On Thu, 1 Feb 2024, Maciej W. Rozycki wrote: > > > There's always an option to store it into pci_dev for later use when the > > > quirk is found to be working for the first time if other solutions don't > > > work. > > Indeed there is. On second thoughts there is not. The thing is if we handle a situation where the downstream device can be replaced while in suspend (which I infer from the discussion that we do), then it could be that one device will require the quirk while the other one will not. So any previous results have to be considered irrelevant and the whole dance with the quirk has to be repeated every time at resume or hot-plug. Maciej
