kernel warning in usb_ep_queue() during suspend/resume

kernel warning in usb_ep_queue() during suspend/resume

[Francesco Dolcini]

Hello all,
while doing some tests on kernel 6.1 the following warning is hit
sporadically:

  ------------[ cut here ]------------
  WARNING: CPU: 0 PID: 1385 at drivers/usb/gadget/udc/core.c:295 usb_ep_queue+0x58/0x6c
  Modules linked in: 8021q bnep mcp251x mwifiex_sdio mwifiex cfg80211 btmrvl_sdio btmrvl imx_sdma fuse
  CPU: 0 PID: 1385 Comm: analogclock Not tainted 6.1.66-6.5.0-devel+git.6c6a6c7e211c #1
  Hardware name: Freescale i.MX6 Ultralite (Device Tree)
   unwind_backtrace from show_stack+0x10/0x14
   show_stack from dump_stack_lvl+0x40/0x4c
   dump_stack_lvl from __warn+0x88/0xc0
   __warn from warn_slowpath_fmt+0x10c/0x1a0
   warn_slowpath_fmt from usb_ep_queue+0x58/0x6c
   usb_ep_queue from eth_start_xmit+0x1ec/0x330
   eth_start_xmit from dev_hard_start_xmit+0x7c/0xf8
   dev_hard_start_xmit from sch_direct_xmit+0x110/0x310
   sch_direct_xmit from __qdisc_run+0x110/0x5d8
   __qdisc_run from net_tx_action+0x170/0x20c
   net_tx_action from __do_softirq+0xc0/0x234
   __do_softirq from __irq_exit_rcu+0x8c/0xb8
   __irq_exit_rcu from irq_exit+0x8/0x10
   irq_exit from call_with_stack+0x18/0x20
   call_with_stack from __irq_usr+0x7c/0xa0
  Exception stack(0xe0fcdfb0 to 0xe0fcdff8)
  dfa0:                                     b2f9e000 b30775a0 0002edf8 00000000
  dfc0: 01493268 014fdd00 0002edf8 0002edf8 000185b8 be8d9734 b3f6f811 0000000c
  dfe0: b2fb4580 be8d95f4 b3f6f7d3 b64006a8 20070010 ffffffff
  ---[ end trace 0000000000000000 ]---


The warning is happening during a suspend/resume test (not sure which
one of the twos), I cannot really tell if this is a regression (and from
which version, eventually), but this is happening since a while with
multiple 6.1 kernel version (this last log is from a recent 6.1.66).

I cannot really tell the current status with mainline or v6.6 given that
it does not seems so trivial to reproduce.

It happened on various arm and arm64 machine.

The ethernet device is a USB NCM/ECM, running in gadget mode.

Any idea?

Francesco

Re: kernel warning in usb_ep_queue() during suspend/resume

[Greg Kroah-Hartman]

On Tue, Dec 12, 2023 at 05:36:14PM +0100, Francesco Dolcini wrote:
> Hello all,
> while doing some tests on kernel 6.1 the following warning is hit
> sporadically:
> 
>   ------------[ cut here ]------------
>   WARNING: CPU: 0 PID: 1385 at drivers/usb/gadget/udc/core.c:295 usb_ep_queue+0x58/0x6c
>   Modules linked in: 8021q bnep mcp251x mwifiex_sdio mwifiex cfg80211 btmrvl_sdio btmrvl imx_sdma fuse
>   CPU: 0 PID: 1385 Comm: analogclock Not tainted 6.1.66-6.5.0-devel+git.6c6a6c7e211c #1
>   Hardware name: Freescale i.MX6 Ultralite (Device Tree)
>    unwind_backtrace from show_stack+0x10/0x14
>    show_stack from dump_stack_lvl+0x40/0x4c
>    dump_stack_lvl from __warn+0x88/0xc0
>    __warn from warn_slowpath_fmt+0x10c/0x1a0
>    warn_slowpath_fmt from usb_ep_queue+0x58/0x6c
>    usb_ep_queue from eth_start_xmit+0x1ec/0x330
>    eth_start_xmit from dev_hard_start_xmit+0x7c/0xf8
>    dev_hard_start_xmit from sch_direct_xmit+0x110/0x310
>    sch_direct_xmit from __qdisc_run+0x110/0x5d8
>    __qdisc_run from net_tx_action+0x170/0x20c
>    net_tx_action from __do_softirq+0xc0/0x234
>    __do_softirq from __irq_exit_rcu+0x8c/0xb8
>    __irq_exit_rcu from irq_exit+0x8/0x10
>    irq_exit from call_with_stack+0x18/0x20
>    call_with_stack from __irq_usr+0x7c/0xa0
>   Exception stack(0xe0fcdfb0 to 0xe0fcdff8)
>   dfa0:                                     b2f9e000 b30775a0 0002edf8 00000000
>   dfc0: 01493268 014fdd00 0002edf8 0002edf8 000185b8 be8d9734 b3f6f811 0000000c
>   dfe0: b2fb4580 be8d95f4 b3f6f7d3 b64006a8 20070010 ffffffff
>   ---[ end trace 0000000000000000 ]---
> 
> 
> The warning is happening during a suspend/resume test (not sure which
> one of the twos), I cannot really tell if this is a regression (and from
> which version, eventually), but this is happening since a while with
> multiple 6.1 kernel version (this last log is from a recent 6.1.66).
> 
> I cannot really tell the current status with mainline or v6.6 given that
> it does not seems so trivial to reproduce.
> 
> It happened on various arm and arm64 machine.

