Stack Overflow Question

With some of you guys with experience in embedded systems, I wanted to ask a question regarding overflowing the stack.

Typically, you define what areas of memory are used for in the linker directive. Now lets say you split up some memory for multiple uses. We currently have it so that if we overflow the stack, we run into memory regions we own already, however we are still getting a watchdog reset. The region that it happens to flow into isn't used for about 5KB so it has some space available.

We have defined however in the linker directive how much space should be used for the stack. Are there internal protection operations going on within the controller that could cause us to get a reset even if we are still using space allocated to us? I ask this because we don't actually have access to the low level code of the microcontroller. We load executable code in other regions, and we have a less than cooperative vendor.

FYI, we are using a MPC5566 if anyone happens to be familiar. I wouldn't even mind just having an explanation of what results are typical in a situation like this since I have no background in this field.

Thanks.

By looking at the reference manual for the MPC5566, I see it has an MMU and thus in theory could implement stack overflow protection if the boundaries of things are page-aligned in a certain way. Basically it's up to the "OS" which, sounds like is controlled by your vendor.

> The region that it happens to flow into isn't used for about 5KB so it has some space available.
So is there any chance of filling this with zeros, and then seeing if it's been overwritten when you've caught the WDR.

Do you have JTAG debuggers for this board?

Is your code massively recursive, have large local arrays, or variable length arrays? If it has none of these things, then how would it cause stack overflow?

Is there an RTOS on the board, or is it just your code on the bare metal?

Quote:

---

So is there any chance of filling this with zeros, and then seeing if it's been overwritten when you've caught the WDR.

---

We have no doubts we are overflowing the stack, we just want to know the cause of the reset.

Quote:

---

Do you have JTAG debuggers for this board?

---

We do have JTAG debuggers, however we may lack the expertise to use them. I'm sure I could read a manual and figure it out. The issue is mostly that we are having problems regarding things outside the scope of our skillset.

Quote:

---

Is your code massively recursive, have large local arrays, or variable length arrays? If it has none of these things, then how would it cause stack overflow?

---

We have identified in the software using stack analysis tools that we are allocating too much stack space. We are fine with this since we have discovered this, but we are really concerned with what is causing the watchdog reset.

Quote:

---

Is there an RTOS on the board, or is it just your code on the bare metal?

---

I'm fairly certain we have an RTOS, but am not entirely certain.