On Tue, 31 Dec 2013, Vi0L0 wrote:
> I've found that problem touches also 3.13 rc6 and that removing "threadirqs"
> kernel option is making USB 2.0 ports fully functional again.
>
> I believe that this problem is much more general and touching many PCs with
> intel (maybe also non-intel) chipsets inside.
>
> Today I had quick access to TOSHIBA SATELLITE PRO R950 laptop with ubuntu
> 13.10 and 3.13 rc3 kernel, i tested how it's working with "threadirqs" option.
> It wasn't able to even start a DM, it was stuck before, throwing these lines
> on a tty:
> BUG: soft lockup - CPU#1 stuck for 23s! [fsck:264]
> BUG: soft lockup - CPU#3 stuck for 22s! [fsck:254]
> INFO: rcu_sched detected stalls on CPU/tasks: { 2}
> But after the reboot (without "threadirqs" it was working perfectly fine) there
> was nothing interesting in the syslog, not even one info about the stalls. As
> said I had a quick access and no time for further investigatiuon. In the
> attachement I'm sending it's lsusb and lspci (toshiba-lspci_lsusb).
>
>
> In the attachements I'm sending my system' logs catched on 3.13 rc6 kernel
> with "threadirqs" kernel option. I decided to switch to open source radeon
> graphics driver to eliminate unnecessary fglrx warnings/errors.
>
> First log - rc6-journalctl-less-debug.log - was catched on a kernel build with
> some basic debugging enabled, while second log - rc6-journalctl-more-
> debug.tar.gz - was catched on a kernel build with much more debugging.
> I'm not sure if it will be helpfull enought. I would be happy to help more,
> just let me know what can I do since I'm no expert in such a debugging...
>
> What I believe is most relevant in the logs are those lines (taken from rc6-
> journalctl-less-debug.log):
> 2105 gru 31 14:40:31 xaos systemd[1]: Started K Display Manager.
> 2106 gru 31 14:40:32 xaos kernel: usb 3-1.2: ep 82: reserve intr @ 0+8 (0.0+1)
> [1/3 us] mask 1c01
> 2107 gru 31 14:40:32 xaos kernel: usb 3-1.2: link qh1-1c01/ffff8800d7cbcf00
> start 0 [1/3 us]
> 2108 gru 31 14:41:40 xaos kernel: ------------[ cut here ]------------
> 2109 gru 31 14:41:40 xaos kernel: WARNING: CPU: 7 PID: 198 at
> kernel/watchdog.c:245 watchdog_overflow_callback+0x9b/0xc0()
> 2110 gru 31 14:41:40 xaos kernel: Watchdog detected hard LOCKUP on cpu 7
> 2114 gru 31 14:41:40 xaos kernel: CPU: 7 PID: 198 Comm: irq/16-ehci_hcd Not
> tainted 3.13.0-2-mainline #1
> 2162 gru 31 14:41:40 xaos kernel: INFO: rcu_preempt detected stalls on
> CPUs/tasks: { 7} (detected by 0, t=60002 jiffies, g=592, c=591, q=0)
> 2165 gru 31 14:41:40 xaos kernel: CPU: 7 PID: 198 Comm: irq/16-ehci_hcd
> Tainted: G W 3.13.0-2-mainline #1
>
> I'm also sending my kernel config (rc6-config.x86_64) used to catch rc6-
> journalctl-less-debug.log
>
>
> Dnia poniedziałek, 30 grudnia 2013 10:15:00 Alan Stern pisze:
> > Since you know that the 3.12 kernel works okay, maybe you can use "git
> > bisect" to find the change that introduced this problem.
> >
> > Alan Stern
>
> I would really like to find that change, but there are so many changes related
> to both irqs and ehci... not to mention that I'm not the developer. It really
> bacame not easy to me. Maybe getting more traces would help me with it, but I
> don't know how to get these, maybe you could help me?
You shouldn't concentrate on USB. It's clear that many other
subsystems are affected too.
You should post a description of the problem on the linux-kernel
mailing list and CC: the people who maintain the threaded IRQs code.
Alan Stern
--
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
