Re: [PATCH] tlb: hugetlb: Add arm64 contiguous hint awareness

From: Peter Zijlstra <peterz@infradead.org>
To: Steve Capper <steve.capper@arm.com>
Cc: linux-mm@kvack.org, David Hildenbrand <david@redhat.com>,
	Anshuman Khandual <anshuman.khandual@arm.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Will Deacon <will@kernel.org>,
	nd@arm.com
Subject: Re: [PATCH] tlb: hugetlb: Add arm64 contiguous hint awareness
Date: Thu, 24 Mar 2022 16:58:52 +0100	[thread overview]
Message-ID: <20220324155852.GI8939@worktop.programming.kicks-ass.net> (raw)
In-Reply-To: <2dbb5cf0-ce41-da93-9a3d-5666601f71c5@arm.com>

On Thu, Mar 24, 2022 at 02:35:44PM +0000, Steve Capper wrote:
> > It occurs to me that perhaps this can be written like:
> > 
> > 	unsigned long _sz = huge_page_size(h);
> > 	if (_sz >= P4D_SIZE)
> > 		tlb_flush_p4d_range(tlb, address, _sz);
> > 	else if (_sz >= PUD_SIZE)
> > 		tlb_flush_pud_range(tlb, address, _sz);
> > 	else if (_sz >= PMD_SIZE)
> > 		tlb_flush_pmd_range(tlb, address, _sz);
> > 	else
> > 		tlb_flush_pte_range(tlb, address, _sz);
> > 	__tlb_remove_tlb_entry(tlb, ptep, address);
> > 
> > And then it can still be generic..
> 
> Thanks Peter,
> My concern with that would be the CONT_PMD_SIZE case would result in a call
> to tlb_flush_pte_range rather than tlb_flush_pmd_range causing some of the
> level parameters to be different.

arch/arm64/include/asm/pgtable-hwdef.h:#define CONT_PMD_SIZE            (CONT_PMDS * PMD_SIZE)

Seems to imply CONT_PMD_SIZE >= PMD_SIZE, and would thus tickle:

> > 	else if (_sz >= PMD_SIZE)
> > 		tlb_flush_pmd_range(tlb, address, _sz);

Or am I confused?