[gcc r14-1896] uiltins: Add support for clang compatible __builtin_{add, sub}c{, l, ll} [PR79173]

[gcc r14-1896] uiltins: Add support for clang compatible __builtin_{add, sub}c{, l, ll} [PR79173]

[Jakub Jelinek]

https://gcc.gnu.org/g:2b4e0415ad664cdb3ce87d1f7eee5ca26911a05b

commit r14-1896-g2b4e0415ad664cdb3ce87d1f7eee5ca26911a05b
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Fri Jun 16 19:47:28 2023 +0200

    uiltins: Add support for clang compatible __builtin_{add,sub}c{,l,ll} [PR79173]
    
    While the design of these builtins in clang is questionable,
    rather than being say
    unsigned __builtin_addc (unsigned, unsigned, bool, bool *)
    so that it is clear they add two [0, 0xffffffff] range numbers
    plus one [0, 1] range carry in and give [0, 0xffffffff] range
    return plus [0, 1] range carry out, they actually instead
    add 3 [0, 0xffffffff] values together but the carry out
    isn't then the expected [0, 2] value because
    0xffffffffULL + 0xffffffff + 0xffffffff is 0x2fffffffd,
    but just [0, 1] whether there was any overflow at all.
    
    It is something used in the wild and shorter to write than the
    corresponding
     #define __builtin_addc(a,b,carry_in,carry_out) \
      ({ unsigned _s; \
         unsigned _c1 = __builtin_uadd_overflow (a, b, &_s); \
         unsigned _c2 = __builtin_uadd_overflow (_s, carry_in, &_s); \
         *(carry_out) = (_c1 | _c2); \
         _s; })
    and so a canned builtin for something people could often use.
    It isn't that hard to maintain on the GCC side, as we just lower
    it to two .ADD_OVERFLOW calls early, and the already committed
    pottern recognization code can then make .UADDC/.USUBC calls out of
    that if the carry in is in [0, 1] range and the corresponding
    optab is supported by the target.
    
    2023-06-16  Jakub Jelinek  <jakub@redhat.com>
    
            PR middle-end/79173
            * builtin-types.def (BT_FN_UINT_UINT_UINT_UINT_UINTPTR,
            BT_FN_ULONG_ULONG_ULONG_ULONG_ULONGPTR,
            BT_FN_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONGPTR): New
            types.
            * builtins.def (BUILT_IN_ADDC, BUILT_IN_ADDCL, BUILT_IN_ADDCLL,
            BUILT_IN_SUBC, BUILT_IN_SUBCL, BUILT_IN_SUBCLL): New builtins.
            * builtins.cc (fold_builtin_addc_subc): New function.
            (fold_builtin_varargs): Handle BUILT_IN_{ADD,SUB}C{,L,LL}.
            * doc/extend.texi (__builtin_addc, __builtin_subc): Document.
    
            * gcc.target/i386/pr79173-11.c: New test.
            * gcc.dg/builtin-addc-1.c: New test.

Diff:
---
 gcc/builtin-types.def                      |   7 ++
 gcc/builtins.cc                            |  53 +++++++++++++++
 gcc/builtins.def                           |   6 ++
 gcc/doc/extend.texi                        |  44 +++++++++++++
 gcc/testsuite/gcc.dg/builtin-addc-1.c      | 101 +++++++++++++++++++++++++++++
 gcc/testsuite/gcc.target/i386/pr79173-11.c |  39 +++++++++++
 6 files changed, 250 insertions(+)

