[ Now with the testfile. ]
On Tue, 27 Oct 2015 22:28:31 +0100, Jan Kratochvil wrote:

Hello,

this is not a regression for FSF GDB; but at least these Fedora versions
	gdb-7.8.2-39.fc21.x86_64
	gdb-7.9.1-17.fc22.x86_64
	gdb-7.10-29.fc24.x86_64
PASS the attached testcase
	p c40pt(1)^M
	$1 = '0-hello', ' ' <repeats 33 times>^M
	(gdb) PASS: gdb.fortran/dwarf-stride.exp: p c40pt(1)
	p c40pt(2)^M
	$2 = '1-hello', ' ' <repeats 33 times>^M
	(gdb) PASS: gdb.fortran/dwarf-stride.exp: p c40pt(2)
while current FSF trunk b80c3053162ec5533e120ee4e4ed30296d4c5fb2 FAILs on:
	p c40pt(1)^M
	$1 = '0-hello', ' ' <repeats 33 times>^M
	(gdb) PASS: gdb.fortran/dwarf-stride.exp: p c40pt(1)
	p c40pt(2)^M
	$2 = '\001\000\000\000\061-hello', ' ' <repeats 29 times>^M
	(gdb) FAIL: gdb.fortran/dwarf-stride.exp: p c40pt(2)

Curiously the Fedora GDB versions are based on:
	[PATCH 00/23] Fortran dynamic array support
	https://sourceware.org/ml/gdb-patches/2014-06/msg00108.html
	https://github.com/intel-gdb/vla/tree/vla-fortran
	commit 511bff520372ffc10fa2ff569c176bdf1e6e475d
	(although I cannot find that commit hash in that github repo now)

Couldn't the Intel patchset regress this testcase during its upstreaming?

I haven't yet tried to really fix/debug it.


Thanks,
Jan