linux-kernel - Re: [PATCH RFC net-next v2 2/7] net: page_pool: place frag

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: <e48185cb-3057-e778-75c4-d266a249088b@intel.com>
Date:   Tue, 18 Jul 2023 15:50:08 +0200
From:   Alexander Lobakin <aleksander.lobakin@...el.com>
To:     Jesper Dangaard Brouer <jbrouer@...hat.com>
CC:     "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>, <brouer@...hat.com>,
        Maciej Fijalkowski <maciej.fijalkowski@...el.com>,
        Larysa Zaremba <larysa.zaremba@...el.com>,
        Yunsheng Lin <linyunsheng@...wei.com>,
        Alexander Duyck <alexanderduyck@...com>,
        Jesper Dangaard Brouer <hawk@...nel.org>,
        "Ilias Apalodimas" <ilias.apalodimas@...aro.org>,
        <netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH RFC net-next v2 2/7] net: page_pool: place frag_* fields
 in one cacheline

From: Jesper Dangaard Brouer <jbrouer@...hat.com>
Date: Fri, 14 Jul 2023 20:37:39 +0200

> 
> 
> On 14/07/2023 19.08, Alexander Lobakin wrote:
>> On x86_64, frag_* fields of struct page_pool are scattered across two
>> cachelines despite the summary size of 24 bytes. The last field,
>> ::frag_users, is pushed out to the next one, sharing it with
>> ::alloc_stats.
>> All three fields are used in pretty much the same places. There are some
>> holes and cold members to move around. Move frag_* one block up, placing
>> them right after &page_pool_params perfectly at the beginning of CL2.
>> This doesn't do any meaningful to the second block, as those are some
>> destroy-path cold structures, and doesn't do anything to ::alloc_stats,
>> which still starts at 200-byte offset, 8 bytes after CL3 (still fitting
>> into 1 cacheline).
>> On my setup, this yields 1-2% of Mpps when using PP frags actively.
>> When it comes to 32-bit architectures with 32-byte CL: &page_pool_params
>> plus ::pad is 44 bytes, the block taken care of is 16 bytes within one
>> CL, so there should be at least no regressions from the actual change.
>>
>> Signed-off-by: Alexander Lobakin <aleksander.lobakin@...el.com>
>> ---
>>   include/net/page_pool.h | 10 +++++-----
>>   1 file changed, 5 insertions(+), 5 deletions(-)
>>
>> diff --git a/include/net/page_pool.h b/include/net/page_pool.h
>> index 829dc1f8ba6b..212d72b5cfec 100644
>> --- a/include/net/page_pool.h
>> +++ b/include/net/page_pool.h
>> @@ -130,16 +130,16 @@ static inline u64
>> *page_pool_ethtool_stats_get(u64 *data, void *stats)
>>   struct page_pool {
>>       struct page_pool_params p;
>>   +    long frag_users;
>> +    struct page *frag_page;
>> +    unsigned int frag_offset;
>> +    u32 pages_state_hold_cnt;
> 
> I think this is okay, but I want to highlight that:
>  - pages_state_hold_cnt and pages_state_release_cnt
> need to be kept on separate cache-lines.

They're pretty far away from each other. I moved hold_cnt here as well
to keep it stacked with frag_offset and avoid introducing 32-bit holes.

> 
> 
>> +
>>       struct delayed_work release_dw;
>>       void (*disconnect)(void *);
>>       unsigned long defer_start;
>>       unsigned long defer_warn;
>>   -    u32 pages_state_hold_cnt;
>> -    unsigned int frag_offset;
>> -    struct page *frag_page;
>> -    long frag_users;
>> -
>>   #ifdef CONFIG_PAGE_POOL_STATS
>>       /* these stats are incremented while in softirq context */
>>       struct page_pool_alloc_stats alloc_stats;
> 

Thanks,
Olek