diff --git a/gcc/builtin-types.def b/gcc/builtin-types.def
index 5cb25ab71d0..43381bc8949 100644
--- a/gcc/builtin-types.def
+++ b/gcc/builtin-types.def
@@ -842,10 +842,17 @@ DEF_FUNCTION_TYPE_4 (BT_FN_PTR_PTR_INT_SIZE_SIZE,
 		     BT_PTR, BT_PTR, BT_INT, BT_SIZE, BT_SIZE)
 DEF_FUNCTION_TYPE_4 (BT_FN_UINT_UINT_UINT_UINT_UINT,
 		     BT_UINT, BT_UINT, BT_UINT, BT_UINT, BT_UINT)
+DEF_FUNCTION_TYPE_4 (BT_FN_UINT_UINT_UINT_UINT_UINTPTR,
+		     BT_UINT, BT_UINT, BT_UINT, BT_UINT, BT_PTR_UINT)
 DEF_FUNCTION_TYPE_4 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT,
 		     BT_UINT, BT_FLOAT, BT_FLOAT, BT_FLOAT, BT_FLOAT)
 DEF_FUNCTION_TYPE_4 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT,
 		     BT_ULONG, BT_ULONG, BT_ULONG, BT_UINT, BT_UINT)
+DEF_FUNCTION_TYPE_4 (BT_FN_ULONG_ULONG_ULONG_ULONG_ULONGPTR,
+		     BT_ULONG, BT_ULONG, BT_ULONG, BT_ULONG, BT_PTR_ULONG)
+DEF_FUNCTION_TYPE_4 (BT_FN_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONGPTR,
+		     BT_ULONGLONG, BT_ULONGLONG, BT_ULONGLONG, BT_ULONGLONG,
+		     BT_PTR_ULONGLONG)
 DEF_FUNCTION_TYPE_4 (BT_FN_STRING_STRING_CONST_STRING_SIZE_SIZE,
 		     BT_STRING, BT_STRING, BT_CONST_STRING, BT_SIZE, BT_SIZE)
 DEF_FUNCTION_TYPE_4 (BT_FN_INT_FILEPTR_INT_CONST_STRING_VALIST_ARG,
diff --git a/gcc/builtins.cc b/gcc/builtins.cc
index 8400adaf5b4..6dff5214ff8 100644
--- a/gcc/builtins.cc
+++ b/gcc/builtins.cc
@@ -9555,6 +9555,51 @@ fold_builtin_arith_overflow (location_t loc, enum built_in_function fcode,
   return build2_loc (loc, COMPOUND_EXPR, boolean_type_node, store, ovfres);
 }
 
+/* Fold __builtin_{add,sub}c{,l,ll} into pair of internal functions
+   that return both result of arithmetics and overflowed boolean
+   flag in a complex integer result.  */
+
+static tree
+fold_builtin_addc_subc (location_t loc, enum built_in_function fcode,
+			tree *args)
+{
+  enum internal_fn ifn;
+
+  switch (fcode)
+    {
+    case BUILT_IN_ADDC:
+    case BUILT_IN_ADDCL:
+    case BUILT_IN_ADDCLL:
+      ifn = IFN_ADD_OVERFLOW;
+      break;
+    case BUILT_IN_SUBC:
+    case BUILT_IN_SUBCL:
+    case BUILT_IN_SUBCLL:
+      ifn = IFN_SUB_OVERFLOW;
+      break;
+    default:
+      gcc_unreachable ();
+    }
+
+  tree type = TREE_TYPE (args[0]);
+  tree ctype = build_complex_type (type);
+  tree call = build_call_expr_internal_loc (loc, ifn, ctype, 2,
+					    args[0], args[1]);
+  tree tgt = save_expr (call);
+  tree intres = build1_loc (loc, REALPART_EXPR, type, tgt);
+  tree ovfres = build1_loc (loc, IMAGPART_EXPR, type, tgt);
+  call = build_call_expr_internal_loc (loc, ifn, ctype, 2,
+				       intres, args[2]);
+  tgt = save_expr (call);
+  intres = build1_loc (loc, REALPART_EXPR, type, tgt);
+  tree ovfres2 = build1_loc (loc, IMAGPART_EXPR, type, tgt);
+  ovfres = build2_loc (loc, BIT_IOR_EXPR, type, ovfres, ovfres2);
+  tree mem_arg3 = build_fold_indirect_ref_loc (loc, args[3]);
+  tree store
+    = fold_build2_loc (loc, MODIFY_EXPR, void_type_node, mem_arg3, ovfres);
+  return build2_loc (loc, COMPOUND_EXPR, type, store, intres);
+}
+
 /* Fold a call to __builtin_FILE to a constant string.  */
 
 static inline tree
@@ -10843,6 +10888,14 @@ fold_builtin_varargs (location_t loc, tree fndecl, tree *args, int nargs)
       ret = fold_builtin_fpclassify (loc, args, nargs);
       break;
 
