On Mon, Mar 24, 2008 at 1:04 PM, John David Anglin <dave@xxxxxxxxxxxxxxxxxx> wrote: > The fundamental issue is there's only one r1 register and its > needed for memory loads and stores. The PA backend has always used > separate rtl for the addil insn. This makes it possible for gcse > to eliminate duplicated addil insns. However, it can't do this > if the two addil insns are separated by a call or an asm which uses > r1 (well it could if reload didn't pick r1 for the addil). It is certainly a difficult problem, and the only thing that appears to trigger this is a function with a lot of register pressure, followed by an asm with a *lot* of clobbers, and a fair number of input/output operands. It's hard to justify fixing this if only a couple of corner cases trigger the bug, and even then it's a "fail-safe" situation. Cheers, Carlos. -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
