Issues on Linux Alpha ...

[S.D. Brown]

I sent this same message to the RedHat Alpha Linux mailing list.

--
I work on a computationally intensive software model and my development
machine is a AlphaPC 164/UX running RedHat 6.2 (kernel 2.2.14-6.0).

I spend a lot of time debugging my code and ever since I went to 6.2
and the distributed kernel I have had "issues" with gdb(1) regarding
floating point numbers.  I could list lots of things that don't work,
but the mini session below should demonstrate the problems fine enough.
These same steps work fine on other non-Linux boxes and on a RedHat 5.1
Sparc machine.

I should note that even though my demo doesn't use an executable, my
daily exposure is using real programs are built with gcc (egcs-2.91.66).
Again, gdb(1) seems to demonstrate this same behavior without an 
executable loaded so I don't think the results would change if I used
ccc(1).  These problems happen with RedHat distributed versions of gdb(1)
(4.18) and my recent build (5.0) using egcs-2.91.66.

The quick example below seems simple but it demonstrates the same problem
I have with doing any operations with floating point values when I am
debugging an actual program.

Has anyone or everyone seen this before, and if so what do I do to fix 
it?  Is this a gdb(1) issue or a kernel issue?  If you are running a
later gdb(1)/kernel combination do you have this behavior?

Thanks,
Scott

GNU gdb 5.0
Copyright 2000 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "alphaev56-unknown-linux-gnu".
Setting up the environment for debugging gdb.
.gdbinit:5: Error in sourced command file:
No symbol table is loaded.  Use the "file" command.
(gdb) p 1.0+1
$1 = 0
(gdb) Erroneous arithmetic operation.
(gdb) p 1.0+1.0
$2 = 0
(gdb) p (float)1.0 + (float)1.0
$3 = 0
(gdb) p (float)1.0 + (double)1.0
$4 = 0
(gdb) Erroneous arithmetic operation.
(gdb) p (double)1.0 + (double)1.0
$5 = 0
(gdb) Erroneous arithmetic operation.
(gdb) 

-- 
Scott D. Brown // Scientist // address@hidden 
Digital Imaging and Remote Sensing Lab // Rochester Institute of Technology
54 Lomb Memorial Dr. Rochester, NY 14623 // 716.475.7194 // FX:716.475.5988