41CL - Warp Core Conflict with Card Reader?

[hth]

It seems the same issue to me.

When you are filling in a prompt you are using the key sequence parser. This sets the calculator in state where the return address of the (next) key sequence handler is kept on the stack while the calculator goes to sleep waiting for the next key press.

During the process of going to sleep, the I/O poll vector is still called. At this point the poll routine is more limited (than usual) on how much it can put on stack (making subroutine calls), as the internal stack is only 4 levels deep and it is partly used in this state.

If some real action is needed, maybe a program card was inserted, the poll routine can decide to abort the key sequence parsing, getting all stack levels back (it terminates the function being entered).

The card reader 1E ROM violates this rule and then weird things happen as the calculator is put in a corrupted state.