| Message ID | 6109c0ad-7256-b591-fb33-04d6ce525da6@gmail.com |
|---|---|
| State | New |
| Headers | show |
| Series | avoid using TYPE_SIZE unless it's constant [PR94131] | expand |
On Tue, 2020-03-17 at 19:35 -0600, Martin Sebor via Gcc-patches wrote: > PR tree-optimization/94131 - ICE on printf with a VLA string and -fno-tree-ccp > > gcc/testsuite/ChangeLog: > > PR tree-optimization/94131 > * gcc.dg/pr94131.c: New test. > > gcc/ChangeLog: > > PR tree-optimization/94131 > * gimple-fold.c (get_range_strlen_tree): Fail for variable-length > types and decls. > * tree-ssa-strlen.c (get_range_strlen_dynamic): Avoid assuming > types have constant sizes. OK jeff
On Tue, 2020-03-17 at 19:35 -0600, Martin Sebor via Gcc-patches wrote: > PR tree-optimization/94131 - ICE on printf with a VLA string and -fno-tree-ccp > > gcc/testsuite/ChangeLog: > > PR tree-optimization/94131 > * gcc.dg/pr94131.c: New test. > > gcc/ChangeLog: > > PR tree-optimization/94131 > * gimple-fold.c (get_range_strlen_tree): Fail for variable-length > types and decls. > * tree-ssa-strlen.c (get_range_strlen_dynamic): Avoid assuming > types have constant sizes. I approved this a week or so ago. I went ahead and committed it to get the P1 resolved. jeff
On 3/25/20 9:40 AM, Jeff Law wrote: > On Tue, 2020-03-17 at 19:35 -0600, Martin Sebor via Gcc-patches wrote: >> PR tree-optimization/94131 - ICE on printf with a VLA string and -fno-tree-ccp >> >> gcc/testsuite/ChangeLog: >> >> PR tree-optimization/94131 >> * gcc.dg/pr94131.c: New test. >> >> gcc/ChangeLog: >> >> PR tree-optimization/94131 >> * gimple-fold.c (get_range_strlen_tree): Fail for variable-length >> types and decls. >> * tree-ssa-strlen.c (get_range_strlen_dynamic): Avoid assuming >> types have constant sizes. > I approved this a week or so ago. I went ahead and committed it to get the P1 > resolved. Thanks. It was on my list of things to do this week, along with the fixes for PR94098 and PR94004. Martin
On Wed, 2020-03-25 at 10:41 -0600, Martin Sebor wrote: > On 3/25/20 9:40 AM, Jeff Law wrote: > > On Tue, 2020-03-17 at 19:35 -0600, Martin Sebor via Gcc-patches wrote: > > > PR tree-optimization/94131 - ICE on printf with a VLA string and -fno-tree- > > > ccp > > > > > > gcc/testsuite/ChangeLog: > > > > > > PR tree-optimization/94131 > > > * gcc.dg/pr94131.c: New test. > > > > > > gcc/ChangeLog: > > > > > > PR tree-optimization/94131 > > > * gimple-fold.c (get_range_strlen_tree): Fail for variable-length > > > types and decls. > > > * tree-ssa-strlen.c (get_range_strlen_dynamic): Avoid assuming > > > types have constant sizes. > > I approved this a week or so ago. I went ahead and committed it to get the > > P1 > > resolved. > > Thanks. It was on my list of things to do this week, along with > the fixes for PR94098 and PR94004. No problem. I'm just whittling down outstanding stuff that potentially affects the release, starting with P1s, of course :-) jeff
