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
