Ref #2622

running the transaction.

almost certainly an error, so make it one.

Fixes #2613

fixes #2639

Not real ids; for consistency with the other platforms.

note on jail_solaris:

ideally, vmods could tell the jails about privileges they require, but
for now (and because vmod_unix lives in varnish-cache), just add the
required privilege to the permitted set.

I have also considered the option to add additional privileges via the
-j argument (and actually would still want to add that somewhen), but
for this purpose, varnish should really DTRT by default.

note on priv_allocset:

Most of the sun folk had proven to be good interface designers, but
an API which requires dynamic allocation/deallocation really does not
play well with my efficiency fetish. So good we got library
constructors/destructors.

note on getpeerucred:

basically the same thing, but this time they at least added
ucred_size(). Not intended for use like this, but anyway....

Negative character constant

Last -need-bytes kept to test that it still works.

Also reappropriate test a00001 to exercise the Teken code with
the vttest (http://invisible-island.net/vttest/) program, if installed.

This allows us to use TERM=xterm.

default options:

For co-existance with libc malloc and potentially other allocators, we
should use mmap() regions in favor of sbrk(). Also, libumem's perthread
cache only makes sense when umem is actually used for all allocations
and, by default, requires an additional massive 1MB per thread, which
is totally out of scale with how we manage memory in varnish.

if libumem is found to already be loaded, we cannot set its options,
so warn about that fact.

As we only use function pointers to libumem, lazy loading of symbols
should be the better option.

libumem's atfork handler will redefine malloc() and friends so if we
load it in the manager, we cannot confine its use to the stevedore

We do this by not linking with libumem, but rather using dlopen/dlsym
to load the slab allocator interface, which is used by the stevedore.