+    case BUILT_IN_ADDC:
+    case BUILT_IN_ADDCL:
+    case BUILT_IN_ADDCLL:
+    case BUILT_IN_SUBC:
+    case BUILT_IN_SUBCL:
+    case BUILT_IN_SUBCLL:
+      return fold_builtin_addc_subc (loc, fcode, args);
+
     default:
       break;
     }
diff --git a/gcc/builtins.def b/gcc/builtins.def
index 4ad95a12f83..76e7200e772 100644
--- a/gcc/builtins.def
+++ b/gcc/builtins.def
@@ -934,6 +934,12 @@ DEF_GCC_BUILTIN        (BUILT_IN_USUBLL_OVERFLOW, "usubll_overflow", BT_FN_BOOL_
 DEF_GCC_BUILTIN        (BUILT_IN_UMUL_OVERFLOW, "umul_overflow", BT_FN_BOOL_UINT_UINT_UINTPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
 DEF_GCC_BUILTIN        (BUILT_IN_UMULL_OVERFLOW, "umull_overflow", BT_FN_BOOL_ULONG_ULONG_ULONGPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
 DEF_GCC_BUILTIN        (BUILT_IN_UMULLL_OVERFLOW, "umulll_overflow", BT_FN_BOOL_ULONGLONG_ULONGLONG_ULONGLONGPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
+DEF_GCC_BUILTIN        (BUILT_IN_ADDC, "addc", BT_FN_UINT_UINT_UINT_UINT_UINTPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
+DEF_GCC_BUILTIN        (BUILT_IN_ADDCL, "addcl", BT_FN_ULONG_ULONG_ULONG_ULONG_ULONGPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
+DEF_GCC_BUILTIN        (BUILT_IN_ADDCLL, "addcll", BT_FN_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONGPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
+DEF_GCC_BUILTIN        (BUILT_IN_SUBC, "subc", BT_FN_UINT_UINT_UINT_UINT_UINTPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
+DEF_GCC_BUILTIN        (BUILT_IN_SUBCL, "subcl", BT_FN_ULONG_ULONG_ULONG_ULONG_ULONGPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
+DEF_GCC_BUILTIN        (BUILT_IN_SUBCLL, "subcll", BT_FN_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONG_ULONGLONGPTR, ATTR_NOTHROW_NONNULL_LEAF_LIST)
 
 /* Category: miscellaneous builtins.  */
 DEF_LIB_BUILTIN        (BUILT_IN_ABORT, "abort", BT_FN_VOID, ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST)
diff --git a/gcc/doc/extend.texi b/gcc/doc/extend.texi
index e01e2444b57..c01cd3fe90c 100644
--- a/gcc/doc/extend.texi
+++ b/gcc/doc/extend.texi
@@ -12839,6 +12839,50 @@ after addition, conditional jump on carry etc.
  
 @enddefbuiltin
 
+@defbuiltin{unsigned int __builtin_addc (unsigned int a, unsigned int b, unsigned int carry_in, unsigned int *carry_out)}
+@defbuiltinx{unsigned long int __builtin_addcl (unsigned long int a, unsigned long int b, unsigned int carry_in, unsigned long int *carry_out)}
+@defbuiltinx{unsigned long long int __builtin_addcll (unsigned long long int a, unsigned long long int b, unsigned long long int carry_in, unsigned long long int *carry_out)}
+
+These built-in functions are equivalent to:
+@smallexample
+  (@{ __typeof__ (@var{a}) s; \
+      __typeof__ (@var{a}) c1 = __builtin_add_overflow (@var{a}, @var{b}, &s); \
+      __typeof__ (@var{a}) c2 = __builtin_add_overflow (s, @var{carry_in}, &s); \
+      *(@var{carry_out}) = c1 | c2; \
+      s; @})
+@end smallexample
+
+i.e.@: they add 3 unsigned values, set what the last argument
+points to to 1 if any of the two additions overflowed (otherwise 0)
+and return the sum of those 3 unsigned values.  Note, while all
+the first 3 arguments can have arbitrary values, better code will be
+emitted if one of them (preferrably the third one) has only values
+0 or 1 (i.e.@: carry-in).
+
+@enddefbuiltin
+
+@defbuiltin{unsigned int __builtin_subc (unsigned int a, unsigned int b, unsigned int carry_in, unsigned int *carry_out)}
+@defbuiltinx{unsigned long int __builtin_subcl (unsigned long int a, unsigned long int b, unsigned int carry_in, unsigned long int *carry_out)}
+@defbuiltinx{unsigned long long int __builtin_subcll (unsigned long long int a, unsigned long long int b, unsigned long long int carry_in, unsigned long long int *carry_out)}
+
+These built-in functions are equivalent to:
+@smallexample
+  (@{ __typeof__ (@var{a}) s; \
+      __typeof__ (@var{a}) c1 = __builtin_sub_overflow (@var{a}, @var{b}, &s); \
+      __typeof__ (@var{a}) c2 = __builtin_sub_overflow (s, @var{carry_in}, &s); \
+      *(@var{carry_out}) = c1 | c2; \
+      s; @})
+@end smallexample
+
+i.e.@: they subtract 2 unsigned values from the first unsigned value,
+set what the last argument points to to 1 if any of the two subtractions
+overflowed (otherwise 0) and return the result of the subtractions.
+Note, while all the first 3 arguments can have arbitrary values, better code
+will be emitted if one of them (preferrably the third one) has only values
+0 or 1 (i.e.@: carry-in).
+
+@enddefbuiltin
+
 @node x86 specific memory model extensions for transactional memory
 @section x86-Specific Memory Model Extensions for Transactional Memory
 
