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: <f1c07b2a-ffdd-0912-139d-d51b34b77bcd@huaweicloud.com>
Date:   Thu, 13 Oct 2022 22:05:53 +0800
From:   Yu Kuai <yukuai1@...weicloud.com>
To:     Jinlong Chen <chenjinlong2016@...look.com>, yukuai1@...weicloud.com
Cc:     axboe@...nel.dk, linux-block@...r.kernel.org,
        linux-kernel@...r.kernel.org, "yukuai (C)" <yukuai3@...wei.com>
Subject: Re: [PATCH] blk-mq: put the reference of the io scheduler module
 after switching back



在 2022/10/13 21:47, Jinlong Chen 写道:
> Hi, Yu Kuai!
> 
>>
>> I'm confused here, cause I do think this patch make sense.
>>
>> blk_mq_update_nr_hw_queues
>>    // for each queue using the tagset
>>    blk_mq_freeze_queue
>>    // if current elevator is not none, swith to none
>>    blk_mq_elv_switch_none
>>     // elevator need to be switched back, got a reference to
>>     // prevent module to be removed.
>>     __module_get
>>     elevator_switch(q, NULL);
>>
>>    // switch back from none elevator
>>    blk_mq_elv_switch_back
>>     -> should release the module reference here
>>    blk_mq_unfreeze_queue
>>
> 
> We need to release the reference only if blk_mq_elv_switch_back got its own
>   reference. However, blk_mq_elv_switch_back (precisely elevator_switch_mq)
>   does not increase the reference of the module it is switching to.
>   Hence, the reference got in blk_mq_elv_switch_none does not need to be released,
>   or we'll have to re-increase the reference count manually.'

But I don't see elevator_switch() release the referenct of the module
it is switching from. It's still not balance to me.

Thanks,
Kuai
> 
>>
>> By the way, I do not test yet, but I think problem can be reporduced:
>>
>>
>> 1. modprobe bfq
>> 2. switch elevator to bfq
>> 3. trigger blk_mq_update_nr_hw_queues
>> 4. switch elevator to none
>> 5. rmmod bfq will fail
>>
> 
> I tried to reproduce the problem but failed, so I found my mistake.
> 
> Sincerely,
> Jinlong Chen
> .
> 

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