@@ -1,3 +1,14 @@
+2018-12-23 Martin Sebor <msebor@redhat.com>
+ Jeff Law <law@redhat.com>
+
+ * builtins.c (unterminated_array): Use empty brace initialization
+ for c_strlen_data.
+ (c_strlen, expand_builtin_strnlen): Likewise.
+ (expand_builtin_stpcpy_1, fold_builtin_strlen): Likewise.
+ * gimple-fold.c (get_range_strlen): Likewise.
+ (gimple_fold_builtin_stpcpy): Likewise.
+ * gimple-ssa-sprintf.c (get_string_length): Likewise.
+
2018-12-23 Alan Modra <amodra@gmail.com>
PR 88346
@@ -575,8 +575,7 @@ unterminated_array (tree exp, tree *size /* = NULL */, bool *exact /* = NULL */)
{
/* C_STRLEN will return NULL and set DECL in the info
structure if EXP references a unterminated array. */
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data = { };
tree len = c_strlen (exp, 1, &data);
if (len == NULL_TREE && data.len && data.decl)
{
@@ -647,8 +646,7 @@ c_strlen (tree src, int only_value, c_strlen_data *data, unsigned eltsize)
/* If we were not passed a DATA pointer, then get one to a local
structure. That avoids having to check DATA for NULL before
each time we want to use it. */
- c_strlen_data local_strlen_data;
- memset (&local_strlen_data, 0, sizeof (c_strlen_data));
+ c_strlen_data local_strlen_data = { };
if (!data)
data = &local_strlen_data;
@@ -3085,8 +3083,7 @@ expand_builtin_strnlen (tree exp, rtx target, machine_mode target_mode)
/* FIXME: Change c_strlen() to return sizetype instead of ssizetype
so these conversions aren't necessary. */
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data { };
tree len = c_strlen (src, 0, &data, 1);
if (len)
len = fold_convert_loc (loc, TREE_TYPE (bound), len);
@@ -4086,8 +4083,7 @@ expand_builtin_stpcpy_1 (tree exp, rtx target, machine_mode mode)
compile-time, not an expression containing a string. This is
because the latter will potentially produce pessimized code
when used to produce the return value. */
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data = { };
if (!c_getstr (src, NULL)
|| !(len = c_strlen (src, 0, &data, 1)))
return expand_movstr (dst, src, target,
@@ -8571,8 +8567,7 @@ fold_builtin_strlen (location_t loc, tree type, tree arg)
return NULL_TREE;
else
{
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data = { };
tree len = c_strlen (arg, 0, &data);
if (len)
@@ -1335,8 +1335,7 @@ get_range_strlen (tree arg, tree length[2], bitmap *visited, int type,
}
else
{
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data = { };
val = c_strlen (arg, 1, &data, eltsize);
/* If we potentially had a non-terminated string, then
@@ -2824,8 +2823,7 @@ gimple_fold_builtin_stpcpy (gimple_stmt_iterator *gsi)
}
/* Set to non-null if ARG refers to an unterminated array. */
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data = { };
tree len = c_strlen (src, 1, &data, 1);
if (!len
|| TREE_CODE (len) != INTEGER_CST)
@@ -2003,8 +2003,7 @@ get_string_length (tree str, unsigned eltsize)
if (!str)
return fmtresult ();
- c_strlen_data data;
- memset (&data, 0, sizeof (c_strlen_data));
+ c_strlen_data data = { };
tree slen = c_strlen (str, 1, &data, eltsize);
if (slen && TREE_CODE (slen) == INTEGER_CST)
{
So I'm working to break down and resubmit Martin's work to fix the over-aggressive string length range computations. There's a fair amount of cleanup and other non-behavorial changes in that work. There's two paths we could take. One would be to drop all the non-behavorial changes and reduce the patch down to just the fixes. Sometimes this is my preferred route, but not this time. With Martin out, I want the end result to look at much like his work as possible to avoid introducing errors. So culling out the non-behavorial stuff will make comparing the end result to Martin's patch from Nov essentially impossible. Additionally, the non-behaviorial changes are generally things we want to fix anyway. So instead I'm going to break out the non-behavorial stuff, get it installed, give the testers a day to build them (beyond my own standard bootstrap/regression testing). Then I'll work my way into the meat of Martin's fixes. Sooo.. This is the first of the non-behaviorial changes from Martin. When I introduced c_strlen_data to make our API more easily extendable, I used memset for the structure initialization. Using empty brace initialization just seems cleaner. I originally considered it, but got it in my head that it wasn't supported in some case. Maybe I was stuck in a C mentality rather than C++.. :-) This patch extracts Martin's work to convert the existing memset calls in this space to instead use empty brace initialization. None of these changes should affect the behavior of the resultant compiler. Bootstrapped and regression tested on x86_64. Installing on the trunk. My tester will pick this up over the next 24hrs. Committed to the trunk. Jeff commit 6f2fa95d8a2b62ffccddd23c6b04cef26b612d2a Author: Jeff Law <law@redhat.com> Date: Fri Dec 21 16:55:19 2018 -0700 2018-12-23 Martin Sebor <msebor@redhat.com> Jeff Law <law@redhat.com> * builtins.c (unterminated_array): Use empty brace initialization for c_strlen_data. (c_strlen, expand_builtin_strnlen): Likewise. (expand_builtin_stpcpy_1, fold_builtin_strlen): Likewise. * gimple-fold.c (get_range_strlen): Likewise. (gimple_fold_builtin_stpcpy): Likewise. * gimple-ssa-sprintf.c (get_string_length): Likewise.