Re: [PATCH 2/3] dmaengine: xilinx: xdma: Fix synchronization issue

From: Miquel Raynal <miquel.raynal@bootlin.com>
To: Lizhi Hou <lizhi.hou@amd.com>
Cc: Louis Chauvet <louis.chauvet@bootlin.com>,
	Brian Xu <brian.xu@amd.com>,
	Raj Kumar Rampelli <raj.kumar.rampelli@amd.com>,
	Vinod Koul <vkoul@kernel.org>,
	Michal Simek <michal.simek@amd.com>,
	Jan Kuliga <jankul@alatek.krakow.pl>, <dmaengine@vger.kernel.org>,
	<linux-arm-kernel@lists.infradead.org>,
	<linux-kernel@vger.kernel.org>, <stable@vger.kernel.org>
Subject: Re: [PATCH 2/3] dmaengine: xilinx: xdma: Fix synchronization issue
Date: Thu, 28 Mar 2024 01:23:13 +0100	[thread overview]
Message-ID: <20240328012257.4a5955f2@xps-13> (raw)
In-Reply-To: <b59dd8cd-fd75-5342-d411-817f33e0ff48@amd.com>

Hi Lizhi,

> > @@ -376,6 +378,8 @@ static int xdma_xfer_start(struct xdma_chan *xchan)
> >   		return ret;  
> >   >   	xchan->busy = true;  
> > +	xchan->stop_requested = false;
> > +	reinit_completion(&xchan->last_interrupt);  
> 
> If stop_requested is true, it should not start another transfer. So I would suggest to add
> 
>       if (xchan->stop_requested)
> 
>                  return -ENODEV;

Maybe -EBUSY in this case?

I thought synchronize() was mandatory in-between. If that's not the
case then indeed we need to block or error-out if a new transfer
gets started.

> 
> at the beginning of xdma_xfer_start().
> 
> xdma_xfer_start() is protected by chan lock.
> 
> >   >   	return 0;  
> >   }

Thanks,
Miquèl