linux-kernel - Re: [PATCH v2 04/11] md/raid5: use mempool to allocate stripe_request

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: <5460e6b3-9b23-49e9-a52b-57e1e6d2490a@fnnas.com>
Date: Sat, 3 Jan 2026 19:12:33 +0800
From: "Yu Kuai" <yukuai@...as.com>
To: "Li Nan" <linan666@...weicloud.com>, <song@...nel.org>, 
	<linux-raid@...r.kernel.org>
Cc: <linux-kernel@...r.kernel.org>, <filippo@...ian.org>, <colyli@...as.com>, 
	<yukuai@...as.com>
Subject: Re: [PATCH v2 04/11] md/raid5: use mempool to allocate stripe_request_ctx

Hi,

在 2025/12/26 16:33, Li Nan 写道:
>
>
> 在 2025/11/24 14:31, Yu Kuai 写道:
>> On the one hand, stripe_request_ctx is 72 bytes, and it's a bit huge for
>> a stack variable.
>>
>> On the other hand, the bitmap sectors_to_do is a fixed size, result in
>> max_hw_sector_kb of raid5 array is at most 256 * 4k = 1Mb, and this will
>> make full stripe IO impossible for the array that chunk_size * 
>> data_disks
>> is bigger. Allocate ctx during runtime will make it possible to get rid
>> of this limit.
>>
>> Signed-off-by: Yu Kuai <yukuai@...as.com>
>> ---
>>   drivers/md/md.h       |  4 +++
>>   drivers/md/raid1-10.c |  5 ----
>>   drivers/md/raid5.c    | 61 +++++++++++++++++++++++++++----------------
>>   drivers/md/raid5.h    |  2 ++
>>   4 files changed, 45 insertions(+), 27 deletions(-)
>>
>
> [...]
>
>> @@ -7374,6 +7380,10 @@ static void free_conf(struct r5conf *conf)
>>       bioset_exit(&conf->bio_split);
>>       kfree(conf->stripe_hashtbl);
>>       kfree(conf->pending_data);
>> +
>> +    if (conf->ctx_pool)
>> +        mempool_destroy(conf->ctx_pool);
>> +
>>       kfree(conf);
>>   }
>>   @@ -8057,6 +8067,13 @@ static int raid5_run(struct mddev *mddev)
>>               goto abort;
>>       }
>>   +    conf->ctx_pool = mempool_create_kmalloc_pool(NR_RAID_BIOS,
>> +                    sizeof(struct stripe_request_ctx));
>> +    if (!conf->ctx_pool) {
>> +        ret = -ENOMEM;
>> +        goto abort;
>> +    }
>> +
>
> What about moving create to setup_conf()? If so, call destroy in
> free_conf() without checks.

No, we can't, this must be done after raid5_set_limits(), which is called
at the end of raid5_run().

>
>>       if (log_init(conf, journal_dev, raid5_has_ppl(conf)))
>>           goto abort;
>>   diff --git a/drivers/md/raid5.h b/drivers/md/raid5.h
>> index eafc6e9ed6ee..6e3f07119fa4 100644
>> --- a/drivers/md/raid5.h
>> +++ b/drivers/md/raid5.h
>> @@ -690,6 +690,8 @@ struct r5conf {
>>       struct list_head    pending_list;
>>       int            pending_data_cnt;
>>       struct r5pending_data    *next_pending_data;
>> +
>> +    mempool_t        *ctx_pool;
>>   };
>>     #if PAGE_SIZE == DEFAULT_STRIPE_SIZE
>
-- 
Thansk,
Kuai