Re: [PATCH V9 1/4] perf/core: Add PERF_SAMPLE_DATA_PAGE_SIZE

[Peter Zijlstra]

On Wed, Nov 11, 2020 at 10:33:44PM +0000, Matthew Wilcox wrote:
> On Wed, Nov 11, 2020 at 09:00:00PM +0100, Peter Zijlstra wrote:
> > On Wed, Nov 11, 2020 at 06:26:20PM +0000, Matthew Wilcox wrote:
> > > On Wed, Nov 11, 2020 at 06:22:53PM +0100, Peter Zijlstra wrote:
> > > > On Wed, Nov 11, 2020 at 04:38:48PM +0000, Matthew Wilcox wrote:
> > > > > 	if (pud_leaf(pud))
> > > > > 		return PUD_SIZE;
> > > > 
> > > > But that doesn't handle non-pagetable aligned hugetlb sizes. Granted,
> > > > that's unlikely at the PUD level, but why be inconsistent..
> > > > 
> > > > So we really want:
> > > > 
> > > > 	if (p*d_leaf(p*d)) {
> > > > 		if (!'special') {
> > > > 			page = p*d_page(p*d);
> > > > 			if (PageHuge(page))
> > > > 				return page_size(compound_head(page));
> > > > 		}
> > > > 		return P*D_SIZE;
> > > > 	}
> > > 
> > > Still doesn't work because pages can be mapped at funny offsets.
> > 
> > Wait, what?! Is there hardware that has unaligned TLB page-sizes?
> 
> No, you can force a 2MB page to be mapped at an address which isn't
> 2MB aligned.

But not a HugeTLB page; AFAICT mmap() will reject if you try and mmap a
hugetlb page out of alignment. So what I wrote above is still valid. If
PageHuge() we can be certain it is aligned properly and using a matching
hardware page size.

You just don't like it because you want me to be purely page-table
based.