linux-kernel - Re: [PATCH v3 3/7] mm/lru: replace pgdat lru

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <0bfa9a03-b095-df83-9cfd-146da9aab89a@linux.alibaba.com>
Date:   Mon, 18 Nov 2019 19:55:43 +0800
From:   Alex Shi <alex.shi@...ux.alibaba.com>
To:     Matthew Wilcox <willy@...radead.org>
Cc:     cgroups@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org, akpm@...ux-foundation.org,
        mgorman@...hsingularity.net, tj@...nel.org, hughd@...gle.com,
        khlebnikov@...dex-team.ru, daniel.m.jordan@...cle.com,
        yang.shi@...ux.alibaba.com, Johannes Weiner <hannes@...xchg.org>,
        Michal Hocko <mhocko@...nel.org>,
        Vladimir Davydov <vdavydov.dev@...il.com>,
        Roman Gushchin <guro@...com>,
        Shakeel Butt <shakeelb@...gle.com>,
        Chris Down <chris@...isdown.name>,
        Thomas Gleixner <tglx@...utronix.de>,
        Vlastimil Babka <vbabka@...e.cz>, Qian Cai <cai@....pw>,
        Andrey Ryabinin <aryabinin@...tuozzo.com>,
        "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
        Jérôme Glisse <jglisse@...hat.com>,
        Andrea Arcangeli <aarcange@...hat.com>,
        David Rientjes <rientjes@...gle.com>,
        "Aneesh Kumar K.V" <aneesh.kumar@...ux.ibm.com>,
        swkhack <swkhack@...il.com>,
        "Potyra, Stefan" <Stefan.Potyra@...ktrobit.com>,
        Mike Rapoport <rppt@...ux.vnet.ibm.com>,
        Stephen Rothwell <sfr@...b.auug.org.au>,
        Colin Ian King <colin.king@...onical.com>,
        Jason Gunthorpe <jgg@...pe.ca>,
        Mauro Carvalho Chehab <mchehab+samsung@...nel.org>,
        Peng Fan <peng.fan@....com>,
        Nikolay Borisov <nborisov@...e.com>,
        Ira Weiny <ira.weiny@...el.com>,
        Kirill Tkhai <ktkhai@...tuozzo.com>,
        Yafang Shao <laoar.shao@...il.com>
Subject: Re: [PATCH v3 3/7] mm/lru: replace pgdat lru_lock with lruvec lock



在 2019/11/16 下午12:38, Matthew Wilcox 写道:
> On Sat, Nov 16, 2019 at 11:15:02AM +0800, Alex Shi wrote:
>> This is the main patch to replace per node lru_lock with per memcg
>> lruvec lock. It also fold the irqsave flags into lruvec.
> 
> I have to say, I don't love the part where we fold the irqsave flags
> into the lruvec.  I know it saves us an argument, but it opens up the
> possibility of mismatched expectations.  eg we currently have:
> 
> static void __split_huge_page(struct page *page, struct list_head *list,
> 			struct lruvec *lruvec, pgoff_t end)
> {
> ...
> 	spin_unlock_irqrestore(&lruvec->lru_lock, lruvec->irqflags);
> 
> so if we introduce a new caller, we have to be certain that this caller
> is also using lock_page_lruvec_irqsave() and not lock_page_lruvec_irq().
> I can't think of a way to make the compiler enforce that, and if we don't,
> then we can get some odd crashes with interrupts being unexpectedly
> enabled or disabled, depending on how ->irqflags was used last.
> 
> So it makes the code more subtle.  And that's not a good thing.

Hi Matthew,

Thanks for comments!

Here, the irqflags is bound, and belong to lruvec, merging them into together helps us to take them as whole, and thus reduce a unnecessary code clues.
The only thing maybe bad that it may take move place in pg_data_t.lruvec, but there are PADDINGs to remove this concern.

As your concern for a 'new' caller, since __split_huge_page is a static helper here, no distub for anyothers.

Do you agree on that?

> 
>> +static inline struct lruvec *lock_page_lruvec_irq(struct page *page,
>> +						struct pglist_data *pgdat)
>> +{
>> +	struct lruvec *lruvec = mem_cgroup_page_lruvec(page, pgdat);
>> +
>> +	spin_lock_irq(&lruvec->lru_lock);
>> +
>> +	return lruvec;
>> +}
> 
> ...
> 
>> +static struct lruvec *lock_page_lru(struct page *page, int *isolated)
>>  {
>>  	pg_data_t *pgdat = page_pgdat(page);
>> +	struct lruvec *lruvec = lock_page_lruvec_irq(page, pgdat);
>>  
>> -	spin_lock_irq(&pgdat->lru_lock);
>>  	if (PageLRU(page)) {
>> -		struct lruvec *lruvec;
>>  
>> -		lruvec = mem_cgroup_page_lruvec(page, pgdat);
>>  		ClearPageLRU(page);
>>  		del_page_from_lru_list(page, lruvec, page_lru(page));
>>  		*isolated = 1;
>>  	} else
>>  		*isolated = 0;
>> +
>> +	return lruvec;
>>  }
> 
> But what if the page is !PageLRU?  What lruvec did we just lock?

like original pgdat->lru_lock, we need the lock from PageLRU racing. And it the lruvec which the page should be.


> According to the comments on mem_cgroup_page_lruvec(),
> 
>  * This function is only safe when following the LRU page isolation
>  * and putback protocol: the LRU lock must be held, and the page must
>  * either be PageLRU() or the caller must have isolated/allocated it.
> 
> and now it's being called in order to find out which LRU lock to take.
> So this comment needs to be updated, if it's wrong, or this patch has
> a race.


Yes, the function reminder is a bit misunderstanding with new patch, How about the following changes:

- * This function is only safe when following the LRU page isolation
- * and putback protocol: the LRU lock must be held, and the page must
- * either be PageLRU() or the caller must have isolated/allocated it.
+ * The caller needs to grantee the page's mem_cgroup is undisturbed during
+ * using. That could be done by lock_page_memcg or lock_page_lruvec.

Thanks
Alex