device: add API to iterate over devices and check their validity #26342
Conversation
Device objects in Zephyr are currently placed into an array by linker scripts, making it easy to iterate over all devices if the array address and size can be obtained. This has applications in device power management, but the existing API for this was available only when that feature was enabled. It also uses a signed type to hold the device count. Provide a new API that is generally available, but marked as internal since normally applications should not iterate over all devices. Mark the PM API approach deprecated. Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
pabigot
requested review from
andrewboie,
andyross,
jakub-uC,
MaureenHelm,
nashif and
nvlsianpu
as code owners
zephyrbot
added
area: Sensors
pabigot
force-pushed
the
nordic/20200622b
branch
from
9182054 to
3077c14
Compare
Several shell modules use cloned code to iterate over all devices and identify the nth instance that meets some criteria. The code was repetitive and included various errors. Abstract to a helper function that performs the check consistently. Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
include/device.h
Outdated
| * | ||
| * @return true if and only if the device is available for use. | ||
| */ | ||
| static inline bool z_device_ok(const struct device *dev) |
There was a problem hiding this comment.
I have a similar patch for device const-ify stuff (name is different and obviously it does not check on driver_api attribute, plus other details). Since that one is not very relevant for that PR, could your remove that commit?
There was a problem hiding this comment.
I wondered if that might conflict, but figured that outside the potential name difference it should be easy to adapt to either. Removed anyway.
I do think we should be consistent in naming, i.e. agree whether the check for validity and fetch of static device pointers should be device_* or z_device_* functions.
There was a problem hiding this comment.
Can't close #24750 without that patch, though.
There was a problem hiding this comment.
indeed, I can pull my patch out for it then
[edit] just a sec
There was a problem hiding this comment.
Hum it's inside a patch that revise the initialization status part.
Well, let's cut the pear in half: call it device_ready() and that will be fine for me then
There was a problem hiding this comment.
I'll do that.
But if this is device_ready then perhaps z_device_get_all_static() should be device_get_all_static(). I don't see applications having a use for device_ready(), as they can't get a device pointer unless it's ready. (In a future Zephyr when they can, e.g. when devices can be made unready, we might have to add more API.)
There was a problem hiding this comment.
It's back, but I kept the internal-use prefix (which I may be mis-using, not sure). This path allows device_ready() to remain available for future situations where somebody might get a device pointer without going through a validating API like device_get_binding(), or where a device may be made unready through power management.
pabigot
force-pushed
the
nordic/20200622b
branch
from
3077c14 to
bf3c1e5
Compare
Currently this is useful only for some internal applications that iterate over the device table, since applications can't get access to a device that isn't ready, and devices can't be made unready. So it's introduced as internal API that may be exposed as device_ready() when those conditions change. Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
The power management infrastructure added a generic API to access the static device list, but it is not generally available. Provide an internal standard API for this.
Replace all the shell infrastructure that iterated over devices with a helper that implements it once. This incidently fixes bugs related to using a pointer before checking it for null, and matching a prefix wherever it occurred in the device name.
Finally add standard API to determine whether a device is OK to use, so we can replace the nulling of the API pointer with something else.
Fixes #24750