Re: [discuss] booting a kernel compiled with -mregparm=0

[Jan Hubicka <jh@suse.cz>]

> Hello,
> 
> I am trying to boot a 2.6.x kernel (on x86_64) compiled with -mregparm=0 
> and it does not boot, i.e. hangs at the very first stage.
> 
> I know this breaks ABI/x86_64 but the reason I need to compile such a 
> kernel is because kdb on x86_64 cannot show the function arguments and the 
> only way to make it work that I found was to pass all arguments on the 
> stack (then kdb works fine and shows correct values for all arguments). 
> But obviously the module and the kernel need to match, otherwise it will 
> panic easily; and so I have to use the same argument passing convention in 
> the kernel. This is obviously for debugging only, nobody would ever ship 
> such "incorrectly" compiled module anywhere.
> 
> So, I have to find a "boundary" between the parts of the kernel that can 
> be safely compiled with -mregparm=0 and which must stay as they are. Any 
> ideas as to what to do in this situation?

Actually -mregparm=0 is not supposed to be even accepted by x86-64
compiler (I've disabled the function attribute but apparently missed
this one) and even if GCC produced valid code by miracle, you will get
into trouble with hand written assembly.

Honza
> 
> Kind regards
> Tigran
>