On Thu, Jul 22, 2021 at 12:17 PM Bjorn Andersson <bjorn.andersson@xxxxxxxxxx> wrote: > > On Jul 21, 2021, 1:45 PM +0200, Krzysztof Kozlowski wrote: > > > I had impression that kernel defines interfaces which should be used and > > > are stable (e.g. syscalls, sysfs and so on). This case is example of > > > user-space relying on something not being marked as part of ABI. Instead > > > they found something working for them and now it is being used in "we > > > cannot break existing systems". Basically, AOSP unilaterally created a > > > stable ABI and now kernel has to stick to it. > > > > > > Really, all normal systems depend on aliases or names and here we have > > > dependency on device address. I proposed way how AOSP should be fixed. > > > Anything happened? Nope. > > > > First time he sent a possible solution for the problem: > > https://lore.kernel.org/lkml/20201221210423.GA2504@kozik-lap/ > > > > To sum up you could have used one of the more portable approaches > > 1. add an udc alias to the controller and use it then to refer to the > > corresponding USB controller > > Is there such a thing as "UDC alias"? Or are you suggesting that we > should add such feature? > > I think it would be wonderful if we could identify the UDCs on our > boards as "USB1" and "USB2", or "the one and only USB-C connector". But > unless that will fall back to the existing naming it would break John's > _existing_ userspace. Well, I'd not hold up the existing userspace I'm using as sacrosanct (AOSP devices still usually don't have to work cross-kernel versions - devboards being the main exception). I'm fine if we can rework userland as proposed, so that the issues can be avoided, but I honestly don't have enough context to really understand what that looks like (no idea what udc aliases are). And whatever we do, the main constraint is that userland has to be able to work with existing LTS kernels and newer kernels. > > 2. search through DT-nodes looking for a specific compatible/reg > > DT-properties. > > > > We could define that this is the way, but again we didn't yesterday so > your proposal is not backwards compatible. It may be backwards compatible, but I'm still not clear exactly how it would work. I guess if we look through all sys/bus/platform/devices/*/of_node/compatbile strings for the desired "snps,dwc3", then find which of the directories have the desired address in the string? (The suggestion for looking at reg seems better, but I don't get a char value out of the dwc3 of_node/reg file). It seems much more straightforward to do an `ls /sys/class/udc/$GADGET_ADDR.*` But again, if folks decide the names can be rearranged to usb.<addr> in the future, that would break too. thanks -john
