Re: [PATCH 0/7] NULL pointer deref fix for stm32-dma

From: Joel Fernandes
Date: Tue Oct 16 2018 - 19:49:36 EST


On Tue, Oct 16, 2018 at 06:11:29PM +0200, Greg KH wrote:
> On Mon, Oct 08, 2018 at 10:47:45PM -0700, Joel Fernandes (Google) wrote:
> > Hi Greg,
> >
> > While looking at android-4.14, I found a NULL pointer deref with
> > stm32-dma driver using Coccicheck errors. I found that upstream had a
> > bunch of patches on stm32-dma that have fixed this and other issues, I
> > applied these patches cleanly onto Android 4.14. I believe these should
> > goto stable and flow into Android 4.14 from there, but I haven't tested
> > this since I have no hardware to do so.
> >
> > Atleast I can say that the coccicheck error below goes away when running:
> > make coccicheck MODE=report
> > ./drivers/dma/stm32-dma.c:567:18-24: ERROR: chan -> desc is NULL but dereferenced.
> >
> > Anyway, please consider this series for 4.14 stable, I have CC'd the
> > author and others, thanks.
> >
> > Pierre Yves MORDRET (7):
> > dmaengine: stm32-dma: threshold manages with bitfield feature
> > dmaengine: stm32-dma: fix incomplete configuration in cyclic mode
> > dmaengine: stm32-dma: fix typo and reported checkpatch warnings
> > dmaengine: stm32-dma: Improve memory burst management
> > dmaengine: stm32-dma: fix DMA IRQ status handling
> > dmaengine: stm32-dma: fix max items per transfer
> > dmaengine: stm32-dma: properly mask irq bits
> >
> > drivers/dma/stm32-dma.c | 287 +++++++++++++++++++++++++++++++++-------
> > 1 file changed, 240 insertions(+), 47 deletions(-)
>
>
> I need git commit ids for all of these for what the patch in mainline
> is.
>
> Can you fix that up and resend them?
>

Vinod raised a concern in the other thread that he does not want all of the
"feature patches" that I posted in these series into -stable, since not
everything is a fix.

So what I'll do is I'll only repost:
dmaengine: stm32-dma: fix incomplete configuration in cyclic mode

which fixes the null pointer issue that raised the coccinelle error I found
and reported.

I'll let Vinod/Pierre decide on posting the other fixes to stable since they
may have more background on whether something is a fix or not.

thanks,

- Joel