https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85788
--- Comment #5 from qinzhao at gcc dot gnu.org ---
with my current patch to PR109071, adding -fdiagnostics-explain-harder, with
an additional option -fno-tree-dominator-opts, the diagnostic becomes:
t_85788.c:5:3: warning: ‘__builtin___memset_chk’ forming offset [64008,
134180872] is out of the bounds [0, 64008] of object ‘m’ with type ‘j’
[-Warray-bounds=]
5 | __builtin___memset_chk(k, b, l, f);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
‘n’: events 1-2
5 | __builtin___memset_chk(k, b, l, f);
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| |
| (2) t_85788.c:20:1: warning: out of array bounds here
......
14 | for (; c; c = e)
| ^
| |
| (1) t_85788.c:20:1: warning: when the condition is evaluated
to true
t_85788.c: In function ‘n’:
t_85788.c:21:3: note: ‘m’ declared here
I think that the above more details will help the user to locate the issue in
their code.
I will improve my current patch to record the code duplication during
"tree-dominator-opts" after "thread" in order to remove the
"-fno-tree-dominator-opts"