https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c1 --- Comment #1 from Stefan Neis 2008-07-01 09:47:16 MDT --- With some help from the USB mailing list (particularly Alan Stern), we collected a bit more information. An "strace" of our program shows this:

ioctl(3, USBDEVFS_SUBMITURB, 0xbf8566d4) = 0 ioctl(3, USBDEVFS_REAPURBNDELAY, 0xbf856718) = -1 EAGAIN (Resource temporarily unavailable) select(4, NULL, [3], NULL, {0, 1000}) = 0 (Timeout) write(1, "test\n", 5test) = 5 gettimeofday({1214911004, 481477}, NULL) = 0 ioctl(3, USBDEVFS_REAPURBNDELAY, 0xbf856718) = -1 EAGAIN (Resource temporarily unavailable) select(4, NULL, [3], NULL, {0, 1000}) = 0 (Timeout) write(1, "test\n", 5test) = 5 gettimeofday({1214911004, 485465}, NULL) = 0 ioctl(3, USBDEVFS_REAPURBNDELAY, 0xbf856718) = -1 EAGAIN (Resource temporarily unavailable) select(4, NULL, [3], NULL, {0, 1000}

i.e. it's hanging in a select call, which Alan commented like this:

It's clear now that the problem exists in the kernel, not in your program on in libusb. To make any more progress you will have to patch the kernel source. The function you want to investigate first is usbdev_poll() in drivers/usb/core/devio.c.

We didn't start that yet, however, we tried a different kernel version and found that we have the same problem with 2.6.25.9 from kernel.org, so it seems (IMHO) to be a regression introduced in 2.6.25 (since a test on an Ubuntu with 2.6.24.whatever showed no problem) and not yet fixed in the current version. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c2 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED --- Comment #2 from Oliver Neukum 2008-07-02 01:26:03 MDT --- 2.6.25 has no change in usbdev_poll() or another obvious change in that regard. This will need debugging patches. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c4 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #4 from Oliver Neukum 2008-07-02 02:01:23 MDT --- Please apply the patch from #3 Then run your test programm while you look at your kernel messages. Please report the last "usbdev_poll - Point" message you see. Secondly, does the caps lock led switch when you hit the caps lock key? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c6 --- Comment #6 from Oliver Neukum 2008-07-02 08:51:08 MDT --- Created an attachment (id=225579) --> (https://bugzilla.novell.com/attachment.cgi?id=225579) additional debugging checkpoints in the internals of sys_select -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c8 Stefan Neis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|stefan.neis@kobil.com | --- Comment #8 from Stefan Neis 2008-07-03 03:01:01 MDT --- Now, the last line is "Jul 2 17:32:07 pc79 kernel: __pollwait - Point 4" BTW, it seems to work better and better the more the kernel is slowed down by all this additional output. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c10 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #10 from Oliver Neukum 2008-07-03 08:22:43 MDT --- Well, I don't see how this can hang. But in #9 is a patch for further debugging output. Please revert the patch in #3 so your kernel isn't slowed down too much. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c12 --- Comment #12 from Oliver Neukum 2008-07-04 04:59:39 MDT --- Created an attachment (id=226018) --> (https://bugzilla.novell.com/attachment.cgi?id=226018) debugging output grasping straws -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c14 Stefan Neis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|stefan.neis@kobil.com | --- Comment #14 from Stefan Neis 2008-07-04 06:39:09 MDT --- Now, I only get the "Point 2" output, seems like the "mb()" call did change at least something... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c16 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #16 from Oliver Neukum 2008-07-04 11:15:14 MDT --- Only with #12 or with #9, too? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c18 Stefan Neis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|stefan.neis@kobil.com | --- Comment #18 from Stefan Neis 2008-07-07 03:58:58 MDT --- It doesn't get any better, it's getting sometimes to point 2 and sometimes to point 3 before hanging - with both versions of the patch. :-( I wonder how multi-threaded the kernel actually is and whether we possibly look at the output from a wrong thread, since all this output from select is essentially generated by everything calling select, not only by our call to some function handling an USB request? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c20 --- Comment #20 from Stefan Neis 2008-07-11 05:38:22 MDT --- Well, at least that select is the only problematic thing that we've encountered so far. That doesn't necessarily prove anything, of course. What pointer are you referring to and how do I get the stack pointer? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c22 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #22 from Oliver Neukum 2008-07-11 06:46:17 MDT --- Please try the patch from #21. It'll show the pointer and prints out the pid of the caller so it should allow us to verify the task assumed to crash indeed does crash. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c24 --- Comment #24 from Stefan Neis 2008-07-11 08:36:30 MDT --- I got it to compile by including linux/sched.h just before the init_waitqueue_entry function, but I can't make sense of the output - if "p->pid" is really supposed to be the PID of the process (i.e. like exactly the same number that you get e.g. from "ps" output, then our process apparently doesn't generate any of the output that's logged... I.e. our test program that's causing the crash has PId=3003, but the last lines of output are all about something with p->pid being 1388 (and when going back through the logfile, various other processes are mentioned but no trace of 3003). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c26 --- Comment #26 from Oliver Neukum 2008-07-14 02:52:25 MDT --- This is simply wrong. LOG_KERN is a facility not level (priority) and has to go to the openlog() function, not to syslog(). To log on the console (stderr), use LOG_PERROR. LOG_CONS is used only when the normal write to /dev/log fails (e.g. syslog daemon not running). openlog("libusb", LOG_CONS | LOG_PID | LOG_PERROR, LOG_KERN); syslog(LOG_CRIT, "..."); -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c28 --- Comment #28 from Stefan Neis 2008-07-14 06:50:36 MDT --- Well, the man-page said something about openlog being optional and syslog taking an "OR" of facility and level values as first argument. Anyway, after an additional sleep(0) right after the syslog, I now at least get to see the syslog output on screen, however, there's still nothing of libusb in /var/log/messages - looks like the kernel causes that hang, before syslog (or the file system?) has a chance to actually write the message? Which makes me wonder how useful those printk's in the kernel code actually are ... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c30 --- Comment #30 from Stefan Neis 2008-07-14 09:12:29 MDT ---

