device: add api to check if device initialized or not

[wentongwu]

Add api to check if device initialized success or not and
power management will create device list based on this API.

Fixes: #24750
Signed-off-by: Wentong Wu wentong.wu@intel.com

[wentongwu]

The only failed case is not related to this PR, re-tun CI

2/2 qemu_x86_64               tests/drivers/ipm/peripheral.mailbox               FAILED: timeout
-------------out-3nd-pass/qemu_x86_64/tests/drivers/ipm/peripheral.mailbox/handler.log--------------
�cSeaBIOS (version rel-1.12.1-0-ga5cab58e9a3f-prebuilt.qemu.org)
Booting from ROM..***** Booting Zephyr OS v1.14.2-1-gad387464460f *****
starting test - Test IPM
ipm_recv0: 'everything is awesome'
ipm_recv0: 'Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor '
ipm_recv0: 'incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nosASSERTION FAIL [z_spin_lock_valid(l)] @ ../../../../../../include/spinlock.h:95
	Recursive spinlock
*** FATAL ERROR vector 6661 code 1016
***  RIP 8:0x100dc7 RSP 16:0x105908 RFLAGS 0x2
***  RAX 0x105017 RCX 0xa RDX 0x3f8 RSI 0x5f RDI 0x104fc9
***  R8 0xffffffff R9 0x0 R10 0x0 R11 0x0
-------------out-3nd-pass/qemu_x86_64/tests/drivers/ipm/peripheral.mailbox/handler.log--------------
1 of 2 tests passed with 0 warnings in 70 seconds 

[pabigot]

Change to use the prefix that all other operations on devices use.
Don't put everything into the loop twice.

[pabigot]

Why is this loop repeated?

[wentongwu]

Why is this loop repeated?

@pabigot we should put the core devices at the first, but we don't know how many core devices are initialized like before. The first loop is to find the core devices and then other devices.

[pabigot]

Also this is targeting v1.14 but doesn't say that anywhere. It needs to be added to master and backported.

[andrewboie]

why is this against the 1.4 branch and not master?

[wentongwu]

why is this against the 1.4 branch and not master?

@andrewboie @pabigot because master is doing the refactor of device module and also PM module. Thanks

[pabigot]

I'm going to say "no" to this until we have guidance on the conditions for changing a stable release to add capability that is not currently supported in master.

In particular while device_is_initialized() is locally adequate, it is inconsistent with device_check_status() which is being proposed in #23589 (+ #24873) in the context of #22941. At this time that API hasn't gotten enough attention to determine whether it's the right approach.

If something needs to be done in 1.14 there should be an issue describing exactly what needs to be done, and why it's important enough to do even when it's not compatible with master. #24750 is generic, and is currently neither classified as a bug nor prioritized.

@nashif @carlescufi please provide programmatic guidance.

[tbursztyka]

I am not commenting the bug and the fix itself, but there is no need to add an API for a bug fix. Just check device->driver_api pointer where relevant directly (as it is being done in various places). It's not "nice" indeed, but adding an API won't fit with 1.14 maintenance I think.