Samuel Holland [Sun, 12 Sep 2021 14:22:42 +0000 (09:22 -0500)]
phy: sun4i-usb: Support VBUS detection via power supply
The device tree binding provides for getting VBUS state from a device
referenced by phandle, as an optional alternative to using a GPIO. In
U-Boot, where there is no power supply class, this VBUS detection will
be implemented using a regulator device and its get_enable method.
Let's hook this up to the PHY driver.
Signed-off-by: Samuel Holland <samuel@sholland.org> Acked-by: Andre Przywara <andre.przywara@arm.com> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Samuel Holland [Sun, 12 Sep 2021 14:22:41 +0000 (09:22 -0500)]
phy: sun4i-usb: Refactor VBUS detection to match Linux
The Linux driver checks the VBUS detection GPIO first; then VBUS power
supply; then finally assumes VBUS is present. When adding VBUS power
supply support, we want to match that order, so we get the same behavior
in case both a GPIO and a power supply are provided in the device tree.
So refactor the function a bit to remove the early return, and use the
same "assume VBUS is present" final fallback.
Signed-off-by: Samuel Holland <samuel@sholland.org> Acked-by: Andre Przywara <andre.przywara@arm.com> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Samuel Holland [Sun, 12 Sep 2021 14:22:40 +0000 (09:22 -0500)]
phy: sun4i-usb: Remove a couple of debug messages
Both of these messages log the GPIO number of the ID detection GPIO,
which is not terribly useful, especially in the VBUS detection function.
Signed-off-by: Samuel Holland <samuel@sholland.org> Reviewed-by: Andre Przywara <andre.przywara@arm.com> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Samuel Holland [Sun, 22 Aug 2021 18:23:53 +0000 (13:23 -0500)]
sunxi: Select SPL_SEPARATE_BSS
sunxi-common.h defines CONFIG_SPL_BSS_START_ADDR to put SPL's BSS in
DRAM. Due to this, we must select SPL_SEPARATE_BSS, or else SPL will
attempt to load its DTB from the wrong address (after BSS in DRAM).
This change fixes booting with SPL_OF_CONTROL=y.
Signed-off-by: Samuel Holland <samuel@sholland.org> Reviewed-by: Andre Przywara <andre.przywara@arm.com> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Kuldeep Singh [Tue, 3 Aug 2021 09:02:58 +0000 (14:32 +0530)]
spi: nxp_fspi: Implement errata workaround for LS1028A
Errata ERR050568 description says that "Flash access by FlexSPI AHB
command may not work with platform frequency equal to 300 MHz" on
LS1028A.
By default, smaller length reads(equal to RX FIFO size) are done by IP
bus and larger length reads using AHB bus. For adding errata workaround,
use IP bus to read entire flash contents and disable AHB path when
platform frequency is 300Mhz.
Signed-off-by: Kuldeep Singh <kuldeep.singh@nxp.com> Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Kuldeep Singh [Tue, 3 Aug 2021 09:02:57 +0000 (14:32 +0530)]
spi: nxp-fspi: Add support for IP read only
Add support for disabling AHB bus and read entire flash contents via IP
bus only. Please note, this enables IP bus read using a quirk which can
be enabled directly in device-type data or in existence of an errata
where AHB bus may need to be disabled.
Signed-off-by: Kuldeep Singh <kuldeep.singh@nxp.com> Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Marek Behún [Tue, 5 Oct 2021 13:56:06 +0000 (15:56 +0200)]
mtd: Remove mtd_erase_callback() entirely
The original purpose of mtd_erase_callback() in Linux at the time it was
imported to U-Boot, was to inform the caller that erasing is done (since
it was an asynchronous operation).
All supplied callback methods in U-Boot do nothing, but the
mtd_erase_callback() function was (until previous patch) grossly abused
in U-Boot's mtdpart implementation for completely different purpose.
Since we got rid of the abusement, remove the mtd_erase_callback()
function and the .callback member from struct erase_info entirely, in
order to avoid such problems in the future.
