They are shared with the cache process but are never used. Only the VCC
process uses them, but they are never set. This specific fchown(2) call
in the VCC process was probably a no-op in the first place: since the
fields are never set this is transferring ownership to root:root and if
that succeeded the process was already root in the first place. If it
failed, we never see the error message since we lacked root privileges.

Both the unix and solaris jails are designed to run VCC (and CC) with
limited privileges, and in the absence of a jail, the outcome should be
the same: VCC creates a file with credentials suitable for the next CC
invocation.

As I suspected, the -Werror setup for libvgz C flags were needed to
properly discard them for suncc. Both warnings are conditionally not
turned into errors, so -Wno-unknown-warning-option should no longer
be needed.

Refs madler/zlib#633

Trying to fix the build for clang 15 actually broke the build for GCC.

The -Werror that was initially set after saving CFLAGS was meant to be
part of NO_VIZ test. We turn warnings into errors later in the configure
script so at this point we shouldn't care about it. If we really do, we
can move this check below the line where -Werror is set.

GCC chokes on -Wno-error=deprecated-non-prototype so instead we add it
conditionally.

To match the naming convention everywhere else, libvgz_extra_cflags was
renamed to VGZ_CFLAGS.

Refs 118fd10c

There are two warnings that we enforce for our own code that zlib does
not. There's also the visibility attribute that we check at configure
time. And regarding the visibility attribute, zlib no longer relies on
a NO_VIZ macro and aligned with the autoconf naming convention and wants
HAVE_HIDDEN instead.

Spotted by: Coverity

Spotted by our Alpine Linux job on CircleCI.

Change CLI cmd debug.xid to set the next XID to be used, rather than "one less than the next XID to be used"

Opportunity noticed during the review of #3872.

This is an enhanced fix for #1777

Fixes #3882

Otherwise a long banner may throw expectations off by one line.

Fixes #3877

Neither Martin nor I can remember why this was uintptr_t in the first
place, and there is no signs of pointers anywhere nearby.

Refs #3874

It makes no sense to emit this just before we end the transaction,
except to help varnishncsa find the client IP address.