diff --git a/gcc/testsuite/gcc.dg/builtin-addc-1.c b/gcc/testsuite/gcc.dg/builtin-addc-1.c
new file mode 100644
index 00000000000..c875d8f66e8
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/builtin-addc-1.c
@@ -0,0 +1,101 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -g" } */
+
+int
+main ()
+{
+  unsigned int c;
+  unsigned long cl;
+  unsigned long long cll;
+  if (__builtin_addc (1, 42, 0, &c) != 43 || c != 0)
+    __builtin_abort ();
+  if (__builtin_addc (1, 42, 15, &c) != 58 || c != 0)
+    __builtin_abort ();
+  if (__builtin_addc (-2U, -3U, -4U, &c) != -9U || c != 1)
+    __builtin_abort ();
+  if (__builtin_addc (-2U, 1, 0, &c) != -1U || c != 0)
+    __builtin_abort ();
+  if (__builtin_addc (-2U, 1, 1, &c) != 0 || c != 1)
+    __builtin_abort ();
+  if (__builtin_addc (-2U, 2, 0, &c) != 0 || c != 1)
+    __builtin_abort ();
+  if (__builtin_addc (-2U, 0, 2, &c) != 0 || c != 1)
+    __builtin_abort ();
+  if (__builtin_addcl (1L, 42L, 0L, &cl) != 43 || cl != 0L)
+    __builtin_abort ();
+  if (__builtin_addcl (1L, 42L, 15L, &cl) != 58 || cl != 0L)
+    __builtin_abort ();
+  if (__builtin_addcl (-2UL, -3UL, -4UL, &cl) != -9UL || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_addcl (-2UL, 1L, 0L, &cl) != -1UL || cl != 0L)
+    __builtin_abort ();
+  if (__builtin_addcl (-2UL, 1L, 1L, &cl) != 0 || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_addcl (-2UL, 2L, 0L, &cl) != 0 || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_addcl (-2UL, 0L, 2L, &cl) != 0 || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_addcll (1LL, 42LL, 0LL, &cll) != 43 || cll != 0LL)
+    __builtin_abort ();
+  if (__builtin_addcll (1LL, 42LL, 15LL, &cll) != 58 || cll != 0LL)
+    __builtin_abort ();
+  if (__builtin_addcll (-2ULL, -3ULL, -4ULL, &cll) != -9ULL || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_addcll (-2ULL, 1LL, 0LL, &cll) != -1ULL || cll != 0LL)
+    __builtin_abort ();
+  if (__builtin_addcll (-2ULL, 1LL, 1LL, &cll) != 0 || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_addcll (-2ULL, 2LL, 0LL, &cll) != 0 || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_addcll (-2ULL, 0LL, 2LL, &cll) != 0 || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_subc (42, 42, 0, &c) != 0 || c != 0)
+    __builtin_abort ();
+  if (__builtin_subc (42, 42, 1, &c) != -1U || c != 1)
+    __builtin_abort ();
+  if (__builtin_subc (1, -3U, -4U, &c) != 8 || c != 1)
+    __builtin_abort ();
+  if (__builtin_subc (-2U, 1, 0, &c) != -3U || c != 0)
+    __builtin_abort ();
+  if (__builtin_subc (-2U, -1U, 0, &c) != -1U || c != 1)
+    __builtin_abort ();
+  if (__builtin_subc (-2U, -2U, 0, &c) != 0 || c != 0)
+    __builtin_abort ();
+  if (__builtin_subc (-2U, -2U, 1, &c) != -1U || c != 1)
+    __builtin_abort ();
+  if (__builtin_subc (-2U, 1, -2U, &c) != -1U || c != 1)
+    __builtin_abort ();
+  if (__builtin_subcl (42L, 42L, 0L, &cl) != 0L || cl != 0L)
+    __builtin_abort ();
+  if (__builtin_subcl (42L, 42L, 1L, &cl) != -1UL || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_subcl (1L, -3UL, -4UL, &cl) != 8L || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_subcl (-2UL, 1L, 0L, &cl) != -3UL || cl != 0L)
+    __builtin_abort ();