Marek Behún [Tue, 5 Oct 2021 13:56:05 +0000 (15:56 +0200)]
mtd: mtdpart: Make mtdpart's _erase method sane
The _erase() method of the mtdpart driver, part_erase(), currently
implements offset shifting (for given mtdpart partition) in a weird way:
1. part_erase() adds partition offset to block address
2. parent driver's _erase() method is called
3. parent driver's _erase() method calls mtd_erase_callback()
4. mtd_erase_callback() subtracts partition offset from block address
so that the callback function is given correct address
The problem here is that if the parent's driver does not call
mtd_erase_callback() in some scenario (this was recently a case for
spi_nor_erase(), which did not call mtd_erase_callback() at all), the
offset is not shifted back.
Moreover the code would be more readable if part_erase() not only added
partition offset before calling parent's _erase(), but also subtracted
it back afterwards. Currently the mtd_erase_callback() is expected to do
this subtracting since it does have to do it anyway.
Add the more steps to this procedure:
5. mtd_erase_callback() adds partition offset to block address so that
it returns the the erase_info structure members as it received them
6. part_erase() subtracts partition offset from block address
This makes the code more logical and also prevents errors in case
parent's driver does not call mtd_erase_callback() for some reason.
(BTW, the purpose of mtd_erase_callback() in Linux is to inform the
caller that it is done, since in Linux erasing is done asynchronously.
We are abusing the purpose of mtd_erase_callback() in U-Boot for
completely different purpose. The callback function itself has empty
implementation in all cases in U-Boot.)
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org> Tested-by: Masami Hiramatsu <masami.hiramatsu@linaro.org>
Marek Behún [Tue, 5 Oct 2021 13:56:03 +0000 (15:56 +0200)]
mtd: spi-nor-core: Call mtd_erase_callback() from spi_nor_erase()
The spi_nor_erase() function does not call mtd_erase_callback() as it
should.
The mtdpart code currently implements the subtraction of partition
offset in mtd_erase_callback().
This results in partition offset being added prior calling
spi_nor_erase(), but not subtracted back on return. The result is that
the `mtd erase` command does not erase the whole partition, only some of
it's blocks:
=> mtd erase "Rescue system"
Erasing 0x00000000 ... 0x006fffff (1792 eraseblock(s))
jedec_spi_nor spi-nor@0: at 0x100000, len 4096
jedec_spi_nor spi-nor@0: at 0x201000, len 4096
jedec_spi_nor spi-nor@0: at 0x302000, len 4096
jedec_spi_nor spi-nor@0: at 0x403000, len 4096
jedec_spi_nor spi-nor@0: at 0x504000, len 4096
jedec_spi_nor spi-nor@0: at 0x605000, len 4096
jedec_spi_nor spi-nor@0: at 0x706000, len 4096
This is obviously wrong.
Add proper calling of mtd_erase_callback() into the spi_nor_erase()
function.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org> Reported-by: Masami Hiramatsu <masami.hiramatsu@linaro.org> Tested-by: Masami Hiramatsu <masami.hiramatsu@linaro.org>
Marek Behún [Tue, 5 Oct 2021 13:56:02 +0000 (15:56 +0200)]
mtd: spi-nor-core: Don't check for zero length in spi_nor_write() / spi_nor_erase()
This check is already done in all callers: mtdcore's mtd_write() /
mtd_erase(), legacy spi_nor_write() / spi_flash_erase(). No reason to do
this here as well.
Marek Behún [Tue, 5 Oct 2021 13:56:01 +0000 (15:56 +0200)]
mtd: spi-flash: Check for zero length in legacy spi_flash_*()
Check for zero length in the legacy spi_flash_read() /
spi_flash_write() / spi_flash_erase() functions.
On zero length, return 0 immediately, don't call the underlying method.
