#70-D and #770-D warnings in working code

Hi Luis,

You're correct of course, that's something I should have picked up on. Anyway I tried changing the memory model but then other parts of my code were inoperable.

I can't use #pragmas to define them in an area because I have an application space and bootloader space and the ISR area is in the bootloader space that I don't want to change when a new application is flashed therefore I would have to define the ISR for each and every interrupt even if it wasn't in use to allow compatibility.

I also cant use 32-bit addresses because of the way that I handle interrupts and so I have simply used:

#pragma diag_suppress 70, 770

...Code here...

#pragma diag_warning 70, 770

I know this is just a workaround but it works the way it is.

I have had a quick look in the past at relocating the vectors into RAM but I shall look into it in more detail upon your recommendation.

Thanks a lot,
Callum

There’s another way that solves multiple problems.

Make each table entry

 0x0080 | ((((uint32_t)Isr)>>8)&0x0F00),  ((uint32_t)Isr)&0xFFFF,

 Your real ISRs can now be anywhere in full 20 bit address range (the proxy table of course must still be in lower 64k).
It should work with large and small code model. And should issue no warnings.

Basically, the 0x4030 in the original code was a BRA #dst (or MOV #dst,PC) instruction, followed by the 16 bit destination address.
The line above instead uses the BRA (MOVA) instruction, which takes a 20 bit immediate value for the destination. The upper 4 bits, however, are fused into the instruction itself.

Hi Jens, thank you for your reply.

After trying what you suggested so I end up with:

#pragma DATA_SECTION(proxyVectorTable, ".APP_PROXY_VECTORS")
#pragma RETAIN(proxyVectorTable)
const uint16_t proxyVectorTable[] =
{
    (0x0080 | ((((uint32_t)Isr)>>8)&0x0F00),  ((uint32_t)Isr)&0xFFFF),
    (0x0080 | ((((uint32_t)Isr)>>8)&0x0F00),  ((uint32_t)Isr)&0xFFFF),
    (0x0080 | ((((uint32_t)Isr)>>8)&0x0F00),  ((uint32_t)Isr)&0xFFFF)
};

I now get two errors for each entry:
1) #28 expression must have a constant value
2) #58 this operator is not allowed in a constant expression

Thanks,
Callum

I see. The compiler doesn’t know the address of the ISR at compile time, and the linker can’t do the math at link time.
Well, if you manipulate the linker script so your ISRs end up in a known memory area (0x1xxxx or 0x2xxxx), you can replace the formula with just these (then known) upper 4 bits of the ISR address:

(0x0080 | 0x01), (uint16_t)(uint32_t)Isr,