Message-ID: <c22e1d58-e16f-fde5-cee7-c13dedbe1656@linux.dev>
Date:   Tue, 21 Mar 2023 15:04:46 +0800
MIME-Version: 1.0
Subject: Re: [PATCH] mm: kfence: fix PG_slab and memcg_data clearing
To:     Peng Zhang <zhangpeng.00@bytedance.com>
Cc:     kasan-dev@googlegroups.com, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, dvyukov@google.com,
        roman.gushchin@linux.dev, jannh@google.com, sjpark@amazon.de,
        akpm@linux-foundation.org, elver@google.com, glider@google.com,
        Muchun Song <songmuchun@bytedance.com>
References: <20230320030059.20189-1-songmuchun@bytedance.com>
 <974ef73e-ab4f-7b24-d070-c981654e8c22@bytedance.com>
From:   Muchun Song <muchun.song@linux.dev>
In-Reply-To: <974ef73e-ab4f-7b24-d070-c981654e8c22@bytedance.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


On 2023/3/21 12:14, Peng Zhang wrote:
>
> 在 2023/3/20 11:00, Muchun Song 写道:
>> It does not reset PG_slab and memcg_data when KFENCE fails to initialize
>> kfence pool at runtime. It is reporting a "Bad page state" message when
>> kfence pool is freed to buddy. The checking of whether it is a compound
>> head page seems unnecessary sicne we already guarantee this when 
>> allocating
>> kfence pool, removing the check to simplify the code.
>>
>> Fixes: 0ce20dd84089 ("mm: add Kernel Electric-Fence infrastructure")
>> Fixes: 8f0b36497303 ("mm: kfence: fix objcgs vector allocation")
>> Signed-off-by: Muchun Song <songmuchun@bytedance.com>
>> ---
>>   mm/kfence/core.c | 30 +++++++++++++++---------------
>>   1 file changed, 15 insertions(+), 15 deletions(-)
>>
>> diff --git a/mm/kfence/core.c b/mm/kfence/core.c
>> index 79c94ee55f97..d66092dd187c 100644
>> --- a/mm/kfence/core.c
>> +++ b/mm/kfence/core.c
>> @@ -561,10 +561,6 @@ static unsigned long kfence_init_pool(void)
>>           if (!i || (i % 2))
>>               continue;
>>   -        /* Verify we do not have a compound head page. */
>> -        if (WARN_ON(compound_head(&pages[i]) != &pages[i]))
>> -            return addr;
>> -
>>           __folio_set_slab(slab_folio(slab));
>>   #ifdef CONFIG_MEMCG
>>           slab->memcg_data = (unsigned long)&kfence_metadata[i / 2 - 
>> 1].objcg |
>> @@ -597,12 +593,26 @@ static unsigned long kfence_init_pool(void)
>>             /* Protect the right redzone. */
>>           if (unlikely(!kfence_protect(addr + PAGE_SIZE)))
>> -            return addr;
>> +            goto reset_slab;
>>             addr += 2 * PAGE_SIZE;
>>       }
>>         return 0;
>> +
>> +reset_slab:
>> +    for (i = 0; i < KFENCE_POOL_SIZE / PAGE_SIZE; i++) {
>> +        struct slab *slab = page_slab(&pages[i]);
>> +
>> +        if (!i || (i % 2))
>> +            continue;
>> +#ifdef CONFIG_MEMCG
>> +        slab->memcg_data = 0;
>> +#endif
>> +        __folio_clear_slab(slab_folio(slab));
>> +    }
> Can this loop be simplified to this?
>
>     for (i = 2; i < KFENCE_POOL_SIZE / PAGE_SIZE; i+=2) {
>         struct slab *slab = page_slab(&pages[i]);
> #ifdef CONFIG_MEMCG
>         slab->memcg_data = 0;
> #endif
>         __folio_clear_slab(slab_folio(slab));
>     }
>

It's a good simplification. The loop setting Pg_slab before this
also can be simplified in the same way. However, I choose a
consistent way to fix this bug. I'd like to send a separate
simplification patch to simplify both two loops instead of
in a bugfix patch.

Thanks.