netdev - Re: [PATCH v2] ravb: Fixed the problem that rmmod can not be done

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <32d3b998-322b-7c0a-b14a-41ca66dc601a@gmail.com>
Date:   Fri, 31 Jul 2020 19:28:38 +0300
From:   Sergei Shtylyov <sergei.shtylyov@...il.com>
To:     "ashiduka@...itsu.com" <ashiduka@...itsu.com>
Cc:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "linux-renesas-soc@...r.kernel.org" 
        <linux-renesas-soc@...r.kernel.org>
Subject: Re: [PATCH v2] ravb: Fixed the problem that rmmod can not be done

Hello!

On 7/31/20 1:18 PM, ashiduka@...itsu.com wrote:

> I understand that the commit log needs to be corrected.

   The subject also could be more concise...

> (Shimoda-san's point is also correct)
> 
> If there is anything else that needs to be corrected, please point it out.

   OK, I'll try to post a proper patch review...

>>    That seems a common pattern, inlluding the Renesas sh_eth
>> driver...
> 
> Yes.

    Not at all so common as I thought! Only 4 drivers use mdio-bitbang, 2 of them are
for the Renesas SoCs...

> If I can get an R-Car Gen2 board, I will also fix sh_eth driver.

    Do yuo have R-Car V3H at hand, by chance? It does have a GEther controler used for
booting up... 

>>    No, the driver's remove() method calls ravb_mdio_release() and
>> that one calls
>> free_mdio_bitbang() that calls module_put(); the actual reason lies
>> somewehre deeper than this...
> 
> No.
> Running rmmod calls delete_module() in kernel/module.c before ravb_mdio_release() is called.
> delete_module()
>    -> try_stop_module()
>      -> try_release_module_ref()
> In try_release_module_ref(), check refcnt and if it is counted up, ravb_mdio_release() is not
> called and rmmod is terminated.

   Yes, after some rummaging in the module support code, I have to agree here. I was
just surprised with you finding such a critical bug so late in the drivers' life cycle.
Well, due to usually using NFS the EtherAVB (and Ether too) driver is probably alwaysbuilt in-kernel...

> Thanks & Best Regards,
> Yuusuke Ashizuka <ashiduka@...itsu.com>

   Trim your messages after your goodbye. That original message stuff typically isn't
tolerated in the Linux mailing lists, nearly the same as top-posting...

[...]

MBR, Sergei