This is mostly the same as the last series rev 3, with the following changes:
1. Rebased onto current master, some trivial merge conflict resolved. 2. R-b's of Tapani and Marek tacked onto all patches. Only the new patch 22/22 is new and unreviewed. 3. Following Tapani's suggestion i moved old patch 1 to patch 6, so the enable for i965 30 bit formats is done after the basic dri and i965 bits are in place. This is more consistent with flipping the on-switch on gallium. 4. Added patch 22/22, so OES read type queries for glReadPixels also handle B10G10R10X2, not only B10G10R10A2, following Marek's advice. I've tested this by quickly hacking up the dEQP-EGL.functional.wide_color.window_1010102_colorspace_default test to ask for zero alpha bits, verifying via apitrace it gets a X2R10G10B10 format and that the precision test fails in the expected way, ie. with a constant alpha == 3 ( ~ 1.0f) returned from a fb that doesn't have an alpha channel when reading via glReadPixels(GL_RGBA, GL_UNSIGNED_INT_10_10_10_2_REV). 5. Finally i've dropped the nouveau patch which added a hacky B10G10R10X2 pixel format for visuals, which isn't actually supported by the driver or NVidia hardware. This means that depth 30 display on NVidia hardware won't work atm., but that needs more work on nouveau-ddx and nouveau-kms anyway to get it working properly. Prime renderoffload from a Intel or AMD display gpu to a NVidia gpu works, as tested on Wayland+Weston and X11, so there's some use to the current nouveau gallium support for depth 30. So this one should be probably good to push after checking patch 22/22. Thanks, -mario _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
