add spinlock_type.h

[hujun260]

Summary

add spinlock_type.h
reason:
Due to incomplete handling of spinlock_t in arch/spinlock.h, it should not be used directly by external code.
Furthermore, because pthread.h and nuttx/spinlock.h have a circular dependency, pthread.h cannot successfully include nuttx/spinlock.h. Therefore, we have split spinlock_type.h from spinlock.h.

Impact

spinlock

Testing

ci

[nuttxpr]

[Experimental Bot, please feedback here]

No. This PR description lacks crucial details required by the NuttX guidelines. Here's why and what's missing:

• Insufficient Summary: While the summary states what changed, it doesn't adequately explain how it works. It mentions circular dependencies and incomplete handling, but lacks specifics. For example:

  • How was the spinlock_t incompletely handled before? Give a concrete example.
  • Precisely what went into spinlock_type.h and what stayed in spinlock.h?
  • What code now includes spinlock_type.h that previously included spinlock.h?

• Incomplete Impact Assessment: "spinlock" is not a sufficient impact description. Address all the impact points individually. At the very least, explicitly state "NO" for those that aren't applicable. For example:

  • Impact on user: Does the user need to change any code due to this split?
  • Impact on build: Does the build system need any modifications?
  • Impact on hardware: Are any architectures or boards affected?
  • Impact on documentation: Does this change necessitate documentation updates?
  • ...and so on for all the impact categories.

• Insufficient Testing: "ci" is not adequate testing verification. Provide:

  • Specific build host details (OS, compiler version, etc.).
  • Specific target details (architecture, board, configuration).
  • Actual before and after logs demonstrating the problem and the fix. Simply stating that CI passed isn't enough; the reviewer needs to see the evidence.

In short, this PR description needs significantly more detail to meet NuttX requirements. Focus on providing concrete information and evidence to support the changes.

[xiaoxiang781216]

add spinlock_type.h reason: Due to incomplete handling of spinlock_t in arch/spinlock.h, it should not be used directly by external code. Furthermore, because pthread.h and nuttx/spinlock.h have a circular dependency, pthread.h cannot successfully include nuttx/spinlock.h. Therefore, we have split spinlock_type.h from spinlock.h.

But the circular dependency doesn't introduce any bad side effect now?

[hujun260]

add spinlock_type.h reason: Due to incomplete handling of spinlock_t in arch/spinlock.h, it should not be used directly by external code. Furthermore, because pthread.h and nuttx/spinlock.h have a circular dependency, pthread.h cannot successfully include nuttx/spinlock.h. Therefore, we have split spinlock_type.h from spinlock.h.

But the circular dependency doesn't introduce any bad side effect now?

Yes, we haven't encountered this situation yet. In Linux, there is a similar concept called "spinlock_types.h".

[xiaoxiang781216]

let's hold this patch until the recursive happen?

[hujun260]

let's hold this patch until the recursive happen?

ok