On Fri, Aug 4, 2023 at 10:52 AM Jan Hubicka <hubi...@ucw.cz> wrote:
>
> Hi,
> so I found the problem. We duplicate multiple paths and end up with:
>
> ;; basic block 6, loop depth 0, count 365072224 (estimated locally, freq
> 0.3400)
> ;; prev block 12, next block 7, flags: (NEW, REACHABLE, VISITED)
> ;; pred: 4 [never (guessed)] count:0 (estimated locally, freq 0.0000)
> (TRUE_VALUE,EXECUTABLE)
> ;; 10 [always] count:365072224 (estimated locally, freq 0.3400)
> (FALLTHRU,EXECUTABLE)
> # _18 = PHI <0(4), 0(10)>
> # d_39 = PHI <d_40(D)(4), d_40(D)(10)>
> if (_18 == 0)
> goto <bb 8>; [97.06%]
> else
> goto <bb 7>; [2.94%]
> ;; succ: 8 [97.1% (guessed)] count:354334801 (estimated locally, freq
> 0.3300) (TRUE_VALUE,EXECUTABLE)
> ;; 7 [2.9% (guessed)] count:10737423 (estimated locally, freq
> 0.0100) (FALSE_VALUE,EXECUTABLE)
>
> Here goto bb 7 is never taken but profile is wrong.
>
> Before threading we have chain of conditionals:
>
> __asm__("pushf{l|d}
> pushf{l|d}
> pop{l} %0
> mov{l} {%0, %1|%1, %0}
> xor{l} {%2, %0|%0, %2}
> push{l} %0
> popf{l|d}
> pushf{l|d}
> pop{l} %0
> popf{l|d}
> " : "=&r" __eax_19, "=&r" __ebx_20 : "i" 2097152);
> _21 = __eax_19 ^ __ebx_20;
> _22 = _21 & 2097152;
> if (_22 == 0)
> goto <bb 4>; [34.00%]
> else
> goto <bb 3>; [66.00%]
>
> <bb 3> [local count: 708669602 freq: 0.660000]:
> __asm__ __volatile__("cpuid
> " : "=a" __eax_24, "=b" __ebx_25, "=c" __ecx_26, "=d" __edx_27 : "0"
> 0);
>
> <bb 4> [local count: 1073741826 freq: 1.000000]:
> # _33 = PHI <0(2), __eax_24(3)>
> _16 = _33 == 0;
> if (_33 == 0)
> goto <bb 6>; [34.00%]
> else
> goto <bb 5>; [66.00%]
>
> <bb 5> [local count: 708669600 freq: 0.660000]:
> __asm__ __volatile__("cpuid
> " : "=a" a_44, "=b" b_45, "=c" c_46, "=d" d_47 : "0" 1, "2" 0);
>
> <bb 6> [local count: 1073741824 freq: 1.000000]:
> # _18 = PHI <0(4), 1(5)>
> # d_39 = PHI <d_40(D)(4), d_47(5)>
> if (_18 == 0)
> goto <bb 8>; [33.00%]
> else
> goto <bb 7>; [67.00%]
>
>
> If first _22 == 0 then also _33 == 0 and _18 == 0 but the last case has
> probability 33% while the first 34%, so the profile guess is not
> consistent with the threaded path. So threading is right to end up with
> profile inconsistency, but it should print reason for doing it.
>
> One option is to disable optimization for the check. Other option is to
> get the first conditional predicted right.
> Would this be OK?
Yeah, that looks OK.
> gcc/ChangeLog:
>
> * config/i386/cpuid.h: Mark CPUs not supporting cpuid as unlikely.
>
> diff --git a/gcc/config/i386/cpuid.h b/gcc/config/i386/cpuid.h
> index 03fd6fc9478..9c768ac0b6d 100644
> --- a/gcc/config/i386/cpuid.h
> +++ b/gcc/config/i386/cpuid.h
> @@ -295,7 +295,7 @@ __get_cpuid_max (unsigned int __ext, unsigned int *__sig)
> : "i" (0x00200000));
> #endif
>
> - if (!((__eax ^ __ebx) & 0x00200000))
> + if (__builtin_expect (!((__eax ^ __ebx) & 0x00200000), 0))
> return 0;
> #endif
>
