[Bug middle-end/54321] [4.8 Regression] ice in tree_low_cst at -O3

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54321

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |jakub at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-08-19 20:22:39 UTC ---
Created attachment 28053
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28053
gcc48-pr54321.patch

The bug is pretty much obvious, val2 is first checked for host_integerp (val2,
0) but then tree_low_cst (val2, 1) is used.  As the middle argument to memset
is int, we should use tree_low_cst (val2, 0).

I'll bootstrap/regtest this momentarily.