On 08/29/2016 02:41 PM, Segher Boessenkool wrote:
On Mon, Aug 29, 2016 at 10:19:25PM +0200, Steven Bosscher wrote:
On Mon, Aug 29, 2016 at 6:50 PM, Segher Boessenkool wrote:
This patch changes that so that that def is only added after
epilogue_completed.
...
Does this work on all other targets?
Guessing: not for targets where INCOMING_RETURN_ADDR_RTX is the stack register?
That'd be at least h8300, rl78, and rx.
The stack pointer is always added (earlier in this function):
/* The always important stack pointer. */
bitmap_set_bit (entry_block_defs, STACK_POINTER_REGNUM);
so that isn't it.
History suggests it was a fix (or workaround) for a problem on ia64,
but I might well have that wrong.
It came in as part of a much larger change where Kenny was trying to
compute the correct set of registers that are live at function entry. I
believe this is the earliest draft of the function in question:
http://permalink.gmane.org/gmane.comp.gcc.devel/75329
Jeff
