On 04/27/2016 10:59 PM, Jeff Law wrote:
PR rtl-optimization/44281
* hard-reg-set.h (struct target_hard_regs): New field
x_fixed_nonglobal_reg_set.
(fixed_nonglobal_reg_set): New macro.
* reginfo.c (init_reg_sets_1): Initialize it.
* ira.c (setup_alloc_regs): Use fixed_nonglobal_reg_set instead
of fixed_reg_set.
PR rtl-optimization/44281
* gcc.target/i386/pr44281.c: New test.
So my recollection of where we left things was that we needed to
dataflow information fixed for implicit uses/sets in asms. With that
infrastructure in place I think we'll be ready to resolve this old
issue. Right?
Yeah. I had prepared the following, and I was planning on getting around
to retesting it with current trunk on top of the other patch, but I
haven't quite got there yet.