Rationale:
- these legacy functions call the _read(), _write() and _erase() methods
of struct mtd
- the DM callers of these methods already check for zero length
- making all callers of these methods check for zero length makes it
possible to remove the check from implementations of these _read(),
_write() and _erase() methods
Marek Behún [Tue, 5 Oct 2021 13:55:59 +0000 (15:55 +0200)]
mtd: spi-nor-core: Don't overwrite return value if it is non-zero
The cleanup code of the spi_nor_erase() function overwrites the ret
variable with return value of clean_bar(), even if the ret variable is
already set. Fix this.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org> Tested-by: Masami Hiramatsu <masami.hiramatsu@linaro.org>
The S25FL256L is a part of the S25FL-L family and has the same feature set
as S25FL128L except the density.
The datasheet can be found in the following link.
https://www.cypress.com/file/316171/download
The S25FL256L is 32MB NOR Flash that does not support Bank Address
Register. This fixup is activated if CONFIG_SPI_FLASH_BAR is enabled and
returns ENOTSUPP in setup() hook to avoid further ops.
Yanhong Wang [Thu, 30 Sep 2021 11:53:01 +0000 (19:53 +0800)]
mtd: spi-nor-ids: Add GD25LQ256D Chip
Add Gigadevice GD25LQ256D SPI NOR chip.
https://www.gigadevice.com/datasheet/gd25lq256d/
Signed-off-by: Yanhong Wang <yanhong.wang@starfivetech.com> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
[jagan: updated commit message] Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Tom Rini [Fri, 22 Oct 2021 01:41:47 +0000 (21:41 -0400)]
Merge tag 'dm-pull-21oct21' of https://source.denx.de/u-boot/custodians/u-boot-dm
Refactoring of env_get_char() etc.
Update buildman to use gcc-11.1.0
Use in-container toolchain for nokia_rx51 CI test
# gpg: Signature made Thu 21 Oct 2021 09:34:07 PM EDT
# gpg: using RSA key B25C0022AF86A7CC1655B6277F173A3E9008ADE6
# gpg: issuer "sjg@chromium.org"
# gpg: Good signature from "Simon Glass <sjg@chromium.org>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B25C 0022 AF86 A7CC 1655 B627 7F17 3A3E 9008 ADE6
Marek Behún [Sun, 17 Oct 2021 15:36:38 +0000 (17:36 +0200)]
env: Move non-cli env functions to env/common.c
Move the following functions from cmd/nvedit.c to env/common.c:
env_set_ulong()
env_set_hex()
env_get_hex()
eth_env_get_enetaddr()
eth_env_set_enetaddr()
env_get()
from_env()
env_get_f()
env_get_ulong()
since these functions are not specific for U-Boot's CLI.
We leave env_set() in cmd/nvedit.c, since it calls _do_env_set(), which
is a static function in that file.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:37 +0000 (17:36 +0200)]
env: Simplify env_match() and inline into env_get_f()
In the past the env_match() function was used to match envs with
- name, i.e. string "name"
- variable assignment, i.e. string "name=other_value"
The latter is not the case anymore, since the env_match() function is
now used only in env_get_f(), and so we can simplify the function into
a simple strncmp() with an additional comparison to '='.
Let's do this, and since the resulting function is quite simple, let's
also inline its code into env_get_f().
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:35 +0000 (17:36 +0200)]
env: Make return value of env_get_f() behave like sprintf() on success
Currently the env_get_f() function's return value behaves weirdly: it
returns the number of bytes written into `buf`, but whether this is
excluding the terminating NULL-byte or including it depends on whether
there was enough space in `buf`.
Change the function to always return the actual length of the value of
the environment variable (excluding the terminating NULL-byte) on
success. This makes it behave like sprintf().
All users of this function in U-Boot are compatible with this change.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:32 +0000 (17:36 +0200)]
env: Use better name for variable in env_get_f()
The `nxt` variable actually points to the terminating null-byte of the
current env var, and the next env var is at `nxt + 1`, not `nxt`. So a
better name for this variable is `end`.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:31 +0000 (17:36 +0200)]
env: Use string pointer instead of indexes in env_get_f()
Since we no longer use env_get_char() to access n-th character of
linearized environment data, but rather access the arrays themselves, we
can convert the iteration to use string pointers instead of position
indexes.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:30 +0000 (17:36 +0200)]
env: Inline env_get_char() into its only user
This function is a relic from the past when environment was read from
underlying device one character at a time.
