[Bug fortran/79230] New: [Regression on 7.0.1] Run time error: malloc on valid code

[juergen.reuter at desy dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79230

            Bug ID: 79230
           Summary: [Regression on 7.0.1] Run time error: malloc on valid
                    code
           Product: gcc
           Version: 7.0.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: juergen.reuter at desy dot de
  Target Milestone: ---

Created attachment 40580
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=40580&action=edit
Unit test suite that experiences the run time error.

The following case (part of our unit test suite) throws run time errors (61 out
of 110 tests), and _all_ of the functional tests. I assume that they are all
related to the same 'feature' (change in the gfortran code). In the tarball, do 
make, and then 
./rt_error --check evaluators
Using r244868, all older gfortran versions since 4.8.5 are working. 
The code is rather lengthy (but still not the complete testsuite). I try to
provide 
a smaller test case if possible, but also a complete test suite. 

This is the error:
Running test: evaluator_1rt_error(56542,0x7fffb05123c0) malloc: *** error for
object 0x7fb4ebf00110: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug

Program received signal SIGABRT: Process abort signal.

Backtrace for this error:
#0  0x10cdfd7c9
#1  0x10cdfcb63
#2  0x7fffa78d2bb9
Abort trap: 6

Up to now I couldn't produce a proper backtrace.