Re: [PATCH v3] forwprop: add simplify_phi_result_op() [PR101179]

On 8/2/2026 11:12 PM, Andrea Pinski wrote:
> On Fri, Jul 31, 2026 at 4:57 AM Daniel Barboza
> <[email protected]> wrote:
>>
>> This new forwprop  step is my attempt to implement Richi's suggestions from
>> v1 of this work [1] where he suggested to push things out of match.pd.
>>
>> The idea is to simplify DIV/MOD into RSHIFT/BIT_AND ops in which the
>> divisor are pow2 integers in a PHI.  E.g.:
>>
>> phi_var = PHI <16,4>
>> _x = _y % phi_var
>>
>> Can be turned into:
>>
>> phi_var = PHI <15,3>
>> _x = _y & phi_var
>>
>> As long as we know that _y is a positive number or '_x' is a single use
>> with a zero comparison.
>>
>> Most of 101179 use cases are solved by this change.
>>
>> Boostrapped and regression tested with x86_64, aarch64 and riscv64.
>>
>> [1] https://gcc.gnu.org/pipermail/gcc-patches/2026-May/716303.html
>>
>>          PR tree-optimization/101179
>>
>> gcc/ChangeLog:
>>
>>          * tree-ssa-forwprop.cc (simplify_phi_result_op): New phiopt step
>>          where MOD/DIV ops with pow2 divisors can be simplified to
>>          BIT_AND/RSHIFT.
>>          (pass_forwprop::execute): Call simplify_phi_result_op.
>>
>> gcc/testsuite/ChangeLog:
>>
>>          * gcc.dg/tree-ssa/pr101179.c: New test.
>> ---
>>
>> Changes from v2:
>> - moved the implementation from phiopt to forwprop.  It resulted in a
>>    smaller and a bit cleaner logic.
>> - v2 link: https://gcc.gnu.org/pipermail/gcc-patches/2026-July/724039.html
>>
>>   gcc/testsuite/gcc.dg/tree-ssa/pr101179.c |  56 ++++++++++++
>>   gcc/tree-ssa-forwprop.cc                 | 110 +++++++++++++++++++++++
>>   2 files changed, 166 insertions(+)
>>   create mode 100644 gcc/testsuite/gcc.dg/tree-ssa/pr101179.c
>>
>> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/pr101179.c b/gcc/testsuite/gcc.dg/tree-ssa/pr101179.c
>> new file mode 100644
>> index 00000000000..478bdc6fcd3
>> --- /dev/null
>> +++ b/gcc/testsuite/gcc.dg/tree-ssa/pr101179.c
>> @@ -0,0 +1,56 @@
>> +/* { dg-do compile } */
>> +/* { dg-options "-O1 -fdump-tree-phiopt1" } */
>> +
>> +typedef unsigned uint;
>> +
>> +int f1 (int y, _Bool x)
>> +{
>> +  return y % (x ? 16 : 4) == 0;
>> +}
>> +
>> +/* We can't turn this into bit_and because there's no
>> +   guarantee 'y' is a positive val.  */
>> +int f2 (int y, _Bool x)
>> +{
>> +  return y % (x ? 16 : 4);
>> +}
>> +
>> +uint f3 (uint y, _Bool x)
>> +{
>> +  return y % (x ? 16 : 4) == 0;
>> +}
>> +
>> +uint f4 (uint y, _Bool x)
>> +{
>> +  return y % (x ? 16 : 4);
>> +}
>> +
>> +int g1 (int y, _Bool x)
>> +{
>> +  return y / (x ? 16 : 4) == 0;
>> +}
>> +
>> +/* We can't turn this into rshift because there's no
>> +   guarantee 'y' is a positive val.  */
>> +int g2 (int y, _Bool x)
>> +{
>> +  return y / (x ? 16 : 4);
>> +}
>> +
>> +/* This will be turned by match.pd into:
>> +   "(X / Y) == 0 -> X < Y if X, Y are unsigned."
>> +   We're adding it here for completioness.  */
>> +uint g3 (uint y, _Bool x)
>> +{
>> +  return y / (x ? 16 : 4) == 0;
>> +}
>> +
>> +uint g4 (uint y, _Bool x)
>> +{
>> +  return y / (x ? 16 : 4);
>> +}
>> +
>> +/* { dg-final { scan-tree-dump-times " \& " 3 "phiopt1" } } */
>> +/* { dg-final { scan-tree-dump-times " \% " 1 "phiopt1" } } */
>> +/* { dg-final { scan-tree-dump-times " >> " 2 "phiopt1" } } */
>> +/* { dg-final { scan-tree-dump-times " \\/ " 1 "phiopt1" } } */
>> diff --git a/gcc/tree-ssa-forwprop.cc b/gcc/tree-ssa-forwprop.cc
>> index 75f06c6ba41..32a9fee560c 100644
>> --- a/gcc/tree-ssa-forwprop.cc
>> +++ b/gcc/tree-ssa-forwprop.cc
>> @@ -3623,6 +3623,108 @@ simplify_count_zeroes (gimple_stmt_iterator *gsi)
>>     return true;
>>   }
>>
>> +/* Verify if we have the following structure:
>> +
>> +   iftmp1 = PHI <pow2a, pow2b>
>> +   _ssa1 = _ssa2 MOD|DIV iftmp1;
>> +   _ssa3 = _ssa1 EQ|NE 0;
>> +
>> +   And, as long as "_ssa2" is either known to be positive or
>> +   "_ssa1" is single use in a zero comparison, change the PHI
>> +   args and "_ssa1" stmt to a cheaper alternative.
>> +
>> +   For MOD:
>> +
>> +   iftmp1 = PHI <(pow2a - 1), (pow2b - 1)>
>> +   _ssa1 = _ssa2 & iftmp1;
>> +
>> +   For DIV:
>> +
>> +   iftmp1 = PHI <log2 (pow2a), log2 (pow2b)>
>> +   _ssa1 = _ssa2 >> iftmp1;  */
>> +static bool
>> +simplify_phi_result_op (gimple *stmt, tree_code code)
>> +{
>> +  gphi *phi = dyn_cast<gphi *> (SSA_NAME_DEF_STMT (gimple_assign_rhs2 (stmt)));
> Need a check for nullptr after dyn_cast.


Hmmmm I thought the nullptr case would be covered by the checks I'm doing before calling
the function:


+                            && SSA_NAME_DEF_STMT (gimple_assign_rhs2 (stmt))
+                            && is_a<gphi*> (SSA_NAME_DEF_STMT (
+                                               gimple_assign_rhs2 (stmt))))
+                     changed |= simplify_phi_result_op (stmt, code);


Maybe it's clearer to fold these checks inside the function ...



> 
>> +  if (gimple_phi_num_args (phi) != 2)
>> +    return false;
> 
> Instead of just handling 2 argument phis. why not handle N argument phis here.


Mostly because I didn't though about that :D I just came up with a simple
C code that generates a PHI with 4 args that we could reduce.

I'll rework the code to support this use case.

> 
> 
>> +
>> +  tree arg0 = gimple_phi_arg_def (phi, 0);
>> +  tree arg1 = gimple_phi_arg_def (phi, 1);
>> +
>> +  if (!integer_pow2p (arg0) || !tree_fits_uhwi_p (arg0)
>> +      || !integer_pow2p (arg1) || !tree_fits_uhwi_p (arg1))
>> +    return false;
> 
> That is 2 loops.
> One to reject if the argument is not a power of 2. I don't think you
> need tree_fits_uhwi_p (see below on why).
> 
>> +
>> +  tree_code new_code;
>> +  switch (code)
>> +    {
>> +      case TRUNC_MOD_EXPR:
>> +      case CEIL_MOD_EXPR:
>> +      case FLOOR_MOD_EXPR:
>> +      case ROUND_MOD_EXPR:
>> +       new_code = BIT_AND_EXPR;
>> +       break;
>> +      case TRUNC_DIV_EXPR:
>> +      case CEIL_DIV_EXPR:
>> +      case FLOOR_DIV_EXPR:
>> +      case ROUND_DIV_EXPR:
>> +       new_code = RSHIFT_EXPR;
>> +       break;
>> +
>> +     default:
>> +       return false;
>> +    }
>> +
>> +  /* If rhs1 is a known positive value we can always apply these
>> +     simplification.  Otherwise see if lhs is single_use with a
>> +     EQ|NE 0 cmp.  */
>> +  tree rhs1 = gimple_assign_rhs1 (stmt);
>> +  if (!tree_expr_nonnegative_p (rhs1))
>> +    {
>> +      gimple *cmp_stmt;
>> +      use_operand_p use_p;
>> +
>> +      if (!single_imm_use (gimple_assign_lhs (stmt), &use_p, &cmp_stmt)
>> +         || !cmp_stmt
>> +         || !is_gimple_assign (cmp_stmt))
>> +       return false;
>> +
>> +      if (!(gimple_assign_rhs_code (cmp_stmt) == NE_EXPR
>> +           || gimple_assign_rhs_code (cmp_stmt) == EQ_EXPR))
>> +       return false;
>> +
>> +      if (!integer_zerop (gimple_assign_rhs2 (cmp_stmt)))
>> +       return false;
>> +    }
> 
> Put this before the phi check loop.
> 
>> +
>> +  tree phires = gimple_phi_result (phi);
>> +  tree type = TREE_TYPE (phires);
>> +  tree new_arg0, new_arg1;
>> +
>> +  if (new_code == RSHIFT_EXPR)
>> +    {
>> +      new_arg0 = build_int_cst (type, wi::exact_log2 (tree_to_uhwi (arg0)));
>> +      new_arg1 = build_int_cst (type, wi::exact_log2 (tree_to_uhwi (arg1)));
> 
> Use wi::to_wide here.
> For RSHIFT_EXPR, add a gimple_convert to integer_type_node for the new
> rhs. That is it might be a long long which we want to convert into an
> integer or a bit_int that we want to convert into an integer.
> 
>> +    }
>> +  else
>> +    {
>> +      new_arg0 = build_int_cst (type, tree_to_uhwi (arg0) - 1);
>> +      new_arg1 = build_int_cst (type, tree_to_uhwi (arg1) - 1);
> 
> Likewise. (this removes the need for the tree_fits_uhwi_p check)
> 
>> +    }
>> +
>> +  SET_PHI_ARG_DEF (phi, 0, new_arg0);
>> +  SET_PHI_ARG_DEF (phi, 1, new_arg1);
>> +  if (SSA_NAME_RANGE_INFO (phires))
>> +    reset_flow_sensitive_info (phires);
> 
> Maybe it is better to insert a new phi rather than reuse the current
> phi for debug info reasons. You might want to fixup the debug info
> too.

If we create a new phi do we need to do stuff w.r.t debug info of the older
phi?


Thanks,
Daniel

> 
> Thanks,
> Andrea
> 
>> +
>> +  gimple_assign_set_rhs1 (stmt, rhs1);
>> +  gimple_assign_set_rhs2 (stmt, phires);
>> +  gimple_assign_set_rhs_code (stmt, new_code);
>> +  update_stmt (stmt);
>> +
>> +  return true;
>> +}
>>
>>   /* Determine whether applying the 2 permutations (mask1 then mask2)
>>      gives back one of the input.  */
>> @@ -5894,6 +5996,14 @@ pass_forwprop::execute (function *fun)
>>                        changed |= simplify_vector_constructor (&gsi);
>>                      else if (code == ARRAY_REF)
>>                        changed |= simplify_count_zeroes (&gsi);
>> +                   else if (get_gimple_rhs_class (code) == GIMPLE_BINARY_RHS
>> +                            && TREE_CODE (
>> +                                   gimple_assign_rhs2 (stmt)) == SSA_NAME
>> +                            && has_single_use (gimple_assign_rhs2 (stmt))
>> +                            && SSA_NAME_DEF_STMT (gimple_assign_rhs2 (stmt))
>> +                            && is_a<gphi*> (SSA_NAME_DEF_STMT (
>> +                                               gimple_assign_rhs2 (stmt))))
>> +                     changed |= simplify_phi_result_op (stmt, code);
>>                      break;
>>                    }
>>
>> --
>> 2.43.0
>>