https://bugs.freedesktop.org/show_bug.cgi?id=89068
--- Comment #5 from Iago Toral <ito...@igalia.com> --- (In reply to Iago Toral from comment #4) > (In reply to Brad King from comment #0) > (...) > > After the change, _mesa_format_convert is called but none of the coe paths > > invoking _mesa_pack_ubyte_rgba_row is taken. Instead the path using > > compute_src2dst_component_mapping is taken and the result is different. The kind of effect you get looks as if you were getting alpha values < 1.0 after the texture upload and then blending against the background. If the texture has a solid alpha channel that should not happen..., however, I wrote a small program to do this same kind of texture upload and it works fine for me, it only blends against the background if I pass alpha values < 65535 in the input, otherwise it won't blend, which is the expected behavior... so I am not sure about what is going on in your case. For reference, this is the output I get in a GDB session for that texture upload (on an Intel GPU): > Brad, it would be super helpful for me to understand what is going on if you > could run a gdb session and print: > > a) the values of the parameters passed to compute_src2dst_component_mapping > _after_ it has been called (they all are 4-element arrays, the last one is > an out parameter so that is why I need the values after the call has been > done) (gdb) p src2rgba $1 = "\000\001\002\003" (gdb) p rgba2dst $2 = "\000\001\002\003" (gdb) p rebase_swizzle $3 = (uint8_t *) 0x0 (gdb) p src2dst $4 = "\000\001\002\003" These are the expected values for my Intel hardware. > b) The following input parameters for the call to _mesa_swizzle_and_convert > right after: src_type, dst_type, dst_num_channels, src_num_channels, > normalized, width. (gdb) p src_type $5 = MESA_ARRAY_FORMAT_TYPE_USHORT (gdb) p src_num_channels $6 = 4 (gdb) p dst_type $7 = MESA_ARRAY_FORMAT_TYPE_UBYTE (gdb) p dst_num_channels $8 = 4 (gdb) p normalized $9 = true Again, these are the expected values. > Could you do this? And also, what GPU are you running on? -- You are receiving this mail because: You are the QA Contact for the bug. You are the assignee for the bug.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev