All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jesper Dangaard Brouer <hawk@kernel.org>
To: "Yunsheng Lin" <linyunsheng@huawei.com>,
	"Toke Høiland-Jørgensen" <toke@redhat.com>,
	davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com
Cc: zhangkun09@huawei.com, fanghaiqing@huawei.com,
	liuyonglong@huawei.com, Robin Murphy <robin.murphy@arm.com>,
	Alexander Duyck <alexander.duyck@gmail.com>,
	IOMMU <iommu@lists.linux.dev>,
	Andrew Morton <akpm@linux-foundation.org>,
	Eric Dumazet <edumazet@google.com>,
	Ilias Apalodimas <ilias.apalodimas@linaro.org>,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	netdev@vger.kernel.org, kernel-team <kernel-team@cloudflare.com>,
	Viktor Malik <vmalik@redhat.com>
Subject: Re: [PATCH net-next v3 3/3] page_pool: fix IOMMU crash when driver has already unbound
Date: Wed, 6 Nov 2024 16:57:25 +0100	[thread overview]
Message-ID: <b8b7818a-e44b-45f5-91c2-d5eceaa5dd5b@kernel.org> (raw)
In-Reply-To: <18ba4489-ad30-423e-9c54-d4025f74c193@kernel.org>

[-- Attachment #1: Type: text/plain, Size: 7688 bytes --]



On 06/11/2024 14.25, Jesper Dangaard Brouer wrote:
> 
> On 26/10/2024 09.33, Yunsheng Lin wrote:
>> On 2024/10/25 22:07, Jesper Dangaard Brouer wrote:
>>
>> ...
>>
>>>
>>>>> You and Jesper seems to be mentioning a possible fact that there might
>>>>> be 'hundreds of gigs of memory' needed for inflight pages, it would 
>>>>> be nice
>>>>> to provide more info or reasoning above why 'hundreds of gigs of 
>>>>> memory' is
>>>>> needed here so that we don't do a over-designed thing to support 
>>>>> recording
>>>>> unlimited in-flight pages if the driver unbound stalling turns out 
>>>>> impossible
>>>>> and the inflight pages do need to be recorded.
>>>>
>>>> I don't have a concrete example of a use that will blow the limit you
>>>> are setting (but maybe Jesper does), I am simply objecting to the
>>>> arbitrary imposing of any limit at all. It smells a lot of "640k ought
>>>> to be enough for anyone".
>>>>
>>>
>>> As I wrote before. In *production* I'm seeing TCP memory reach 24 GiB
>>> (on machines with 384GiB memory). I have attached a grafana screenshot
>>> to prove what I'm saying.
>>>
>>> As my co-worker Mike Freemon, have explain to me (and more details in
>>> blogposts[1]). It is no coincident that graph have a strange "sealing"
>>> close to 24 GiB (on machines with 384GiB total memory).  This is because
>>> TCP network stack goes into a memory "under pressure" state when 6.25%
>>> of total memory is used by TCP-stack. (Detail: The system will stay in
>>> that mode until allocated TCP memory falls below 4.68% of total memory).
>>>
>>>   [1] 
>>> https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/
>>
>> Thanks for the info.
> 
> Some more info from production servers.
> 
> (I'm amazed what we can do with a simple bpftrace script, Cc Viktor)
> 
> In below bpftrace script/oneliner I'm extracting the inflight count, for
> all page_pool's in the system, and storing that in a histogram hash.
> 
> sudo bpftrace -e '
>   rawtracepoint:page_pool_state_release { @cnt[probe]=count();
>    @cnt_total[probe]=count();
>    $pool=(struct page_pool*)arg0;
>    $release_cnt=(uint32)arg2;
>    $hold_cnt=$pool->pages_state_hold_cnt;
>    $inflight_cnt=(int32)($hold_cnt - $release_cnt);
>    @inflight=hist($inflight_cnt);
>   }
>   interval:s:1 {time("\n%H:%M:%S\n");
>    print(@cnt); clear(@cnt);
>    print(@inflight);
>    print(@cnt_total);
>   }'
> 
> The page_pool behavior depend on how NIC driver use it, so I've run this 
> on two prod servers with drivers bnxt and mlx5, on a 6.6.51 kernel.
> 
> Driver: bnxt_en
> - kernel 6.6.51
> 
> @cnt[rawtracepoint:page_pool_state_release]: 8447
> @inflight:
> [0]             507 |                                        |
> [1]             275 |                                        |
> [2, 4)          261 |                                        |
> [4, 8)          215 |                                        |
> [8, 16)         259 |                                        |
> [16, 32)        361 |                                        |
> [32, 64)        933 |                                        |
> [64, 128)      1966 |                                        |
> [128, 256)   937052 |@@@@@@@@@                               |
> [256, 512)  5178744 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
> [512, 1K)     73908 |                                        |
> [1K, 2K)    1220128 |@@@@@@@@@@@@                            |
> [2K, 4K)    1532724 |@@@@@@@@@@@@@@@                         |
> [4K, 8K)    1849062 |@@@@@@@@@@@@@@@@@@                      |
> [8K, 16K)   1466424 |@@@@@@@@@@@@@@                          |
> [16K, 32K)   858585 |@@@@@@@@                                |
> [32K, 64K)   693893 |@@@@@@                                  |
> [64K, 128K)  170625 |@                                       |
> 
> Driver: mlx5_core
>   - Kernel: 6.6.51
> 
> @cnt[rawtracepoint:page_pool_state_release]: 1975
> @inflight:
> [128, 256)         28293 |@@@@                               |
> [256, 512)        184312 |@@@@@@@@@@@@@@@@@@@@@@@@@@@        |
> [512, 1K)              0 |                                   |
> [1K, 2K)            4671 |                                   |
> [2K, 4K)          342571 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
> [4K, 8K)          180520 |@@@@@@@@@@@@@@@@@@@@@@@@@@@        |
> [8K, 16K)          96483 |@@@@@@@@@@@@@@                     |
> [16K, 32K)         25133 |@@@                                |
> [32K, 64K)          8274 |@                                  |
> 
> 
> The key thing to notice that we have up-to 128,000 pages in flight on
> these random production servers. The NIC have 64 RX queue configured,
> thus also 64 page_pool objects.
> 

