linux-kernel - Re: Potential race condition in drivers/ata/sata

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] [day] [month] [year] [list]

Message-ID: <20160811162650.GB2695@mtj.duckdns.org>
Date:	Thu, 11 Aug 2016 12:26:50 -0400
From:	Tejun Heo <tj@...nel.org>
To:	Pavel Andrianov <andrianov@...ras.ru>
Cc:	linux-ide@...r.kernel.org, linux-kernel@...r.kernel.org,
	ldv-project@...uxtesting.org,
	Vaishali Thakkar <vaishali.thakkar@...cle.com>
Subject: Re: Potential race condition in drivers/ata/sata_mv.ko

On Thu, Aug 11, 2016 at 05:18:31PM +0300, Pavel Andrianov wrote:
> Hi!
> 
> I have found such example:
> 
> ... ->
> 
> ata_exec_internal_sg ->
> 
>     ata_qc_issue ->
> 
>         mv_qc_issue ->
> 
>             mv_clear_and_enable_port_irqs ->
> 
>                 mv_enable_port_irqs ->
> 
>                     mv_set_main_irq_mask
> 
> 
> ata_exec_internal_sg acquires spin_lock(ap->lock) and call of the last
> function mv_set_main_irq_mask is with this lock. mv_interrupt acquires
> spin_lock(host->lock) before call of the same function. I am not sure is it
> correct to add one more spin_lock or move a call of request_irq in
> ata_host_activate, thus I can not easily fix the issue.

ap->lock and host->lock point to the the same lock.  The only reason
ap->lock is a pointer is for SAS.

Thanks.

-- 
tejun