Re: drivers/dma/xilinx/xilinx_dpdma.c:1172 xilinx_dpdma_synchronize() error: double unlocked 'chan->vchan.lock' (orig line 1170)

[Vinod Koul]

HI Laurent,

On 12-08-20, 01:50, Laurent Pinchart wrote:
> On Wed, Aug 12, 2020 at 06:37:59AM +0800, kernel test robot wrote:
> > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
> > head:   c636eef2ee3696f261a35f34989842701a107895
> > commit: 7cbb0c63de3fc218fd06ecfedb477772a4d12f76 dmaengine: xilinx: dpdma: Add the Xilinx DisplayPort DMA engine driver
> > date:   4 weeks ago
> > config: h8300-randconfig-m031-20200811 (attached as .config)
> > compiler: h8300-linux-gcc (GCC) 9.3.0
> > 
> > If you fix the issue, kindly add following tag as appropriate
> > Reported-by: kernel test robot <lkp@xxxxxxxxx>
> > 
> > smatch warnings:
> > drivers/dma/xilinx/xilinx_dpdma.c:1172 xilinx_dpdma_synchronize() error: double unlocked 'chan->vchan.lock' (orig line 1170)
> 
> Unless I'm mistaken, this is a false positive. I'd appreciate if someone
> could check if I'm missing something obvious.
> 
> > vim +1172 drivers/dma/xilinx/xilinx_dpdma.c
> > 
> >   1140	
> >   1141	/**
> >   1142	 * xilinx_dpdma_synchronize - Synchronize callback execution
> >   1143	 * @dchan: DMA channel
> >   1144	 *
> >   1145	 * Synchronizing callback execution ensures that all previously issued
> >   1146	 * transfers have completed and all associated callbacks have been called and
> >   1147	 * have returned.
> >   1148	 *
> >   1149	 * This function waits for the DMA channel to stop. It assumes it has been
> >   1150	 * paused by a previous call to dmaengine_terminate_async(), and that no new
> >   1151	 * pending descriptors have been issued with dma_async_issue_pending(). The
> >   1152	 * behaviour is undefined otherwise.
> >   1153	 */
> >   1154	static void xilinx_dpdma_synchronize(struct dma_chan *dchan)
> >   1155	{
> >   1156		struct xilinx_dpdma_chan *chan = to_xilinx_chan(dchan);
> >   1157		unsigned long flags;
> >   1158	
> >   1159		xilinx_dpdma_chan_stop(chan);
> >   1160	
> >   1161		spin_lock_irqsave(&chan->vchan.lock, flags);

Lock is held here

> >   1162		if (chan->desc.pending) {
> >   1163			vchan_terminate_vdesc(&chan->desc.pending->vdesc);

This should be called with lock held

> >   1164			chan->desc.pending = NULL;
> >   1165		}
> >   1166		if (chan->desc.active) {
> >   1167			vchan_terminate_vdesc(&chan->desc.active->vdesc);

same here

> >   1168			chan->desc.active = NULL;
> >   1169		}
> > > 1170		spin_unlock_irqrestore(&chan->vchan.lock, flags);

Dropped

> >   1171	
> > > 1172		vchan_synchronize(&chan->vchan);

This should be called without lock

So I agree with you, looks like a false positive to me as well

-- 
~Vinod