On Fri, Dec 13, 2024 at 04:00:38PM -0700, Jeff Law wrote:
> On 12/13/24 8:20 AM, Jakub Jelinek wrote:
> > The following testcase is miscompiled on s390x-linux with -O2 -march=z15.
> > The problem happens during cse2, which sees in an extended basic block
> > (jump_insn 217 78 216 10 (parallel [
> > (set (pc)
> > (if_then_else (ne (reg:SI 165)
> > (const_int 1 [0x1]))
> > (label_ref 216)
> > (pc)))
> > (set (reg:SI 165)
> > (plus:SI (reg:SI 165)
> > (const_int -1 [0xffffffffffffffff])))
> > (clobber (scratch:SI))
> > (clobber (reg:CC 33 %cc))
> > ]) "t.c":14:17 discrim 1 2192 {doloop_si64}
> > (int_list:REG_BR_PROB 955630228 (nil))
> > -> 216)
> > ...
> > (insn 99 98 100 12 (set (reg:SI 138)
> > (const_int 1 [0x1])) "t.c":9:31 1507 {*movsi_zarch}
> > (nil))
> > (insn 100 99 103 12 (parallel [
> > (set (reg:SI 137)
> > (minus:SI (reg:SI 138)
> > (subreg:SI (reg:HI 135 [ a ]) 0)))
> > (clobber (reg:CC 33 %cc))
> > ]) "t.c":9:31 1904 {*subsi3}
> > (expr_list:REG_DEAD (reg:SI 138)
> > (expr_list:REG_DEAD (reg:HI 135 [ a ])
> > (expr_list:REG_UNUSED (reg:CC 33 %cc)
> > (nil)))))
> I don't really see the connection between (reg 165) and (reg 138), but I
> don't think it matters enough to dive into.
The ebb continues after jump_insn 217 when the branch is not taken,
i.e. if (ne (reg:SI 165) (const_int 1 [0x1])) is false. As it ignored the
r165 -= 1 part of the insn, it recorded (reg:SI 165) must be const1_rtx
and (reg:SI 138) is set to the same value, so let's use the older holder
of SImode 1 constant instead of a new one...
Except that (reg:SI 165) is actually 0 after the doloop insn stops
iterating.
Jakub
