Ben, does this resolve your issues with the confusing zone selection?
On Mon, Oct 01, 2018 at 01:10:16PM -0700, Christoph Hellwig wrote: > What we are doing here isn't quite obvious, so add a comment explaining > it. > > Signed-off-by: Christoph Hellwig <[email protected]> > --- > kernel/dma/direct.c | 9 ++++++++- > 1 file changed, 8 insertions(+), 1 deletion(-) > > diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c > index ba6f5956a291..14b966e2349a 100644 > --- a/kernel/dma/direct.c > +++ b/kernel/dma/direct.c > @@ -84,7 +84,14 @@ static gfp_t __dma_direct_optimal_gfp_mask(struct device > *dev, u64 dma_mask, > else > *phys_mask = dma_to_phys(dev, dma_mask); > > - /* GFP_DMA32 and GFP_DMA are no ops without the corresponding zones: */ > + /* > + * Optimistically try the zone that the physicall address mask falls > + * into first. If that returns memory that isn't actually addressable > + * we will fallback to the next lower zone and try again. > + * > + * Note that GFP_DMA32 and GFP_DMA are no ops without the corresponding > + * zones. > + */ > if (*phys_mask <= DMA_BIT_MASK(ARCH_ZONE_DMA_BITS)) > return GFP_DMA; > if (*phys_mask <= DMA_BIT_MASK(32)) > -- > 2.19.0 > > _______________________________________________ > iommu mailing list > [email protected] > https://lists.linuxfoundation.org/mailman/listinfo/iommu ---end quoted text--- _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
