On Sat, Oct 13, 2018 at 11:26 AM tedheadster <tedheadster@xxxxxxxxx> wrote: > > > it seems like we do for some reason never actually enable swiotlb > > for 32-bit x86. Before my commit the block bounce buffering papered > > over that for networking, Please try this patch: > > > > diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c > > index 661583662430..71c0b01d93b1 100644 > > --- a/arch/x86/kernel/pci-swiotlb.c > > +++ b/arch/x86/kernel/pci-swiotlb.c > > @@ -42,10 +42,8 @@ IOMMU_INIT_FINISH(pci_swiotlb_detect_override, > > int __init pci_swiotlb_detect_4gb(void) > > { > > /* don't initialize swiotlb if iommu=off (no_iommu=1) */ > > -#ifdef CONFIG_X86_64 > > if (!no_iommu && max_possible_pfn > MAX_DMA32_PFN) > > swiotlb = 1; > > -#endif > > > > /* > > * If SME is active then swiotlb will be set to 1 so that bounce > > Christoph, > your patch fixed it nicely. No more error messages when I boot with > 16GiB enabled on a 32-bit PAE-enabled system. Christoph, can you please submit this patch to be backported to the long-term kernels (4.14, 4.9, 4.4, 3.18, and 3.16)? - Matthew