I realized that we primarily want to know the maximum in-flight pages.

So, I modified the bpftrace oneliner to track the max for each page_pool 
in the system.

sudo bpftrace -e '
  rawtracepoint:page_pool_state_release { @cnt[probe]=count();
   @cnt_total[probe]=count();
   $pool=(struct page_pool*)arg0;
   $release_cnt=(uint32)arg2;
   $hold_cnt=$pool->pages_state_hold_cnt;
   $inflight_cnt=(int32)($hold_cnt - $release_cnt);
   $cur=@inflight_max[$pool];
   if ($inflight_cnt > $cur) {
     @inflight_max[$pool]=$inflight_cnt;}
  }
  interval:s:1 {time("\n%H:%M:%S\n");
   print(@cnt); clear(@cnt);
   print(@inflight_max);
   print(@cnt_total);
  }'

I've attached the output from the script.
For unknown reason this system had 199 page_pool objects.

The 20 top users:

$ cat out02.inflight-max | grep inflight_max | tail -n 20
@inflight_max[0xffff88829133d800]: 26473
@inflight_max[0xffff888293c3e000]: 27042
@inflight_max[0xffff888293c3b000]: 27709
@inflight_max[0xffff8881076f2800]: 29400
@inflight_max[0xffff88818386e000]: 29690
@inflight_max[0xffff8882190b1800]: 29813
@inflight_max[0xffff88819ee83800]: 30067
@inflight_max[0xffff8881076f4800]: 30086
@inflight_max[0xffff88818386b000]: 31116
@inflight_max[0xffff88816598f800]: 36970
@inflight_max[0xffff8882190b7800]: 37336
@inflight_max[0xffff888293c38800]: 39265
@inflight_max[0xffff888293c3c800]: 39632
@inflight_max[0xffff888293c3b800]: 43461
@inflight_max[0xffff888293c3f000]: 43787
@inflight_max[0xffff88816598f000]: 44557
@inflight_max[0xffff888132ce9000]: 45037
@inflight_max[0xffff888293c3f800]: 51843
@inflight_max[0xffff888183869800]: 62612
@inflight_max[0xffff888113d08000]: 73203

