[Mesa-dev] [Bug 84566] Unify the format conversion code

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Fri Oct 24 01:38:36 PDT 2014


https://bugs.freedesktop.org/show_bug.cgi?id=84566

--- Comment #40 from Iago Toral <itoral at igalia.com> ---
Jason, I think we may want to be a bit more conservative with the pack/unpack
fast paths when we consider the internal base format.

Right now, if the src or dst formats are RGBA we will try the pack/unpack fast
paths. However, these paths do not consider the internal base format and we can
run into cases where simply packing or unpacking is not enough.

Consider this example: the client executes a texture upload of RGBA data and
selects a Luminance format for the texture. The driver does not support the
luminance format, so decides to use BGRA instead. Since the source format is
RGBA, we will directly pack to BGRA, but since the internal base format is
luminance, that won't match the semantics we expect (luminance requires to copy
R on all components).

I could try to add conditions to protect these paths in situations like this.
For example, in this particular case, I could check if the base format of the
destination matches the base internal format provided by the client, and if
they don't and the internal base format is luminance or intensity, then avoid
the fast path. However, I think this might hit other formats too, since the
problem, in general, is that pack/unpack fast paths do not consider that the
dst format and the dst base internal format might be different and that might
involve more than a simples pack or unpack to dst, so I am thinking that maybe
we want to simply avoid the pack/unpack paths completely when the dst format is
not the same as the dst base internal format.

What do you think?

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20141024/a10ef409/attachment-0001.html>


More information about the mesa-dev mailing list