MIPS to ICC endian problem

Hi all. I have a program that I have been compiling using MIPS compiler. I have recently been tasked to port this to our new Intel system using icc.

It compiles fine, but when I run it, I get "Segmentation fault" with no other info.

Someone suggested it may be an endian problem. I turned on warnings and it seems like every one of my macro definitions has a problem.

Here is an example of the first one I have come to:

Code:

---

#define DEBUG_ENTER(name) {tdepth++; 
 if(tnames[tdepth] == NULL) tnames[tdepth] = memalign(8, sizeof(char)*MAXLEN);
 strcopy(tnames[tdepth],name);
 FU_DEBUG("Entering \n");}

---

What is actually going wrong here? How do I fix this in my program? It is a really large piece of software with many many macros and such. Is there a compiler "switch" that can correct this problem?

> Someone suggested it may be an endian problem.
Nothing you've posted so far suggests that.

> I turned on warnings and it seems like every one of my macro definitions has a problem.
Nice - and the warnings were what?
Insufficient cheese on mouse trap perhaps?

> What is actually going wrong here?
Who knows - maybe your tdepth is out of control and off in the weeds somewhere.
what about tnames - is that allocated with sufficient space?

> How do I fix this in my program?
Well you could start with
#define DEBUG_ENTER(name)
so that the whole trickery goes away.

The short answer is compile it with debug, then use a debugger to trap the segfault.
Then go looking at what has the bad address, then figure out why it's bad.

I think I figured it out. I thought memalign was something standard (it's in stdlib) but it is only for the IRIX platform. It is made to allocated the space with a certain alignment. I am going to try to switch them to plain alloc statements and see what happens. memalign is not supported in icc..but for some reason doesnt throw an error?