On Thu, Aug 12, 2021 at 09:56:14AM -0600, Alex Williamson wrote: > On Thu, 12 Aug 2021 09:26:17 +0200 > Christoph Hellwig <hch@xxxxxx> wrote: > > > On Wed, Aug 11, 2021 at 04:03:41PM -0600, Alex Williamson wrote: > > > > + dev_warn(dev, "Adding kernel taint for vfio-noiommu group on device\n"); > > > > + return vfio_noiommu_group_alloc(dev); > > > > > > Nit, we taint regardless of the success of this function, should we > > > move the tainting back into the function (using the flags to skip for > > > mdev in subsequent patches) or swap the order to check the return value > > > before tainting? Thanks, > > > > Does it really matter to have the extra thread if a memory allocation > > failed when going down this route? > > Extra thread? In practice this is unlikely to ever fail, but if we've > chosen the point at which we have a no-iommu group as where we taint, > then let's at least be consistent and not move that back to the point > where we tried to make a no-iommu group, regardless of whether it was > successful. Thanks, Sorry, the mental spell checker kicked in. Thread should have read taint instead. But if you don't want to tain in the failure case I'll need to refactor this a bit.
