On Mon, Apr 20, 2015 at 12:17:08PM -0600, Jeff Law wrote: > So just one question. Why handle the split case differently? In the split > case you effectively move the debug insn to the new block. In the !split > case, you just delete the debug insn.
The idea was that when split we have a new block with no debug info, but !split the block already has debug info about the var/reg. Now that I've looked at a lot more .debug_loc sections than I ever wanted, and the debug_insns that generate that info, I realize my assumption about the !split case is wrong. Working on a new patch to fix some of the things I've discovered. -- Alan Modra Australia Development Lab, IBM
