Hi Dan
On Tue, Jan 06, 2015 at 10:04:15PM +0300, Dan Carpenter wrote:
> Hello Maxime Ripard,
>
> This is a semi-automatic email about new static checker warnings.
>
> The patch bcd1b0b9015b: "dmaengine: pl08x: Split device_control" from
> Nov 17, 2014, leads to the following Smatch complaint:
>
> drivers/dma/amba-pl08x.c:1745 pl08x_pause()
> error: we previously assumed 'plchan->phychan' could be null (see line 1740)
>
> drivers/dma/amba-pl08x.c
> 1739 spin_lock_irqsave(&plchan->vc.lock, flags);
> 1740 if (!plchan->phychan && !plchan->at) {
> ^^^^^^^^^^^^^^^
> Old test. Maybe the && should be ||?
>
> 1741 spin_unlock_irqrestore(&plchan->vc.lock, flags);
> 1742 return 0;
> 1743 }
> 1744
> 1745 pl08x_pause_phy_chan(plchan->phychan);
> ^^^^^^^^^^^^^^^
> New dereference inside the call to pl08x_pause_phy_chan()
>
> 1746 plchan->state = PL08X_CHAN_PAUSED;
> 1747
>
> See also:
> drivers/dma/amba-pl08x.c:412 pl08x_start_next_txd() warn: this loop depends on readl() succeeding
> (readl() return (uint)-1 on failure/hotplug).
> drivers/dma/amba-pl08x.c:1359 pl08x_burst() error: buffer overflow 'burst_sizes' 8 <= 8
> drivers/dma/amba-pl08x.c:1745 pl08x_pause() error: we previously assumed 'plchan->phychan' could be null (see line 1740)
> drivers/dma/amba-pl08x.c:1768 pl08x_resume() error: we previously assumed 'plchan->phychan' could be null (see line 1763)
I don't really know what to do here, I just moved things around in
this driver. Linus?
Maxime
--
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com
Attachment:
signature.asc
Description: Digital signature