Adding all values together:

  grep inflight_max out02.inflight-max | awk 'BEGIN {tot=0} {tot+=$2; 
printf "total:" tot "\n"}' | tail -n 1

total:1707129

Worst case we need a data structure holding 1,707,129 pages.
Fortunately, we don't need a single data structure as this will be split
between 199 page_pool's.

--Jesper

[-- Attachment #2: out02.inflight-max --]
[-- Type: text/plain, Size: 8017 bytes --]

15:07:05
@cnt[rawtracepoint:page_pool_state_release]: 6344
@inflight_max[0xffff88a1512d9800]: 318
@inflight_max[0xffff88a071703800]: 318
@inflight_max[0xffff88a151367800]: 318
@inflight_max[0xffff88a1512fc000]: 318
@inflight_max[0xffff88a151365000]: 318
@inflight_max[0xffff88a1512d8000]: 318
@inflight_max[0xffff88a1512fe800]: 318
@inflight_max[0xffff88a1512dd800]: 318
@inflight_max[0xffff88a1512de800]: 318
@inflight_max[0xffff88a071707800]: 318
@inflight_max[0xffff88a1512ff000]: 318
@inflight_max[0xffff88a1512dd000]: 318
@inflight_max[0xffff88a1512dc800]: 318
@inflight_max[0xffff88a151366800]: 318
@inflight_max[0xffff88a071701000]: 318
@inflight_max[0xffff88a1512f9800]: 318
@inflight_max[0xffff88a071706000]: 318
@inflight_max[0xffff88a1512fb000]: 318
@inflight_max[0xffff88a071700000]: 318
@inflight_max[0xffff88a151360800]: 318
@inflight_max[0xffff88a1512fa800]: 318
@inflight_max[0xffff88a151360000]: 318
@inflight_max[0xffff88a151361800]: 318
@inflight_max[0xffff88a1512da000]: 318
@inflight_max[0xffff88a1512de000]: 318
@inflight_max[0xffff88a1512db800]: 318
@inflight_max[0xffff88a1512fb800]: 318
@inflight_max[0xffff88a1512fe000]: 318
@inflight_max[0xffff88a151364800]: 318
@inflight_max[0xffff88a071706800]: 318
@inflight_max[0xffff88a151364000]: 318
@inflight_max[0xffff88a1512fd000]: 318
@inflight_max[0xffff88a151366000]: 318
@inflight_max[0xffff88a071701800]: 318
@inflight_max[0xffff88a1512da800]: 318
@inflight_max[0xffff88a071700800]: 318
@inflight_max[0xffff88a1512fd800]: 318
@inflight_max[0xffff88a071702000]: 318
@inflight_max[0xffff88a151365800]: 318
@inflight_max[0xffff88a151361000]: 318
@inflight_max[0xffff88a1512f8000]: 318
@inflight_max[0xffff88a071705000]: 318
@inflight_max[0xffff88a151363800]: 318
@inflight_max[0xffff88a151362000]: 318
@inflight_max[0xffff88a1512d8800]: 318
@inflight_max[0xffff88a071704800]: 318
@inflight_max[0xffff88a1512db000]: 318
@inflight_max[0xffff88a1512fc800]: 318
@inflight_max[0xffff88a1512df000]: 318
@inflight_max[0xffff88a1512f8800]: 318
@inflight_max[0xffff88a1512df800]: 318
@inflight_max[0xffff88a071707000]: 318
@inflight_max[0xffff88a1512dc000]: 318
@inflight_max[0xffff88a071704000]: 318
@inflight_max[0xffff88a071702800]: 318
@inflight_max[0xffff88a071703000]: 318
@inflight_max[0xffff88a1512d9000]: 318
@inflight_max[0xffff88a151362800]: 318
@inflight_max[0xffff88a151367000]: 318
@inflight_max[0xffff88a1512f9000]: 318
