Dear Maintainer,

    * What led up to the situation?

I tried to compile cosmdist commit 3ea40c144d [1] from source with
'./configure && make' on my Pinephone.
[1] https://codeberg.org/boud/cosmdist/commit/3ea40c144dc6a5095ce9e46802aacc7c5a41a557

    * What exactly did you do (or not do) that was effective (or
      ineffective)?

My minimal working example [2] is:
--------------------------------------------------
$ cat x.f90

subroutine myfun(i,y)
   integer i
   logical y
   if(i.eq.1)then
      y = .true.
   endif
   return
end subroutine myfun

$ gfortran  -g -O2 -c -o x.o x.f90
--------------------------------------------------

[2] https://paste.debian.net/1377105/

    * What was the outcome of this action?

On repeated tries, the outcome was [3] about
20% of the time:

--------------------------------------------------
during GIMPLE pass: fre
x.f90:1:16:
#
     1 | subroutine myfun(i,y)
      |                ^
internal compiler error: Segmentation fault
0xffffxxxxxxxb __libc_start_call_main
         ../sysdeps/nptl/libc_start_call_main.h:58
0xffffxxxxxxxb __libc_start_main_impl
        ../csu/libc-start.c:360
--------------------------------------------------

and another 20% of the time it was the same, but
with 'pre' instead of 'fre'. About 60% of the time
'x.o' was created with no reported errors.

[3] https://paste.debian.net/1377107/

    * What outcome did you expect instead?

I expected 'x.o' to have been created with no reported errors.

Kernel: Linux 6.12-sunxi64 (SMP w/4 CPU threads)
Kernel taint flags: TAINT_CRAP

Locale: LANG=C.UTF-8, LC_CTYPE=C.UTF-8 (charmap=UTF-8), LANGUAGE not set
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)

Versions of packages gfortran depends on:
ii  cpp                         4:14.2.0-1
ii  cpp-aarch64-linux-gnu       4:14.2.0-1
ii  gcc                         4:14.2.0-1
ii  gfortran-14                 14.2.0-19
ii  gfortran-aarch64-linux-gnu  4:14.2.0-1

gfortran recommends no packages.

Versions of packages gfortran suggests:
pn  gfortran-doc  <none>
----------------------------------------------------------------------

Cheers
Boud

hi again,

There's a workaround. On the same system, compiling with


gfortran  -g -O1 -c -o x.o x.f90


gives no error at all on many repeated tries.


For the file

https://codeberg.org/boud/cosmdist/src/branch/main/cosm_f90.f90

on which I discovered the error,

gfortran  -g -O0 -c -o x.o x.f90

is needed to compile with no error on the same system; the
error still often occurs with '-O1'.

Cheers
Boud