| Age | Commit message (Collapse) | Author |
|---|
|
Follow the correct path in device_probe() when and event handler fails.
This avoids getting into a strange state where the device appears to be
activated but is not.
Signed-off-by: Simon Glass <[email protected]>
|
|
This patch introduces EVT_DM_POST_INIT_R event type for handling hooks
after relocation.
Fixes: 55171aedda88 ("dm: Emit the arch_cpu_init_dm() even only before relocation")
Suggested-by: Simon Glass <[email protected]>
Cc: Bin Meng <[email protected]>
Signed-off-by: Chanho Park <[email protected]>
Tested-by: Milan P. Stanić <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Tested-by: Roland Ruckerbauer <[email protected]>
Tested-by: Roland Ruckerbauer <[email protected]>
Fixed missing event name in event.c:
Signed-off-by: Simon Glass <[email protected]>
|
|
The "oftree_count" is the number of entries which have been set in
the oftree_list[] array. If all the entries have been initialized then
this off by one would result in reading one element beyond the end
of the array.
Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Add dev_read_addr_size_index_ptr function with the same functionality as
dev_read_addr_size_index, but instead a return pointer is given.
Use map_sysmem() function as cast for the return.
Signed-off-by: Jonas Karlman <[email protected]>
Reviewed-by: Kever Yang <[email protected]>
|
|
The behaviour of dev_read_addr_size() is surprising as it does not
handle #address-cells and #size-cells but instead hardcodes the values
based on sizeof(fdt_addr_t).
This is different from dev_read_addr_size_index() and
dev_read_addr_size_name() both of which do read the cell sizes from the
device tree.
Since dev_read_addr_size() is only used by a single driver and this
driver is broken when CONFIG_FDT_64BIT does not match the address size
in the device tree, fix the function to behave like all of the other
similarly named functions. Drop the property name argument as the only
caller passes "reg" and this is the expected property name matching the
other similarly named functions.
Signed-off-by: John Keeping <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Tested-by: Simon Glass <[email protected]> # chromebook_jerry
Tested-by: Simon Glass <[email protected]> # chromebook_bob
|
|
of_property_match_string calls of_find_property to search for the
string property.
If the device node does not exist, of_find_property returns NULL, and
of_property_match_string returns -EINVAL, which is correct.
However, if the device node exists, but the property is not found,
of_find_property still returns NULL, but it will place -FDT_ERR_NOTFOUND
in the *lenp variable.
of_property_match_string does not use the lenp parameter, thus this error
case is being lost, and treated as if the node is NULL, and returns
-EINVAL, which is incorrect.
The callers of of_property_match_string treat the error differently if
the return value is -EINVAL or -ENOENT, e.g. in dwc3 driver:
ret = generic_phy_get_by_name(dev, "usb3-phy", &phy);
if (!ret) {
ret = generic_phy_init(&phy);
if (ret)
return ret;
} else if (ret != -ENOENT && ret != -ENODATA) {
debug("could not get phy (err %d)\n", ret);
return ret;
} else {
phy.dev = NULL;
}
So the caller drivers will just consider the property missing if -ENOENT
is returned, versus the case of -EINVAL, which means something else.
To fix this situation, changed the code to call the of_find_property
with the right third argument to catch this error code and treat it
accordingly.
Signed-off-by: Eugen Hristev <[email protected]>
|
|
At present livetree can only be used for the control FDT. It is useful
to be able to use the ofnode API for other FDTs, e.g. those used by
the upcoming configuration editor.
We already have most of the support present, and tests can be marked with
the UT_TESTF_OTHER_FDT flag to use another FDT as a special case. But
with this change, the functionality becomes more generally available.
Plumb in the require support.
Signed-off-by: Simon Glass <[email protected]>
|
|
Reword this so it is easier to understand.
