Off-by-one regression with deferred-length character string in structure constructor

[jwmwalrus]

Hi.

Although I'm still using the 2023.2.4 version of ifort (and icc), I decided to evaluate the viability of newest ifx.

After the shock of the false positives caused by "-check uninit", I noticed something that seems like a regression.

The following code

module mod1
    implicit none

contains
    subroutine parse_args(args)
        character(*), intent(inout) :: args

        integer :: i, iline, nline, ndash
        character(:), allocatable :: line

        type :: token_type
            logical :: is_flag = .false.
            character(:), allocatable :: val
        end type

        type(token_type), allocatable :: tokens(:)

        line = trim(args)//new_line('')
        tokens = [token_type ::]
        print*,'args=',line

        iline = 0
        nline = len(line)
        ndash = 0
        do while (iline < nline); iline = iline+1
            select case (line(iline:iline))
            case ('-')
                ndash = ndash+1
            case (new_line(''))
                exit
            case default
                if (ndash == 1) then
                    do i = iline, nline
                       if (line(i:i) == new_line('')) exit

                       print*,'line(i:i)=',line(i:i)
                       tokens = [tokens, token_type(.true., line(i:i))]
                    enddo
                    ndash = 0
                endif
            end select
        enddo

        do i = 1, size(tokens)
            print*,'token #', i, ', is_flag=',tokens(i)%is_flag,&
                ', val=',tokens(i)%val
        enddo
    end subroutine
end module mod1

use mod1

implicit none

character(7) :: args
args = '-0'
call parse_args(args)
end

Has the correct output when compiled with ifort 2023.2.4:

(ins)$ ifort -V main2.f90 && ./a.out 
Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.10.0 Build 20230609_000000
Copyright (C) 1985-2023 Intel Corporation.  All rights reserved.

 Intel(R) Fortran 2021.10.0-1769.02
GNU ld (GNU Binutils for Debian) 2.42
 args=-0

 line(i:i)=0
 token #           1 , is_flag= T , val=0

But with the latest ifx, the value assigned to tokens(1)%val through the structure constructor is incorrect:

$ ifx -V main2.f90 && ./a.out 
Intel(R) Fortran Compiler for applications running on Intel(R) 64, Version 2024.1.0 Build 20240308
Copyright (C) 1985-2024 Intel Corporation. All rights reserved.

 Intel(R) Fortran 24.0-1472.3
GNU ld (GNU Binutils for Debian) 2.42
 args=-0

 line(i:i)=0
 token #           1 , is_flag= T , val=-

And when trying to use the latest ifort, I get an ICE:

$ ifort -V main2.f90 && ./a.out 
Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.12.0 Build 20240222_000000
Copyright (C) 1985-2024 Intel Corporation.  All rights reserved.

ifort: remark #10448: Intel(R) Fortran Compiler Classic (ifort) is now deprecated and will be discontinued late 2024. Intel recommends that customers transition now to using the LLVM-based Intel(R) Fortran Compiler (ifx) for continued Windows* and Linux* support, new language support, new language features, and optimizations. Use '-diag-disable=10448' to disable this message.
 Intel(R) Fortran 2021.12.0-1472
0_1167

catastrophic error: **Internal compiler error: internal abort** Please report this error along with the circumstances in which it occurred in a Software Problem Report.  Note: File and line given may not be explicit cause of this error.
compilation aborted for main2.f90 (code 1)

That's it.

[Barbara_P_Intel]

Thanks for reporting these issues. I see the same thing with the latest internal build of the ifx and ifort, too. I filed a bug report, CMPLRLLVM-57421. We'll let you know when it's fixed.

[Barbara_P_Intel]

WOW! The Fortran compiler engineers are cranking out the bug fixes!

Look for the fix for this one in the 2024.2 release that is planned for mid-2024.