Brian Norris [Fri, 26 Jul 2024 19:02:33 +0000 (12:02 -0700)]
patman: Resolve python string vs. regex escaping syntax
Python strings have their own notion of backslash-escaping, and that can
conflict with the intentions for strings passed to the 're' module. In
particular, I get warnings like this:
dma: ti: k3-udma: Add support for native configuration of chan/flow
In absence of Device Manager (DM) services such as at R5 SPL stage,
driver will have to natively setup TCHAN/RCHAN/RFLOW cfg registers.
Existing UDMA driver performed the above mentioned configuration
for UDMA. Add similar configuration for PKTDMA here.
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@siemens.com> Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com> Signed-off-by: Chintan Vankar <c-vankar@ti.com>
Chintan Vankar [Mon, 26 Aug 2024 10:25:09 +0000 (15:55 +0530)]
soc: ti: k3-navss-ringacc: Fix reconfiguration of qmode API
Function "k3_ringacc_ring_reconfig_qmode_raw()" should reset qmode to
requested value and should not update other fields in ring configuration
register.
soc: ti: k3-navss-ringacc: Initialize base address of ring cfg registers
Initialize base address of ring config registers required to natively
setup ring cfg registers in the absence of Device Manager (DM) services
at R5 SPL stage. Since register property is defined as "ring" for PKTDMA
and "cfg" for UDMA, configure base address of ring configuration register
accordingly.
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@siemens.com> Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com> Signed-off-by: Chintan Vankar <c-vankar@ti.com>
RX_FL_CFG message should not be forwarded to TIFS and should be
handled within R5 SPL (when DM services are not available). Add
a no-op function to not handle RX_FL_CFG messages.
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@siemens.com> Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com> Signed-off-by: Chintan Vankar <c-vankar@ti.com>
Jerome Forissier [Wed, 28 Aug 2024 15:36:37 +0000 (17:36 +0200)]
mach-imx: do not use if_changed more than once per target
doc/develop/makefiles.rst has the following note:
if_changed should not be used more than once per target.
It stores the executed command in a corresponding .cmd
file and multiple calls would result in overwrites and
unwanted results when the target is up to date and only the
tests on changed commands trigger execution of commands.
The mach-imx Makefile does not follow this recommandation, so fix it
by implementing a single command that performs both the cpp_cfg
and imx9_check actions.
This change fixes an issue with "tools/buildman/buildman imx8ulp_evk"
failing every other time [1].
Jonas Karlman [Sat, 3 Aug 2024 12:41:44 +0000 (12:41 +0000)]
bootstage: Fix unstash of records from SPL
The commit b81e31a1e6c5 ("bootstash: Do not provide a default address
for all") changed a bootstage unstash call to bootstage stash, this
has resulted in bootstage records stashed in SPL no longer get unstaged
in U-Boot proper. Fix this by changing back to a unstage call.
Fixes: b81e31a1e6c5 ("bootstash: Do not provide a default address for all") Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Tom Rini <trini@konsulko.com>
Jonas Karlman [Sat, 3 Aug 2024 12:41:44 +0000 (12:41 +0000)]
bootstage: Fix unstash of records from SPL
The commit b81e31a1e6c5 ("bootstash: Do not provide a default address
for all") changed a bootstage unstash call to bootstage stash, this
has resulted in bootstage records stashed in SPL no longer get unstaged
in U-Boot proper. Fix this by changing back to a unstage call.
Fixes: b81e31a1e6c5 ("bootstash: Do not provide a default address for all") Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Reviewed-by: Simon Glass <sjg@chromium.org> Reviewed-by: Tom Rini <trini@konsulko.com>
mmc: fix signed vs unsigned compare in read check in _spl_load()
Fix signed vs unsigned compare in read check in _spl_load()
Issue: when info->read() returns a negative value because of an error,
the comparison of 'read' (signed) with 'sizeof(*header)'
(unsigned silently converts the negative value into a very
large unsigned value and the check on the error condition
always return false, i.e. the error is not detected
Symptoms: if spl_load_image_fat() is unable to find the file 'uImage',
the SPL phase of the boot process just hangs after displaying
the following line:
Trying to boot from MMC1
Fix: cast 'sizeof(*header)' to int so the compare is now between
signed types
Reference: https://stackoverflow.com/questions/17293749/sizeof-operator-in-if-statement
Signed-off-by: Franco Venturi <fventuri@comcast.net> Reviewed-by: Sean Anderson <seanga2@gmail.com> Reviewed-by: Peng Fan <peng.fan@nxp.com>
Billy Tsai [Tue, 20 Aug 2024 07:27:36 +0000 (15:27 +0800)]
gpio: Add G7 Aspeed gpio controller driver
In the 7th generation of the SoC from Aspeed, the control logic of the
GPIO controller has been updated to support per-pin control. Each pin now
has its own 32-bit register, allowing for individual control of the pin’s
value, direction, interrupt type, and other settings.
