On Tue, 19 Nov 2019, I wrote: > > But this is a theoretical problem so far. You may need to use sg_utils > to generate a SCSI command like that. > > (For a DMA controller subject to, say, 24-bit boundaries, there are > additional ways to end up with short transfers.) > On reflection these probably can't happen in practice. I was being over-cautious. So I'll leave this patch out-of-tree until there's a need for it (i.e. some driver that uses both DMA and PIO in data phases). --
