https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83528
--- Comment #6 from Markus Trippelsdorf <trippels at gcc dot gnu.org> --- (In reply to Michael_S from comment #3) > Well, the guidline here https://gcc.gnu.org/bugs/ specifically tells me that > it's one of the things that you don't want ;) The rule is there to discourage folks from posting assembly output of huge C++ compile units. For small testcases like yours, it is easier to see what is going on directly in assembly rather than parsing prose descriptions. Anyway, gcc trunk generates: .file "foo.i" .section .text .align 2 .global foo .type foo, @function foo: movhi r10, %hiadj(lottery) movhi r9, %hiadj(bars+220) addi r10, r10, %lo(lottery) addi r9, r9, %lo(bars+220) movi r11, 42 .L3: movhi r2, %hiadj(bars) addi r2, r2, %lo(bars) .L2: movhi r3, %hiadj(lottery) ldw r5, 0(r2) ldw r7, %lo(lottery)(r3) ldw r3, 4(r2) ldw r4, 8(r2) add r7, r7, r5 cmpeq r6, r7, zero add r6, r6, r3 cmpeq r5, r6, zero ldw r3, 12(r2) add r5, r5, r4 ldw r8, 16(r2) cmpeq r4, r5, zero add r4, r4, r3 cmpeq r3, r4, zero add r3, r3, r8 sub r8, r11, r3 stw r7, 0(r2) stw r6, 4(r2) stw r5, 8(r2) stw r4, 12(r2) stw r3, 16(r2) stw r8, 4(r10) addi r2, r2, 20 bne r2, r9, .L2 br .L3 .size foo, .-foo .comm lottery,8,4 .section .bss .type bars, @object .size bars, 220 .align 2 bars: .zero 220 .ident "GCC: (GNU) 8.0.0 20171221 (experimental)" So the issue looks to be fixed already. The only question is for which version it was fixed.
