Fix leaks and test for getpagesize() returning == -1

[ngie]

Summary

Fix leaks and test for getpagesize() returning == -1

• close file descriptors after use.
• Always munmap memory regions after mmap'ing them.
• Make sure getpagesize() returns a value greater than 0 and use a cached value instead of always calling getpagesize(3).

MFC after: 2 weeks
Reported by: Coverity
CID: 1331374-1331377, 1331653-1331662
Sponsored by: EMC / Isilon Storage Division

Test Plan

sudo kyua test -k /usr/tests/sys/vm/Kyuafile

[cem]

tests/sys/vm/mmap_test.c

257–258

What about munmaping p1, p2, p3?

[ngie]

tests/sys/vm/mmap_test.c

257–258

Good catch -- fixed :).

[jhb]

Ok I guess?

I've yet to see the ATF stuff ever run more than one test in a process. It seems to always want to start a new process to run each test, so doing the cleanups here is just slowing things down (which seems somewhat ironic given that you are micro-optimizing getpagesize() which will never fail).

[asomers]

In D6011#128048, @jhb wrote:

Ok I guess?

I've yet to see the ATF stuff ever run more than one test in a process. It seems to always want to start a new process to run each test, so doing the cleanups here is just slowing things down (which seems somewhat ironic given that you are micro-optimizing getpagesize() which will never fail).

True, there is a separate process for each ATF test. But I think the cleanup is worth it just to make Coverity shut up. It helps to find the real Coverity bugs, like the buffer overflow that cem fixed yesterday.