Add dtrace probe support for zfs SET_ERROR(..)

[smh]

ATM this is the zfs module which I believe makes it the same as upstream illumos.

Should we keep that or move to the opensolaris module?

[markj]

sys/cddl/compat/opensolaris/kern/opensolaris_dtrace.c

3

Shouldn't this just be 2014?

sys/cddl/compat/opensolaris/sys/sdt.h

35

The SDT_PROBE_DECLARE can be lifted out of the #ifdef, I think.

sys/sys/sdt.h

85 ↗

(On Diff #1651)

Is this just needed for the opensolaris_dtrace.c you added? If so, could you move these includes there instead?

[smh]

Corrected date in license of new file based on feedback from markj.

[smh]

Thanks for the review Mark, if you have any more concerns after reading my responses to your last batch LMK, otherwise if you can approve and I'll get this into the tree.

sys/cddl/compat/opensolaris/kern/opensolaris_dtrace.c

3

Yep I copied from the license I was pointed too ;-)

sys/cddl/compat/opensolaris/sys/sdt.h

35

Nope it if you do that it causes build failures when KDTRACE_HOOKS are removed from the kernel. If it where using the macros then yes but as we're hardcoding it due to static requirements then no.

sys/sys/sdt.h

85 ↗

(On Diff #1651)

Yes but this fixes it to match the documented required includes.

Its not showed up previously because where its included before already had these for other reasons.

[markj]

sys/sys/sdt.h

85 ↗

(On Diff #1651)

I'd say that that's a bug in the documentation then (assuming you're referring to SDT(9)). :)

I still prefer moving the #includes to opensolaris_dtrace.c and updating the man page, but it's up to you.

[smh]

Added missing includes to SDT (9) man page and added them to the new kernel file instead of adding them to sdt.h as per markj's feedback.

[markj]

In D790#11, @smh wrote:

Added missing includes to SDT (9) man page and added them to the new kernel file instead of adding them to sdt.h as per markj's feedback.

Thanks! LGTM.