Signed-off-by: Simon Glass <[email protected]>
|
|
At present there are various restrictions on the use of livetree:
- It is only available once the tree is unflattened, i.e. after relocation
- It is designed to be used with the control FDT
- It can (in principle) be used with other FDTs, but only if they are
unflattened first; this is not supported
Add a few checks to make sure that any tree that is created is actually
valid. Otherwise it can be confusing when nodes and properties cannot
actually be accessed.
Signed-off-by: Simon Glass <[email protected]>
|
|
For the devfdt_get_addr_size_index_ptr() function use
map_sysmem() function as cast for the return for use in
sandbox.
Signed-off-by: Johan Jonker <[email protected]>
|
|
In case where a single timing resolution is implemented in the
device-tree, the property is named "panel-timing", as specify
in Linux kernel binding file:
Documentation/devicetree/bindings/display/panel/panel-common.yaml
# Display Timings
panel-timing:
description:
Most display panels are restricted to a single resolution and
require specific display timings. The panel-timing subnode expresses those
timings.
$ref: panel-timing.yaml#
display-timings:
description:
Some display panels support several resolutions with different timings.
The display-timings bindings supports specifying several timings and
optionally specifying which is the native mode.
$ref: display-timings.yaml#
Fixes: 0347cc773270 ("drivers: core: ofnode: Add panel timing decode.")
Signed-off-by: Raphael Gallais-Pou <[email protected]>
Signed-off-by: Patrick Delaunay <[email protected]>
|
|
The original function was only called once, before relocation. The new
one is called again after relocation. This was not the intent of the
original call. Fix this by renaming and updating the calling logic.
With this, chromebook_link64 makes it through SPL.
Fixes: 7fe32b3442f0 ("event: Convert arch_cpu_init_dm() to use events")
Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
|
|
We have several Kconfig options for ACPI, but all relate to specific
functions, such as generating tables and AML code.
Add a new option which controls including basic ACPI library code,
including the lib/acpi directory. This will allow us to add functions
which are available even if table generation is not supported.
Adjust the command to avoid a build error when ACPIGEN is not enabled.
Signed-off-by: Simon Glass <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
|
|
Add dev_read_addr_index_ptr function with the
same functionality as dev_read_addr_index,
but instead a return pointer is given.
Use map_sysmem() function as cast for the return.
Make same fix for dev_read_addr_ptr() function.
Signed-off-by: Johan Jonker <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Add devfdt_get_addr_size_index_ptr function with the same
functionality as devfdt_get_addr_size_index, but instead
a return pointer is given.
Suggested-by: Michael Nazzareno Trimarchi <[email protected]>
Signed-off-by: Johan Jonker <[email protected]>
Reviewed-by: Michael Trimarchi <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
The fdt_addr_t and phys_addr_t size have been decoupled.
A 32bit CPU can expect 64-bit data from the device tree parser,
so convert regmap_init_mem_plat() input to handel both. The
syscon class driver also makes use of the regmap_init_mem_plat()
function, but has no way of knowing the format of the
device-specific platform data. In case of odd reg structures other
then that the syscon class driver assumes the regmap must be
filled in the individual syscon driver before pre-probe.
Also fix the ARRAY_SIZE divider in the syscon class driver.
Signed-off-by: Johan Jonker <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
For the devfdt_get_addr_index_ptr() and devfdt_get_addr_size_index_ptr()
function use map_sysmem() function as cast for the return for use in
sandbox. Also fix sandbox test.
Signed-off-by: Johan Jonker <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Drop second hunk:
Signed-off-by: Simon Glass <[email protected]>
|
|
There's quite a few instances of board-specific code doing
off = fdt_path_offset(gd->fdt_blob, ...);
...
ret = uclass_get_device_by_of_offset(..., off, &dev);
looking for an eeprom or a pmic via some alias. Such code can be
simplified a little if we have a helper for directly getting a device
via device tree path (including being given as an alias).
Implement it in terms of ofnode rather than raw offsets so that this
will work whether live tree is enabled or not.
