| Re: [photran] Re: Intel Fortran 9.0.021 compiler and Photran |
Hi Luud,
If you haven't already, try explicitly setting –O0 with –g in order to make sure that all optimizations are turned off.
Regards,
Leo
From: photran-bounces@xxxxxxxxxxx [mailto: photran-bounces@xxxxxxxxxxx] On Behalf Of Luud Heck
Sent: Friday, December 02, 2005 9:17 AM
To: photran@xxxxxxxxxxx
Subject: [photran] Re: Intel Fortran 9.0.021 compiler and Photran
Hi,
I tried removing -save and -zero for the few files that are compiled as debug. The reset of the code is compiled without debug but with -save and -zero (apart from the program main routine, which has debug turned on as well).
I see the following messages from the debugger:
Setting environment variable "TC_STAT" to null value.
Setting environment variable "GCC_SPECS" to null value.
Setting environment variable "TC_MULTIPLE_CACHES" to null value.
Stopped due to shared library event
Stopped due to shared library event
Stopped due to shared library event
Current language: auto; currently fortran
mi_cmd_stack_list_frames: No stack.
Previous frame inner to this frame (corrupt stack?)
Previous frame inner to this frame (corrupt stack?)
Previous frame inner to this frame (corrupt stack?)
< loads of these >
After stepping through the code for a while the debugger looses it and stepping no longer works.
Another observation is that in the Variables display, the variables do not get updated when they change, but clicking one that has changed does display the new value in the bottom part. However, the value in the upper part does not change.
I'm will try to create a simple program to see if that works.
If anyone has any tips, please let me know.
Regards,
Luud
On 12/2/05, Luud Heck <luud.heck@xxxxxxxxx> wrote:
Hi Jeffery,
Thanks for replying. I try to get this to the right list now.
I managed to get a bit further with it. Still not 100% though.
I compiled our software paacke completely with the intel 8.1 compiler suite (that's both the C and the Fortran parts).
I also detected that using -O3 and -g gives problems (although I would have expected that -g would override -O3). Also we need to use -save and -zero to make sure that uninitialized variables are at least zero (I know these are actually coding bugs, but the package is too large and old to change all that now).
I have to see if removing those options improves the debugging capabilities of gdb.
Cheers,
Luud
On 12/2/05, Jeffrey Overbey <overbey2@xxxxxxxx > wrote:
Hi Luud,
Sorry I forgot to respond to this. All of the traffic is on the Photran
list at eclipse.org; I'm not sure what happened. Try subscribing again,
and send a test post if necessary. If you have problems subscribing,
try mailing webmaster@xxxxxxxxxxx.
If you are using the -g switch, I'm not sure why it wouldn't work. I
will forward this to the list in case someone else has any advice.
Otherwise, after things wind down in a couple of weeks, I will download
Intel Fortran 9 and try it myself. Strange...
Jeff
Luud Heck wrote:
> Hi Jeffrey,
>
> I first sent this mail to the photran mailing list, but that one
> doesn;t seem to work well. I did subscribe wehn you opened with the
> migration to eclipse.org < http://eclipse.org>. however, nobody seems
> to post there, and if I try to post there I complains that I have to
> subscribe first. Strange.
>
> So, maybe you can help me out with the issues below or redirect me to
> the correct support mailing list. To be sure I CC-ed the old list.
>
>
> I managed to get the Intel Fortran 9.0.021 compiler to work on my
> 32bit Linux machine and am trying to do some debugging with it from
> Photran.
>
> This does not really work as expected and I'm not sure why.
>
> For one thing, the source line numbers are wrong, so the line
> indicator while debugging is a few lines off. Also the variables do
> net get shown correctly or do not update when their values are changed.
>
> Are there any tips on getting this fixed? Do I need to add specific
> compiler options for debugging information to the ifort compiler, or
> some additional options for gdb?
>
> On a related note, you claim to support the Intel 8.x Fortran
> compilers. So it might be worth trying this one out. Is the following
> file correct for a 32 bit setup?
>
> l_fce_pc_8.1.032.tar.gz
>
> It is a fraction of the size of the l_fc_p_9.0.021.tar.gz file for
> version 9.
>
>
> Thanks for your time.
> Luud
>
>
> P.S.
> These are the compiler options I use now:
>
> ifort -I. -I../include -o example.o -O3 -c -zero -save -g example.f
>
> i also tried with the addtional option "-debug externded"
_______________________________________________
photran mailing list
photran@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/photran