vfs:add nxsched_foreach to sched_lock avoid crash

[anjiahao1]

Summary

vfs:add sched_lock to sync avoid crash
If scheduling occurs in file_fsync,
fl_lock may be released, and an error may
occur when calling nxmutex_unlock

Impact

call nxsched_foreach

Testing

sim:nsh

[pkarashchenko]

I'm missing detailed description of the issue that this change is intended to fix. This is needed to justify why locking scheduler is needed inside the nxsched_foreach interface compared to case when nxsched_foreach caller should lock scheduler.

[patacongo]

This is a special case of the more general Issue: #1138. I don't think we would want such a specific solution if a more general solution is possible.

But there are also may be places where we want to be suspended while we are in a critical section. Perhaps some logic depends on this current behavior. Perhaps this should still be in enter/leave_critical_section with an parameter to permit suspension or not.

[anjiahao1]

If scheduling occurs in the handle, then some system resources are likely to be corrupted

[xiaoxiang781216]

I'm missing detailed description of the issue that this change is intended to fix. This is needed to justify why locking scheduler is needed inside the nxsched_foreach interface compared to case when nxsched_foreach caller should lock scheduler.

@pkarashchenko as the discussion in #1138, the thread holds the critical section will be scheduled out if it wakes up other high priority thread, sched lock could prevent this happen.

[xiaoxiang781216]

@pkarashchenko do you have more queston about this patch?

[pkarashchenko]

I do not see @patacongo 's comment is answered. Currently I'm fine with this if it does not bring any regression.