+  if (__builtin_subcl (-2UL, -1UL, 0L, &cl) != -1UL || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_subcl (-2UL, -2UL, 0L, &cl) != 0L || cl != 0L)
+    __builtin_abort ();
+  if (__builtin_subcl (-2UL, -2UL, 1L, &cl) != -1UL || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_subcl (-2UL, 1L, -2UL, &cl) != -1UL || cl != 1L)
+    __builtin_abort ();
+  if (__builtin_subcll (42LL, 42LL, 0LL, &cll) != 0LL || cll != 0LL)
+    __builtin_abort ();
+  if (__builtin_subcll (42LL, 42LL, 1LL, &cll) != -1ULL || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_subcll (1LL, -3ULL, -4ULL, &cll) != 8LL || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_subcll (-2ULL, 1LL, 0LL, &cll) != -3ULL || cll != 0LL)
+    __builtin_abort ();
+  if (__builtin_subcll (-2ULL, -1ULL, 0LL, &cll) != -1ULL || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_subcll (-2ULL, -2ULL, 0LL, &cll) != 0LL || cll != 0LL)
+    __builtin_abort ();
+  if (__builtin_subcll (-2ULL, -2ULL, 1LL, &cll) != -1ULL || cll != 1LL)
+    __builtin_abort ();
+  if (__builtin_subcll (-2ULL, 1LL, -2ULL, &cll) != -1ULL || cll != 1LL)
+    __builtin_abort ();
+  return 0;
+}
diff --git a/gcc/testsuite/gcc.target/i386/pr79173-11.c b/gcc/testsuite/gcc.target/i386/pr79173-11.c
new file mode 100644
index 00000000000..ee45afa1ebd
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/pr79173-11.c
@@ -0,0 +1,39 @@
+/* PR middle-end/79173 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -fno-stack-protector -masm=att" } */
+/* { dg-final { scan-assembler-times "addq\t%r\[^\n\r]*, \\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "adcq\t%r\[^\n\r]*, 8\\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "adcq\t%r\[^\n\r]*, 16\\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "adcq\t%r\[^\n\r]*, 24\\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "subq\t%r\[^\n\r]*, \\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "sbbq\t%r\[^\n\r]*, 8\\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "sbbq\t%r\[^\n\r]*, 16\\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "sbbq\t%r\[^\n\r]*, 24\\\(%rdi\\\)" 1 { target lp64 } } } */
+/* { dg-final { scan-assembler-times "addl\t%e\[^\n\r]*, \\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "adcl\t%e\[^\n\r]*, 4\\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "adcl\t%e\[^\n\r]*, 8\\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "adcl\t%e\[^\n\r]*, 12\\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "subl\t%e\[^\n\r]*, \\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "sbbl\t%e\[^\n\r]*, 4\\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "sbbl\t%e\[^\n\r]*, 8\\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+/* { dg-final { scan-assembler-times "sbbl\t%e\[^\n\r]*, 12\\\(%e\[^\n\r]*\\\)" 1 { target ia32 } } } */
+
+void
+foo (unsigned long *p, unsigned long *q)
+{
+  unsigned long c;
+  p[0] = __builtin_addcl (p[0], q[0], 0, &c);
+  p[1] = __builtin_addcl (p[1], q[1], c, &c);
+  p[2] = __builtin_addcl (p[2], q[2], c, &c);
+  p[3] = __builtin_addcl (p[3], q[3], c, &c);
+}
+
+void
+bar (unsigned long *p, unsigned long *q)
+{
+  unsigned long c;
+  p[0] = __builtin_subcl (p[0], q[0], 0, &c);
+  p[1] = __builtin_subcl (p[1], q[1], c, &c);
+  p[2] = __builtin_subcl (p[2], q[2], c, &c);
+  p[3] = __builtin_subcl (p[3], q[3], c, &c);
+}