Signed-off-by: Rasmus Villemoes <[email protected]>
|
|
The functions uclass_find_device_by_phandle() and
uclass_get_device_by_phandle_id() both loop over a given uclass
looking for a device with a given phandle. Factor that out to a common
helper.
For now, there are no (known potential) users of the new helper
outside uclass.c, so make it static.
Signed-off-by: Rasmus Villemoes <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Fix warning on sandbox_spl; fix code style:
Signed-off-by: Simon Glass <[email protected]>
|
|
aliases_lookup is only referenced in of_access.c
Signed-off-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
This is used in some x86 code, so add a symbol for it.
Signed-off-by: Simon Glass <[email protected]>
|
|
Update various build and test components to use the new schema.
Signed-off-by: Simon Glass <[email protected]>
|
|
Add a CONFIG option to deal with this automatically, printing a warning
when U-Boot starts up. This can be useful if the device tree comes from
another project.
We will maintain this through the 2023.07 release, providing 6 months
for people to notice.
Signed-off-by: Simon Glass <[email protected]>
Version 4:
Acked-by: Michal Simek <[email protected]>
|
|
Tracing is typically enabled by the time driver model starts up, so there
is no point in adding a 'notrace' to the timer-init function. However,
once the driver model timer is enabled, we do need to be able to access
the timer's private data when reading the timer, so add it to the core
function needed for that.
Update the function's documentation while we are here.
Signed-off-by: Simon Glass <[email protected]>
|
|
ofnode_decode_display_timing supports reading timing parameters from
subnode of display-timings node, for displays supporting multiple
resolution, in case if a display supports single resolution, it fails
reading directly from display-timings node, to support it
ofnode_decode_panel_timing is added.
Signed-off-by: Nikhil M Jain <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Use a more accurate check for determining if the full format string will
be handled correctly, since SPL_USE_TINY_PRINTF can be disabled.
Signed-off-by: Samuel Holland <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Add a -s flag to sort the top-level devices in order of uclass ID.
Signed-off-by: Simon Glass <[email protected]>
|
|
The event framework is just that, a framework. Enabling it by itself
does nothing, so we shouldn't ask the user about it. Reword (and correct
typos) around this the option and help text. This also applies to
DM_EVENT and EVENT_DYNAMIC. Only EVENT_DEBUG and CMD_EVENT should be
visible to the user to select, when EVENT is selected.
With this, it's time to address the larger problems. When functionality
uses events, typically via EVENT_SPY, the appropriate framework then
must be select'd and NOT imply'd. As the functionality will cease to
work (and so, platforms will fail to boot) this is non-optional and
where select is appropriate. Audit the current users of EVENT_SPY to
have a more fine-grained approach to select'ing the framework where
used. Also ensure the current users of event_register and also select
EVENT_DYNAMIC.
Cc: AKASHI Takahiro <[email protected]>
Cc: Heinrich Schuchardt <[email protected]>
Reported-by: Oliver Graute <[email protected]>
Reported-by: Francesco Dolcini <[email protected]>
Fixes: 7fe32b3442f0 ("event: Convert arch_cpu_init_dm() to use events")
Fixes: 42fdcebf859f ("event: Convert misc_init_f() to use events")
Fixes: c5ef2025579e ("dm: fix DM_EVENT dependencies")
Signed-off-by: Tom Rini <[email protected]>
Tested-by: Simon Glass <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>
|
|
Currently, dm_probe_devices checks that the flags of the device contains
DM_FLAG_PRE_RELOC. However DM_FLAG_PRE_RELOC is a driver - and not a
device - flag. This means that the check in pre_reloc_only mode would
always fail.
Instead, what was aimed to be checked is that either the driver of the
device has the flag set, or that the device has the u-boot,dm-pre-reloc
Device Tree property set.
So let's fix the check to allow u-boot,dm-pre-reloc devices to be
probed.
