-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 Hello Paul,
On Sat, 2016-11-26 at 17:28 +0100, Paul Menzel wrote: > > > I've had similar issues on my box. I worked with the USB folks but really > > couldn't come up with a convincing resolution. > > Is that discussion public? Could you please give me the thread subject > or URL? > Yes. Part of it we root caused. But not what was bothering me. The conclusion was that I have a broken device. I'm not convinced, given that under Windows it works perfect, but then I don't have the resource to challenge the statement. You can access the thread at: https://www.mail-archive.com/[email protected]/msg79514.html It should lead you to the rest of the thread. > > Nov 25 18:46:54 learner kernel: usb 2-4: Product: USB2.0-CRW > > Nov 25 18:46:54 learner kernel: usb 2-4: Manufacturer: Generic > > Are you able to reproduce the crash? > I guess you mean the erratic USB disconnects. Yes. And once they trigger, the erratic USB device vanishes from the `lsusb` listing. > > But what do you mean by "crashed" ? The particular USB port becomes > > inaccessible. > > Indeed, it does not crash the Linux kernel, but the USB port is not > usable anymore. > Yes. IN my case, it is not the port which is reflecting the disconnects, but rather one of the USB devices (SDCard Reader). But I do very very occasionally get other devices also to reset, so it could be a much more severe problem, yet to root cause. > > The kernel still works, maybe inefficient, but it does not crash > > the kernel. > > Indeed. What do you mean by inefficient? > Well. The USB device resets very frequently, like every 10 seconds. This overall has the side-effect of drawing too much power on my laptop. > I noticed, that `uptime`, `top`, or `htop` display a high usage. For > each `lsusb` I started (for debugging) it looks like one CPU core gets > used 100 %? > > ``` > $ uptime > 17:17:37 up x days, 9:34, x users, load average: 7,29, 7,64, 7,82 > ``` > I haven't monitored this aspect, but it is possible that this too may be happening, though for a small amount of time. > I can find no processes in `htop` or `top` using that much CPU. I have > no idea how to find out what is causing this, but I’d guess it’s > something in the Linux kernel. > Yes. Because it is in the kernel. If it were related to I/O, you may have noticed per-BDI threads, but in this case, these are device resets and disconnects. So there's not much to capture in process listings. You could gather some additional information from udev, because your USB disconnects are equivalent to physical device plug/unplug, which triggers in the hotplug mechanism. At least, it may tell you how frequent your disconnect errors are happening. And if you have heavy (USB) rules processing, then it could be the cause of your sudden load increase. I hope this information helps you. - -- Ritesh Raj Sarraf | http://people.debian.org/~rrs Debian - The Universal Operating System -----BEGIN PGP SIGNATURE----- iQIzBAEBCgAdFiEEQCVDstmIVAB/Yn02pjpYo/LhdWkFAlg5wAQACgkQpjpYo/Lh dWkNeA/7BNaIkRnobzGPzLjKnbw5X5uEt/MHmRRakDsgDB6232elJC/8ON0BkGYv zYp4vGVkr+hyx6WSBoB/3bH4DijOuZmWJNN5YPrL/dBBGmWCDTeptmj9adSyO1C5 chX+wLMnqV07dz55GAkz2y1CiClUp/bxO1Qtx5eBJNod94i0KVWbrK4bYSP0vRuv 6R0JOVjOJHqKK0RpHmNqHKroFXVDsJ0hugCRVCz9ksrSPQ6NWrFn/cphmX3KNXLC kpEyhwIb3x4G/31OrTRL7HErb2CabIvdX6NCyNm2c6QA+7aLBIJxaos0nuH8hKQ0 MmS+KbCLtL600+zitIO16av7Ercd3P49PTcnqjGWiUaASD2uYcdvl6UiIn45b2dL aAWf3ZeuFRKt6coFl0bgvDoMoJ1hOqS3daY3itlnkXhRNb7+gXXBqw3rat3RpQOz oFaKgExJqvvzYPwmQyAuwopsWezsTTEJCDfBXNsbM7y77uAUzoeuRNry16dbk7gF BI0F5YB2BbJYHoNzyhzFkEKEEjEud9YX6SpnzeuIuWwx7zqvr24PRnQfZm0N5Db2 61ycsFOZDYxgoG2928qo/TE6jZIfqtZ3S25FSs1qSVZYiy/oDSguU85bzSBSue1o 2ZGDsCir8N4jtx1TOFVrsZP2k9cNqZwWO/wIKn7t1XozkxXPTPw= =1aXQ -----END PGP SIGNATURE-----
