Re: [PATCH -V7 18/18] powerpc: Update tlbie/tlbiel as per ISA doc

[David Gibson <dwg@au1.ibm.com>]

[-- Attachment #1: Type: text/plain, Size: 3088 bytes --]

On Tue, Apr 30, 2013 at 10:51:00PM +0530, Aneesh Kumar K.V wrote:
> David Gibson <dwg@au1.ibm.com> writes:
> 
> > On Mon, Apr 29, 2013 at 01:07:39AM +0530, Aneesh Kumar K.V wrote:
> >> From: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
> >> 
> >> Encode the actual page correctly in tlbie/tlbiel. This make sure we handle
> >> multiple page size segment correctly.
> >
> > As mentioned in previous comments, this commit message needs to give
> > much more detail about what precisely the existing implementation is
> > doing wrong.
> >
> >> 
> >> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
> >> ---
> >>  arch/powerpc/mm/hash_native_64.c | 32 ++++++++++++++++++++++++++++++--
> >>  1 file changed, 30 insertions(+), 2 deletions(-)
> >> 
> >> diff --git a/arch/powerpc/mm/hash_native_64.c b/arch/powerpc/mm/hash_native_64.c
> >> index bb920ee..6a2aead 100644
> >> --- a/arch/powerpc/mm/hash_native_64.c
> >> +++ b/arch/powerpc/mm/hash_native_64.c
> >> @@ -61,7 +61,10 @@ static inline void __tlbie(unsigned long vpn, int psize, int apsize, int ssize)
> >>  
> >>  	switch (psize) {
> >>  	case MMU_PAGE_4K:
> >> +		/* clear out bits after (52) [0....52.....63] */
> >> +		va &= ~((1ul << (64 - 52)) - 1);
> >>  		va |= ssize << 8;
> >> +		va |= mmu_psize_defs[apsize].sllp << 6;
> >>  		asm volatile(ASM_FTR_IFCLR("tlbie %0,0", PPC_TLBIE(%1,%0), %2)
> >>  			     : : "r" (va), "r"(0), "i" (CPU_FTR_ARCH_206)
> >>  			     : "memory");
> >> @@ -69,9 +72,20 @@ static inline void __tlbie(unsigned long vpn, int psize, int apsize, int ssize)
> >>  	default:
> >>  		/* We need 14 to 14 + i bits of va */
> >>  		penc = mmu_psize_defs[psize].penc[apsize];
> >> -		va &= ~((1ul << mmu_psize_defs[psize].shift) - 1);
> >> +		va &= ~((1ul << mmu_psize_defs[apsize].shift) - 1);
> >>  		va |= penc << 12;
> >>  		va |= ssize << 8;
> >> +		/* Add AVAL part */
> >> +		if (psize != apsize) {
> >> +			/*
> >> +			 * MPSS, 64K base page size and 16MB parge page size
> >> +			 * We don't need all the bits, but rest of the bits
> >> +			 * must be ignored by the processor.
> >> +			 * vpn cover upto 65 bits of va. (0...65) and we need
> >> +			 * 58..64 bits of va.
> >
> > I can't understand what this comment is saying.  Why do we need to do
> > something different in the psize != apsize case?
> >
> >> +			 */
> >> +			va |= (vpn & 0xfe);
> >> +		}
> 
> That is as per ISA doc. It says if base page size == actual page size,
> (RB)56:62 must be zeros, which must be ignored by the processor.
> Otherwise it should be filled with the selected bits of VA as explained above.

What you've just said here makes much more sense than what's written
in the comment in the code.

> We only support MPSS with base page size = 64K and actual page size = 16MB.

Is that actually relevant to this code though?

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]