printk will always hit the console if the log level is high enough

That was the hint I needed. After some googling, I found how to see the kernel messages on the console (instead of looking into /var/log/messages after rebooting) - and now the last line of log-output is: ehci_hcd 0000:00:0c.2: HC died; cleaning up which at least finally does sound like it's USB-related... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c32 Stefan Neis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|stefan.neis@kobil.com | --- Comment #32 from Stefan Neis 2008-07-15 06:05:06 MDT --- Yes, after unloading ehci_hcd, not only the simplified test program, but even the real application works - although, (of course) very slow. BTW, we've collected some more data, using another machine as well: On the first machine, we get the initially described complete hang with kernel 2.6.25.9. With kernel 2.6.26, the USB-2.0 extension card apparently isn't even found, nothing that's connected to it, will work (but that way, we at least don't get that nasty hang). On the second machine, the behaviour is a bit more friendly: With 2.6.25.9 the application hangs (and can't even be killed), but the machine isn't completely dead, i.e. you can switch to a different virtual console, but typically that console will hang as well as soon as you try to execute a command. With 2.6.26, ehci_hcd still says "HC died; cleaning up" and our application dies, but the machine fully recovers from that problem, i.e. you can at least continue to work with other applications. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c34 --- Comment #34 from Stefan Neis 2008-07-15 07:59:04 MDT --- Created an attachment (id=227869) --> (https://bugzilla.novell.com/attachment.cgi?id=227869) Interesting part of /var/log/messages with kernel 2.6.26 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c36 --- Comment #36 from Oliver Neukum 2008-07-16 05:21:45 MDT --- Jul 15 15:37:37 linux-t2fx kernel: ehci_hcd 0000:00:1a.7: HC died; cleaning up This does not contain any information about why this happened. Please compile your test kernel with CONFIG_USB_DEBUG and provide dmesg. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c38 --- Comment #38 from Oliver Neukum 2008-07-18 02:37:46 MDT --- You have to reinstall usbcore. Please verify that you actually load the new modules. You may have to rebuild your initrd. That happened to many people. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c40 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #40 from Oliver Neukum 2008-07-20 23:48:29 MDT --- According to this log you keep plugging and unplugging a device. Do you? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c41 --- Comment #41 from Stefan Neis 2008-07-21 06:05:58 MDT --- Well, I apparently didn't remove enough of the irrelevant output at the beginning of the file I attached, so yes, according to the time stamps I apparently plug and unplug the device at 18:14:18 and 18:15:03 and a second time at 18:15:13 and 18:15:49 and only then, there's finally the interesting case. I plug the device at 18:16:05, then detach the HID driver to access the device via libusb at 18:16:34 and try to send commands via libusb, resulting in ehci_hcd dying immediately(see this message: Jul 18 18:16:34 linux-t2fx kernel: ehci_hcd 0000:00:1a.7: HC died; cleaning up ) And that's the place, were I'm _not_ unplugging the device. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c43 --- Comment #43 from Oliver Neukum 2008-07-21 06:14:06 MDT --- I had to ask because some bugs cause spurious disconnects. Can you also get an strace of your test programm with timestamps so the log can be correlated with it? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c45 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #45 from Oliver Neukum 2008-07-21 12:57:20 MDT --- This [pid 3549] 17:06:50.296531 ioctl(29, USBDEVFS_CONTROL, 0xffae35e0) = -1 ESHUTDOWN (Cannot send after transport endpoint shutdown) and this correlate Jul 21 17:06:50 linux-t2fx kernel: ehci_hcd 0000:00:1a.7: irq status 0003 ERR INT Jul 21 17:06:50 linux-t2fx kernel: ehci_hcd 0000:00:1a.7: dev4 ep1in qtd token 800d8d50 --> status -75 Jul 21 17:06:50 linux-t2fx kernel: usb 2-6: unlink qh16-0001/ffff81003d0f2140 start 15 [2/0 us] Jul 21 17:06:50 linux-t2fx kernel: ehci_hcd 0000:00:1a.7: reused qh ffff81003d0f2140 schedule Jul 21 17:06:50 linux-t2fx kernel: usb 2-6: link qh16-0001/ffff81003d0f2140 start 15 [2/0 us] Jul 21 17:06:50 linux-t2fx kernel: usb 2-6: unlink qh16-0001/ffff81003d0f2140 start 15 [2/0 us] Jul 21 17:06:50 linux-t2fx kernel: ehci_hcd 0000:00:1a.7: HC died; cleaning up is this repeatable and which control request did you send? PS: Don't zip. Handling storage problems is the system's problem, not ours. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c47 --- Comment #47 from Oliver Neukum 2008-07-22 04:55:31 MDT --- You can bisect it with git. That's tedious but reliable. In addition you can do a test with the patch from comment #42. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c51 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #51 from Oliver Neukum 2008-07-31 00:27:47 MDT --- Any progress? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c53 --- Comment #53 from Stefan Neis 2008-08-05 11:13:50 MDT --- Okay, it doesn't make any sense to me (which probably is part of why it took so long ...), but here is what I found: After applying the patch from #49, the problem on x86 goes away, _if_ in addition I issue an "rmmod ehci-hcd ; modprobe -a ehci-hcd" after booting. However, if I don't do that before starting my test program, it still hangs. After realising this, I retested on x86_64 and found that it behaves the same way, i.e. when I reported that it works there, I apparently had only unloaded the old module and loaded the new one but had not tested how it behaves after rebooting. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c55 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #55 from Oliver Neukum 2008-08-06 00:45:04 MDT --- Have you tested whether the patch from #49 is needed at all? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c57 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #57 from Oliver Neukum 2008-08-06 04:45:20 MDT --- OK, please check whether your initrd contains ehci_hcd. If so, did you make sure it was replaced by the versions of the module you had compiled? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c59 Stefan Neis changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|stefan.neis@kobil.com | --- Comment #59 from Stefan Neis 2008-08-07 04:08:33 MDT --- Regarding #57, I suppose my initrd is containing ehci_hcd, I gunzipped the initrd and "strings <unpacked file> | grep ehci" gave a number of matches, among them e.g. lib/modules/2.6.25.5-1.1-default/kernel/drivers/usb/host/ehci-hcd.ko and drivers/usb/host/. Is there a cleaner way to check and an easy way to update initrd? I'm looking for an easy fix for users and telling them "patch the kernel source, recompile and reinstall the kernel" is soemthing I don't really consider "easy". "Just exchange ehci-hcd.ko against the fixed one" would have neen something I like, but unfortunately it's obviously not sufficient. Anyway, I guess that explains the reloading issue. Regarding #58 do I understand it correctly that you agree with the suggested fix (reverting that optimization)? And if so, what does that mean in practice? Can we hope for a fixed kernel in the next release or is it already to late for e.g. SuSE 11.1 ? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c61 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #61 from Oliver Neukum 2008-08-07 06:26:02 MDT --- Yes, it does explain the reloading issue and it had to be tested to make sure this is a fairly ordinary bug. This bug will be fixed in 11.0 and newer releases for sure. The question is whether it needs to be fixed in mainline, too. Can you replicate this bug only with your special device? If so I have to ask to replicate this bug in vanilla 2.6.27-rc2 otherwise I'll try to replicate it. To understand what exactly is going wrong you need to be an ehci specialists. But it is highly plausible that a bug in that area would have consequences you experience. I've tried to describe the mechanism in #58 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c63 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|stefan.neis@kobil.com | --- Comment #63 from Oliver Neukum 2008-08-07 12:19:26 MDT --- OK, that's not good, but the problem seems to be specific to the 11.0 kernel. I'll fix that. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c65 --- Comment #65 from Oliver Neukum 2008-08-10 00:43:29 MDT --- Sorry, if I cannot replicate it, I cannot report it to linux-usb@vger.kernel.org You have to do it. Please CC gregkh@suse.de and the author of the commit you found by bisecting. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User stefan.neis@kobil.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c67 --- Comment #67 from Stefan Neis 2008-10-07 08:00:27 MDT --- So that means that it should work with OpenSuSE 11.1 Beta 3 (or whatever comes next)? At least, it still does not yet work with Beta 2. :-( -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c69 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | --- Comment #69 from Oliver Neukum 2008-10-14 04:09:24 MDT --- On Monday 13 October 2008, Oliver Neukum wrote:

Hi David,

you suggest a patch in this thread: http://article.gmane.org/gmane.linux.usb.general/8804/match=usb+ehci+problem... which never went in. Do you have a better alternative? Should it get in?

Better alternatives to that quick hack are in 2.6.27 already: - fix error handling -- 0bcfeb3ec9467a8dfae3e906925ec28fe788afcf - fix root cause -- 01c17142659c02d8cd4c67acdd8358180c5c4fb6 And ISTR seeing them go to the -stable tree too. - Dave -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c72 --- Comment #72 from Oliver Neukum 2008-10-14 05:55:19 MDT --- In this the preliminary patch should not be needed. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c74 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |NEEDINFO Info Provider| |stefan.neis@kobil.com --- Comment #74 from Oliver Neukum 2008-10-14 06:27:12 MDT --- OK, I have removed it. Please test beta3 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=403346 User oneukum@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=403346#c76 Oliver Neukum changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |RESOLVED Info Provider|stefan.neis@kobil.com | Resolution| |FIXED --- Comment #76 from Oliver Neukum 2008-10-24 12:14:10 MDT --- Tested and found working -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.