On Mon, Apr 06, 2020 at 06:09:27AM -0700, Christoph Hellwig wrote:> Pleae just drop the code - we should not add new drivers with custom > DMA ops.I'm not sure there's another way to do this: this not a driver, it's a software emulator that pretends to be an actual device. We can't have the platform supply DMA ops here since the platform is a regular x86 or whatever. -- MST
On 2020/4/6 ??10:02, Michael S. Tsirkin wrote:> On Mon, Apr 06, 2020 at 06:09:27AM -0700, Christoph Hellwig wrote: >> Pleae just drop the code - we should not add new drivers with custom >> DMA ops. > I'm not sure there's another way to do this: this not a driver, it's a > software emulator that pretends to be an actual device. We can't > have the platform supply DMA ops here since the platform is a regular > x86 or whatever.Yes, actually vdpa-sim does not depends on HAS_DMA. Maybe we can do shortcut like dev->dma_ops = vdpasim_dma_ops; Thanks>
On Mon, Apr 06, 2020 at 10:12:56PM +0800, Jason Wang wrote:> > On 2020/4/6 ??10:02, Michael S. Tsirkin wrote: > > On Mon, Apr 06, 2020 at 06:09:27AM -0700, Christoph Hellwig wrote: > > > Pleae just drop the code - we should not add new drivers with custom > > > DMA ops. > > I'm not sure there's another way to do this: this not a driver, it's a > > software emulator that pretends to be an actual device. We can't > > have the platform supply DMA ops here since the platform is a regular > > x86 or whatever. > > > Yes, actually vdpa-sim does not depends on HAS_DMA. Maybe we can do shortcut > like dev->dma_ops = vdpasim_dma_ops; > > Thanks >We can do that - does it them actually work - and not just compile? -- MST