Re: [PATCH v3 3/3] mm/memory.c: Add memory read privilege before filling PTE entry

From: Andrew Morton <akpm@linux-foundation.org>
To: Bibo Mao <maobibo@loongson.cn>
Cc: "Thomas Bogendoerfer" <tsbogend@alpha.franken.de>,
	"Jiaxun Yang" <jiaxun.yang@flygoat.com>,
	"Huacai Chen" <chenhc@lemote.com>,
	"Paul Burton" <paulburton@kernel.org>,
	"Dmitry Korotin" <dkorotin@wavecomp.com>,
	"Philippe Mathieu-Daudé" <f4bug@amsat.org>,
	"Stafford Horne" <shorne@gmail.com>,
	"Steven Price" <steven.price@arm.com>,
	"Anshuman Khandual" <anshuman.khandual@arm.com>,
	linux-mips@vger.kernel.org, linux-kernel@vger.kernel.org,
	"Mike Rapoport" <rppt@linux.ibm.com>,
	"Sergei Shtylyov" <sergei.shtylyov@cogentembedded.com>,
	"Maciej W. Rozycki" <macro@wdc.com>,
	linux-mm@kvack.org, "David Hildenbrand" <david@redhat.com>
Subject: Re: [PATCH v3 3/3] mm/memory.c: Add memory read privilege before filling PTE entry
Date: Mon, 18 May 2020 13:57:47 -0700	[thread overview]
Message-ID: <20200518135747.d8837ba6742b2d193e14fbb0@linux-foundation.org> (raw)
In-Reply-To: <1589778529-25627-3-git-send-email-maobibo@loongson.cn>

On Mon, 18 May 2020 13:08:49 +0800 Bibo Mao <maobibo@loongson.cn> wrote:

> On mips platform, hw PTE entry valid bit is set in pte_mkyoung
> function, it is used to set physical page with readable privilege.

pte_mkyoung() seems to be a strange place to set the pte's valid bit. 
Why is it done there?  Can it be done within mips's mk_pte()?

> Here add pte_mkyoung function to make page readable on MIPS platform
> during page fault handling. This patch improves page fault latency
> about 10% on my MIPS machine with lmbench lat_pagefault case.
> 
> ...
>
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -2704,6 +2704,7 @@ static vm_fault_t wp_page_copy(struct vm_fault *vmf)
>  		}
>  		flush_cache_page(vma, vmf->address, pte_pfn(vmf->orig_pte));
>  		entry = mk_pte(new_page, vma->vm_page_prot);
> +		entry = pte_mkyoung(entry);
>  		entry = maybe_mkwrite(pte_mkdirty(entry), vma);

What is the effect on non-mips machines?  If it's only "additional
overhead" then it would be better to add a new pte_mkvalid() (or
whatever) and arrange for it to be a no-op on all but mips?

>  		/*
>  		 * Clear the pte entry and flush it first, before updating the
> @@ -3378,6 +3379,7 @@ static vm_fault_t do_anonymous_page(struct vm_fault *vmf)
>  	__SetPageUptodate(page);
>  
>  	entry = mk_pte(page, vma->vm_page_prot);
> +	entry = pte_mkyoung(entry);
>  	if (vma->vm_flags & VM_WRITE)
>  		entry = pte_mkwrite(pte_mkdirty(entry));
>  
> @@ -3660,6 +3662,7 @@ vm_fault_t alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg,
>  
>  	flush_icache_page(vma, page);
>  	entry = mk_pte(page, vma->vm_page_prot);
> +	entry = pte_mkyoung(entry);
>  	if (write)
>  		entry = maybe_mkwrite(pte_mkdirty(entry), vma);
>  	/* copy-on-write page */
> diff --git a/mm/mprotect.c b/mm/mprotect.c
> index 494192ca..673f1cd 100644
> --- a/mm/mprotect.c
> +++ b/mm/mprotect.c
> @@ -131,6 +131,8 @@ static unsigned long change_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
>  				ptent = pte_clear_uffd_wp(ptent);
>  			}
>  
> +			if (vma->vm_flags & VM_READ)
> +				ptent = pte_mkyoung(ptent);
>  			/* Avoid taking write faults for known dirty pages */
>  			if (dirty_accountable && pte_dirty(ptent) &&
>  					(pte_soft_dirty(ptent) ||