Re: GCC 4.5.3 breaks sandpoint's altboot

Frank Wille <[email protected]> Sun, 23 Oct 2011 14:12:26 +0200
Newsgroups gmane.os.netbsd.ports.powerpc,gmane.os.netbsd.devel.toolchain,gmane.os.netbsd.ports.sandpoint
Organization Privates Internet Ostwestfalen/Lippe e.V.
Message-ID <[email protected]>
matthew green wrote:

>> Any idea where to look? What has changed for 4.5.3?
>> src/sys/lib/libkern/arch/powerpc/gprsavrest.S, which also caused
>> problems with altboot, was already fixed some weeks ago, and seems ok
>> now.
>
> first thing i would look at is asm() calls.  newer GCC traditionally
> has enforced stricter asm calls, in particular, marking which values
> and or registers are input/output values.

Ok, thanks. I didn't know that. Although in the altboot source itself there
is no asm().


> one thing you can do is check which file breaks, by compiling each
> file used with each compiler and seeing when it breaks,

Yes, I already had this idea yesterday. Will do that now.


> or also by using -O0 vs -O2.

I doubt that the codegen has a serious bug, otherwise we would see more
problems, also in kernels and userland.


-- 
Frank Wille