Signed-off-by: Billy Tsai <billy_tsai@aspeedtech.com>
Hiago De Franco [Mon, 19 Aug 2024 16:24:49 +0000 (13:24 -0300)]
tools/scripts/Kconfig: Update references from kconfig-language.txt to rst
The Linux kernel documentation has transitioned from using
kconfig-language.txt to kconfig-language.rst. Therefore update all
occurrences of kconfig-language.txt.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
Maxim Moskalets [Sat, 17 Aug 2024 11:02:22 +0000 (14:02 +0300)]
lib: move phdr increment to for loop heading
Shifting this pointer in the loop will be more logical when working
with the code later, because you can see at a glance what exactly
changes at each iteration. Moreover, the code remains equivalent
because this variable is not used after the loop.
Signed-off-by: Maxim Moskalets <maximmosk4@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Tom Rini [Tue, 27 Aug 2024 20:32:54 +0000 (14:32 -0600)]
Merge patch series "dm: core: Avoid multiple calls to fdt_parent_offset()"
Jonas Karlman <jonas@kwiboo.se> says:
Use of fdt_parent_offset() is very expensive as detailed by the function
documentation:
NOTE: This function is expensive, as it must scan the device tree
structure from the start to nodeoffset, *twice*.
This series remove multiple calls to fdt_parent_offset() or
ofnode_get_parent() when instead a single call can be made and the
returned value can be reused.
This series help reduce boot time by around:
- ~137ms on a Radxa ROCK Pi 4 (RK3399)
- ~33ms on a Radxa ZERO 3W (RK3566)
Jonas Karlman [Sun, 4 Aug 2024 15:05:51 +0000 (15:05 +0000)]
dm: core: regmap: Avoid multiple calls to ofnode_get_parent()
Until a live tree is used in U-Boot proper after relocation, use of
ofnode_get_parent() will trigger a call to the very expensive
fdt_parent_offset(), as detailed by the function documentation:
NOTE: This function is expensive, as it must scan the device tree
structure from the start to nodeoffset, *twice*.
Re-use the returned value from a single call instead of having to make
multiple calls for same node.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Reviewed-by: Simon Glass <sjg@chromium.org>
Jonas Karlman [Sun, 4 Aug 2024 15:05:50 +0000 (15:05 +0000)]
dm: core: ofnode: Avoid multiple calls to ofnode_get_parent()
Until a live tree is used in U-Boot proper after relocation, use of
ofnode_get_parent() will trigger a call to the very expensive
fdt_parent_offset() as detailed by the function documentation:
NOTE: This function is expensive, as it must scan the device tree
structure from the start to nodeoffset, *twice*.
Re-use the returned value from a single call instead of having to make
multiple calls for same node.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Reviewed-by: Simon Glass <sjg@chromium.org>
Love Kumar [Wed, 31 Jul 2024 09:24:12 +0000 (14:54 +0530)]
test/py: spi: Add tests for SPI flash device
Add test cases for sf commands to verify various SPI flash operations
such as erase, write and read. It also adds qspi lock unlock cases.
This test relies on boardenv_* configurations to run it for different
SPI flash family such as single SPI, QSPI, and OSPI.
Zixun LI [Fri, 2 Aug 2024 09:28:13 +0000 (11:28 +0200)]
cmd: bind: Use device sequence instead for driver bind/unbind
Currently uclass index is used for bind/unbind which ignores alias
sequence numbering. Use device sequence number instead as it's
the number explicitly set in the DT.
Also update documentation to use sequence number.
Signed-off-by: Zixun LI <admin@hifiphile.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Zixun LI [Fri, 2 Aug 2024 09:28:12 +0000 (11:28 +0200)]
dm: core: Show device sequence instead in dm_dump_tree()
Currently uclass index is shown in DM tree dump which ignores alias
sequence numbering. The result could be confusing since these 2 numbers
could be different. Show device sequence number instead as it's more
meaningful.
