Merge pull request #39 from DataCoreSoftware/zfs-454

Periodically calling 'vmem_update' function to dynamically update the vmem->vm_hash_table size.

Inno change to x64 program files default (openzfsonwindows#181)

Correct WIN32 paths after bulk git merge

fix ZVOL blocksize

decrease impact of delete delay

Silence mze warning in zap_micro.c

Clean up Security methods

Incorrect IOCTL defined after merge

cv_wait to use literal waiters instead of address

Since address would never == 1, the broadcast would never clear
and txg would continuously sync.

Work around for deltree race

If we have pending deletes, wait for them to complete before we
deny directory deletion

FsTeardown should be in CLOSE and zil_commit iocount

Following fastfat, we should not FsRtlTeardownPerStreamContexts in
CLEANUP but rather in CLOSE.

zil_commit via zfs_fsync() / FLUSH_BUFFERS calls zget with WITHOUT_VNODE
and such should not release zp in zfs_get_data().

make some fileobjects spinlock tighter, to avoid calling kmem_alloc
in between. Should consider rwlocks, since spinlocks are expensive.

NULL zfsvfs checks

Got a call to getpartitionex and even though the zmo tests passed,
zfsvfs was NULL, causing panic. Handle it with more grace.