Re: [PR81611] improve auto-inc

2018-02-27 Thread Jeff Law
On 02/27/2018 04:18 PM, Alexandre Oliva wrote: > On Feb 14, 2018, Jeff Law wrote: > >>> + regno = REGNO (inc_insn.reg0); >>> + int luid = DF_INSN_LUID (mem_insn.insn); >>> + mem_insn.insn = get_next_ref (regno, bb, reg_next_use);

Re: [PR81611] improve auto-inc

2018-02-27 Thread Alexandre Oliva
On Feb 14, 2018, Jeff Law wrote: >> + regno = REGNO (inc_insn.reg0); >> + int luid = DF_INSN_LUID (mem_insn.insn); >> + mem_insn.insn = get_next_ref (regno, bb, reg_next_use); > So I think a comment is warranted right as we

Re: [PR81611] improve auto-inc

2018-02-13 Thread Jeff Law
On 01/23/2018 08:42 PM, Alexandre Oliva wrote: > These two patches fix PR81611. > > The first one improves forwprop so that we avoid adding SSA conflicting > by forwpropping the iv increment, which may cause both the incremented > and the original value to be live, even when the iv is copied

Re: [PR81611] improve auto-inc

2018-01-30 Thread Alexandre Oliva
On Jan 25, 2018, Alexandre Oliva wrote: > Thanks, I'll retest with the simplified test (just in case; for I can't > recall why I ended up with all those redundant conditions), As suspected, removing the redundant tests didn't regress anything. I suppose they mattered in some

Re: [PR81611] improve auto-inc

2018-01-25 Thread Alexandre Oliva
On Jan 24, 2018, Richard Biener wrote: > given gimple_assign_ssa_name_copy checks it is an assign *nod* > the lhs != preinc check is always false given you got to phi via > SSA_NAME_DEF_STMT of preinc. *nod* > The simple_iv_increment_p change is ok with that

Re: [PR81611] improve auto-inc

2018-01-24 Thread Richard Biener
On Wed, Jan 24, 2018 at 4:42 AM, Alexandre Oliva wrote: > These two patches fix PR81611. > > The first one improves forwprop so that we avoid adding SSA conflicting > by forwpropping the iv increment, which may cause both the incremented > and the original value to be live,

[PR81611] improve auto-inc

2018-01-23 Thread Alexandre Oliva
These two patches fix PR81611. The first one improves forwprop so that we avoid adding SSA conflicting by forwpropping the iv increment, which may cause both the incremented and the original value to be live, even when the iv is copied between the PHI node and the increment. We already handled