@inflight_max[0xffff88a151363000]: 318
@inflight_max[0xffff88a1512fa000]: 318
@inflight_max[0xffff88a071705800]: 318
@inflight_max[0xffff88991d969000]: 331
@inflight_max[0xffff8899ca7b6800]: 336
@inflight_max[0xffff88991d96c000]: 339
@inflight_max[0xffff8899ca7b6000]: 340
@inflight_max[0xffff8899ca7b3000]: 342
@inflight_max[0xffff8899ca7a1800]: 342
@inflight_max[0xffff8899ca7b2800]: 342
@inflight_max[0xffff8899ca7a7800]: 343
@inflight_max[0xffff88991d96e000]: 343
@inflight_max[0xffff88991d96a000]: 344
@inflight_max[0xffff88991d96b000]: 344
@inflight_max[0xffff8898b3c92800]: 345
@inflight_max[0xffff8899ca7a4800]: 345
@inflight_max[0xffff8899ca7b4000]: 346
@inflight_max[0xffff8898b3c93800]: 347
@inflight_max[0xffff88991d968000]: 347
@inflight_max[0xffff88991d96f800]: 348
@inflight_max[0xffff8898b3c92000]: 348
@inflight_max[0xffff88991d96b800]: 350
@inflight_max[0xffff8898b3c90800]: 350
@inflight_max[0xffff8898b3c96800]: 351
@inflight_max[0xffff8899ca7b5000]: 351
@inflight_max[0xffff8898b3c97800]: 351
@inflight_max[0xffff8899ca7b4800]: 352
@inflight_max[0xffff8898b3c93000]: 353
@inflight_max[0xffff88991d969800]: 353
@inflight_max[0xffff8899ca7b5800]: 354
@inflight_max[0xffff8899ca7a2000]: 357
@inflight_max[0xffff8898b3c91000]: 357
@inflight_max[0xffff8898b3c97000]: 357
@inflight_max[0xffff8899ca7b1000]: 358
@inflight_max[0xffff8898b3c94000]: 359
@inflight_max[0xffff88991d96d800]: 362
@inflight_max[0xffff8899ca7b0800]: 363
@inflight_max[0xffff8899ca7a1000]: 364
@inflight_max[0xffff8899ca7a0000]: 364
@inflight_max[0xffff8898b3c95800]: 364
@inflight_max[0xffff8899ca7b7800]: 364
@inflight_max[0xffff8898b3c94800]: 365
@inflight_max[0xffff88991d96d000]: 365
@inflight_max[0xffff88991d968800]: 365
@inflight_max[0xffff8898b3c90000]: 365
@inflight_max[0xffff8899ca7a5800]: 365
@inflight_max[0xffff8899ca7a5000]: 366
@inflight_max[0xffff8899ca7a2800]: 366
@inflight_max[0xffff8899ca7b0000]: 366
@inflight_max[0xffff8899ca7a7000]: 367
@inflight_max[0xffff88991d96a800]: 368
@inflight_max[0xffff88991d96c800]: 368
@inflight_max[0xffff8899ca7a0800]: 368
@inflight_max[0xffff8899ca7a3800]: 370
@inflight_max[0xffff88991d96f000]: 371
@inflight_max[0xffff88991d96e800]: 372
@inflight_max[0xffff8899ca7b3800]: 373
@inflight_max[0xffff8899ca7b2000]: 373
@inflight_max[0xffff8899ca7b7000]: 373
@inflight_max[0xffff8899ca7a6800]: 373
@inflight_max[0xffff8899ca7a4000]: 374
@inflight_max[0xffff8899ca7a6000]: 377
@inflight_max[0xffff8899ca7a3000]: 378
@inflight_max[0xffff8898b3c95000]: 379
@inflight_max[0xffff8898b3c91800]: 379
@inflight_max[0xffff8898b3c96000]: 389
@inflight_max[0xffff888111079800]: 4201
@inflight_max[0xffff888111205000]: 4203
@inflight_max[0xffff888111079000]: 4393
@inflight_max[0xffff8881134fb800]: 4519
@inflight_max[0xffff88811107d800]: 4520
@inflight_max[0xffff8881134fc800]: 4586
