[Mesa-dev] [PATCH] i965/skl: Ignore the vertical alignment for the qpitch of 1D textures

[Neil Roberts]

Ben Widawsky <ben at bwidawsk.net> writes:

> I guess I'd appreciate a comment about how the total_width is
> guaranteed to be a multiple of 64, and therefore is a multiple of all
> possible H_ALIGNS. This is required to meet the qpitch restraint in
> the surface format, "This field must be set to an integer multiple of
> the Surface Horizontal Alignment."

My interpretation of that comment is that for 1D surfaces 64 *is* the
horizontal alignment so other possible alignments don't matter. I think
it's expected that total_width is aligned to the horizontal alignment,
and in this case that is enforced in gen9_miptree_layout_1d. I will add
a comment to make that clearer.

> Note, I don't know anything about compressed textures and what the
> block widths can be, but just doing the math, if block size > 16 and
> not a multiple of 16, this constraint will not hold.

Currently I think all of the block sizes are either 4x4 or 8x4. None of
these would make any sense for a 1D texture and wouldn't make the
texture any smaller. The bspec explicitly disallows compressed formats
for 1D surfaces. If someone were to eventually invent a compressed
format with a block height of 1 then I guess the block alignment would
be handled in gen9_miptree_layout_1d in the unlikely event that the size
isn't a factor of 64.

Note that I'm putting off landing this patch and the other qpitch one
until someone can review patch 6 in the series because without that
patch the qpitch patch seems to cause a GPU hang which makes the kernel
panic.

Thanks for the review.

Regards,
- Neil
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 472 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20150302/a033d4b3/attachment.sig>