It is used only in the case when getting an environemnt variable prior
relocation, and the function is simple enough to be inlined there.
Since env_get_char() is being changed to simple access to an array, we
can drop the failing cases and simplify the code (this could have been
done before, since env_get_char() did not fail even before).
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:27 +0000 (17:36 +0200)]
env: Drop env_get_char_spec() and old, unused .get_char() implementations
Commit b2cdef4861be ("env: restore old env_get_char() behaviour")
dropped the .get_char() method from struct env_driver, but left the two
existing implementations (eeprom and nvram) in case someone would use
them by overwriting weak function env_get_char_spec().
Since this was never done in the 3.5 years, let's drop these methods and
simplify the code.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Marek Behún [Sun, 17 Oct 2021 15:36:26 +0000 (17:36 +0200)]
env: Fix documentation for env_get_f()
This function actually returns:
- the number of bytes written into @buf excluding the terminating
NULL-byte, if there was enough space in @buf
- the number of bytes written into @buf including the terminating
NULL-byte, if there wasn't enough space in @buf
- -1 if the variable is not found
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Simon Glass <sjg@chromium.org>
Fix up the header's include guard to contain the definition of
dm_priv_to_rw(), which was erroneously added outside of it, by moving
its #endif to the end of the file (i.e. where it belongs). This removes
the risk of compilation errors resulting from the redefinition of that
function where the header might have been (indirectly) included more
than once.
Fixes: cfb9c9b77c2 ("dm: core: Use separate priv/plat data region") Signed-off-by: Pierre-Clément Tosi <ptosi@google.com> Cc: Simon Glass <sjg@chromium.org>
Tom Rini [Fri, 15 Oct 2021 14:54:41 +0000 (10:54 -0400)]
arm: mvebu: Add missing "if SPL"
We can only select SPL_SKIP_LOWLEVEL_INIT if SPL is enabled, otherwise
we get a warning about unmet dependencies on platforms that don't use
SPL.
Fixes: cf47a8cf8f7e ("arm: mvebu: Select SPL_SKIP_LOWLEVEL_INIT on ARMADA_32BIT") Cc: Stefan Roese <sr@denx.de> Signed-off-by: Tom Rini <trini@konsulko.com> Reviewed-by: Stefan Roese <sr@denx.de>
Pali Rohár [Tue, 12 Oct 2021 11:19:19 +0000 (13:19 +0200)]
arm: a37xx: pci: Do not allow setting bars on PCI Bridge
PCI Bridge which represents Aardvark PCIe Root Port does not have
configurable bars.
So ensure that write operation to bars registers on PCI Bridge is noop and
bars registers always contain zero address which indicates that bars are
unsupported.
After this change U-Boot 'pci bar 0.0.0' command does not show any
allocated bars for PCI Bridge device.
Signed-off-by: Pali Rohár <pali@kernel.org> Fixes: cb056005dc67 ("arm: a37xx: pci: Add support for accessing PCI Bridge on root bus") Reviewed-by: Stefan Roese <sr@denx.de>
Marek Behún [Sat, 9 Oct 2021 17:33:43 +0000 (19:33 +0200)]
arm: mvebu: turris_omnia: Overwrite ethaddr only if invalid
Currently we always overwrite ethaddrs with those from EEPROM.
In order to allow user to use a cloned MAC address in U-Boot, change the
code so that it sets ethaddr variables only if they aren't set or are
invalid.
Signed-off-by: Marek Behún <marek.behun@nic.cz> Reviewed-by: Stefan Roese <sr@denx.de>
There are two Linux ioctls which implements tcsendbreak() functionality:
TCSBRK and TCSBRKP
TCSBRK with non-zero parameter implements tcdrain() and with zero parameter
implements tcsendbreak() for duration of 0.25s.
