mul_32_32
fix_up -= mul_32_32(0x898cc517, argSqrd.msw) / 6;
fix_up = mul_32_32(fix_up, LL_MSW(fixed_arg));
fix_up.msw -= mul_32_32(0x898cc517, argSqrd.msw) / 2;
fix_up.msw += mul_32_32(0x898cc517, argTo4.msw) / 24;
adj = mul_32_32(adj, adj); /* tan^2 */
adj = mul_32_32(0x898cc517, adj); /* delta * tan^2 */