Add sync_file_range(2) to linux(4).

[trasz]

Summary

Add sync_file_range(2) implementation to linux(4); underneath it's a wrapper over the usual fsync(2). This silences some warnings when running "apt-get upgrade".

[emaste]

sys/compat/linux/linux_file.c

916 ↗

(On Diff #64319)

This is a perfectly reasonable way to implement this in the linuxulator, but we should have some way to track this in the future, maybe a probe as done elsewhere in the linuxulator?

sys/compat/linux/linux_file.h

137 ↗

(On Diff #64319)

Should have a comment sync_file_range flags?

[brooks]

sys/arm/linux/syscalls.master

1701 ↗

(On Diff #64319)

Remove the ); on this line. It's interesting that this worked at all...

sys/arm64/linux/syscalls.master

457 ↗

(On Diff #64319)

Another extra );. Note that the sys calls above have the wrong style (); should be on it's own line).

[trasz]

sys/compat/linux/linux_file.c

916 ↗

(On Diff #64319)

I’m not quite sure what you mean?

[trasz]

Fix most of the problems.

[emaste]

sys/compat/linux/linux_file.c

916 ↗

(On Diff #64319)

sorry I didn't provide enough context.

From linux sync_file_range manpage:

offset is the starting byte of the file range to be synchronized.
nbytes specifies the length of the range to be synchronized, in
bytes

so for the purpose of the linuxulator it's reasonable to translate this to plain kern_fsync, but we should have some reminder or other way to know we should revisit this in the future if we grow a FreeBSD interface for fine-grained sync

[brooks]

The syscalls.master bits look correct now.

[trasz]

sys/compat/linux/linux_file.c

916 ↗

(On Diff #64319)

Ah, so an XXX comment would do?

[trasz]

(Tinderboxed.)

[emaste]

sys/compat/linux/linux_file.c

916 ↗

(On Diff #64319)

Sure, I'd say ideally a LIN_SDT_PROBE* but at least an XXX comment.