Hi Michael,

On Tue, Feb 16, 2021 at 08:36:02PM +1100, Michael Ellerman wrote:
> Feng Tang <feng.tang@intel.com> writes:
> > Hi Christophe and Michael,
> >
> > On Mon, Jan 18, 2021 at 10:24:08PM +0800, Christophe Leroy wrote:
> >> 
> >> Le 05/01/2021 ? 11:58, kernel test robot a 閏rit :
> >> > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> >> > head:   e71ba9452f0b5b2e8dc8aa5445198cd9214a6a62
> >> > commit: 8b8319b181fd9d6821703fef1228b4dcde613a16 powerpc/44x: Don't support 440 when CONFIG_PPC_47x is set
> >> 
> >> I see no link with that commit. Looks like the problem has been existing for some time.
> >> It exists on the commit before that one, it exists on v5.9 and it exists on v5.10 with that commit
> >> reverted.
> >  
> > Yes, this seems to be a long-standing issue, and we just double checked
> > this compile error. 
> >
> > It happend when compiling arch/powerpc/platforms/44x/fsp2.c, macro
> > 'mfdcr' requirs an instant number as parameter, while is not met by
> > show_plbopb_regs(). Changing show_plbopb_regs() from function to
> > a macro fixes the error, as the patch below:
> >
> > Thanks,
> > Feng
> >
> >
> > From 3bcb9638afc873d0e803aea1aad4f77bf1c2f6f6 Mon Sep 17 00:00:00 2001
> > From: Feng Tang <feng.tang@intel.com>
> > Date: Fri, 5 Feb 2021 16:08:43 +0800
> > Subject: [PATCH] powerpc/44x/fsp2: fix a compiling error regarding macro
> >  'mdfcr'
> >
> > 0day's kbuild test found error:
> >
> > "
> >   CC      arch/powerpc/platforms/44x/fsp2.o
> >
> >   {standard input}:577: Error: unsupported relocation against base
> >   {standard input}:580: Error: unsupported relocation against base
> >   {standard input}:583: Error: unsupported relocation against base
> > "
> >
> > The reason is macro 'mfdcr' requirs an instant number as parameter,
> > which is not met by show_plbopb_regs().
> 
> It doesn't require a constant, it checks if the argument is constant:

Aha, seems my grep found the wrong target: arch/powerpc/boot/dcr.h,
which has  

#define mfdcr(rn) \
	({	\
		unsigned long rval; \
		asm volatile("mfdcr %0,%1" : "=r"(rval) : "i"(rn)); \
		rval; \
	})

> #define mfdcr(rn)						\
> 	({unsigned int rval;					\
> 	if (__builtin_constant_p(rn) && rn < 1024)		\
> 		asm volatile("mfdcr %0," __stringify(rn)	\
> 		              : "=r" (rval));			\
> 	else if (likely(cpu_has_feature(CPU_FTR_INDEXED_DCR)))	\
> 		rval = mfdcrx(rn);				\
> 	else							\
> 		rval = __mfdcr(rn);				\
> 	rval;})
> 
> But the error you're seeing implies the compiler is choosing the first
> leg of the if, even when rn == "base + x", which is surprising.

Yes, it might be related to compiler (though myself isn't faimiliar
with it). As show_plbopb_regs() was introduced by commit 7813043e1bbc
("powerpc/44x/fsp2: Add irq error handlers") back in 2017, while it
was just reported.

> We've had cases in the past of __builtin_constant_p() returning false
> for things that a human can see are constant at build time, but I've
> never seen the reverse.
> 
> cheers


Thanks,
Feng