TCSBRKP with zero parameter is same as TCSBRK and with non-zero parameter
implements tcsendbreak() for duration in deciseconds specified by
parameter. TCSBRKP does not have to be provided by older toolchain
versions.
So tcsendbreak() has to either use TCSBRK with zero parameter or TCSBRKP
with any parameter.
Fix code to use TCSBRKP and fallback to TCSBRK with 0.
Signed-off-by: Pali Rohár <pali@kernel.org> Reviewed-by: Stefan Roese <sr@denx.de>
Alistair Delva [Wed, 20 Oct 2021 21:31:31 +0000 (21:31 +0000)]
efi_loader: Fix link of EFI apps with ld.lld
When compiling U-Boot with ld.lld as the linker, the helloworld EFI app
example fails to link:
LD lib/efi_loader/helloworld_efi.so
ld.lld: error: section: .dynamic is not contiguous with other relro
sections
LLD will always create RELRO program header regardless of target
emulation, whereas BFD may automatically disable it for unsupported
targets. Add -znorelro to disable it explicitly in all cases.
Signed-off-by: Alistair Delva <adelva@google.com> Cc: Heinrich Schuchardt <xypron.glpk@gmx.de> Cc: Nick Desaulniers <ndesaulniers@google.com> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Ilias Apalodimas [Mon, 11 Oct 2021 12:10:23 +0000 (15:10 +0300)]
efi_loader: Fix loaded image alignment
We are ignoring the alignment communicated via the PE/COFF header.
Starting 5.10 the Linux kernel will loudly complain about it. For more
details look at [1] (in linux kernel).
So add a function that can allocate aligned EFI memory and use it for our
relocated loaded image.
[1] c32ac11da3f83 ("efi/libstub: arm64: Double check image alignment at entry")
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Tested-by: Vincent Stehlé <vincent.stehle@arm.com> Acked-by: Ard Biesheuvel <ardb@kernel.org> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
efi_loader: don't load Shim's MOK database from file
When using a file to store UEFI variables we must make sure that secure
boot related variables are not loaded from this file. With commit 9ef82e29478c ("efi_loader: don't load signature database from file")
this has already been implemented for variables defined in the UEFI
specification. As most Linux distributions use Shim we should do the same
for Shim's MOK database.
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Simon Glass [Sat, 25 Sep 2021 00:30:21 +0000 (18:30 -0600)]
x86: Show some EFI info with the bdinfo command
It is useful to see some basic EFI info with the command as it forms part
of the information about a board.
Add a hook for this and show the table address as a start.
While here, fix an invalid cast in setup_efi_info(). Note that this
function is using a data structure defined by Linux so we cannot change
it. Also note that ulong is used since this is the standard in U-Boot
(>6k uses), despite there being quite a bit of the more verbose uintptr_t
(930 uses).
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Simon Glass [Sat, 25 Sep 2021 00:30:20 +0000 (18:30 -0600)]
x86: Create a new header for EFI
The setup routines are called from zimage but don't really belong in the
zimage header. Add a new EFI header to house these. Add comments so it is
clear what the functions do.
Note that these functions are x86-specific. The zimage business is not
used on other architectures.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
efi_selftest: Receive the packets until the receive buffer is empty
Repeatedly receive the packets until the receive buffer is empty.
If the buffer is empty, EFI_SIMPLE_NETWORK_PROTOCOL::Receive()
returns EFI_NOT_READY. We don't need to use the wait_for_event()
every time.
Signed-off-by: Masami Hiramatsu <masami.hiramatsu@linaro.org> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
efi_selftest: Do not check EFI_SIMPLE_NETWORK_RECEIVE_INTERRUPT
Do not check EFI_SIMPLE_NETWORK_RECEIVE_INTERRUPT in packet
receiving loop. This depends on the implementation and not
related to whether the packet can be received or not.
Whether the received packets are available or not is ensured
by wait_for_packet, and that is already done in the loop.
