Re: PR63633: May middle-end come up width hard regs for insn expanders?

[Georg-Johann Lay <avr at gjlay dot de>]

Am 10/27/2014 08:43 PM, schrieb Jeff Law:
> On 10/27/14 13:33, Georg-Johann Lay wrote:
> > Am 10/24/2014 08:29 PM, schrieb Jakub Jelinek:
> > > On Fri, Oct 24, 2014 at 08:19:57PM +0200, Georg-Johann Lay wrote:
> > > > Yes, that's the straight forward approach which works so far.  Bit
> > > > tedious,
> > > > but well...
> > > >
> > > > In one case expmed generated different code, though: divmodhi instead of
> > > > mulhi_highpart for HI division by const_int.  Cheating with costs did
> > > > not
> > > > help.  However for now I am mostly after correct, ICE-less code.
> > > >
> > > > What I am concerned about is:
> > > >
> > > > 1) May it happen that a value lives in a hard-reg across the
> > > > expander? The
> > > > expander has no means to detect that situation and interfere, e.g.
> > > >
> > > > hard-reg = source_value // middle-end
> > > > expand-code             // back-end
> > > > sink_value = hard-reg   // middle-end
> > > >
> > > > where "expand-code" is one defind_expand / define_insn that clobbers /
> > > > changes (parts of) hard-reg but does *not* get hard-reg as operand.
> > > > This is
> > > > wrong code obviously.
> > >
> > > It can happen, but if it happens, that would mean user code bug, like
> > > using
> > > register asm with an register that is unsuitable for use as global or
> > > local
> > > register var on the target, or it could be backend bug (expansion of some
> > > pattern clobbering register that has other fixed uses).
> > > You shouldn't ICE on it, but what happens is undefined.
> >
> > Now I have a test case where exact that happens:
> >
> > - An expander needs to clobber a hard register.
> > - That hard reg is live at that time and *not* used as operand to the
> > expander
> > - That hard reg is neither fixed nor a global register.
> > - That hard reg is neither used for argument passing nor for frame pointer
> >    or stack pointer and not for any exotic use like static chain etc.
> So can you give more details about the live hard reg?  How did a value get into
> that hard reg, how did it get used later? That's where I'd focus my efforts
> since that probably shouldn't be happening except in very special circumstances.

The C test case is


__attribute__((noinline,noclone))
unsigned bug_mulhi_highpart (unsigned x)
{
    register unsigned reg26 asm ("26") = 10;
    a = x / 10;
    __asm volatile ("; %0 " : "+r" (reg26));
    return reg26;
}

int main (void)
{
    if (10 != bug_mulhi_highpart (0))
        __builtin_abort();
    return 0;
}


The specification guarantees reg26 to be allocated to R26/27 only at the time 
it is used as operand to the inline asm.

I am fine if the code above is no legal use of local register variables. 
However the question arises how a user writing such code can know that the 
generated code will be wrong?

As you told the compiler has no means to detect the situation as reg26 is not 
used as operand for divmod.  Sometimes the middle-end generates mul_highpart 
instead so that a user cannot tell what registers are to be avoided even if he 
is familiar with GCC internals...

Johann