Also update documentation to use sequence number.
Signed-off-by: Zixun LI <admin@hifiphile.com> Reviewed-by: Simon Glass <sjg@chromium.org>
mtd: nand: raw: omap_gpmc: Check return value of gpmc_nand_init
If the function is called with no NAND device attached, then this
function can return error value, proceeding further ignoring the same
can cause system crash. This is seen when "mtd list" is run with no NAND
addon cards connected.
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com> Acked-by: Roger Quadros <rogerq@kernel.org>
Sean Anderson [Fri, 16 Aug 2024 02:38:22 +0000 (22:38 -0400)]
sandbox: Fix pinmux warnings with non-test devicetrees
The sandbox pinmux driver is used in the non-test devicetree as well as
the test one. I didn't realize this when I modified the driver for
tests, and so broke the regular use case (which only resulted in
warnings). First, making the pinmux and the UART group available
pre-relocation to avoid ENODEV errors. Then, convert the pin groups and
functions to the new style, adding onewire group as well.
Fixes: 7f0f1806e3a ("test: pinmux: Add test for pin muxing") Closes: https://source.denx.de/u-boot/u-boot/-/issues/2 Signed-off-by: Sean Anderson <seanga2@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
A system may have multiple SATA controller. Removing the controller with
the lowest sequence number before probing all SATA controllers makes no
sense.
In sata_rescan we remove all block devices which are children of SATA
controllers. We also have to remove the bootdev devices as they will be
created when scanning for block devices.
After probing all SATA controllers we must scan for block devices otherwise
we end up without any SATA block device.
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com> Reviewed-by: Tony Dinh <mibodhi@gmail.com> Reviewed-by: Simon Glass <sjg@chromium.org>
pci: mediatek: add support for upstream split PCIe node
Add support for upstream linux split PCIe node.
Upstream linux have an alternative way to declare PCIe nodes that splits
them in dedicated nodes for each line instead of putting them all in one
node.
Detect this by checking if the mediatek,generic-pciecfg node is passed
as it's used to reference the common address for all the PCIe lines.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
remoteproc: uclass: Modify uc_pdata->name to use combination of device name and device's parent name
uc_pdata->name is populated from device tree property "remoteproc-name".
For those devcices that don't set "remoteproc-name", uc_pdata->name
falls back to dev->name.
If two devices have same name, this will result into uc_pdata->name not
being unique and rproc_init() will fail.
Fix this by using combination of dev->name and dev->parent->name instead
of using just the dev->name to populate uc_pdata->name.
Signed-off-by: MD Danish Anwar <danishanwar@ti.com> Reviewed-by: Roger Quadros <rogerq@kernel.org> Reviewed-by: Andrew Davis <afd@ti.com>
Upstream DTS added explicit ranges to the fss node. It did not include
the 32 bit memory space needed by the R5 to access OSPI. With the
upstream DTS sync, OSPI boot no longer works.
Adding the missing range here. It is also being added in the upstream DTS,
so after the next upstream DTS sync, these patches can be removed.
Fixes: 5024a96db8e ("Subtree merge tag 'v6.10-dts' of devicetree-rebasing repo [1] into dts/upstream")
Simon Glass [Thu, 15 Aug 2024 20:30:22 +0000 (14:30 -0600)]
bootstd: Make bootdev_next_prio() continue after failure
When a device fails to probe, the next device should be tried, until
either we find a suitable device or run out of devices. A device
should never be tried twice.
When we run out of devices of a particular priority, the hunter should
be used to generate devices of the next priority. Only if all attempts
fail should this function return an error.
Update the function to use the latent 'found' boolean to determine
whether another loop iteration is warranted, rather than setting 'dev'
to NULL, which creates confusion, suggesting that no devices have been
scanned and the whole process is starting from the beginning.
Note that the upcoming bootflow_efi() test is used to test this
behaviour.
Signed-off-by: Simon Glass <sjg@chromium.org> Fixes: https://source.denx.de/u-boot/custodians/u-boot-dm/-/issues/17
Simon Glass [Thu, 15 Aug 2024 20:30:21 +0000 (14:30 -0600)]
Revert "bootdev: avoid infinite probe loop"
This turns out to be insufficient to fix the problem, since when
bootdev_next_prio() exits, the caller has no idea that this really
is the end. Nor is it, since there may be other devices which should
be checked.
