optimization/9037: [3.3 regression] -fssa-dce crashes with simple code

[Steven Bosscher]

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9037

A patch for this from Jose Renau (who is also the
reporter of this problem) was applied almost three
months ago on mainline.  Here's a backport of the
patch.  OK for 3.3 if it passes regression tests?

I don't know what the correct ChangeLog entry should
be, so I just copy-pasted the entry from the mainline
ChangeLog.  Is there a convention for backported
changes?

Greetz
Steven


2002-12-26  Jose Renau <renau@cs.uiuc.edu>

        * ssa-dce.c (EXECUTE_IF_UNNECESSARY): Verify INSN is an
        INSN_P before checking to see if it is dead.
        (mark_all_insn_unnecessary): Similarly.
        (ssa_eliminate_dead_code): Similarly.
        * rtl.h (struct rtx_def): Update comments for in_struct usage
        in dead code elimination pass.
        (INSN_DEAD_CODE_P): Allow JUMP_INSN and CALL_INSN as well.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch
Type: application/x-java-vm
Size: 3808 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-bugs/attachments/20030324/8aa5023f/attachment.bin>