the standard crypt
format.
The files crypt-sha256.c and crypt-sha512.c originate from libxcrypt and
their formatting is therefor retained.
The integration is done via a crypt_compare() function in crypt.c.
Nandor Han [Thu, 10 Jun 2021 13:56:44 +0000 (16:56 +0300)]
reboot-mode: read the boot mode from GPIOs status
A use case for controlling the boot mode is when the user wants
to control the device boot by pushing a button without needing to
go in user-space.
Add a new backed for reboot mode where GPIOs are used to control the
reboot-mode. The driver is able to scan a predefined list of GPIOs
and return the magic value. Having the modes associated with
the magic value generated based on the GPIO values, allows the
reboot mode uclass to select the proper mode.
Signed-off-by: Nandor Han <nandor.han@vaisala.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Nandor Han [Thu, 10 Jun 2021 13:56:43 +0000 (16:56 +0300)]
reboot-mode: add support for reboot mode control
A new driver uclass is created to handle the reboot mode control.
The new uclass driver is updating an environment variable with the
configured reboot mode. The mode is extracted from a map provided
at initialization time. The map contains a list of modes
and associated ids.
Signed-off-by: Nandor Han <nandor.han@vaisala.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Michal Simek [Wed, 14 Jul 2021 07:12:41 +0000 (09:12 +0200)]
net: remove CONFIG_NET_MULTI (again)
CONFIG_NET_MULTI has been removed long time ago by commit 795428fc679f
("net: remove CONFIG_NET_MULTI") but 4 boards were added later which wasn't
found. The patch is removing this reference from 4 boards.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
The ARM Juno boards can be used as somewhat decent machines to run
off-the-shelf distributions, with USB, SATA, GBit Ethernet and 8GB of
DRAM.
With stable DTs in the board's NOR flash this would work really nicely,
however the default boot command is to fetch a kernel and an initrd from
the on-board NOR flash, which sounds somewhat embedded.
Include the config_distro_bootcmd.h header and define the available
devices (starting with USB, to catch USB installer sticks) to make
distributions and UEFI work out of the box.
The NOR flash kernel functionality is still preserved as the last
resort, should all other methods fail.
Signed-off-by: Andre Przywara <andre.przywara@arm.com> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Andre Przywara [Sun, 11 Jul 2021 23:07:09 +0000 (00:07 +0100)]
distro_boot: Fix block device after DHCP boot sequence
The generic distro boot sequence iterates over several types of devices,
trying one after another. In doing do, it starts with setting the
"devtype" variable, then uses this later in more generic commands.
Now most (block) device types use a local variable for that, but DHCP
puts the type into the environment, where it shadows any local variables.
As a consequence any boot attempt after bootcmd_dhcp has been run fails:
===========================
VExpress64# run bootcmd_dhcp
...
VExpress64# run bootcmd_sata0
SATA#0:
(3.0 Gbps)
SATA#1:
(No RDY)
Device 0: Model: 16GB SATA Flash Drive ....
... is now current device
Couldn't find partition dhcp 0:1
===========================
This problem typically doesn't show, because DHCP is mostly the last
command to try, but is a problem when this order is different, or when
distro_bootcmd or bootcmd_xxx are run separately or multiple times.
Let bootcmd_dhcp use a local variable, as the other kids do, to make the
order of boot commands irrelevant, and allow repeated calls.
Signed-off-by: Andre Przywara <andre.przywara@arm.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Patrick Delaunay [Fri, 21 May 2021 07:47:31 +0000 (09:47 +0200)]
cmd: pinmux: update result of do_status
Update the result of do_status and always returns a CMD_RET_ value
(-ENOSYS was a possible result of show_pinmux).
This patch also adds pincontrol name in error messages (dev->name)
and treats correctly the status sub command when pin-controller device is
not selected.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com> Reviewed-by: Simon Glass <sjg@chromium.org>
John Keeping [Wed, 19 May 2021 14:05:53 +0000 (15:05 +0100)]
spl: use CONFIG_IS_ENABLED(LOAD_FIT_FULL)
It's possible that LOAD_FIT_FULL will have different values for TPL and
SPL, in which case just checking CONFIG_SPL_LOAD_FIT_FULL causes this to
be compiled in to the TPL even though functions and struct members it
depends on are not.
Use CONFIG_IS_ENABLED() to ensure the correct TPL/SPL variant is
checked.
Signed-off-by: John Keeping <john@metanate.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Patrick Delaunay [Tue, 20 Jul 2021 18:09:54 +0000 (20:09 +0200)]
net: dwc_eth_qos: remove the field phyaddr of the struct eqos_priv
Since the commit commit 6a895d039ba7 ("net: Update eQos driver and FEC
driver to use eth phy interfaces") the field phyaddr of driver private data
struct eqos_priv is no more used in eqos_start() for the phy_connect()
parameter.