Signed-off-by: Masami Hiramatsu <masami.hiramatsu@linaro.org> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
efi_selftest: Use EFI_SIMPLE_NETWORK_PROTOCOL::GetStatus() for media check
According to the UEF specification v2.9, the main purpose of the
EFI_SIMPLE_NETWORK_PROTOCOL::GetStatus() is for checking the link
status via EFI_SIMPLE_NETWORK_MODE::MediaPresent.
So this uses net->get_status() for checking the link status before
running network test.
Signed-off-by: Masami Hiramatsu <masami.hiramatsu@linaro.org> Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
mkeficapsule is used to create capsules for UEFI firmware update.
To ease inclusion into U-Boot tools packages of Linux distributions we
should add it to the tools-only_defconfig.
Provide dummy values for CONFIG_AVB_BUF_ADDR, CONFIG_AVB_BUF_SIZE to
satisfy Kconfig.
Suggested-by: Vagrant Cascadian <vagrant@debian.org> Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
For enums documented according to the requirements in chapter
"Structure, union, and enumeration documentation" of
https://www.kernel.org/doc/html/latest/doc-guide/kernel-doc.html
errors occur with Sphinx 2.4.4 which disappear with Sphinx 3.4.3,
e.g.
/builds/u-boot/custodians/u-boot-efi/doc/api/sysreset:6:
./include/sysreset.h:60:Unknown interpreted text role "enum".
Sphinx 3.4.3 is the version used by Debian in the current release.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Stefano Babic [Wed, 20 Oct 2021 10:13:44 +0000 (12:13 +0200)]
imx8mm-cl-iot-gate-optee: align config with Kconfig
Due to missing configs, CI goes in deadlock until an OOM is tracked. Add
CONFIG_SYS_LOAD_ADDR and replace CONFIG_SYS_EXTRA_OPTIONS with
CONFIG_IMX_CONFIG.
Stefano Babic [Wed, 20 Oct 2021 09:55:25 +0000 (11:55 +0200)]
kontron-sl-mx8mm: fix missing configs and deadlock in CI
Even if board can be successfuly built, CI goes in deadlock (see thread
on https://www.mail-archive.com/u-boot@lists.denx.de/msg419663.html).
This is caused by SYS_CONFIG set in header file and because defconfig
for the board is out of sync with Kconfig. As result, buildman goes on
to read from stdin until an OOM is reached.
Simon Glass [Wed, 20 Oct 2021 03:43:24 +0000 (21:43 -0600)]
buildman: Detect Kconfig loops
Hex and int Kconfig options are supposed to have defaults. This is so we
can configure U-Boot without having to enter particular values for the
items that don't have specific values in the board's defconfig file.
If this rule is not followed, then introducing a new Kconfig can produce
a loop like this:
Break things (BREAK_ME) [] (NEW)
Error in reading or end of file.
Break things (BREAK_ME) [] (NEW)
Error in reading or end of file.
The continues forever since buildman passes /dev/null to 'conf', and
the build system just tries again. Eventually there is so much output that
buildman runs out of memory.
We can detect this situation by looking for a symbol (like 'BREAK_ME')
which has no default (the '[]' above) and is marked as new. If this
appears multiple times in the output, we know something is wrong.
Add a filter function for the output which detects this situation. Allow
it to return True to terminate the process. Implement this termination in
cros_subprocess.
With this we get a nice message:
buildman --board sandbox -T0
Building current source for 1 boards (0 threads, 32 jobs per thread)
sandbox: w+ sandbox
+.config:66:warning: symbol value '' invalid for BREAK_ME
+
+Error in reading or end of file.
+make[3]: *** [scripts/kconfig/Makefile:75: syncconfig] Terminated
+make[2]: *** [Makefile:569: syncconfig] Terminated
+make: *** [Makefile:177: sub-make] Terminated
+(** did you define an int/hex Kconfig with no default? **)
Simon Glass [Wed, 20 Oct 2021 03:43:23 +0000 (21:43 -0600)]
buildman: Write output even on fatal error
At present buildman does not write any output (to the 'out' and 'err)
files if the build terminates with a fatal error. This is to avoid adding
lots of spam to the logs.
