On Wed, Dec 18, 2024 at 06:13:35PM +0800, Markus Rechberger wrote: > On Mon, 2024-11-18 at 22:11 -0500, Alan Stern wrote: > > On Mon, Nov 18, 2024 at 04:13:29PM +0800, Markus Rechberger wrote: > > > Hi, > > > > > > we repeatedly received issues from customers with usb autosuspend, > > > our > > > devices are randomly disconnected on some systems, it did not > > > happen > > > many times but if more than 4-5 customers report this there's > > > usually > > > some real issue behind that. > > > > > > I'm asking here because the Libreelec distribution recently got my > > > attention that they seem to set usb autosuspend to -1 during boot > > > time > > > so they probably figured that out much earlier with other devices. > > > I have never experienced that issue on an Ivy or Haswell based > > > system. > > > > > > Were there some issues with particular kernel versions in the past > > > or > > > are there any issues known in general with USB autosuspend? > > > > I'm not aware of any such issues. > > > > Do you think you can get kernel debugging logs from your customers > > showing examples of these disconnections? > > > > one customer just replied his problem is gone since he disabled auto > suspend (that was the latest one). He tested it since the last mail to > the mailing list. > > Do you want any particular logfiles / logging from him? The only interesting logs would be ones showing an example of a disconnect during autosuspend. If your customer has disabled autosuspend then there won't be any examples like this. But maybe there are some old kernel logs that do show the problem? Regardless, disabling autosuspend is a perfectly good way of dealing with these problems (unless you're in a very power-constrained setting, such as running off a battery). Alan Stern