PR tree-optimization/94131 - ICE on printf with a VLA string and -fno-tree-ccp gcc/testsuite/ChangeLog: PR tree-optimization/94131 * gcc.dg/pr94131.c: New test. gcc/ChangeLog: PR tree-optimization/94131 * gimple-fold.c (get_range_strlen_tree): Fail for variable-length types and decls. * tree-ssa-strlen.c (get_range_strlen_dynamic): Avoid assuming types have constant sizes. diff --git a/gcc/gimple-fold.c b/gcc/gimple-fold.c index fa7a396a361..c85b4f2662b 100644 --- a/gcc/gimple-fold.c +++ b/gcc/gimple-fold.c @@ -1372,7 +1372,9 @@ get_range_strlen_tree (tree arg, bitmap *visited, strlen_range_kind rkind, /* Fail when the array bound is unknown or zero. */ val = TYPE_SIZE_UNIT (optype); - if (!val || integer_zerop (val)) + if (!val + || TREE_CODE (val) != INTEGER_CST + || integer_zerop (val)) return false; val = fold_build2 (MINUS_EXPR, TREE_TYPE (val), val, @@ -1406,7 +1408,9 @@ get_range_strlen_tree (tree arg, bitmap *visited, strlen_range_kind rkind, /* Fail when the array bound is unknown or zero. */ val = TYPE_SIZE_UNIT (optype); - if (!val || integer_zerop (val)) + if (!val + || TREE_CODE (val) != INTEGER_CST + || integer_zerop (val)) return false; val = fold_build2 (MINUS_EXPR, TREE_TYPE (val), val, integer_one_node); @@ -1442,7 +1446,9 @@ get_range_strlen_tree (tree arg, bitmap *visited, strlen_range_kind rkind, /* Fail if the offset is out of bounds. Such accesses should be diagnosed at some point. */ val = DECL_SIZE_UNIT (ref); - if (!val || integer_zerop (val)) + if (!val + || TREE_CODE (val) != INTEGER_CST + || integer_zerop (val)) return false; poly_offset_int psiz = wi::to_offset (val); diff --git a/gcc/testsuite/gcc.dg/pr84131.c b/gcc/testsuite/gcc.dg/pr84131.c new file mode 100644 index 00000000000..19de5685224 --- /dev/null +++ b/gcc/testsuite/gcc.dg/pr94131.c @@ -0,0 +1,29 @@ +/* PR 94131 - ICE on printf with a VLA string and -fno-tree-ccp + -fno-tree-forwprop + { dg-do compile } + { dg-options "-O1 -fno-tree-ccp -fno-tree-forwprop" } */ + +void rv1 (int n) +{ + char a[n]; + __INTPTR_TYPE__ i = (__INTPTR_TYPE__ )&a[0]; + i &= 3; + + __builtin_memset (a, '\0', sizeof a); + __builtin_printf ("%s", i ? &a[0] : ""); +} + + +void sink (void*); + +void rv2 (int n) +{ + char a[n]; + __INTPTR_TYPE__ i = (__INTPTR_TYPE__)&a[0]; + i &= 3; + + __builtin_memset (a, '\0', sizeof a); + __builtin_printf ("%s", i ? &a[0] : ""); + + sink (a); +} diff --git a/gcc/tree-ssa-strlen.c b/gcc/tree-ssa-strlen.c index 8815cdbc9ca..e1589f68806 100644 --- a/gcc/tree-ssa-strlen.c +++ b/gcc/tree-ssa-strlen.c @@ -1139,10 +1139,16 @@ get_range_strlen_dynamic (tree src, c_strlen_data *pdata, bitmap *visited, { tree basetype = TREE_TYPE (base); tree size = TYPE_SIZE_UNIT (basetype); - ++off; /* Increment for the terminating nul. */ - pdata->maxlen = fold_build2 (MINUS_EXPR, size_type_node, size, - build_int_cst (size_type_node, off)); - pdata->maxbound = pdata->maxlen; + if (TREE_CODE (size) == INTEGER_CST) + { + ++off; /* Increment for the terminating nul. */ + tree toffset = build_int_cst (size_type_node, off); + pdata->maxlen = fold_build2 (MINUS_EXPR, size_type_node, size, + toffset); + pdata->maxbound = pdata->maxlen; + } + else + pdata->maxlen = build_all_ones_cst (size_type_node); } else pdata->maxlen = build_all_ones_cst (size_type_node);