After the warning happens, does the machine still work?  That warning is
a big "something went wrong" but it should still recover and keep on
going.

thanks,

greg k-h

Re: kernel warning in usb_ep_queue() during suspend/resume

[Francesco Dolcini]

Hello Greg,

On Tue, Dec 12, 2023 at 05:44:43PM +0100, Greg Kroah-Hartman wrote:
> On Tue, Dec 12, 2023 at 05:36:14PM +0100, Francesco Dolcini wrote:
> >   ------------[ cut here ]------------
> >   WARNING: CPU: 0 PID: 1385 at drivers/usb/gadget/udc/core.c:295 usb_ep_queue+0x58/0x6c
> >   Modules linked in: 8021q bnep mcp251x mwifiex_sdio mwifiex cfg80211 btmrvl_sdio btmrvl imx_sdma fuse
> >   CPU: 0 PID: 1385 Comm: analogclock Not tainted 6.1.66-6.5.0-devel+git.6c6a6c7e211c #1
> >   Hardware name: Freescale i.MX6 Ultralite (Device Tree)
> >    unwind_backtrace from show_stack+0x10/0x14
> >    show_stack from dump_stack_lvl+0x40/0x4c
> >    dump_stack_lvl from __warn+0x88/0xc0
> >    __warn from warn_slowpath_fmt+0x10c/0x1a0
> >    warn_slowpath_fmt from usb_ep_queue+0x58/0x6c
> >    usb_ep_queue from eth_start_xmit+0x1ec/0x330
> >    eth_start_xmit from dev_hard_start_xmit+0x7c/0xf8
> >    dev_hard_start_xmit from sch_direct_xmit+0x110/0x310
> >    sch_direct_xmit from __qdisc_run+0x110/0x5d8
> >    __qdisc_run from net_tx_action+0x170/0x20c
> >    net_tx_action from __do_softirq+0xc0/0x234
> >    __do_softirq from __irq_exit_rcu+0x8c/0xb8
> >    __irq_exit_rcu from irq_exit+0x8/0x10
> >    irq_exit from call_with_stack+0x18/0x20
> >    call_with_stack from __irq_usr+0x7c/0xa0
> >   Exception stack(0xe0fcdfb0 to 0xe0fcdff8)
> >   dfa0:                                     b2f9e000 b30775a0 0002edf8 00000000
> >   dfc0: 01493268 014fdd00 0002edf8 0002edf8 000185b8 be8d9734 b3f6f811 0000000c
> >   dfe0: b2fb4580 be8d95f4 b3f6f7d3 b64006a8 20070010 ffffffff
> >   ---[ end trace 0000000000000000 ]---
> > 
> > 
> > The warning is happening during a suspend/resume test (not sure which
> > one of the twos), I cannot really tell if this is a regression (and from
> > which version, eventually), but this is happening since a while with
> > multiple 6.1 kernel version (this last log is from a recent 6.1.66).
> > 
> > I cannot really tell the current status with mainline or v6.6 given that
> > it does not seems so trivial to reproduce.
> > 
> > It happened on various arm and arm64 machine.
> 
> After the warning happens, does the machine still work?
yes, the machine keeps working. I cannot tell much more since this is
happening sporadically while doing some automated test and a few seconds
after it is just switched off.


Thanks,
Francesco