Cc: Quentin Schulz <[email protected]>
Signed-off-by: Quentin Schulz <[email protected]>
|
|
The rest of the unmigrated CONFIG symbols in the CONFIG_SYS namespace do
not easily transition to Kconfig. In many cases they likely should come
from the device tree instead. Move these out of CONFIG namespace and in
to CFG namespace.
Signed-off-by: Tom Rini <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
We should only perform additional iteration steps when needed to
initialize the parent of a device. Other binding errors (such as a
missing driver) should not lead to additional iteration steps.
Unnecessary iteration steps can cause issues when memory is tightly
constrained (such as in the TPL/SPL) since device_bind_by_name()
unconditionally allocates memory for a struct udevice. On the SanCloud
BBE this led to boot failure caused by memory exhaustion in the SPL
when booting from SPI flash.
Signed-off-by: Paul Barker <[email protected]>
|
|
Use dot instead of comma. The fix doesn't affect anything but it is good to
be aligned with used pattern. The first is used only for string size
calculation and the second change is in the comment.
Fixes: db681d4929ca ("net: phy: Add new read ethernet phy id function")
Signed-off-by: Michal Simek <[email protected]>
|
|
When probing a device fails NULL pointer is returned, and following
devices in uclass list cannot be iterated. Skip to next device on error
instead.
With that the only condition under which these simple iteration
functions return error is when the dm is not initialized at uclass_get
time. This is not all that interesting, change return type to void.
Fixes: 6494d708bf ("dm: Add base driver model support")
Signed-off-by: Michal Suchanek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
This is needed in the platforms that use "ranges" node property for
address translation in their dts for TPL.
Signed-off-by: William Zhang <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
The call to device_bind_with_driver_data() passes id->data but if
the entry has no of_match then the id has not been set by the selected
driver.
Normally this passes unnoticed since a previous driver likely had an
of_match value, so the id is set to that. Of course it is not correct
to pass the id->data from a different driver.
With clang-14 the driver ordering is such that the id is never actually
set in the 'bind /usb@1 usb_ether' line in test_bind_unbind_with_node()
thus causing a crash.
Fix this by passing 0 if the of_match for a driver does not exist.
Signed-off-by: Simon Glass <[email protected]>
|
|
There are no longer any platforms which do not enable DM, move this to a
def_bool y and remove the check in the Makefile.
Cc: Simon Glass <[email protected]>
Signed-off-by: Tom Rini <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
The _err variant iterators use the simple iterators without suffix as
basis.
However, there is no user that uclass_next_device_err for iteration,
many users of uclass_first_device_err use it to get the first and
(assumed) only device of an uclass, and a couple that use
uclass_next_device_err to get the device following a known device in the
uclass list.
While there are some truly singleton device classes in which more than
one device cannot exist these are quite rare, and most classes can have
multiple devices even if it is not the case on the SoC's EVB.
In a later patch the simple iterators will be updated to not stop on
error and return next device instead. With this in many cases the code
that expects the first device or an error if it fails to probe may get
the next device instead. Use the _check iterators as the basis of _err
iterators to preserve the old behavior.
Signed-off-by: Michal Suchanek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
uclass_probe_all uses uclass_first_device/uclass_next_device assigning
the return value.
The interface for getting meaningful error is
uclass_first_device_check/uclass_next_device_check, use it.
Also do not stop iteration when an error is encountered. Probing all
devices includes those that happen to be after a failing device in the
uclass order.
Fixes: a59153dfeb ("dm: core: add function uclass_probe_all() to probe all devices")
Signed-off-by: Michal Suchanek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
A recent change to regmap breaks building of phycore-rk3288 for me. The
difference is only a few bytes. Somehow CI seems to pass, even though it
fails when I run docker locally. But it prevents me from sending any more
pull requests.
In any case this board is clearly near the limit. We could revert the
offending change, but it is needed for sandbox tests.
Instead, add a way to drop the range checks in SPL, since they end up
doing nothing if everything is working as expected.
This makes phycore-rk3288 build again for me and reduces the size of SPL
slightly for a number of boards.
