Suggest to create a separate patch for this renaming.

how about zfs? or other filesystem doesn't support suspend?

hv_kvp_daemon and hv_vss_daemon has something common to share. Error code is one of them. Another is negotiation with host. That is why this patch impacts kvp daemon. Of course, I can separate them to another patch.

As I know, ZFS does not support it. I need to double check it. If we want to extend this driver to support the other filesystem, what we need to modify is the daemon. So, the effort is not big.

This path is already defiined in paths.h.

Why do you need additional define for 'unopened handle' ? Common unix practice is to mark closed file descriptor with the -1 value, which is very convenient since error return from open(2) is also -1. Below you test for -1 and UNOPENED_HANDLE, which would remove several lines of code if testing only for -1.

Use ANSI C definitions (freeze(void)).

space between for and '('.

This is excessive. The suspend ioctl does race-free sync when needed (after the writes are blocked), and only for the suspended volume.

This is rather useless label. Why not write return (1); instead of all error = 1; goto handle_error; places ?

Use ANSI C.

You silently ignore invalid options or stray arguments there.

Excessive ().

Do not do that. errno values and exit(3) codes are different value spaces. Please see sysexits(3) for common exit(3) codes.

This return is not needed, and it is in fact strange to see use of exit() and return mixed in the main() function.

Yeah, I'll remove UNOPEND_HANDLE.

Yeah.

Thanks. That is what I want to confirm.

Sure.

Thanks for your example.

Oh, I need to print usage information for wrong options.

Ok.

Thanks for pointing this preferable exit.

Yes.

It was whacked.

Add this after dev/hwpmc. This list is sorted in alphabetic order.

Probably after evdev, not acpica. ditto

After evdev.

After evdev.

Use __packed.

callout_reset(..., hz * sec, );

Better comment it as: reserved 0.