[Bug other/69432] [4.9/5/6 Regression] ICE in connect_traces, at dwarf2cfi.c with -O3 -m32 -minline-stringops-dynamically

[jakub at gcc dot gnu.org]

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

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2016-01-22
                 CC|                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |4.9.4
            Summary|ICE in connect_traces, at   |[4.9/5/6 Regression] ICE in
                   |dwarf2cfi.c with -O3 -m32   |connect_traces, at
                   |-minline-stringops-dynamica |dwarf2cfi.c with -O3 -m32
                   |lly                         |-minline-stringops-dynamica
                   |                            |lly
     Ever confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
This started to ICE with r203937.
pr69432.C: In function ‘void LZMADeCompress()’:
pr69432.C:63:33: error: wrong number of branch edges after unconditional jump
in bb 8
 void LZMADeCompress() { F b(a); }
                                 ^
and got fixed with r206242 and then started to ICE in connect_traces with
r206947.  With additional -mno-accumulate-outgoing-args this started to ICE
with r203171 in connect_traces and then for r203937-r206241 ICEd with wrong
number of branch edges.