[Bug other/69432] New: ICE in connect_traces, at dwarf2cfi.c with -O3 -m32 -minline-stringops-dynamically

["jamborm at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69432

            Bug ID: 69432
           Summary: ICE in connect_traces, at dwarf2cfi.c with -O3 -m32
                    -minline-stringops-dynamically
           Product: gcc
           Version: 6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: other
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jamborm at gcc dot gnu.org
  Target Milestone: ---
              Host: x86_64-linux
            Target: i686-linux

Created attachment 37434
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=37434&action=edit
C-reduced testcase

The attached testcase is reduced from cromfs-1.5.10.2.
On my x86_64-linux desktop, compiling with

path_to_trunk/inst/bin/g++ -O3 -m32 -minline-stringops-dynamically -S
creduced.C

leads to:

  creduced.C: In function âvoid LZMADeCompress()â:
  creduced.C:63:33: internal compiler error: in connect_traces, at
dwarf2cfi.c:2752
   void LZMADeCompress() { F b(a); }
                                 ^

  0xaa8a6a connect_traces
          /home/mjambor/gcc/trunk/src/gcc/dwarf2cfi.c:2752
  0xaac866 execute_dwarf2_frame
          /home/mjambor/gcc/trunk/src/gcc/dwarf2cfi.c:2975

The ICE happens with trunk and branches for gcc 5 and 4.9 but does not
with 4.8.

(Please change the component if you can think of a more suitable one.
As you can see, -g is not necessary so I suspect this is an EH bug,
rather than a debug issue.  I also can't tell to what extent this is a
target bug.)