Compiling in 64 bits

Compiling in 64 bits

[Walther]

De: Walther [mailto:WaltherC@att.net] 


Hi 
I am receiving an error compiling in a 64 AMD linux machine
Did you have a problem before?
The problem is with a casting of a pointer to integer (I think instead of a
long in 64 bits)
Instead of start changing the software I would like to know if you have
compiled it in 64 bits before..
Do you have the source code for the 64 bits edition?
Thanks

Walther Contreras
Walther@wmiats.com

Re: Compiling in 64 bits

[David Haworth]

Hi Walther,

On Thu, Mar 06, 2008 at 09:24:11AM -0600, Walther wrote:
> I am receiving an error compiling in a 64 AMD linux machine
> Did you have a problem before?
> The problem is with a casting of a pointer to integer (I think instead of a
> long in 64 bits)
> Instead of start changing the software I would like to know if you have
> compiled it in 64 bits before..
> Do you have the source code for the 64 bits edition?

I had the same problem a while ago (compiling on amd64 for an ARM
target). The problem seems to be with how some of the GUI stuff passes
parameters around using the wrong types. I tried to fix the problem
by changing a few local types and typecasting. I managed to get
insight to compile but it still doesn't run properly, so I fear there
might be other problems in the GUI or even in the tcl subsystem.

I posted to the list but didn't get a reply. At the time the list was
very quiet - it seems a lot more active now, so maybe you'll have
better luck.

FWIW the underlying gdb appears to work fine (at least, for the
qemu ARM target I was playing with). It's just the GUI.
I haven't had much time to play recently, but I planned to work around the
problem by copying the build from a 32-bit machine, where insight
builds and runs without problems.

Sorry I can't be more help.

Dave

Re: Compiling in 64 bits

[Keith Seitz]

David Haworth wrote:

> I posted to the list but didn't get a reply. At the time the list was
> very quiet - it seems a lot more active now, so maybe you'll have
> better luck.

That is, of course, my fault (and I apologize): I do tend to make random 
disappearances depending on what is happening with my work load. But I 
now have access to an x86_64 box of my (wife's) very own, so I'll try to 
take a look at this tonight.

Keith

Re: Compiling in 64 bits

[David Haworth]

Hi Keith,

On Thu, Mar 06, 2008 at 10:53:10AM -0800, Keith Seitz wrote:
> That is, of course, my fault (and I apologize): I do tend to make random 
> disappearances depending on what is happening with my work load. But I now 
> have access to an x86_64 box of my (wife's) very own, so I'll try to take a 
> look at this tonight.

What I said wasn't intended as criticism, so I'm sorry if it sounded like
that. I think those of us who run 64-bit Linux are aware that it's
still bleeding-edge. Yes, the mainstream stuff works OK, but stray a little
from the beaten track and you start finding programs that won't run, or that
appreas to run perfectly well but refuse to open perfectly good files,
claiming that they are not in the correct format. The usual culprit for
the former is an assumption that a pointer is 32-bits, for the latter that
a long is 32-bits.
I'm also aware that not everyone has access to an amd64 machine, and even
those who do may find that it has a 32-bit linux installed.

I did try (briefly) to fix the problems. I got rid of all the
compiler errors by careful type-changes and casts, but the GUI part
simply won't run. I fear there may be deeper problems in the tcl/tk
package. I recall you mentioning recently that insight has an old verison
of tcl/tk built-in and that you intended to try to get rid of it
and use a normal version. I have a feeling that supporting amd64
might force the issue :-O

Dave

Re: Compiling in 64 bits

[Keith Seitz]

David Haworth wrote:
> What I said wasn't intended as criticism, so I'm sorry if it sounded like
> that.

I never thought that, so your apology is NOT accepted! O:-)

> I recall you mentioning recently that insight has an old verison
> of tcl/tk built-in and that you intended to try to get rid of it
> and use a normal version. I have a feeling that supporting amd64
> might force the issue :-O

Yeah, we do lug around a lot of baggage. I've been playing with ripping 
insight out of gdb (as a separate toplevel project), but itcl/itk are 
giving me severe problems with configuration... I'm not even sure if 
itcl, itk, and iwidgets are supported any more!

I'll see what I can do tonight.

Keith

Re: Compiling in 64 bits

[Keith Seitz]

Keith Seitz wrote:
> I'll see what I can do tonight.

Wow, not much, I'm afraid. All I was able to accomplish tonight was 
eliminating the pointer conversion warnings from src/gdb/gdbtk. Alas, 
when I tested insight on itself, all #&@! broke loose. I believe that 
tcl/tk/itcl/libgui is our problem. All of those modules showed a ton of 
these warnings.

I've been playing around with ripping tcl et al out of our build, so 
that the user is responsible for downloading the "proper" versions, but 
I didn't get too far. I'll pick up on this again this weekend or next week.

Tcl/Tk 8.5 looks a lot better in terms of the warnings situation, but I 
have a real fear of libgui's tkTable widget. But I haven't tried the 2.9 
build yet...

Keith