[Bug debug/57481] New: [4.8 Regression] LTO VTA compile time hog

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57481

            Bug ID: 57481
           Summary: [4.8 Regression] LTO VTA compile time hog
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Keywords: compile-time-hog
          Severity: normal
          Priority: P3
         Component: debug
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jakub at gcc dot gnu.org
                CC: aoliva at gcc dot gnu.org
            Target: x86_64-linux

Created attachment 30233
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30233&action=edit
mksh/misc.c

With the attached testcase,
gcc -g -m32 -O2 -flto -r -nostdlib -o mksh.o mksh.c
spends extreme amount of time in var-tracking (10+ minutes, killed it
afterwards), while without -flto, or with -flto -fno-var-tracking-assignments
it is compiled almost instantly.
Works fine with gcc 4.7 or current trunk, but as it involves LTO, haven't
bisected it (my bisect stuff doesn't include LTO).
No idea whether this is a LTO bug that it breaks the debug stmts some way, or
just a var-tracking bug.