Re: [PATCH 0/5] Rewrite Motorola MMU page-table layout

[Will Deacon]

On Fri, Jan 31, 2020 at 12:14:59PM +0100, Peter Zijlstra wrote:
> On Fri, Jan 31, 2020 at 11:22:39AM +0100, Peter Zijlstra wrote:
> > On Fri, Jan 31, 2020 at 09:38:13AM +0000, Will Deacon wrote:
> > 
> > > > This series breaks compilation for the ColdFire (with MMU) variant of
> > > > the m68k family:
> > 
> > That's like the same I had reported by the build robots for sun3, which
> > I fixed by frobbing pgtable_t. That said, this is probably a more
> > consistent change.
> > 
> > One note below:
> > 
> > 
> > > -static inline struct page *pte_alloc_one(struct mm_struct *mm)
> > > +static inline pgtable_t pte_alloc_one(struct mm_struct *mm)
> > >  {
> > >  	struct page *page = alloc_pages(GFP_DMA, 0);
> > >  	pte_t *pte;
> > > @@ -54,20 +55,19 @@ static inline struct page *pte_alloc_one(struct mm_struct *mm)
> > >  		return NULL;
> > >  	}
> > >  
> > > -	pte = kmap(page);
> > > -	if (pte) {
> > > -		clear_page(pte);
> > > -		__flush_page_to_ram(pte);
> > > -		flush_tlb_kernel_page(pte);
> > > -		nocache_page(pte);
> > > -	}
> > > -	kunmap(page);
> > > +	pte = page_address(page);
> > > +	clear_page(pte);
> > > +	__flush_page_to_ram(pte);
> > > +	flush_tlb_kernel_page(pte);
> > > +	nocache_page(pte);
> > 
> > See how it does the nocache dance ^
> 
> > So either, alloc_one() shouldn't either, or it's all buggered.
> 
> Damn, we weren't going to touch coldfire! :-))
> 
> So now I found the coldfire docs, and it looks like this thing is a
> software tlb-miss arch, so there is no reason what so ever for this to
> be nocache. I'll 'fix' that.

Does that mean we can drop the GFP_DMA too? If so, this all ends up
looking very similar to the sun3 code wrt alloc/free and they could
probably use the same implementation (since the generic code doesn't
like out pgtable_t definition).

Will