Re: [PATCH] spi: Assign dummy scatterlist to unidirectional transfers

[Nícolas F. R. A. Prado <nfraprado@xxxxxxxxxxxxx>]

On Tue, Jul 23, 2024 at 12:05:52AM +0800, Yongqin Liu wrote:
> On Thu, 30 May 2024 at 05:04, Andy Shevchenko <andy.shevchenko@xxxxxxxxx> wrote:
> >
> > Wed, May 29, 2024 at 11:42:35AM -0400, Nícolas F. R. A. Prado kirjoitti:
> > > From: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
> > >
> > > Commit 8cc3bad9d9d6 ("spi: Remove unneded check for orig_nents")
> > > introduced a regression: unmapped data could now be passed to the DMA
> > > APIs, resulting in null pointer dereferences. Commit 9f788ba457b4 ("spi:
> > > Don't mark message DMA mapped when no transfer in it is") and commit
> > > da560097c056 ("spi: Check if transfer is mapped before calling DMA sync
> > > APIs") addressed the problem, but only partially. Unidirectional
> > > transactions will still result in null pointer dereference. To prevent
> > > that from happening, assign a dummy scatterlist when no data is mapped,
> > > so that the DMA API can be called and not result in a null pointer
> > > dereference.
> >
> > I feel that with this the da560097c056 ("spi: Check if transfer is mapped
> > before calling DMA sync APIs") can be reverted as unneeded. Nícolas, can
> > you check that? If it works, we better revert the unneeded checks.
> 
> FYI, just tested based on the Android Common Kernel android-mainline branch,
> with only the following two changes, the issue is not reported too:
>     9dedabe95b49 spi: Assign dummy scatterlist to unidirectional transfers
>     9f788ba457b4 spi: Don't mark message DMA mapped when no transfer in it is

Hi Yongqin,

Simply reverting commit da560097c056 ("spi: Check if transfer is mapped before
calling DMA sync APIs") caused issues on the sc7180-limozeen platform as I
mentioned in
https://lore.kernel.org/all/1ea41944-a107-4528-8e8d-559c06907e3f@notapiano/.

Instead of that, Andy landed this commit reworking the flag, which got rid of
that check anyway and provided a cleaner solution:
https://lore.kernel.org/all/20240531194723.1761567-9-andriy.shevchenko@xxxxxxxxxxxxxxx/

Thanks,
Nícolas