On Tue, Jan 19, 2016 at 4:24 AM, Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> wrote: > On Tue, Jan 19, 2016 at 12:47:32AM +0100, Markus Rechberger wrote: >> Who's going to unload the usb controller stack while a USBFS driver is active? > > yanking out a USB host controller is quite common on some systems (pci > hotplug, docking stations, pcmcia, thunderbolt usb controller, etc.) If > this code oopses with that configuration, we need to fix it. > I'm following this in detail, our customers are usually left alone with those reports and if it comes to thousands of them I see the common pattern that the head is just put into the ground and the issue will be left over. Example? Exactly the mmap issue which we are describing here! Customers were complaining over years already. We even had customers returning devices because of that. >> Practically this issue is absolutely irrelevant. > > To you, but not to us who have to maintain and debug and support it. > I can forward every single issue that comes up to you if you want and play translator? But I have little to no confidence that this will help anything, as the initial request was just ignored as well here. Alternatively you can look for some kernel oopses at http://support.sundtek.de and pay attention when they came up. Now those issues affect certain platforms more than other ones - (just as you mentioned, some controllers freak out .. with USB 3.0 likely 90%). I see no support from you or anyone else to be serious (look at some of my mails to the lkml in the past including backtrace) - and you have to agree here that you have failed with that as well. So your argument is pretty much void. >> I agree it should be fixed but no one is going to use it while the >> patch is just floating around on the Mailinglist for years. > > It has to be correct, for everyone, just like any other kernel feature > that any operating system adds, this isn't a new thing so please stop > complaining. We're practically affected of this and do business in that area for many years and heavily depend on this interface. If something does not work customers will complain immediately, the chance that they will find some support on various mailinglists is low. Why? First of all they mainly speak german, the level of linux knowledge varies between beginner to expert. But how many want or are able to recompile their kernel? That number is even lower. I don't want to fight with you but I seriously advice that this behaviour has to improve. I gain nothing from that but I just fear that everything sent to this mailinglist is just worthless. It should be in your interest to cooperate with people like me who have many users behind this around the whole world (even if you don't like to). Instead of replying and discussing why all this has failed so utterly (this patch has been out there for several years) put out a schedule when this patch will be added and everyone will be happy and there won't be any further discussion needed. Markus -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
