netdev - Re: [PATCH net] net: dsa: microchip: Added the condition for scheduling ksz_mib_read

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: <20211008113402.0aed1d2b@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Fri, 8 Oct 2021 11:34:02 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     Arun Ramadoss <arun.ramadoss@...rochip.com>,
        linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
        George McCollister <george.mccollister@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Vladimir Oltean <olteanv@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        Vivien Didelot <vivien.didelot@...il.com>,
        UNGLinuxDriver@...rochip.com,
        Woojung Huh <woojung.huh@...rochip.com>
Subject: Re: [PATCH net] net: dsa: microchip: Added the condition for
 scheduling ksz_mib_read_work

On Fri, 8 Oct 2021 15:58:16 +0200 Andrew Lunn wrote:
> On Fri, Oct 08, 2021 at 02:13:48PM +0530, Arun Ramadoss wrote:
> > When the ksz module is installed and removed using rmmod, kernel crashes
> > with null pointer dereferrence error. During rmmod, ksz_switch_remove
> > function tries to cancel the mib_read_workqueue using
> > cancel_delayed_work_sync routine.
> > 
> > At the end of  mib_read_workqueue execution, it again reschedule the
> > workqueue unconditionally. Due to which queue rescheduled after
> > mib_interval, during this execution it tries to access dp->slave. But
> > the slave is unregistered in the ksz_switch_remove function. Hence
> > kernel crashes.  
> 
> Something not correct here.
> 
> https://www.kernel.org/doc/html/latest/core-api/workqueue.html?highlight=delayed_work#c.cancel_delayed_work_sync
> 
> This is cancel_work_sync() for delayed works.
> 
> and
> 
> https://www.kernel.org/doc/html/latest/core-api/workqueue.html?highlight=delayed_work#c.cancel_work_sync
> 
> This function can be used even if the work re-queues itself or
> migrates to another workqueue.
> 
> Maybe the real problem is a missing call to destroy_worker()?

Also the cancel_delayed_work_sync() is suspiciously early in the remove
flow. There is a schedule_work call in ksz_mac_link_down() which may 
schedule the work back in. That'd also explain why the patch helps since
ksz_mac_link_down() only schedules if (dev->mib_read_interval).