i have some simple changes that address most of the (currently suppressed) compiler warnings in mac, but there are a few bits i'm unsure of (and might require more digging, depending on time)... would it be worth throwing up what I have that's pretty straightforward (mostly just initializing values)

even if it's not entirely complete?

I'm all in favour of removing warning suppression, yes please, and I can work through it - I'm always wary of changes that just initialise values :)

i did go through and look at the relevant code for these

that's why there some that i wasn't sure about and left them

(well for some unrelated work, i do have a separate change i'm not planning to put up for now that just quiets them only so i can build my changes to make sure they don't add any)

I didn't mean to imply you didn't, bad choice of words there. I just meant that I always give them a bit more scrutiny and I'm happy to review.

illumos/illumos-gate 8e1b7a3 if you want a preview

there's actually one (not in that change) that I'm completely baffled by: src.illumos.org/source/xref/illumos….c?r=119d61cc&mo=87558&fi=3057#3057

it's complaining that on line 3207, flent is uninitialized (which it is)

oh wait.. i guess i just rubber ducked it

that argument is completely ignored in mac_client_datapath_teardown

(or i guess overwritten before any passed-in value is ever used)

.... which i guess brings up why it's a parameter in the first place if it's not used

it doesn't look (from first glance) to be something that's used as a callback

I agree with what you've said. The other call site for mac_client_datapath_teardown passes in what will be used anyway. Looks to me like the parameter could be removed.

yeah..

let me update that branch with that, plus a few other things I missed

now the only thing left is this: src.illumos.org/source/xref/illumos…c?r=2aaafd60&mo=179446&fi=6756#6882

we're ignoring the return value

but i'm not sure if we should just continue to do that, or if we should try to 'undo' what was done earlier

like some of the earlier code in the function

though it's all for functionality that seems to only be present on some NICs that I think were sparc-only

and i don't want to just silence it if that's a legitimate bug

that's been masked

[illumos-gate] 15986 svcs can stumble over file dependencies -- Mike Zeller <mike⊙mn>

[illumos-gate] 15995 2GB arc_c_min ceiling should be 1GB -- Dan McDonald <danmcd⊙mi>