Re: [V4 00/21] Fortran dynamic array support

[Bernhard Heckel <bernhard.heckel@intel.com>]

On 04/09/2016 19:14, Jan Kratochvil wrote:
> On Fri, 02 Sep 2016 15:44:15 +0200, Bernhard Heckel wrote:
>> dynamic properties in GDB is done via CORE_ADDR which is unsigned.
>> This causes the issue that the type length seems to be way to big.
> Maybe that explains also a regression visible only on i686 (32-bit host).
> It still PASSes on x86_64 (64-bit host) with either native od 32-bit inferior
> (-m32):
>
> The length is bogus (it is %rax or %eax) but it should be some number.
>
> i686:
> ptype reg_string^M
> type = char [variable length]^M
>               ^^^^^^^^^^^^^^^
> (gdb) FAIL: gdb.dwarf2/dw2-bound-loclist.exp: ptype reg_string
> p reg_string^M
> Insufficient memory in host GDB for object of size 4160138653 bytes, maximum allowed 536870911 bytes.^M
>
> x86_64 or x86_64 -m32:
> ptype reg_string^M
> type = char [4160048541]^M
> (gdb) PASS: gdb.dwarf2/dw2-bound-loclist.exp: ptype reg_string
>
> Attaching the Fedora testcase.
>
>
> Thanks,
> Jan
Hi Jan,

I just recreated my two user branches.
- Fixed should be "Werror indentation" on the vla-string branch
- Stride issues you reported in general (stride branch)
- Stride issues on 32bit appl. running on 64bit system (gdb) (stride branch)
- Fixed also neg. bounds on 32bit appl. running on 64bit system (stride 
branch)

Let me know your results for the two branches.


Intel Deutschland GmbH
Registered Address: Am Campeon 10-12, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de
Managing Directors: Christin Eisenschmid, Christian Lamprechter
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928