linux-kernel - Re: [PATCH 00/46] Use dmaenginem_async_device

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: <20180803084814.GA30216@hsj-Precision-5520>
Date:   Fri, 3 Aug 2018 16:48:14 +0800
From:   Huang Shijie <sjhuang@...vatar.ai>
To:     Peter Ujfalusi <peter.ujfalusi@...com>
Cc:     vkoul@...nel.org, dmaengine@...r.kernel.org,
        linux-kernel@...r.kernel.org, dave.jiang@...el.com,
        radhey.shyam.pandey@...inx.com, appana.durga.rao@...inx.com,
        jmkrzyszt@...il.com, gomonovych@...il.com, keescook@...omium.org,
        horms+renesas@...ge.net.au, geert+renesas@...der.be,
        shawnguo@...nel.org, baoyou.xie@...aro.org,
        michal.simek@...inx.com, baohua@...nel.org,
        ludovic.desroches@...rochip.com, linus.walleij@...aro.org,
        david.brown@...aro.org
Subject: Re: [PATCH 00/46] Use dmaenginem_async_device_register to simplify
 code

On Fri, Aug 03, 2018 at 11:20:49AM +0300, Peter Ujfalusi wrote:
> Hi,
> 
> On 2018-08-03 10:19, Huang Shijie wrote:
> > All the patches are using dmaenginem_async_device_register to simplify code
> > except the last one:
> >   dmaengine: add COMPILE_TEST for the drivers
> > 
> > I use the last one to do the compiler test.  
> > There are still 20 drivers which do not use the dmaenginem_async_device_register.
> > Let me take a rest, if this patch set is accepted, I will do the rest.
> 
> I think for most of the drivers this series is going to open a race
> which is essentially:
> 
> prior:
> _remove()
> {
> ...
> 	dma_async_device_register()
> 	/* Free resources, disable HW, etc */
> 	return 0;
> }
> 
> after:
> _remove()
> {
> ...
> 	/* Free resources, disable HW, etc */
> 	/*
> 	 * The dma device is still registered and a channel can be
> 	 * requested
> 	 */
> 
> 	dma_async_device_register()
> 	return 0;
> }
> 
> It might be theoretical, but conversion to managed device resources is
> not straight forward in some cases. 
Yes. 

okay, let me think again..

@Vinod, please just ignore this patch set.

Thanks
Huang Shijie