Signed-off-by: Simon Glass <[email protected]>
Fixes: 947d4f132b4 ("regmap: fix range checks")
|
|
On the 32bit ARM sandbox 'dm ut dm_test_devm_regmap' fails with an abort.
This is due to incorrect range checks.
On 32-bit systems the size of size_t and int is both 32 bit. The expression
(offset + val_len) is bound to overflow if offset == -1. Add an overflow
check.
Signed-off-by: Heinrich Schuchardt <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Add a function to copy properties from one node to another.
Signed-off-by: Simon Glass <[email protected]>
|
|
At present ofnode_write_prop() is inconsistent between livetree and
flattree, in that livetree requires the caller to ensure the property
value is stable (e.g. in rodata or allocated) but flattree does not, since
it makes a copy.
This makes the API call a bit painful to use, since the caller must do
different things depending on OF_LIVE.
Add a new 'copy' argument which tells the function to make a copy if
needed. Add some tests to cover this behaviour.
Signed-off-by: Simon Glass <[email protected]>
|
|
We need to be able to look up phandles in any FDT, not just the control
FDT. Use the 'other' FDT to test this, with a helper function which gets
this as an oftree that can then we used as needed.
Add a few more tests and some comments at the top of the file, to explain
what is going on.
Signed-off-by: Simon Glass <[email protected]>
|
|
Add the logic to redirect requests for the device tree through a function
which can look up the tree ID. This works by using the top bits of
ofnode.of_offset to encode a tree.
It is assumed that there will only be a few device trees used at runtime,
typically the control FDT (always tree ID 0) and possibly a separate FDT
to be passed the OS.
The maximum number of device trees supported at runtime is 8, with this
implementation. That would use bits 30:28 of the node-offset value,
meaning that the positive offset range is limited to bits 27:0, versus
30:1 with this feature disabled. That still allows a device tree of up
to 256MB, which should be enough for most FITs. Larger ones can be
supported by using external data with the FIT, or by enabling OF_LIVE.
Update the documentation a little and fix up the comment for
ofnode_valid().
Signed-off-by: Simon Glass <[email protected]>
|
|
At present, unless OF_LIVE is enabled, ofnode only supports access to one
device tree, the control FDT. This is because only the node offset is
encoded in ofnode, with the tree being implicit.
This makes ofnode (without OF_LIVE) unsuitable for device tree fixups, as
implemented by ft_board_setup() and other such functions.
To solve this, we can use the top bits of the node offset to hold a tree
ID.
Add the definitions for this.
Signed-off-by: Simon Glass <[email protected]>
|
|
This function turns out to be a little confusing since it looks up a path
and also registers the tree. Split it into two, one that gets the root
node and one that looks up a path, so the purpose is clear.
Registering the tree will happen in a function to be added in a later
patch, called oftree_from_fdt().
Signed-off-by: Simon Glass <[email protected]>
|
|
In some cases we want to obtain an ofnode in the same tree as a different
ofnode, such as when looking up a subnode. At present this is trivial,
since there is only one tree. When there are multiple trees, this
implementation will change.
Also move the ofnode_to_offset() function up higher in the header file,
since we will need to provide a different implementation with multiple
trees.
Signed-off-by: Simon Glass <[email protected]>
|
|
When we have multiple trees, the ofnode logic needs to be told which one
to use. Create a new function which takes an oftree argument, along with
a helper to obtain the FDT pointer from an oftree.
Signed-off-by: Simon Glass <[email protected]>
|
|
The flat device tree is assumed to be the control FDT but this is not
always the case. Update the ofnode implementation to obtain the node via
an function call so we can eventually add support for selecting different
trees.
Signed-off-by: Simon Glass <[email protected]>
|
|
The current tests do not cover all the behaviour. Add some more.
Tidy up a few inconsistencies between livetree and flattree which come to
light with these tests. Also drop the -ENODATA error since it is never
actually returned.
Signed-off-by: Simon Glass <[email protected]>
|
