[Sdcc-user] SDCC 2.4.0 vs 2.6.0 Interrupt Vector Table Generation....

[Jon H. <jha...@re...>]

Hello all SDCC users & developers, being that the 8051 is in your area of
expertise, I wanted to ask your thoughts on the following issue I have been
having.... I have poured through the message boards to no avail.  My
question is very unique.
 
I built a boot loader using SDCC 2.4.0 for the Nordic nRF24E1.  I did this
to eliminate the 4K memory barrier, which allows me to swap firmware loads
in/out of memory on the fly.
 
Part of my boot loader updates the interrupt vector table for the firmware
load of interest.  It seemed despite SDCC 2.4.0 claim that it automatically
adjusted the table based on the ISR requirements it seemed to always adjust
to a full size interrupt 0-5 (0x0003->0x002B) table no matter what, with a
total length of 41 bytes.  This was advantageous to me, since all the
firmware loads I would swap in and out of RAM would always have the same
length interrupt vector table and layout irregardless to my ISR usage for
that particular firmware load.  My boot loader vector table update process
could could work reliably with consistent table lengths.
 
Now, since I upgraded to 2.6.0, SDCC does automatically adjust the ISR table
length based on ISR usage for each firmware load, which are not consistent
from firmware load to firmware load because of different ISR usage, and my
boot loader is not happy with this.
 
Do you know of a way with SDCC 2.6.0 to make it behave as 2.4.0 as far as
the interrupt vector table is concerned, but yet still use SDCC 2.6.0 for
its other improvements and bug fixes?
 
Thanks in advance for your help on this.
 
Excellent job on SDCC, it is a remarkable compiler!
 
Thanks.
Jon Hallsten