Re: [avr-libc-dev] -mint8 problem with setjmp/longjmp

[Nils Kristian Strom]

Joerg Wunsch wrote:

> I don't think the fix is quite right.  Syntactically, you cannot
> simply replace an integer by a pointer type.

setjmp.h is *broken*.  The integer type should never have been there in 
the first place.  A pointer is not - and never will be, an integer.

My fix corrects this obvious flaw in the header, replacing addresses 
(pointers) with pointers.

How can you argue with that ?


With regards to the inttypes.h file, I do not use this as it is 
-mint8-broken.  As you say, this needs patching for -mint8 compatibility.

This is a different case, as this file is *correct*, but not compatible 
with -mint8.

Point:
  1. setjmp.h is *broken*, uses unknown-sized integers as pointers.
  2. inttypes.h is *correct*, but not compatible with -mint8.




The -mint8 option gave me 6% code size reduction (source is about 20K 
lines of code).  The *only* problem arising from -mint8 was the broken 
'setjmp.h'.


Regards,
  Nils