The following program gives a segfault with Intel 15.0.2 on Linux:
PROGRAM read_empty_substing
CHARACTER(LEN=20) :: c_in = "hello"
CHARACTER(LEN=20) :: c_out
INTEGER :: i_iostat
! These statements give a segfault with Intel 15.0.2
READ(c_in(2:1), '(A)') c_out
READ(c_in(2:1), '(A)', IOSTAT=i_iostat) c_out
! The following statements give end-of-file as expected:
!READ(c_in(2:1), *) c_out
!READ(c_in(2:1), *, IOSTAT=i_iostat) c_out
print *, TRIM(c_out)
END PROGRAM
As you can see, we try to read from an empty substring of c_in, and I do expect an error condition, i.e. i_iostat/=0. Instead I get a segfault. The read statements with format string segfaults, while the list directed read gives end-of-file. Am I supposed to check the length of the internal file before I read from it, or could this be a compiler bug?
Øyvind
------------------------
(gdb) run
Starting program: /fs/auto/home/oyvindj/tmp/read_segfault/a.out
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Catchpoint -2 (signal SIGSEGV), 0x00000000004804c6 in __intel_ssse3_rep_memcpy ()
(gdb) bt
#0 0x00000000004804c6 in __intel_ssse3_rep_memcpy ()
#1 0x0000000000409546 in for_read_int_fmt_xmit ()
#2 0x000000000040783f in for_read_int_fmt ()
#3 0x0000000000402e01 in read_empty_substing () at read.f90:7
#4 0x0000000000402d5e in main ()
