This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug c/16409] [3.5 regression] ICE in size_binop, at fold-const.c


------- Additional Comments From rth at gcc dot gnu dot org  2004-07-07 17:15 -------
I have no idea what this is supposed to mean, if anything.

But for the record, with a build from July 4, we get 

      __builtin_memset (&a, 0, T.1);
      T.1 = (long unsigned int)i.0;
      __builtin_memcpy (&a, "", T.1);

and with a build from July 7 (the ICE) we get

      i.0 = i;
      i.1 = (unsigned int)i.0;
      a[0]{lb: 0 sz: NON_LVALUE_EXPR <i.1>} = "";

The abort is, unsurprisingly, in translating the array assignment to rtl.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16409


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]