The caller iterates which calls iter_incr() which calls
bootdev_next_prio() again, which finds the same device and the loop
continues.
We never did create a test for this[1], which makes it hard to be
sure which problem was fixed.
The original code had the virtue of staying in the loop looking for a
bootdev, so let's go back to that and try to fix this another way.
A future patch will make bootdev_next_prio() continue after failure
which should provide same effect.
arch: arm: dts: k3-j721e-r5-sk: Add missing fss range
Upstream DTS added explicit ranges to the fss node. It did not include the
32 bit memory space needed by the R5 to access OSPI. With the upstream DTS
sync, OSPI boot no longer works.
Adding the missing range here. It is also being added in the upstream DTS,
so after the next upstream DTS sync, this patch can be removed. See 0c0e03ec22 (arm64: dts: ti: k3-j721e: Use exact ranges for FSS node)
Fixes: 5024a96db8e ("Subtree merge tag 'v6.10-dts' of devicetree-rebasing repo [1] into dts/upstream") Signed-off-by: Jonathan Humphreys <j-humphreys@ti.com>
arch: arm: dts: k3-j721e-r5-evm: Add missing fss range
Upstream DTS added explicit ranges to the fss node. It did not include the
32 bit memory space needed by the R5 to access OSPI. With the upstream DTS
sync, OSPI boot no longer works.
Adding the missing range here. It is also being added in the upstream DTS,
so after the next upstream DTS sync, this patch can be removed. See 0c0e03ec22 (arm64: dts: ti: k3-j721e: Use exact ranges for FSS node)
Fixes: 5024a96db8e ("Subtree merge tag 'v6.10-dts' of devicetree-rebasing repo [1] into dts/upstream") Signed-off-by: Jonathan Humphreys <j-humphreys@ti.com>
arch: arm: dts: k3-j7200-r5-evm: Add missing fss range
Upstream DTS added explicit ranges to the fss node. It did not include the
32 bit memory space needed by the R5 to access OSPI. With the upstream DTS
sync, OSPI boot no longer works.
Adding the missing range here. It is also being added in the upstream DTS,
so after the next upstream DTS sync, this patch can be removed. See f00e626085 (arm64: dts: ti: k3-j7200: Use exact ranges for FSS node)
Fixes: 5024a96db8e ("Subtree merge tag 'v6.10-dts' of devicetree-rebasing repo [1] into dts/upstream") Signed-off-by: Jonathan Humphreys <j-humphreys@ti.com> Reviewed-by: Aniket Limaye <a-limaye@ti.com>
arch: arm: dts: k3-am69-r5-sk: Add missing fss range
Upstream DTS added explicit ranges to the fss node. It did not include the
32 bit memory space needed by the R5 to access OSPI. With the upstream DTS
sync, OSPI boot no longer works.
Adding the missing range here. It is also being added in the upstream DTS,
so after the next upstream DTS sync, this patch can be removed. See f062a015f4 (arm64: dts: ti: k3-j784s4: Use exact ranges for FSS node)
Fixes: 5024a96db8e ("Subtree merge tag 'v6.10-dts' of devicetree-rebasing repo [1] into dts/upstream") Signed-off-by: Jonathan Humphreys <j-humphreys@ti.com>
arch: arm: dts: k3-j784s4-r5-evm: Add missing fss range
Upstream DTS added explicit ranges to the fss node. It did not include the
32 bit memory space needed by the R5 to access OSPI. With the upstream DTS
sync, OSPI boot no longer works.
Adding the missing range here. It is also being added in the upstream DTS,
so after the next upstream DTS sync, this patch can be removed. See f062a015f4 (arm64: dts: ti: k3-j784s4: Use exact ranges for FSS node)
Fixes: 5024a96db8e ("Subtree merge tag 'v6.10-dts' of devicetree-rebasing repo [1] into dts/upstream") Signed-off-by: Jonathan Humphreys <j-humphreys@ti.com> Reviewed-by: Andrew Davis <afd@ti.com>
Tom Rini [Tue, 27 Aug 2024 00:52:18 +0000 (18:52 -0600)]
Merge patch series "Tidy up console recording in tests"
Simon Glass <sjg@chromium.org> says:
This series started as a small fix for checking for an empty line,
but in the process several other problems were found and fixed:
- fix tests which use console recording but don't set the flag
- drop unnecessary resetting of the console in tests
- drop unnecessary blank line before MMC output
- update the docs a little
- fix buildman test failure on newer Pythons
- a few other minor things
This series also renames the confusing flag names, so that they are
easier to remember - just a UTF_ (unit-test flags) prefix.