Now this variable is only initialized in eqos_probe_resources_stm32()
it can be removed.
Reviewed-by: Ramon Fried <rfried.dev@gmail.com> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Patrick Delaunay [Tue, 20 Jul 2021 18:09:52 +0000 (20:09 +0200)]
net: eth-phy: use dev_dbg and log_notice
Replace debug trace and printf to log macros:
- debug() replaced by dev_dbg() when device is available, this macro
indicate the device name since commit ceb70bb870ac ("dm: Print device
name in dev_xxx like Linux")
- printf() replaced by log_notice() to allow dispatch to log backends.
Reviewed-by: Ramon Fried <rfried.dev@gmail.com> Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>
Patrick Delaunay [Tue, 20 Jul 2021 18:09:51 +0000 (20:09 +0200)]
net: eth-phy: add support of device tree configuration for gpio reset
The gpio reset and the assert or deassert delay are defined in generic
binding of the ethernet phy in Linux:
Documentation/devicetree/bindings/net/ethernet-phy.yaml
reset-gpios:
maxItems: 1
description:
The GPIO phandle and specifier for the PHY reset signal.
reset-assert-us:
description:
Delay after the reset was asserted in microseconds. If this
property is missing the delay will be skipped.
reset-deassert-us:
description:
Delay after the reset was deasserted in microseconds. If
this property is missing the delay will be skipped.
See also U-Boot: doc/device-tree-bindings/net/phy.txt
This patch adds the parsing of this common DT properties in the
u-class "eth_phy_generic", used by default in the associated driver
"eth_phy_generic_drv"
This parsing function eth_phy_of_to_plat can be reused by other
ethernet phy drivers for this uclass UCLASS_ETH_PHY.
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com> Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
dm: core: fix no null pointer detection in ofnode_get_addr_size_index()
Fixed a defect of a null pointer being discovered by Coverity Scan:
CID 331544: Null pointer dereferences (REVERSE_INULL)
Null-checking "size" suggests that it may be null, but it has already been
dereferenced on all paths leading to the check.
The sandbox can handle signals. Due to a damaged global data pointer
additional exceptions in the signal handler may occur leading to an endless
loop. In this case leave the handling of the secondary exception to the
operating system.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Simon Glass <sjg@chromium.org>
Simon Glass [Tue, 6 Jul 2021 16:36:41 +0000 (10:36 -0600)]
binman: Add basic support for debugging performance
One of binman's attributes is that it is extremely fast, at least for a
Python program. Add some simple timing around operations that might take
a while, such as reading an image and compressing it. This should help
to maintain the performance as new features are added.
Simon Glass [Tue, 6 Jul 2021 16:36:36 +0000 (10:36 -0600)]
binman: Put compressed data into separate files
At present compression uses the same temporary file for all invocations.
With multithreading this causes the data to become corrupted. Use a
different filename each time.
Simon Glass [Mon, 5 Jul 2021 22:33:00 +0000 (16:33 -0600)]
log: Allow padding of the function name
At present when function names are logged, the output is a little hard to
read since every function is a different length. Add a way to pad the
names so that the log messages line up vertically. This doesn't work if
the function name is very long, but it makes a big difference in most
cases.
Use 20 characters as a default since this covers the vast majority of
functions.
Simon Glass [Mon, 5 Jul 2021 22:32:59 +0000 (16:32 -0600)]
blk: Support iteration
It is useful to be able to iterate over block devices. Typically there
are fixed and removable devices. For security reasons it is sometimes
useful to ignore removable devices since they are under user control.
Add iterators which support selecting the block-device type.
Simon Glass [Mon, 5 Jul 2021 22:32:58 +0000 (16:32 -0600)]
sandbox: mmc: Support fixed MMC devices
Add support for reading devicetree flags for MMC devices. With this we
can distinguish between fixed and removable drives. Note that this
information is only available when the device is probed, not when it is
bound, since it is read in the of_to_plat() method. This could be changed
if needed later.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Simon Glass [Mon, 5 Jul 2021 22:32:57 +0000 (16:32 -0600)]
spl: Provide more information on boot failure
If SPL fails to boot, try to provide an error code to indicate what is
wrong. For example, if a uclass is missing, this can return -EPFNOSUPPORT
(-96) which provides useful information.
Add a helper for accessing the image-loader name so we can drop the use
of #ifdefs in this code.
Put this feature behind a CONFIG_SHOW_ERRORS option to avoid increasing
the code size.
