Re: [PATCH] MIPS: check return value of pgtable_pmd_page_ctor

From: Huang Pei <huangpei@loongson.cn>
To: Joshua Kinard <kumba@gentoo.org>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>,
	ambrosehua@gmail.com, Bibo Mao <maobibo@loongson.cn>,
	linux-mips@vger.kernel.org, Jiaxun Yang <jiaxun.yang@flygoat.com>,
	Paul Burton <paulburton@kernel.org>,
	Li Xuefeng <lixuefeng@loongson.cn>,
	Yang Tiezhu <yangtiezhu@loongson.cn>,
	Gao Juxin <gaojuxin@loongson.cn>,
	Huacai Chen <chenhuacai@loongson.cn>,
	Jinyang He <hejinyang@loongson.cn>
Subject: Re: [PATCH] MIPS: check return value of pgtable_pmd_page_ctor
Date: Wed, 21 Jul 2021 16:13:50 +0800	[thread overview]
Message-ID: <20210721081350.ygtezwksqfy6cbsq@ambrosehua-HP-xw6600-Workstation> (raw)
In-Reply-To: <d08a98c9-e696-57b0-5837-b8a640e156fa@gentoo.org>

On Wed, Jul 21, 2021 at 12:23:39AM -0400, Joshua Kinard wrote:
> On 7/20/2021 23:43, Huang Pei wrote:
> > +. According to Documentation/vm/split_page_table_lock, handle failure
> > of pgtable_pmd_page_ctor
> > 
> > +. use GFP_KERNEL_ACCOUNT instead of GFP_KERNEL|__GFP_ACCOUNT
> > 
> > Reported-by: Joshua Kinard <kumba@gentoo.org>
> > Signed-off-by: Huang Pei <huangpei@loongson.cn>
> > ---
> >  arch/mips/include/asm/pgalloc.h | 12 ++++++++----
> >  1 file changed, 8 insertions(+), 4 deletions(-)
> > 
> > diff --git a/arch/mips/include/asm/pgalloc.h b/arch/mips/include/asm/pgalloc.h
> > index d0cf997b4ba8..5c9597a6c60c 100644
> > --- a/arch/mips/include/asm/pgalloc.h
> > +++ b/arch/mips/include/asm/pgalloc.h
> > @@ -62,11 +62,15 @@ static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
> >  	pmd_t *pmd = NULL;
> >  	struct page *pg;
> >  
> > -	pg = alloc_pages(GFP_KERNEL | __GFP_ACCOUNT, PMD_ORDER);
> > +	pg = alloc_pages(GFP_KERNEL_ACCOUNT, PMD_ORDER);
> >  	if (pg) {
> > -		pgtable_pmd_page_ctor(pg);
> > -		pmd = (pmd_t *)page_address(pg);
> > -		pmd_init((unsigned long)pmd, (unsigned long)invalid_pte_table);
> > +		if(pgtable_pmd_page_ctor(pg)) {
> > +			pmd = (pmd_t *)page_address(pg);
> > +			pmd_init((unsigned long)pmd, (unsigned long)invalid_pte_table);
> > +		} else {
> > +			__free_pages(pg, PMD_ORDER);
> > +		}
> > +
> >  	}
> >  	return pmd;
> >  }
> > 
> Instead of the nested if statements, why not go with something that looks more
> like what is in arch/x86/include/asm/pgalloc.h?
> 
> Note, I don't have a full kernel tree in front of me at the moment, so this is
> just the refactored function of pmd_alloc_one:
> 
> static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
> {
> 	pmd_t *pmd;
> 	struct page *page;
> 
> 	page = alloc_pages(GFP_KERNEL_ACCOUNT, PMD_ORDER);
> 	if (!page)
> 		return NULL;
> 
> 	if (!pgtable_pmd_page_ctor(page)) {
> 		__free_pages(page, PMD_ORDER);
> 		return NULL;
> 	}
> 
> 	pmd = (pmd_t *)page_address(page);
> 	pmd_init((unsigned long)pmd, (unsigned long)invalid_pte_table);
> 
> 	return pmd;
> }
> 

Much more readable, V2 just resend, thank you!

PS:

Latest Gentoo/MIPS stage3 is only available for big endian, is there any
little endian stage3 available? 
> -- 
> Joshua Kinard
> Gentoo/MIPS
> kumba@gentoo.org
> rsa6144/5C63F4E3F5C6C943 2015-04-27
> 177C 1972 1FB8 F254 BAD0 3E72 5C63 F4E3 F5C6 C943
> 
> "The past tempts us, the present confuses us, the future frightens us.  And
> our lives slip away, moment by moment, lost in that vast, terrible in-between."
> 
> --Emperor Turhan, Centauri Republic