However there are times when this is actually useful, such as when the
build fails for an obscure reason such as a Kconfig loop.
Update the logic to always write the output, so that the user gets a clue
as to what is happening.
Bin Meng [Sun, 12 Sep 2021 03:15:12 +0000 (11:15 +0800)]
dm: core: Add a new API devfdt_get_addr_index_ptr()
At present there is only devfdt_get_addr_ptr() which only returns
the first <addr, size> pair in the 'reg' property. Add a new API
devfdt_get_addr_index_ptr() to return the indexed pointer.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Bin Meng [Sun, 12 Sep 2021 03:15:10 +0000 (11:15 +0800)]
gpio: sifive: Fix -Wint-to-pointer-cast warning
dev_read_addr() returns a value of type fdt_addr_t which is a 64-bit
address and plat->base is a pointer. In a 32-bit build, this causes the
following warning seen when building sifive-gpio.c:
warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
Change to use dev_read_addr_ptr().
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Bin Meng [Sun, 12 Sep 2021 03:15:09 +0000 (11:15 +0800)]
clk: sifive: Fix -Wint-to-pointer-cast warning
dev_read_addr() returns a value of type fdt_addr_t which is a 64-bit
address and pd->va is a pointer. In a 32-bit build, this causes the
following warning seen when building sifive-prci.c:
warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
Change to use dev_read_addr_ptr().
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Reviewed-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Bin Meng [Sat, 11 Sep 2021 14:31:23 +0000 (22:31 +0800)]
board: sifive: Fix a potential build warning in board_fdt_blob_setup()
Commit 47d73ba4f4a4 ("board: sifive: overwrite board_fdt_blob_setup in u-boot proper")
added a board-specific implementation of board_fdt_blob_setup() which
takes a pointer as the return value, but it does not return anything
if CONFIG_OF_SEPARATE is not enabled. This will cause a build warning
seen when testing booting S-mode U-Boot directly from QEMU, per the
instructions in [1]:
board/sifive/unleashed/unleashed.c: In function ‘board_fdt_blob_setup’:
board/sifive/unleashed/unleashed.c:125:1: warning: control reaches end of non-void function [-Wreturn-type]
Samuel Holland [Wed, 13 Oct 2021 00:40:29 +0000 (19:40 -0500)]
clk: fixed_rate: add dummy disable() function
commit 6bf6d81c1112 ("clk: fixed_rate: add dummy enable() function")
implemented .enable, so fixed rate clocks can be used where drivers
might call clk_enable(). Implement the .disable op for the same reason;
some drivers, e.g. USB PHYs, may attempt to disable clocks at runtime.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Tom Rini [Mon, 11 Oct 2021 15:11:41 +0000 (11:11 -0400)]
tools: Stop re-defining -std= when building tools
While we intentionally set -std=gnu11 for building host tools, and have
for quite some time, we never dropped -std=gnu99 from tools/Makefile.
This resulted in passing -std=gnu11 ... -std=gnu99 when building, and
gnu99 would win. This in turn would result now in warnings such as:
tools/mkeficapsule.c:25:15: warning: redefinition of typedef 'u32' is a C11 feature [-Wtypedef-redefinition]
typedef __u32 u32;
^
This enables the clock controller driver support on TI's SoCs. This will
fix this GPIO issue at boot time:
request_and_set_gpio: Unable to request GPIO_PR1_MII_CTRL
request_and_set_gpio: Unable to request GPIO_MUX_MII_CTRL
request_and_set_gpio: Unable to request GPIO_FET_SWITCH_CTRL
request_and_set_gpio: Unable to request GPIO_PHY_RESET
This issue comes from the fact that the clock controller is not probed.
Enable the TI's clock controller driver support to solve this.
projects / u-boot.git / log