RESHAPE with zero size input and padding blows stack

[IanH]

The following:
 

[fortran]PROGRAM BlowMyStack
  IMPLICIT NONE
  CALL TestProc
CONTAINS
  SUBROUTINE TestProc  
    CALL Proc2(RESHAPE([INTEGER::], [10], [1]))
  END SUBROUTINE
  
  SUBROUTINE Proc2(arr)
    INTEGER, INTENT(IN) :: arr(:)
    PRINT *, arr
  END SUBROUTINE Proc2

END PROGRAM BlowMyStack

[/fortran]

dies with a stack overflow when run after compiling with 12.1.1 with /Od on ia32.
 

[plain]H:\projects\FortranMisc>ifort /check:all /warn:all /standard-semantics /Od /traceback stack.f90
Intel Visual Fortran Compiler XE for applications running on IA-32, Version 12.1.1.258 Build 20111011
Copyright (C) 1985-2011 Intel Corporation.  All rights reserved.

Microsoft  Incremental Linker Version 10.00.40219.01
Copyright (C) Microsoft Corporation.  All rights reserved.

-out:stack.exe
-subsystem:console
-incremental:no
stack.obj

H:\projects\FortranMisc>stack
forrtl: severe (170): Program Exception - stack overflow
Image              PC        Routine            Line        Source
stack.exe          00E0E957  Unknown               Unknown  Unknown
stack.exe          00DE1021  _MAIN__                     3  stack.f90
stack.exe          00E288F3  Unknown               Unknown  Unknown
stack.exe          00E0EA8F  Unknown               Unknown  Unknown
kernel32.dll       761AD309  Unknown               Unknown  Unknown
ntdll.dll          76F716C3  Unknown               Unknown  Unknown
ntdll.dll          76F71696  Unknown               Unknown  Unknown[/plain]

[Steven_L_Intel1]

Very interesting. The compiler is using an uninitialized value when computing the size of the zero-sized array constructor. Escalated as DPD200175839. Incidentally, I found this very difficult to reproduce (as in get an error) from the command line - even using the same command you did. But I could reproduce it in the debugger and saw what was causing the problem.

[Steven_L_Intel1]

This has been fixed for a release later this year.