Re: [PATCH v2] s390/pci: Fix dev.dma_range_map missing sentinel element

[Vasily Gorbik]

On Thu, Mar 27, 2025 at 10:45:40AM +0100, Niklas Schnelle wrote:
> On Wed, 2025-03-12 at 11:32 +0100, Niklas Schnelle wrote:
> > The fixed commit sets up dev.dma_range_map but missed that this is
> > supposed to be an array of struct bus_dma_region with a sentinel element
> > with the size field set to 0 at the end. This would lead to overruns in
> > e.g. dma_range_map_min(). It could also result in wrong translations
> > instead of DMA_MAPPING_ERROR in translate_phys_to_dma() if the paddr
> > were to not fit in the aperture. Fix this by using the
> > dma_direct_set_offset() helper which creates a sentinel for us.
> > 
> > Fixes: d236843a6964 ("s390/pci: store DMA offset in bus_dma_region")
> > Reviewed-by: Matthew Rosato <mjrosato@xxxxxxxxxxxxx>
> > Signed-off-by: Niklas Schnelle <schnelle@xxxxxxxxxxxxx>
> > ---
> > Note: Based on iommu/next
> > 
> > v1 -> v2:
> > - Fixed typo, added trailers
> > - Link to v1: https://lore.kernel.org/r/20250306-fix_dma_map_alloc-v1-1-b4fa44304eac@xxxxxxxxxxxxx
> > ---
> >  arch/s390/pci/pci_bus.c | 24 +++++++++++-------------
> >  1 file changed, 11 insertions(+), 13 deletions(-)
> > 
> 
> With the IOMMU pull request for Linux v6.15 the fixed commit is now in
> Linus tree (with the SHA as cited) but this fix isn't. Since this fix
> only touches s390 specific code I think it could go through the s390
> tree but that currently doesn't carry the fixed commit so would have to
> wait until the s390 tree pulls in v6.15-rc1. 
> 
> I don't think any breakage from the missing sentinel is visible outside
> of guests using the new IOMMU passthrough mode so I guess waiting until
> after v6.15-rc1 would be okay.
> 
> Heiko, Vasily, Alexander what do you think?

Sure, I'll pick it up in the s390 tree for rc1 or rc2, depending on
whether there will be a second s390 tree pull request during the merge
window.