Simon Glass [Thu, 22 Aug 2024 13:57:55 +0000 (07:57 -0600)]
test: fdt: Check internal-function return values
Some functions are using asserts but the result of the functions
themselves is not checked. This means that if a test fails, the result
is not noticed until later, which can be confusing to debug.
Simon Glass [Thu, 22 Aug 2024 13:57:52 +0000 (07:57 -0600)]
test: Update NAND test to avoid extra macros
Write out the tests in full to allow the test to be found more easily
when there is a failure. We could use a single test function with a
for() loop but this would stop at the first failure, and some variations
might while other pass.
Simon Glass [Thu, 22 Aug 2024 13:57:49 +0000 (07:57 -0600)]
test: Drop the blank line before test macros
Most tests don't have this. It helps to keep the test declaration
clearly associated with the function it relates to, rather than the next
one in the file. Remove the extra blank line and mention this in the
docs.
Simon Glass [Thu, 22 Aug 2024 13:57:47 +0000 (07:57 -0600)]
test: Fail when an empty line is expected but not present
The existing implementation of ut_assert_nextline_empty() cannot
distinguish between an empty line and no line at all. It can in fact be
called at the end of the recorded output and will happily return
success.
Adjust the logic so that this condition is detected. Show a failure
message in this case.
Fix the one test which falls foul of this fix.
Signed-off-by: Simon Glass <sjg@chromium.org> Fixes: 400175b0a7d ("test: Add a way to check each line of console...") Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
Tom Rini [Mon, 26 Aug 2024 20:06:16 +0000 (14:06 -0600)]
Merge patch series "global_data: Reduce size of struct global_data"
Simon Glass <sjg@chromium.org> says:
The global data structure has grown quite a lot over the years, being
the best place to put an important pointer or something that must be
accessed before and after relocation.
This series attempts to reduce the size a little, by moving some things
out and shrinking and aligning some fields.
Some fields are needed during init but not afterwards. To deal with this
a new 'boardf' structure is created, which sits on the stack and is only
present during board_init_f(). It is possible that more fields could
move to this struct, but for now only 4 are moved.
An assumption is made that an int is 32-bits wide on all architectures,
which seems to be true, but maintainers should be able to confirm.
Mostly the code-size impact is neutral, but the patch
'Use less space for environment fields' does increase U-Boot's size by
about 30 bytes on aarch64.
For firefly-rk3399 (64-bit) the size of global reduces from 456 to 368
bytes. For SPL it reduces from 416 to 272 bytes.
There are other things which could be attempted, for example:
- Using hlist instead of list for some lists
- Checking that only necessary fields are present in SPL
Simon Glass [Wed, 21 Aug 2024 16:19:26 +0000 (10:19 -0600)]
doc: Move generic-board documentation into rst
This information is useful for people looking at how U-Boot has changed
over the years and the design decisions which led to it. Move it into
doc/ in an 'historical' section.
Simon Glass [Fri, 23 Aug 2024 20:27:04 +0000 (14:27 -0600)]
global_data: Reduce size of early-malloc vars
The early malloc region is normally quite small and is certainly less
than 4GB, so use a 32-bit value for the limit and pointer. Update the
comments for clarity while we are here.
Simon Glass [Wed, 21 Aug 2024 16:19:16 +0000 (10:19 -0600)]
global_data: Reduce the size of mon_len
This is the length of the U-Boot binary, which is typically 200-800KB
and certainly not larger than 4GB. Use a 32-bit value to save space in
global_data and move it up to be with fields of the same alignment.
Simon Glass [Wed, 21 Aug 2024 16:19:09 +0000 (10:19 -0600)]
board_f: Add a new struct to hold pre-relocation info
Quite a few of the members of struct global_data are only used before
reloction, or have little meaning afterwards, yet they hang around in
struct global_data for the lifetime of U-Boot. This uses up precious
pre-relocation SRAM on many boards.
To help with this, start a new struct which exists only before
relocation. Move new_fdt into this new struct. Drop the display of it
in the 'bdinfo' command as it is probably not very useful.
projects / u-boot.git / log