Simon Glass [Mon, 5 Jul 2021 22:32:56 +0000 (16:32 -0600)]
image: Allow @ in node names when not using signatures
If signature verification is not in use we don't need to worry about
the risk of using @ in node names. Update fit_image_verify() to allow
it if the function is not enabled.
Simon Glass [Mon, 5 Jul 2021 22:32:55 +0000 (16:32 -0600)]
bloblist: Correct condition in bloblist_addrec()
It is possible to add a blob that ends at the end of the bloblist, but at
present this is not supported. Fix it and add a regression test for this
case.
Simon Glass [Mon, 5 Jul 2021 22:32:54 +0000 (16:32 -0600)]
bloblist: Tidy up a few API comments
Some comments for struct bloblist_hdr are a bit ambiguous. Update them to
clarify the meaning more precisely. Also document bloblist_get_stats()
properly.
Simon Glass [Mon, 5 Jul 2021 22:32:53 +0000 (16:32 -0600)]
bloblist: Support resizing a blob
Sometimes a blob needs to expand, e.g. because it needs to hold more log
data. Add support for this. Note that the bloblist must have sufficient
spare space for this to work.
Simon Glass [Mon, 5 Jul 2021 22:32:52 +0000 (16:32 -0600)]
cros_ec: Use standard calls for recovery-request checking
Rather than calling directly into the sandbox SDL code, we can use the
normal U-Boot console handling for this feature. Update the code, to make
it more generic.
Simon Glass [Sun, 4 Jul 2021 18:19:49 +0000 (12:19 -0600)]
dtoc: Detect drivers which do not parse correctly
At present if a driver is missing a uclass or compatible stirng, this
is silently ignored. This makes sense in most cases, particularly for
the compatible string, since it is not required except when the driver
is used with of-platdata.
But it is also not very helpful. When there is some sort of problem
with a driver, the missing compatible string (for example) may be the
cause.
Add a warning in this case, showing it only for drivers which are used
by the build.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Walter Lozano <walter.lozano@collabora.com>
Simon Glass [Sun, 4 Jul 2021 18:19:48 +0000 (12:19 -0600)]
dtoc: Detect unexpected suffix on .of_match
Some rockchip drivers use a suffix on the of_match line which is not
strictly valid. At present this causes the parsing to fail. Fix this
and offer a warning.
Simon Glass [Sun, 4 Jul 2021 18:19:45 +0000 (12:19 -0600)]
dtoc: Allow multiple warnings for a driver
At present we show when a driver is missing but this is not always that
useful. There are various reasons why a driver may appear to be missing,
such as a parse error in the source code or a missing field in the driver
declaration.
Update the implementation to record all warnings for each driver, showing
only those which relate to drivers that are actually used. This avoids
spamming the user with warnings related to a driver for a different board.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Walter Lozano <walter.lozano@collabora.com>
Vincent Chen [Thu, 8 Jul 2021 01:08:20 +0000 (09:08 +0800)]
board: sifive: unmatched: refine GEMGXL initialized function in SPL
Create a new function spl_reset_device_by_gpio to reset the device
whose reset pin is connected to the GPIO. Then, using this function
to initialize GEMGXL.
Signed-off-by: Vincent Chen <vincent.chen@sifive.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Zong Li [Tue, 20 Jul 2021 06:26:08 +0000 (14:26 +0800)]
board: sifive: drop stuff related to unmatched revision 1
This patch reverts the following commits:
- 4b4159d0f3 ("riscv: dts: add dts for unmatched rev1")
- ffe9a394df ("board: sifive: support spl multi-dtb on unmatched board")
We won't plan to support unmatched that the revision below 3 in u-boot,
so they can be dropped because they might be useless.
Changed in v2:
- rebase codebase to the latest master branch
Signed-off-by: Zong Li <zong.li@sifive.com> Suggested-by: David Abdurachmanov <david.abdurachmanov@sifive.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
riscv: booti: do not force relocation if force_reloc is not set
If force_reloc flag is not set and booti is called for an address
ouside RAM (i. e. QSPI NOR flash), we should honor that and not try
to force relocation in a bogus fashion.
Signed-off-by: Vitaly Wool <vitaly.wool@konsulko.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Zong Li [Fri, 9 Jul 2021 08:26:35 +0000 (16:26 +0800)]
board: sifive: remove the command for setting serial number
We wouldn't like to allow user to change the serial number, so remove
the command for changing serial number in EEPROM.
Signed-off-by: Zong Li <zong.li@sifive.com> Suggested-by: David Abdurachmanov <david.abdurachmanov@sifive.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Filter out dmas dma-names interrupt-parent interrupts interrupts-extended
interrupt-names interrupt-map interrupt-map-mask iommus DT properties on
R-Car Gen3, since they are not used by U-Boot and only take space.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
projects / u-boot.git / log
