Fix various Coverity-detected errors in nvmecontrol

[dab]

Summary

This fixes several Coverity-detected errors in
nvmecontrol. While in here, a couple additional errors with shift/mask
confusion that were not diagnosed by Coverity are also fixed.

CIDs addressed: 1040299, 1040300, 1403972, 1403973, 1403985,
1403988, 1403990, 1404374, 1404427, 1404469, 1404510, 1404534, 1418118

CID 1403657 (resource leak of shared library handle) was marked "intentional" in the Coverity scan database.

Test Plan

Run affected nvmecontrol commands to ensure code still
functions as intended.

[vangyzen]

sbin/nvmecontrol/identify.c

97

You probably want _SHIFT instead of _MASK.

sbin/nvmecontrol/passthru.c

194–199

This message will be misleading for a short read.

258

I wouldn't bother with the condition.

[dab]

sbin/nvmecontrol/identify.c

97

Egad! Can't believe I fixed the PIT1 but missed the SHIFT/MASK confusion on this one. Thanks.

[imp]

Honestly, don't bother to unload the sharee libraries. It's a total waste of effort.

sbin/nvmecontrol/comnd.c

293

Why?

sbin/nvmecontrol/nvmecontrol.c

187

This makes 0 sense to me.

sbin/nvmecontrol/passthru.c

260

Not needed either. Just close it unconditionally.

[dab]

Fix a couple issues pointed out in review comments.

[dab]

contrib/tcpdump/README

1

Ignore this file; I had to rebuild my environment and somehow this got marked as changed. I'll fix it before committing.

[dab]

Correct diff; the previous one had some unintended files included.

[dab]

Address comment about check on fd > -1 not being needed in passthru.c.

[dab]

sbin/nvmecontrol/comnd.c

293

Loading the library leaks the handle, so I implemented the list of handles and this function to address that issue. I know that it isn't strictly needed as exiting the program will unload the libraries. I didn't experiment to see if just stashing the handles in the list is enough to quiet Coverity or if it would complain about them not being deallocated by calling dlclose as this does. Assuming it works, would that approach be more acceptable?

[imp]

sbin/nvmecontrol/comnd.c

293

No other program on the planet cleans up for things like this, at least not that I've ever encountered...
I'd rather we tag the error in Coverity as a false positive. This code buys us nothing but extra run time for no benefit.

[dab]

Remove the shared library unloading code at imp@'s suggestion.