Re: [avr-gcc-list] Avr-gcc Produces Incorrect Code with -Os

[hutchinsonandy]

Eric,

The problem with the function is typical housekeeping overhead gets 
wrapped up.

To take a silly example:

Say you want to send some  message to ISR (via memory) using critical 
function

critical void function(int *buffer, int *cValue)
{
*cValue++ = *buffer++;
}

Getting *buffer++ does not need ISR disabled. Perhaps neither does the 
cValue++;

So this method, although not wrong,  is something I would normally 
reject for  time critical systems, where latency is key.
Inline coding has less overhead but even then its not perfect - though 
perhaps more palatable than assembler.

I do not know much about the gcc atomic operations. But it might be 
worth reviewing. There may be some hooks that make
it work well with optimisations without adding a bunch of code hacks. 
Also it makes code more portable (Atmel ARM or course!)

Andy



-----Original Message-----
From: Weddington, Eric <address@hidden>
To: address@hidden; address@hidden; address@hidden; 
address@hidden
Sent: Thu, 22 May 2008 2:33 pm
Subject: RE: [avr-gcc-list] Avr-gcc Produces Incorrect Code with -Os





> -----Original Message-----
> From: address@hidden [mailto:address@hidden
> Sent: Thursday, May 22, 2008 11:48 AM
> To: address@hidden; Weddington, Eric; address@hidden;
> address@hidden
> Subject: Re: [avr-gcc-list] Avr-gcc Produces Incorrect Code with -Os
>
> You can do it this way but you won't have precise control over
> disable/restore timing. So you will typically disable
> interrupts longer
> than is needed.

If the disable can happen right after prologue and the enable right
before the epilogue then I don't see an issue. Caveat emptor. Maybe I'm
missing something?

> Also disabling all interrupts is often not what you need. Disabling
> specfic interrupts is more palatable on real time systems.

True, but it can't be easily done in the compiler. Disabling all
interrupts is a common enough approach, and while it may not ideal for
every application, I think that having this function attribute would be
a useful tool available to the programmer.


> NMI is a feature not a limitation - it supposed to  stop such
>  software
> hanging up the CPU.
>
> BTW gcc provides atomic operations on other ports.

What kinds of different mechanisms do they provide? Would any of them be
useful for the AVR port?


_______________________________________________
AVR-GCC-list mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/avr-gcc-list