arch/powerpc/math-emu/mtfsf.c - incorrect mask?

[Gabriel Paubert]

	Hi James,

On Mon, Feb 10, 2014 at 11:03:07AM -0600, James Yang wrote:

[snipped]
> > Ok, if you have measured that method1 is faster than method2, let us go for it.
> > I believe method2 would be faster if you had a large out-of-order execution
> > window, because more parallelism can be extracted from it, but this is probably
> > only true for high end cores, which do not need FPU emulation in the first place.
> 
> Yeah, 8548 can issue 2 SFX instructions per cycle which is what the 
> compiler generated.   
> 

Then it is method1.

>  
> > The other place where we can optimize is the generation of FEX. Here is 
> > my current patch:
> > 
> > 
> > diff --git a/arch/powerpc/math-emu/mtfsf.c b/arch/powerpc/math-emu/mtfsf.c
> > index dbce92e..b57b3fa8 100644
> > --- a/arch/powerpc/math-emu/mtfsf.c
> > +++ b/arch/powerpc/math-emu/mtfsf.c
> > @@ -11,48 +11,35 @@ mtfsf(unsigned int FM, u32 *frB)
> >  	u32 mask;
> >  	u32 fpscr;
> >  
> > -	if (FM == 0)
> > +	if (likely(FM == 0xff))
> > +		mask = 0xffffffff;
> > +	else if (unlikely(FM == 0))
> >  		return 0;
> > -
> > -	if (FM == 0xff)
> > -		mask = 0x9fffffff;
> >  	else {
> > -		mask = 0;
> > -		if (FM & (1 << 0))
> > -			mask |= 0x90000000;
> > -		if (FM & (1 << 1))
> > -			mask |= 0x0f000000;
> > -		if (FM & (1 << 2))
> > -			mask |= 0x00f00000;
> > -		if (FM & (1 << 3))
> > -			mask |= 0x000f0000;
> > -		if (FM & (1 << 4))
> > -			mask |= 0x0000f000;
> > -		if (FM & (1 << 5))
> > -			mask |= 0x00000f00;
> > -		if (FM & (1 << 6))
> > -			mask |= 0x000000f0;
> > -		if (FM & (1 << 7))
> > -			mask |= 0x0000000f;
> > +		mask = (FM & 1);
> > +		mask |= (FM << 3) & 0x10;
> > +		mask |= (FM << 6) & 0x100;
> > +		mask |= (FM << 9) & 0x1000;
> > +		mask |= (FM << 12) & 0x10000;
> > +		mask |= (FM << 15) & 0x100000;
> > +		mask |= (FM << 18) & 0x1000000;
> > +		mask |= (FM << 21) & 0x10000000;
> > +		mask *= 15;
> 
> 
> Needs to also mask out bits 1 and 2, they aren't to be set from frB.
> 
> 		mask &= 0x9FFFFFFF;
> 
> 

Look at the following lines:

> 
> 
> >  	}
> >  
> > -	__FPU_FPSCR &= ~(mask);
> > -	__FPU_FPSCR |= (frB[1] & mask);
> > +	fpscr = ((__FPU_FPSCR & ~mask) | (frB[1] & mask)) &
> > +		~(FPSCR_VX | FPSCR_FEX);
 
It's here (masking FPSCR_VX and FPSCR_FEX).

Actually the previous code was redundant, it cleared FEX and VX in the
mask computation and later again when recomputing them. Clearing them
once should be enough.

> >  
> > -	__FPU_FPSCR &= ~(FPSCR_VX);
> > -	if (__FPU_FPSCR & (FPSCR_VXSNAN | FPSCR_VXISI | FPSCR_VXIDI |
> > +	if (fpscr & (FPSCR_VXSNAN | FPSCR_VXISI | FPSCR_VXIDI |
> >  		     FPSCR_VXZDZ | FPSCR_VXIMZ | FPSCR_VXVC |
> >  		     FPSCR_VXSOFT | FPSCR_VXSQRT | FPSCR_VXCVI))
> > -		__FPU_FPSCR |= FPSCR_VX;
> > -
> > -	fpscr = __FPU_FPSCR;
> > -	fpscr &= ~(FPSCR_FEX);
> > -	if (((fpscr & FPSCR_VX) && (fpscr & FPSCR_VE)) ||
> > -	    ((fpscr & FPSCR_OX) && (fpscr & FPSCR_OE)) ||
> > -	    ((fpscr & FPSCR_UX) && (fpscr & FPSCR_UE)) ||
> > -	    ((fpscr & FPSCR_ZX) && (fpscr & FPSCR_ZE)) ||
> > -	    ((fpscr & FPSCR_XX) && (fpscr & FPSCR_XE)))
> > -		fpscr |= FPSCR_FEX;
> > +		fpscr |= FPSCR_VX;
> > +
> > +	/* The bit order of exception enables and exception status
> > +	 * is the same. Simply shift and mask to check for enabled
> > +	 * exceptions.
> > +	 */
> > +	if (fpscr & (fpscr >> 22) &  0xf8) fpscr |= FPSCR_FEX;
> >  	__FPU_FPSCR = fpscr;
> >  
> >  #ifdef DEBUG
> >  mtfsf.c |   57 ++++++++++++++++++++++-----------------------------------
> >  1 file changed, 22 insertions(+), 35 deletions(-)
> > 
> > 
> > Notes: 
> > 
> > 1) I'm really unsure on whether 0xff is frequent or not. So the likely()
> > statement at the beginning may be wrong. Actually, if it is not very likely,
> > it might be better to remove the special casef for FM==0xff. A look at 
> > GCC sources shows that it never generates a mask of 0xff. From glibc
> > sources, there vast majority of cases uses 0x1, only isnan() uses 0xff.
> 
> Can't handle all cases here.  

That's why I would go for the simplest possible code. Conditionals are
expensive and minimizing cache footprint is often the best measure of 
performance for infrequently used code. With this in mind, I would get 
rid of all the tests for special FM values and rely on the optimized
generic case.


>  
> > 2) it may be better to remove the check for FM==0, after all, the instruction
> > effectively becomes a nop, and generating the instruction in the first place
> > would be too stupid for words.
> 
> Hmm a heavy no-op.  I wonder if it is heavier than a sync.

In theory not. It contains the equivalent of several isync (taking an exception
and returning from it), but not any synchronization wrt the memory accesses.

	Gabriel