[Mesa-dev] GL_EXT_texture_array: mesa cleanups, and intel implementation

[Eric Anholt]

This patch series implements GL_EXT_texture_array for Intel, and
partially cleans up Mesa core support for it in the process.  It
passes piglit here, plus some internal tests (except for the one
that's broken and ignores the minimum maximum layers).

The first patch you've seen before.  I'm just noting that it's relied
on here.  Brian said he was working on state_tracker fixes for it, and
I've ignored state_tracker in this series as a result.  I think it
will be able to dump some of its magic for 1D array textures, too.

I'm quite happy with the 2D array mipmap gen cleanup.  I think bugs
were fixed in both the 2D and 1D paths related to border width, where
it was being counted towards number of slices when it shouldn't.
Actually, I think the spec has a bug related to border for 1D array
textures -- surely they didn't mean to have border width apply to
number of slices for 1D, but not number of slices for 2D.  They meant
to skip border width for number of slices on both, right?