> From: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx> > Sent: Tuesday, March 28, 2023 11:40 PM > > Hi Kevin, > > On Tue, 28 Mar 2023 07:47:45 +0000, "Tian, Kevin" <kevin.tian@xxxxxxxxx> > wrote: > > > > From: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx> > > > Sent: Tuesday, March 28, 2023 7:22 AM > > > > > > Devices that use ENQCMDS to submit work on buffers mapped by DMA > API > > > must attach a PASID to the default domain of the device. In preparation > > > for this use case, this patch implements set_dev_pasid() for the > > > default_domain_ops. Besides PASID attachment, device will also be > > > attached to the domain as the result of this call if the device has not > > > been attached before. > > > > > > > I didn't get the last point. PASID attach should only have the scope > > for the pasid. RID of the device might be attached to another domain > > which shouldn't be changed by this call. > I meant if the RID context has not been set up before attaching this PASID, > this call will also set up the context, PASID dir etc. In the end, we > eliminated ordering requirement of attaching device, RID_PASID first, then > other PASIDs. > How about: > "If the device context has not been set up prior to this call, this will > set up the device context in addition to PASID attachment." > this is clearer.
