On Wed, Apr 7, 2021 at 5:51 AM Greg KH <greg@xxxxxxxxx> wrote: > > On Mon, Apr 05, 2021 at 09:02:22PM +0000, Jianxiong Gao wrote: > > Hi all, > > > > This series of backports fixes the SWIOTLB library to maintain the > > page offset when mapping a DMA address. The bug that motivated this > > patch series manifested when running a 5.4 kernel as a SEV guest with > > an NVMe device. However, any device that infers information from the > > page offset and is accessed through the SWIOTLB will benefit from this > > bug fix. > > But this is 5.10, not 5.4, why mention 5.4 here? Oops. The cover letter shouldn't mention the kernel version. The bug is present in both 5.4 and 5.10. Sorry for the confusion.> > And you are backporting a 5.12-rc feature to 5.10, what happened to > 5.11? No. The goal is to backport a bug fix to the LTS releases. > Why not just use 5.12 to get this new feature instead of using an older > kernel? It's not like this has ever worked before, right? > It's true, that a new feature (SEV virtualization) is what motivated the bug fix. However, I still think this makes sense to backport to the LTS releases because it does fix a pre-existing bug that may be impacting pre-existing setups. In particular, while working on these patches, I got the following feedback: "There are plenty of other hardware designs that rely on dma mapping not adding offsets that did not exist, e.g. ahci and various RDMA NICs." [1] https://lkml.org/lkml/2020/11/24/520 > thanks, > > greg k-h -- Jianxiong Gao
