Alexey Kardashevskiy wrote: > > > On 2/9/25 09:49, Alexey Kardashevskiy wrote: > > > > > > On 30/8/25 12:37, dan.j.williams@xxxxxxxxx wrote: > >> Alexey Kardashevskiy wrote: > >> [..] > >>>>> We have pdev in pci_tdi, pci_tsm and pci_tsm_pf0 (via .base), using > >>>>> these in pci_tsm_ops will document better which call is allowed on > >>>>> what entity - DSM or TDI. Or may be ditch those back "pdev" > >>>>> references? > >>>> > >>>> Not immediately understanding what change you want here. Do you want > >>>> iommufd to track the pci_tdi? > >>> > >>> I'd like to either: > >>> > >>> - get rid of pdev back refs in pci_tsm/pci_tdi since we pass pci_dev > >>> everywhere as if a pdev from pci_tsm/pci_tdi is used in, say, 1-2 > >>> places, then it is just cleaner to pass pdev to those places > >>> explicitly > >> > >> Maybe if we see that that are unused then they are easy to delete later. > > > > It is way easier to do now than later when it grows. I'll dig a bit. > > So far it appears so that the only use for these backrefs is > pci_tsm_ops's hooks which take pci_tsm/pci_tdi instead of pci_dev. So > the backrefs are only needed because unbind()/remove() do not take > pci_dev. > > My problem with these backrefs is that for a new reader of the code > it won't be immediately obvious whether we need > pci_dev_get/pci_dev_put for those, are pci_tsm/pci_tdi ever detached > from pci_dev, etc. Dunno, I won't be nak-ing of this though. Thanks, Why would the new reader audit that the core is taking references on the back pointers it provides? The to_pci_tsm_pf0() object casting path has safety checks based on type which can be inferred by walking the backref.