@inflight_max[0xffff88811107a000]: 4650
@inflight_max[0xffff8881111b1800]: 5674
@inflight_max[0xffff88811107d000]: 6314
@inflight_max[0xffff888293c3d800]: 11714
@inflight_max[0xffff88818386c000]: 12302
@inflight_max[0xffff888293c3c000]: 12393
@inflight_max[0xffff888132cea800]: 12500
@inflight_max[0xffff888165966000]: 12940
@inflight_max[0xffff888113d0d800]: 13370
@inflight_max[0xffff88818386c800]: 13510
@inflight_max[0xffff888113d0c800]: 14027
@inflight_max[0xffff8882190b0800]: 15149
@inflight_max[0xffff888132ceb800]: 15405
@inflight_max[0xffff888132ceb000]: 15633
@inflight_max[0xffff888183869000]: 15684
@inflight_max[0xffff88818386b800]: 16142
@inflight_max[0xffff888132ced000]: 16450
@inflight_max[0xffff888165964800]: 17007
@inflight_max[0xffff8882190b5800]: 17879
@inflight_max[0xffff8882190b6000]: 17915
@inflight_max[0xffff88819ee80800]: 17977
@inflight_max[0xffff88819ee84000]: 18132
@inflight_max[0xffff888118680000]: 18204
@inflight_max[0xffff888118680800]: 18514
@inflight_max[0xffff88819ee83000]: 18546
@inflight_max[0xffff888183868000]: 18552
@inflight_max[0xffff8881076f1800]: 18706
@inflight_max[0xffff88819ee87000]: 18801
@inflight_max[0xffff888165965800]: 19556
@inflight_max[0xffff888293c3d000]: 20675
@inflight_max[0xffff88818386d000]: 20749
@inflight_max[0xffff88818386a800]: 21226
@inflight_max[0xffff888183868800]: 21559
@inflight_max[0xffff8881076f3000]: 21933
@inflight_max[0xffff888293c3a000]: 22086
@inflight_max[0xffff88819ee82800]: 22975
@inflight_max[0xffff88818386a000]: 23600
@inflight_max[0xffff88816598c000]: 24092
@inflight_max[0xffff888293c39800]: 24093
@inflight_max[0xffff88818386f000]: 24438
@inflight_max[0xffff888113d0e800]: 24882
@inflight_max[0xffff888293c39000]: 25218
@inflight_max[0xffff88818386d800]: 25276
@inflight_max[0xffff888293c3a800]: 25292
@inflight_max[0xffff888293c3e800]: 25429
@inflight_max[0xffff888293c38000]: 25794
@inflight_max[0xffff8881076f6800]: 26030
@inflight_max[0xffff88829133d800]: 26473
@inflight_max[0xffff888293c3e000]: 27042
@inflight_max[0xffff888293c3b000]: 27709
@inflight_max[0xffff8881076f2800]: 29400
@inflight_max[0xffff88818386e000]: 29690
@inflight_max[0xffff8882190b1800]: 29813
@inflight_max[0xffff88819ee83800]: 30067
@inflight_max[0xffff8881076f4800]: 30086
@inflight_max[0xffff88818386b000]: 31116
@inflight_max[0xffff88816598f800]: 36970
@inflight_max[0xffff8882190b7800]: 37336
@inflight_max[0xffff888293c38800]: 39265
@inflight_max[0xffff888293c3c800]: 39632
@inflight_max[0xffff888293c3b800]: 43461
@inflight_max[0xffff888293c3f000]: 43787
@inflight_max[0xffff88816598f000]: 44557
@inflight_max[0xffff888132ce9000]: 45037
@inflight_max[0xffff888293c3f800]: 51843
@inflight_max[0xffff888183869800]: 62612
@inflight_max[0xffff888113d08000]: 73203
@cnt_total[rawtracepoint:page_pool_state_release]: 67263129

