On Thu, Dec 24, 2015 at 12:33:31AM +0100, Gianluca Renzi wrote: > In this case I suppose even the userspace code (dynamic linking library) > could not to be used > with this linker bug and not only the kernel module loader tool. The > undefined symbols are discarded in every way. The only way to ensure a > correct code buildup is to use a -static linking of any external library in > every piece of code. So no dynamic linking is usable in that way. This bug > seems so unreal to me... > Are the gcc team people so silly not to try any created code (even a simple > checking for an external library) in every supported platform for binutils > and gcc??? > I suppose not. Are you sure of the bug? > Can you ask the gnu gcc people to test the ld in a kvm with ppc64 machine?
This is a binutils change, NOT a gcc change. The only package I changed on the system to get the different result was binutils. -- Len Sorensen