linux-kernel - Re: [PATCH] thp: use mm_file_counter to determine update which rss counter

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] [day] [month] [year] [list]

Message-ID: <88cf1585-bddf-0f49-ee6c-e30473654d97@linux.alibaba.com>
Date:   Wed, 20 Jun 2018 09:42:33 -0700
From:   Yang Shi <yang.shi@...ux.alibaba.com>
To:     Vlastimil Babka <vbabka@...e.cz>, hughd@...gle.com,
        kirill.shutemov@...ux.intel.com, akpm@...ux-foundation.org
Cc:     linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        Jerome Marchand <jmarchan@...hat.com>
Subject: Re: [PATCH] thp: use mm_file_counter to determine update which rss
 counter



On 6/20/18 3:57 AM, Vlastimil Babka wrote:
> On 06/19/2018 11:08 PM, Yang Shi wrote:
>> Since commit eca56ff906bdd0239485e8b47154a6e73dd9a2f3 ("mm, shmem: add
>> internal shmem resident memory accounting"), MM_SHMEMPAGES is added to
>> separate the shmem accounting from regular files. So, all shmem pages
>> should be accounted to MM_SHMEMPAGES instead of MM_FILEPAGES.
>>
>> And, normal 4K shmem pages have been accounted to MM_SHMEMPAGES, so
>> shmem thp pages should be not treated differently. Accouting them to
>> MM_SHMEMPAGES via mm_counter_file() since shmem pages are swap backed
>> to keep consistent with normal 4K shmem pages.
>>
>> This will not change the rss counter of processes since shmem pages are
>> still a part of it.
> <Andrew>So what are the user-visible effects of the patch then?</Andrew>
>
> Let's add this?:
>
> The /proc/pid/status and /proc/pid/statm counters will however be more
> accurate wrt shmem usage, as originally intended.

Thanks. Yes, other than this as commit 
eca56ff906bdd0239485e8b47154a6e73dd9a2f3 mentioned, oom also could 
report more accurate "shmem-rss"

>
>> Signed-off-by: Yang Shi <yang.shi@...ux.alibaba.com>
>> Cc: Hugh Dickins <hughd@...gle.com>
>> Cc: Kirill A. Shutemov <kirill.shutemov@...ux.intel.com>
> Good catch, thanks. I guess the discrepancy happened due to the
> thp-tmpfs patchset existing externally for a long time and number of
> iterations, while commit eca56ff906bdd was merged, and nobody noticed
> that the thp case added more MM_FILEPAGES users.
>
> Acked-by: Vlastimil Babka <vbabka@...e.cz>

Thanks.

Yang

>
>> ---
>>   mm/huge_memory.c | 4 ++--
>>   mm/memory.c      | 2 +-
>>   2 files changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>> index 1cd7c1a..2687f7c 100644
>> --- a/mm/huge_memory.c
>> +++ b/mm/huge_memory.c
>> @@ -1740,7 +1740,7 @@ int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
>>   		} else {
>>   			if (arch_needs_pgtable_deposit())
>>   				zap_deposited_table(tlb->mm, pmd);
>> -			add_mm_counter(tlb->mm, MM_FILEPAGES, -HPAGE_PMD_NR);
>> +			add_mm_counter(tlb->mm, mm_counter_file(page), -HPAGE_PMD_NR);
>>   		}
>>   
>>   		spin_unlock(ptl);
>> @@ -2088,7 +2088,7 @@ static void __split_huge_pmd_locked(struct vm_area_struct *vma, pmd_t *pmd,
>>   			SetPageReferenced(page);
>>   		page_remove_rmap(page, true);
>>   		put_page(page);
>> -		add_mm_counter(mm, MM_FILEPAGES, -HPAGE_PMD_NR);
>> +		add_mm_counter(mm, mm_counter_file(page), -HPAGE_PMD_NR);
>>   		return;
>>   	} else if (is_huge_zero_pmd(*pmd)) {
>>   		/*
>> diff --git a/mm/memory.c b/mm/memory.c
>> index 7206a63..4a2f2a8 100644
>> --- a/mm/memory.c
>> +++ b/mm/memory.c
>> @@ -3372,7 +3372,7 @@ static int do_set_pmd(struct vm_fault *vmf, struct page *page)
>>   	if (write)
>>   		entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
>>   
>> -	add_mm_counter(vma->vm_mm, MM_FILEPAGES, HPAGE_PMD_NR);
>> +	add_mm_counter(vma->vm_mm, mm_counter_file(page), HPAGE_PMD_NR);
>>   	page_add_file_rmap(page, true);
>>   	/*
>>   	 * deposit and withdraw with pmd lock held
>>