vm_phys: Factor out some calls to vm_freelist_add()

[markj]

Summary

A subsequent patch will make this factoring more worthwhile.

No functional change intended.

[markj]

Rebase.

[dougm]

When I compile GENERIC-NODEBUG on amd64, I see that this change makes some functions grow larger. In particular, vm_phys_alloc_pages and vm_phys_enqueue_contig.

0000000000000930 <vm_phys_alloc_npages>:                        0000000000000930 <vm_phys_alloc_npages>:
0000000000000ea0 <vm_phys_alloc_pages>:                       | 0000000000000ec0 <vm_phys_alloc_pages>:
0000000000000f10 <vm_phys_alloc_freelist_pages>:              | 0000000000000f30 <vm_phys_alloc_freelist_pages>:
0000000000001180 <vm_phys_paddr_to_vm_page>:                  | 00000000000011a0 <vm_phys_paddr_to_vm_page>:
0000000000001200 <vm_phys_fictitious_to_vm_page>:             | 0000000000001220 <vm_phys_fictitious_to_vm_page>:
0000000000001270 <vm_phys_fictitious_reg_range>:              | 0000000000001290 <vm_phys_fictitious_reg_range>:
0000000000001780 <vm_phys_fictitious_unreg_range>:            | 00000000000017a0 <vm_phys_fictitious_unreg_range>:
0000000000001b40 <vm_phys_free_pages>:                        | 0000000000001b60 <vm_phys_free_pages>:
0000000000001cf0 <vm_phys_enqueue_contig>:                    | 0000000000001d10 <vm_phys_enqueue_contig>:
0000000000001e70 <vm_phys_free_contig>:                       | 0000000000001ea0 <vm_phys_free_contig>:

[markj]

Tolerate some code duplication to avoid making vm_phys.o larger.

Fix handling of "tail".

[dougm]

sys/vm/vm_phys.c

686

What is the advantage of this new function over a change that just adds the KASSERT to vm_freelist_add?

[markj]

sys/vm/vm_phys.c

686

The subsequent patch adds more callers and further extends the function.