devicetree-based device definitions and dependency representations reboot #32127
Conversation
pabigot
requested review from
andyross,
dcpleung,
jakub-uC,
nashif,
nordic-krch,
tbursztyka and
tejlmand
as code owners
github-actions
bot
added
area: API
zephyrbot
requested review from
carlescufi,
ceolin,
galak and
mbolivar-nordic
|
Added to dev-review to re-introduce to the community and solicit reviews. |
pabigot
force-pushed
the
nordic/20210124a
branch
2 times, most recently
from
3c3a9d5 to
48291ac
Compare
pabigot
force-pushed
the
nordic/20210124a
branch
from
48291ac to
3e2430a
Compare
Yes, they're useful for diagnostics and shell identification. However, if we're going to do that, then we should reverse the in-progress course of removing The |
include/device.h
Outdated
| * pointer if @p dh does not provide dependency information. | ||
| */ | ||
| static inline const device_handle_t * | ||
| device_get_requires_handles(const struct device *dev, |
There was a problem hiding this comment.
how about device_dependency_get(..)
There was a problem hiding this comment.
The problem is that when we support dynamic dependencies this API won't work, because we won't have all devices in a contiguous array. We'll need a callback-based API to iterate over both static and dynamic dependencies.
But I can make this device_requires_handles_get() which would handle the static fetch with naming closer to the proposed standard.
There was a problem hiding this comment.
@nashif Changed to device_required_handles_get(). Turns out when this was reverted the test code was inadvertently left in tree because it had been patched subsequent to the original merge.
There was a problem hiding this comment.
the description of the API is Get the set of devices on which a given device depends, so dependency_get would have reflected that.
I see you changed that to device_required_handles_get, this is fine I guess (better than device_requires_handles_get).
There was a problem hiding this comment.
pabigot
force-pushed
the
nordic/20210124a
branch
from
2b9a09f to
117f79e
Compare
Generate arrays of dependency information for each device. If a device definition is being constructed from devicetree these come from the devicetree dependency information. Additional dependencies may be passed through using the DT_ macros. Define flag values for device handles so we can partition the dependency array into distinct sets, which include things it requires, things it supports (may not be needed), and child nodes (not implemented, may not be needed). Signed-off-by: Peter Bigot <[email protected]>
Following the idiom used for system calls, add script support to read the initial application binary to identify which devices are defined, and to use their offset in the device array as their unique handle rather than the externally-defined ordinal from devicetree. The device dependency arrays are updated to use these handles. Signed-off-by: Peter Bigot <[email protected]>
Verify that device dependencies are encoded into and retrievable from the device structure. Signed-off-by: Peter Bigot <[email protected]>
Refactor the output of device list to use standard API to retrieve the list of devices, and to always display a status rather than hiding disabled/failed devices. Add API to associate a distinct identifier with any "device" that does not have a name. Where a device has requires dependencies display the devices on which it depends. Signed-off-by: Peter Bigot <[email protected]>
Fix the order so that it reflects the actual initialization order, rather than putting PRE_KERNEL initializations after APPLICATION. Add SMP in the proper location. Use the helper function to provide unique identifiers for "devices" that don't have a device pointer (so don't have a name). Signed-off-by: Peter Bigot <[email protected]>
pabigot
force-pushed
the
nordic/20210124a
branch
from
117f79e to
e337bd8
Compare
|
Recent build failures are still due to #32344. |
The script gen_handles.py was introduce in zephyrproject-rtos#32127 but relies on ZEPHYR_BASE being set in environment. However, it is not a requirement to set Zephyr base in the users environment, therefore this is changed to an argument `-z` or `--zephyr-base` which will be passed from the build system to the script. If `-z` or `--zephyr-base` is not provided, the environment will be checked for a ZEPHYR_BASE setting there. Signed-off-by: Torsten Rasmussen <[email protected]>
The script gen_handles.py was introduce in #32127 but relies on ZEPHYR_BASE being set in environment. However, it is not a requirement to set Zephyr base in the users environment, therefore this is changed to an argument `-z` or `--zephyr-base` which will be passed from the build system to the script. If `-z` or `--zephyr-base` is not provided, the environment will be checked for a ZEPHYR_BASE setting there. Signed-off-by: Torsten Rasmussen <[email protected]>
... argument The script gen_handles.py was introduce in zephyrproject-rtos#32127 but relies on ZEPHYR_BASE being set in environment. However, it is not a requirement to set Zephyr base in the users environment, therefore this is changed to an argument `-z` or `--zephyr-base` which will be passed from the build system to the script. If `-z` or `--zephyr-base` is not provided, the environment will be checked for a ZEPHYR_BASE setting there. Signed-off-by: Torsten Rasmussen <[email protected]>
This PR resurrects the features of #29644 that were reverted in #31548 with a few changes: