[RFA,(fold/symtab)] c++: constexpr, fold, weak redecl, fp/0 [PR103310]

  For PR61825, honza changed tree_single_nonzero_warnv_p to prevent a later
declaration from marking a function as weak after we've determined that it
wasn't weak before.  But we shouldn't do that for speculative folding; we
should only do it when we actually need a constant value.  In C++, such a
context is called "manifestly constant-evaluated".  In fold, this seems to
correspond to the folding_initializer flag, since in C this situation only
occurs in static initializers.

This change makes nonzero-1.c well-formed; I've added a nonzero-1a.c to
verify that we delete the null check eventually if there is no weak
redeclaration.

The varasm.c change is so that if we do get the weak redeclaration error, we
get it at the position of the weak declaration rather than the previous
declaration.

Using the FOLD_INIT paths also affects floating point arithmetic: notably,
this makes floating point division by zero in a manifestly
constant-evaluated context constant, as in a C static initializer.  I've had
some success convincing CWG that this is the right direction; C++ should
follow C's floating point semantics more than we have been doing, and Joseph
says that the C policy is that Annex F overrides other parts of the standard
that say that some operations are undefined.  But since we're in stage 3,
I'm only making this change with the new flag -fconstexpr-fp-except.  It may
turn on by default in a future release.

I think this distinction is only relevant for binary operations; arithmetic
for the floating point case, comparison for possibly non-zero addresses.

Tested x86_64-pc-linux-gnu, OK for trunk?

	PR c++/103310

gcc/ChangeLog:

	* fold-const.c (maybe_nonzero_address): Use get_create or get
	depending on folding_initializer.
	(fold_binary_initializer_loc): New.
	* fold-const.h (fold_binary_initializer_loc): Declare.
	* varasm.c (mark_weak): Don't use the decl location.
	* doc/invoke.texi: Document -fconstexpr-fp-except.

gcc/c-family/ChangeLog:

	* c.opt: Add -fconstexpr-fp-except.

gcc/cp/ChangeLog:

	* constexpr.c (cxx_eval_binary_expression): Use
	fold_binary_initializer_loc if manifestly cxeval.

gcc/testsuite/ChangeLog:

	* g++.dg/cpp0x/constexpr-fp-except1.C: New test.
	* g++.dg/cpp1z/constexpr-if36.C: New test.
	* gcc.dg/tree-ssa/nonzero-1.c: Now well-formed.
	* gcc.dg/tree-ssa/nonzero-1a.c: New test.
---
 gcc/doc/invoke.texi                           | 14 ++++++++++
 gcc/c-family/c.opt                            |  4 +++
 gcc/fold-const.h                              |  1 +
 gcc/cp/constexpr.c                            |  9 ++++++-
 gcc/fold-const.c                              | 26 ++++++++++++++++---
 .../g++.dg/cpp0x/constexpr-fp-except1.C       |  4 +++
 gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C   | 19 ++++++++++++++
 gcc/testsuite/gcc.dg/tree-ssa/nonzero-1.c     |  5 ++--
 gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c    | 11 ++++++++
 gcc/varasm.c                                  |  2 +-
 10 files changed, 88 insertions(+), 7 deletions(-)
 create mode 100644 gcc/testsuite/g++.dg/cpp0x/constexpr-fp-except1.C
 create mode 100644 gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C
 create mode 100644 gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c


base-commit: a3e75c1491cd2d501081210925a89a65b1c1e5e5
  

On Wed, Dec 1, 2021 at 3:10 AM Jason Merrill via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> For PR61825, honza changed tree_single_nonzero_warnv_p to prevent a later
> declaration from marking a function as weak after we've determined that it
> wasn't weak before.  But we shouldn't do that for speculative folding; we
> should only do it when we actually need a constant value.  In C++, such a
> context is called "manifestly constant-evaluated".  In fold, this seems to
> correspond to the folding_initializer flag, since in C this situation only
> occurs in static initializers.
>
> This change makes nonzero-1.c well-formed; I've added a nonzero-1a.c to
> verify that we delete the null check eventually if there is no weak
> redeclaration.
>
> The varasm.c change is so that if we do get the weak redeclaration error, we
> get it at the position of the weak declaration rather than the previous
> declaration.
>
> Using the FOLD_INIT paths also affects floating point arithmetic: notably,
> this makes floating point division by zero in a manifestly
> constant-evaluated context constant, as in a C static initializer.  I've had
> some success convincing CWG that this is the right direction; C++ should
> follow C's floating point semantics more than we have been doing, and Joseph
> says that the C policy is that Annex F overrides other parts of the standard
> that say that some operations are undefined.  But since we're in stage 3,
> I'm only making this change with the new flag -fconstexpr-fp-except.  It may
> turn on by default in a future release.
>
> I think this distinction is only relevant for binary operations; arithmetic
> for the floating point case, comparison for possibly non-zero addresses.
>
> Tested x86_64-pc-linux-gnu, OK for trunk?

OK.

Thanks,
Richard.

>         PR c++/103310
>
> gcc/ChangeLog:
>
>         * fold-const.c (maybe_nonzero_address): Use get_create or get
>         depending on folding_initializer.
>         (fold_binary_initializer_loc): New.
>         * fold-const.h (fold_binary_initializer_loc): Declare.
>         * varasm.c (mark_weak): Don't use the decl location.
>         * doc/invoke.texi: Document -fconstexpr-fp-except.
>
> gcc/c-family/ChangeLog:
>
>         * c.opt: Add -fconstexpr-fp-except.
>
> gcc/cp/ChangeLog:
>
>         * constexpr.c (cxx_eval_binary_expression): Use
>         fold_binary_initializer_loc if manifestly cxeval.
>
> gcc/testsuite/ChangeLog:
>
>         * g++.dg/cpp0x/constexpr-fp-except1.C: New test.
>         * g++.dg/cpp1z/constexpr-if36.C: New test.
>         * gcc.dg/tree-ssa/nonzero-1.c: Now well-formed.
>         * gcc.dg/tree-ssa/nonzero-1a.c: New test.
> ---
>  gcc/doc/invoke.texi                           | 14 ++++++++++
>  gcc/c-family/c.opt                            |  4 +++
>  gcc/fold-const.h                              |  1 +
>  gcc/cp/constexpr.c                            |  9 ++++++-
>  gcc/fold-const.c                              | 26 ++++++++++++++++---
>  .../g++.dg/cpp0x/constexpr-fp-except1.C       |  4 +++
>  gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C   | 19 ++++++++++++++
>  gcc/testsuite/gcc.dg/tree-ssa/nonzero-1.c     |  5 ++--
>  gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c    | 11 ++++++++
>  gcc/varasm.c                                  |  2 +-
>  10 files changed, 88 insertions(+), 7 deletions(-)
>  create mode 100644 gcc/testsuite/g++.dg/cpp0x/constexpr-fp-except1.C
>  create mode 100644 gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C
>  create mode 100644 gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c
>
> diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
> index 3bddfbaae6a..d6858d834f9 100644
> --- a/gcc/doc/invoke.texi
> +++ b/gcc/doc/invoke.texi
> @@ -3035,6 +3035,20 @@ users are likely to want to adjust it, but if your code does heavy
>  constexpr calculations you might want to experiment to find which
>  value works best for you.
>
> +@item -fconstexpr-fp-except
> +@opindex fconstexpr-fp-except
> +Annex F of the C standard specifies that IEC559 floating point
> +exceptions encountered at compile time should not stop compilation.
> +C++ compilers have historically not followed this guidance, instead
> +treating floating point division by zero as non-constant even though
> +it has a well defined value.  This flag tells the compiler to give
> +Annex F priority over other rules saying that a particular operation
> +is undefined.
> +
> +@smallexample
> +constexpr float inf = 1./0.; // OK with -fconstexpr-fp-except
> +@end smallexample
> +
>  @item -fconstexpr-loop-limit=@var{n}
>  @opindex fconstexpr-loop-limit
>  Set the maximum number of iterations for a loop in C++14 constexpr functions
> diff --git a/gcc/c-family/c.opt b/gcc/c-family/c.opt
> index 4b8a094b206..5654f044ae4 100644
> --- a/gcc/c-family/c.opt
> +++ b/gcc/c-family/c.opt
> @@ -1615,6 +1615,10 @@ fconstexpr-cache-depth=
>  C++ ObjC++ Joined RejectNegative UInteger Var(constexpr_cache_depth) Init(8)
>  -fconstexpr-cache-depth=<number>       Specify maximum constexpr recursion cache depth.
>
> +fconstexpr-fp-except
> +C++ ObjC++ Var(flag_constexpr_fp_except) Init(0)
> +Allow IEC559 floating point exceptions in constant expressions
> +
>  fconstexpr-loop-limit=
>  C++ ObjC++ Joined RejectNegative UInteger Var(constexpr_loop_limit) Init(262144)
>  -fconstexpr-loop-limit=<number>        Specify maximum constexpr loop iteration count.
> diff --git a/gcc/fold-const.h b/gcc/fold-const.h
> index 56e9d399c0d..01bfe5df9b5 100644
> --- a/gcc/fold-const.h
> +++ b/gcc/fold-const.h
> @@ -77,6 +77,7 @@ extern tree fold_build_call_array_loc (location_t, tree, tree, int, tree *);
>  #define fold_build_call_array_initializer(T1,T2,N,T4)\
>     fold_build_call_array_initializer_loc (UNKNOWN_LOCATION, T1, T2, N, T4)
>  extern tree fold_build_call_array_initializer_loc (location_t, tree, tree, int, tree *);
> +extern tree fold_binary_initializer_loc (location_t, tree_code, tree, tree, tree);
>  extern tree get_array_ctor_element_at_index (tree, offset_int,
>                                              unsigned * = NULL);
>  extern bool fold_convertible_p (const_tree, const_tree);
> diff --git a/gcc/cp/constexpr.c b/gcc/cp/constexpr.c
> index d66a56583ae..b4b8a96c6af 100644
> --- a/gcc/cp/constexpr.c
> +++ b/gcc/cp/constexpr.c
> @@ -3366,7 +3366,14 @@ cxx_eval_binary_expression (const constexpr_ctx *ctx, tree t,
>      }
>
>    if (r == NULL_TREE)
> -    r = fold_binary_loc (loc, code, type, lhs, rhs);
> +    {
> +      if (ctx->manifestly_const_eval
> +         && (flag_constexpr_fp_except
> +             || TREE_CODE (type) != REAL_TYPE))
> +       r = fold_binary_initializer_loc (loc, code, type, lhs, rhs);
> +      else
> +       r = fold_binary_loc (loc, code, type, lhs, rhs);
> +    }
>
>    if (r == NULL_TREE
>        && (code == LSHIFT_EXPR || code == RSHIFT_EXPR)
> diff --git a/gcc/fold-const.c b/gcc/fold-const.c
> index 90d82257ae7..0b9a42f764a 100644
> --- a/gcc/fold-const.c
> +++ b/gcc/fold-const.c
> @@ -85,8 +85,8 @@ along with GCC; see the file COPYING3.  If not see
>  #include "asan.h"
>  #include "gimple-range.h"
>
> -/* Nonzero if we are folding constants inside an initializer; zero
> -   otherwise.  */
> +/* Nonzero if we are folding constants inside an initializer or a C++
> +   manifestly-constant-evaluated context; zero otherwise.  */
>  int folding_initializer = 0;
>
>  /* The following constants represent a bit based encoding of GCC's
> @@ -9924,8 +9924,15 @@ pointer_may_wrap_p (tree base, tree offset, poly_int64 bitpos)
>  static int
>  maybe_nonzero_address (tree decl)
>  {
> +  /* Normally, don't do anything for variables and functions before symtab is
> +     built; it is quite possible that DECL will be declared weak later.
> +     But if folding_initializer, we need a constant answer now, so create
> +     the symtab entry and prevent later weak declaration.  */
>    if (DECL_P (decl) && decl_in_symtab_p (decl))
> -    if (struct symtab_node *symbol = symtab_node::get_create (decl))
> +    if (struct symtab_node *symbol
> +       = (folding_initializer
> +          ? symtab_node::get_create (decl)
> +          : symtab_node::get (decl)))
>        return symbol->nonzero_address ();
>
>    /* Function local objects are never NULL.  */
> @@ -13991,6 +13998,19 @@ fold_build_call_array_initializer_loc (location_t loc, tree type, tree fn,
>    return result;
>  }
>
> +tree
> +fold_binary_initializer_loc (location_t loc, tree_code code, tree type,
> +                            tree lhs, tree rhs)
> +{
> +  tree result;
> +  START_FOLD_INIT;
> +
> +  result = fold_binary_loc (loc, code, type, lhs, rhs);
> +
> +  END_FOLD_INIT;
> +  return result;
> +}
> +
>  #undef START_FOLD_INIT
>  #undef END_FOLD_INIT
>
> diff --git a/gcc/testsuite/g++.dg/cpp0x/constexpr-fp-except1.C b/gcc/testsuite/g++.dg/cpp0x/constexpr-fp-except1.C
> new file mode 100644
> index 00000000000..3887a5ba743
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/cpp0x/constexpr-fp-except1.C
> @@ -0,0 +1,4 @@
> +// { dg-do compile { target c++11 } }
> +// { dg-additional-options -fconstexpr-fp-except }
> +
> +constexpr double inf = 1./0.;
> diff --git a/gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C b/gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C
> new file mode 100644
> index 00000000000..4a1b134cc19
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/cpp1z/constexpr-if36.C
> @@ -0,0 +1,19 @@
> +// PR c++/103310
> +// Test that only manifestly-constant-evaluated comparisons lock a symbol's
> +// weakness.
> +
> +// { dg-do compile { target c++17 } }
> +
> +extern void weakfn1 (void);
> +extern void weakfn2 (void);
> +
> +void call_weakfn (void)
> +{
> +  if (weakfn1)
> +    weakfn1 ();
> +  if constexpr (weakfn2)
> +    weakfn2 ();
> +}
> +
> +extern void weakfn1 (void)  __attribute__((weak));
> +extern void weakfn2 (void)  __attribute__((weak)); // { dg-error "declared weak after being used" }
> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1.c b/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1.c
> index c9d438e9374..aa21b71c730 100644
> --- a/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1.c
> +++ b/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1.c
> @@ -1,12 +1,13 @@
>  /* { dg-do compile } */
> -/* { dg-options "-O2 -fdelete-null-pointer-checks" } */
> +/* { dg-options "-O2 -fdelete-null-pointer-checks -fdump-tree-optimized" } */
>  /* { dg-require-weak "" } */
>
>  /* { dg-skip-if "" keeps_null_pointer_checks } */
> -extern int a; /* { dg-error "declared weak after being used" } */
> +extern int a;
>  int
>  t()
>  {
> +  /* { dg-final { scan-tree-dump "&a != 0" "optimized" } } */
>    return &a!=0;
>  }
>  extern int a __attribute__ ((weak));
> diff --git a/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c b/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c
> new file mode 100644
> index 00000000000..3d1eb97f151
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/tree-ssa/nonzero-1a.c
> @@ -0,0 +1,11 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fdelete-null-pointer-checks -fdump-tree-optimized" } */
> +
> +/* { dg-skip-if "" keeps_null_pointer_checks } */
> +extern int a;
> +int
> +t()
> +{
> +  /* { dg-final { scan-tree-dump-not "&a != 0" "optimized" } } */
> +  return &a!=0;
> +}
> diff --git a/gcc/varasm.c b/gcc/varasm.c
> index d6031d6f1a2..9315e2c6936 100644
> --- a/gcc/varasm.c
> +++ b/gcc/varasm.c
> @@ -5869,7 +5869,7 @@ mark_weak (tree decl)
>
>    struct symtab_node *n = symtab_node::get (decl);
>    if (n && n->refuse_visibility_changes)
> -    error ("%+qD declared weak after being used", decl);
> +    error ("%qD declared weak after being used", decl);
>    DECL_WEAK (decl) = 1;
>
>    if (DECL_RTL_SET_P (decl)
>
> base-commit: a3e75c1491cd2d501081210925a89a65b1c1e5e5
> --
> 2.27.0
>
  

On Tue, Nov 30, 2021 at 09:09:41PM -0500, Jason Merrill via Gcc-patches wrote:
> For PR61825, honza changed tree_single_nonzero_warnv_p to prevent a later
> declaration from marking a function as weak after we've determined that it
> wasn't weak before.  But we shouldn't do that for speculative folding; we
> should only do it when we actually need a constant value.  In C++, such a
> context is called "manifestly constant-evaluated".  In fold, this seems to
> correspond to the folding_initializer flag, since in C this situation only
> occurs in static initializers.
> 
> This change makes nonzero-1.c well-formed; I've added a nonzero-1a.c to
> verify that we delete the null check eventually if there is no weak
> redeclaration.
> 
> The varasm.c change is so that if we do get the weak redeclaration error, we
> get it at the position of the weak declaration rather than the previous
> declaration.
> 
> Using the FOLD_INIT paths also affects floating point arithmetic: notably,
> this makes floating point division by zero in a manifestly
> constant-evaluated context constant, as in a C static initializer.  I've had
> some success convincing CWG that this is the right direction; C++ should
> follow C's floating point semantics more than we have been doing, and Joseph
> says that the C policy is that Annex F overrides other parts of the standard
> that say that some operations are undefined.  But since we're in stage 3,
> I'm only making this change with the new flag -fconstexpr-fp-except.  It may
> turn on by default in a future release.
> 
> I think this distinction is only relevant for binary operations; arithmetic
> for the floating point case, comparison for possibly non-zero addresses.
> 
> Tested x86_64-pc-linux-gnu, OK for trunk?

[...]

> diff --git a/gcc/c-family/c.opt b/gcc/c-family/c.opt
> index 4b8a094b206..5654f044ae4 100644
> --- a/gcc/c-family/c.opt
> +++ b/gcc/c-family/c.opt
> @@ -1615,6 +1615,10 @@ fconstexpr-cache-depth=
>  C++ ObjC++ Joined RejectNegative UInteger Var(constexpr_cache_depth) Init(8)
>  -fconstexpr-cache-depth=<number>	Specify maximum constexpr recursion cache depth.
>  
> +fconstexpr-fp-except
> +C++ ObjC++ Var(flag_constexpr_fp_except) Init(0)
> +Allow IEC559 floating point exceptions in constant expressions

Sorry about nitpicking but there's a missing period after 'expressions', which
will break a test.

Marek