On Wed, Jul 15, 2015 at 2:52 PM, Oded Gabbay <[email protected]> wrote: > No, and it is of course erroneous. The reason it slipped under my > radar is because this function is not used at all. > Thanks for catching that :) > I removed it from the original patch.
Sorry about this email as I checked the wrong branch and that lead me to think this function is not used. In fact it is used, in vmx_composite_over_n_8_8888(). However, because I removed that function as it caused regressions, the helper function is no longer needed. The second mistake in the email is that its implementation is indeed correct. That's because we are expanding a 32-bit value to 128, so there is no issue with BE/LE (regarding vmx instrinc implementation that is). Sorry again for the mixup. Oded _______________________________________________ Pixman mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/pixman
