| Message ID | Y4cmlqFFMt3p7Nz8@tucnak |
|---|---|
| State | New |
| Headers | show |
| Series | tree-chrec: Fix up ICE on pointer multiplication [PR107835] | expand |
On Wed, 30 Nov 2022, Jakub Jelinek wrote: > Hi! > > r13-254-gdd3c7873a61019e9 added an optimization for {a, +, a} (x-1), > but as can be seen on the following testcase, the way it is written > where chrec_fold_multiply is called with type doesn't work for pointers: > res = build_int_cst (TREE_TYPE (x), 1); > res = chrec_fold_plus (TREE_TYPE (x), x, res); > res = chrec_convert_rhs (type, res, NULL); > res = chrec_fold_multiply (type, chrecr, res); > while what we were doing before and what is still used if the condition > doesn't match is fine: > res = chrec_convert_rhs (TREE_TYPE (chrecr), x, NULL); > res = chrec_fold_multiply (TREE_TYPE (chrecr), chrecr, res); > res = chrec_fold_plus (type, CHREC_LEFT (chrec), res); > because it performs chrec_fold_multiply on TREE_TYPE (chrecr) and converts > only afterwards. > > I think the easiest fix is to ignore the new path for pointer types. > > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? OK. Thanks, Richard. > 2022-11-30 Jakub Jelinek <jakub@redhat.com> > > PR tree-optimization/107835 > * tree-chrec.cc (chrec_apply): Don't handle "{a, +, a} (x-1)" > as "a*x" if type is a pointer type. > > * gcc.c-torture/compile/pr107835.c: New test. > > --- gcc/tree-chrec.cc.jj 2022-05-10 18:33:14.641029951 +0200 > +++ gcc/tree-chrec.cc 2022-11-29 15:24:41.810400368 +0100 > @@ -622,7 +622,8 @@ chrec_apply (unsigned var, > /* "{a, +, b} (x)" -> "a + b*x". */ > else if (operand_equal_p (CHREC_LEFT (chrec), chrecr) > && TREE_CODE (x) == PLUS_EXPR > - && integer_all_onesp (TREE_OPERAND (x, 1))) > + && integer_all_onesp (TREE_OPERAND (x, 1)) > + && !POINTER_TYPE_P (type)) > { > /* We know the number of iterations can't be negative. > So {a, +, a} (x-1) -> "a*x". */ > --- gcc/testsuite/gcc.c-torture/compile/pr107835.c.jj 2022-11-29 15:31:32.565382590 +0100 > +++ gcc/testsuite/gcc.c-torture/compile/pr107835.c 2022-11-29 15:31:15.795628304 +0100 > @@ -0,0 +1,11 @@ > +/* PR tree-optimization/107835 */ > + > +int * > +foo (void) > +{ > + int *x = 0; > + unsigned n = n; > + for (; n; --n, ++x) > + ; > + return x; > +} > > Jakub > >
--- gcc/tree-chrec.cc.jj 2022-05-10 18:33:14.641029951 +0200 +++ gcc/tree-chrec.cc 2022-11-29 15:24:41.810400368 +0100 @@ -622,7 +622,8 @@ chrec_apply (unsigned var, /* "{a, +, b} (x)" -> "a + b*x". */ else if (operand_equal_p (CHREC_LEFT (chrec), chrecr) && TREE_CODE (x) == PLUS_EXPR - && integer_all_onesp (TREE_OPERAND (x, 1))) + && integer_all_onesp (TREE_OPERAND (x, 1)) + && !POINTER_TYPE_P (type)) { /* We know the number of iterations can't be negative. So {a, +, a} (x-1) -> "a*x". */ --- gcc/testsuite/gcc.c-torture/compile/pr107835.c.jj 2022-11-29 15:31:32.565382590 +0100 +++ gcc/testsuite/gcc.c-torture/compile/pr107835.c 2022-11-29 15:31:15.795628304 +0100 @@ -0,0 +1,11 @@ +/* PR tree-optimization/107835 */ + +int * +foo (void) +{ + int *x = 0; + unsigned n = n; + for (; n; --n, ++x) + ; + return x; +}
Hi! r13-254-gdd3c7873a61019e9 added an optimization for {a, +, a} (x-1), but as can be seen on the following testcase, the way it is written where chrec_fold_multiply is called with type doesn't work for pointers: res = build_int_cst (TREE_TYPE (x), 1); res = chrec_fold_plus (TREE_TYPE (x), x, res); res = chrec_convert_rhs (type, res, NULL); res = chrec_fold_multiply (type, chrecr, res); while what we were doing before and what is still used if the condition doesn't match is fine: res = chrec_convert_rhs (TREE_TYPE (chrecr), x, NULL); res = chrec_fold_multiply (TREE_TYPE (chrecr), chrecr, res); res = chrec_fold_plus (type, CHREC_LEFT (chrec), res); because it performs chrec_fold_multiply on TREE_TYPE (chrecr) and converts only afterwards. I think the easiest fix is to ignore the new path for pointer types. Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? 2022-11-30 Jakub Jelinek <jakub@redhat.com> PR tree-optimization/107835 * tree-chrec.cc (chrec_apply): Don't handle "{a, +, a} (x-1)" as "a*x" if type is a pointer type. * gcc.c-torture/compile/pr107835.c: New test. Jakub