  reply	other threads:[~2024-11-06 15:57 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-10-22  3:22 [PATCH net-next v3 0/3] fix two bugs related to page_pool Yunsheng Lin
2024-10-22  3:22 ` [Intel-wired-lan] " Yunsheng Lin
2024-10-22  3:22 ` [PATCH net-next v3 1/3] page_pool: introduce page_pool_to_pp() API Yunsheng Lin
2024-10-22  3:22   ` [Intel-wired-lan] " Yunsheng Lin
2024-10-22  3:22 ` [PATCH net-next v3 2/3] page_pool: fix timing for checking and disabling napi_local Yunsheng Lin
2024-11-07  6:17   ` Xuan Zhuo
2024-10-22  3:22 ` [PATCH net-next v3 3/3] page_pool: fix IOMMU crash when driver has already unbound Yunsheng Lin
2024-10-22 16:40   ` Simon Horman
2024-10-22 18:14   ` Jesper Dangaard Brouer
2024-10-23  8:59     ` Yunsheng Lin
2024-10-24 14:40       ` Toke Høiland-Jørgensen
2024-10-25  3:20         ` Yunsheng Lin
2024-10-25 11:16           ` Toke Høiland-Jørgensen
2024-10-25 14:07             ` Jesper Dangaard Brouer
2024-10-26  7:33               ` Yunsheng Lin
2024-11-06 13:25                 ` Jesper Dangaard Brouer
2024-11-06 15:57                   ` Jesper Dangaard Brouer [this message]
2024-11-06 19:55                     ` Alexander Duyck
2024-11-07 11:10                       ` Yunsheng Lin
2024-11-07 11:09                     ` Yunsheng Lin
2024-11-11 11:31                 ` Yunsheng Lin
2024-11-11 18:51                   ` Toke Høiland-Jørgensen
2024-11-12 12:22                     ` Yunsheng Lin
2024-11-12 14:19                       ` Jesper Dangaard Brouer
2024-11-13 12:21                         ` Yunsheng Lin
2024-11-18  9:08                         ` Yunsheng Lin
2024-11-18 15:11                           ` Jesper Dangaard Brouer
2024-10-26  7:32             ` Yunsheng Lin
2024-10-29 13:58               ` Toke Høiland-Jørgensen
2024-10-30 11:30                 ` Yunsheng Lin
2024-10-30 11:57                   ` Toke Høiland-Jørgensen
2024-10-31 12:17                     ` Yunsheng Lin
2024-10-31 16:18                       ` Toke Høiland-Jørgensen
2024-11-01 11:11                         ` Yunsheng Lin
2024-11-05 20:11                           ` Jesper Dangaard Brouer
2024-11-06 10:56                             ` Yunsheng Lin
2024-11-06 14:17                               ` Robin Murphy
2024-11-07  8:41                               ` Christoph Hellwig

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=b8b7818a-e44b-45f5-91c2-d5eceaa5dd5b@kernel.org \
    --to=hawk@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=alexander.duyck@gmail.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=fanghaiqing@huawei.com \
    --cc=ilias.apalodimas@linaro.org \
    --cc=iommu@lists.linux.dev \
    --cc=kernel-team@cloudflare.com \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linyunsheng@huawei.com \
    --cc=liuyonglong@huawei.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=robin.murphy@arm.com \
    --cc=toke@redhat.com \
    --cc=vmalik@redhat.com \
    --cc=zhangkun09@huawei.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.