search for: elt_type

...30: 00000001 00000000 7fbaf29c 7fa87652 0x7fbaf240: 7fd48210 7fd68a08 00000000 00000000 0x7fbaf250: 00000004 00000000 7fd48478 7faa94b4 0200: sel=1007 base=7fe6c000 limit=00001fff 32-bit rw- Backtrace: =>1 0x7faa1654 symt_add_udt_element+0x84(module=0x7fd48210, udt_type=0x7fd68a08, name=0x0, elt_type=0x0, offset=0x4, size=0x0) [/home/tom/build/wine-0.9.2/dlls/dbghelp/type.c:222] in dbg help (0x7faa1654) 2 0x7fa87652 dwarf2_parse_udt_members+0x202(symt=0x7fd68a08) [/home/tom/build/wine-0.9.2/dlls/dbghelp/dwarf.c:1272] in dbghelp (0x7fa87652) 3 0x7fa87b62 dwarf2_parse_struct_type [/hom...

...registers assuming a vector width of 128 bits. For option A, we can directly specify this in the new ABI. For option B and C however we need a new parameter attribute to enforce this. For example, ‘split_vec(N)’ below would specify that the vector argument is passed split up in chunks of N x <elt_type> in argument register where <elt_type> is the same as the element type of the vector argument. For example, this is a function taking two and returning a 3x3 matrix (with option B syntax): define <12 x float> split_vec(4) @f(<12 x float> split_vec(4) %a, <12 x float> s...