From mboxrd@z Thu Jan 1 00:00:00 1970 From: Wolfgang Grandegger Subject: Re: Kernel-Warning when stopping the CAN interface Date: Thu, 12 Mar 2020 08:05:44 +0100 Message-ID: <38dccda1-ca41-bc5d-01fb-70050d16730d@grandegger.com> References: <4c5cd8bd-3d4d-4826-9333-0d0f23b30f5a@schneidersoft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: Received: from mailproxy03.manitu.net ([217.11.48.67]:40518 "EHLO mailproxy03.manitu.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S2387959AbgCLHFr (ORCPT ); Thu, 12 Mar 2020 03:05:47 -0400 In-Reply-To: <4c5cd8bd-3d4d-4826-9333-0d0f23b30f5a@schneidersoft.net> Content-Language: en-GB Sender: linux-can-owner@vger.kernel.org List-ID: To: Maximilian Schneider Cc: linux-can@vger.kernel.org Hello Maximilian, I think, the problem is not gs_usb specific. The following patch seems to help: --- a/drivers/net/can/usb/gs_usb.c +++ b/drivers/net/can/usb/gs_usb.c @@ -321,7 +321,7 @@ static void gs_usb_receive_bulk_callback(struct urb *urb) netdev = dev->netdev; stats = &netdev->stats; - if (!netif_device_present(netdev)) + if (!netif_device_present(netdev) || !netif_running(netdev)) return; if (hf->echo_id == -1) { /* normal rx */ Wolfgang Am 12.03.20 um 07:06 schrieb Maximilian Schneider: > I guess the naive solution would be to test dev->state in the RX > callback. I doubt that's the best solution thought. Is it known if this > warning affects only the gs_usb driver? > On 11 Mar 2020, at 16:53, Wolfgang Grandegger > wrote: > > Hello, > > I can easily trigger the kernel warning below with a "gs_usb" device when > I send messages with "cangen -g1 can0" and then stop the interface with > "ifconfig can0 down". The warning comes from here: > > https://elixir.bootlin.com/linux/latest/source/net/core/skbuff.c#L651 > > What happens is that the "ifconfig down" command first sets > > clear_bit(__LINK_STATE_START, &dev->state); > > before it calls the device specific stop/close routine. In the meantime, > TX done events may still occur, calling can_get_echo_skb(), which calls > netif_rx() then netif_rx_internal() and finally enqueue_to_backlog(). > Because the interface is not running any longer, the skb is dropped in > there: > > https://elixir.bootlin.com/linux/latest/source/net/core/dev.c#L4394 > > causing the warning below, because the kfree_skb(skb) is called from > the interrupt context. > > How can I avoid that warning? Actually, I do not see a simple method. > Maybe I have missed something. > > Wolfgang > > [ 1694.131475] ------------[ cut here ]------------ > [ 1694.139911] WARNING: CPU: 0 PID: 0 at skb_release_head_state+0x9e/0xb0 > [ 1694.147206] Modules linked in: intel_telemetry_pltdrv intel_pmc_ipc intel_punit_ipc intel_telemetry_core x86_pkg_temp_thermal igb tpm_tis tpm_tis_core btusb btrtl dca tpm lpc_ich btbcm spi_pxa2xx_platform mei_me btintel mei thermal rfcomm bluetooth ecdh_generic ntngdm_i2c ntngdm_core ntngdm_bl ntngdm_tsi ntngdm_irq ntngdm_hwmon i915 video > [ 1694.180782] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G U W 4.14.51apollolake #1 > [ 1694.189828] Hardware name: Default string Default string/AGCO_NT-NG, BIOS X3.00c#AGCO BETA 09/19/2019 > [ 1694.200136] task: ffffffff840104c0 task.stack: ffffffff84000000 > [ 1694.206753] RIP: 0010:skb_release_head_state+0x9e/0xb0 > [ 1694.212494] RSP: 0018:ffff8b2b3fc03cb0 EFLAGS: 00010006 > [ 1694.218333] RAX: ffffffff835a4fb0 RBX: ffff8b2b3a20b100 RCX: 0000000000000018 > [ 1694.226310] RDX: 0000000000010000 RSI: 0000000000000000 RDI: 0000000000000000 > [ 1694.234284] RBP: 0000000000020ac0 R08: 0000000000000014 R09: 0000000000000014 > [ 1694.242259] R10: 0000000000000000 R11: ffff8b2b3a7a2180 R12: ffff8b2b3fc03d40 > [ 1694.250235] R13: ffff8b2b3a20b100 R14: ffff8b2b3fc20bcc R15: 0000000000000086 > [ 1694.258211] FS: 0000000000000000(0000) GS:ffff8b2b3fc00000(0000) knlGS:0000000000000000 > [ 1694.267256] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 1694.273678] CR2: 00007f83a2db5130 CR3: 0000000179c6a000 CR4: 00000000003406f0 > [ 1694.281653] Call Trace: > [ 1694.284383] > [ 1694.286629] skb_release_all+0x9/0x20 > [ 1694.290720] kfree_skb+0x19/0x40 > [ 1694.294327] enqueue_to_backlog+0x7e/0x230 > [ 1694.298906] ? recalibrate_cpu_khz+0x10/0x10 > [ 1694.303677] ? ktime_get_with_offset+0x4f/0xb0 > [ 1694.308644] netif_rx_internal+0x56/0x150 > [ 1694.313126] can_get_echo_skb+0x37/0x60 > [ 1694.317415] gs_usb_receive_bulk_callback+0x111/0x320 > [ 1694.323063] __usb_hcd_giveback_urb+0x7b/0xf0 > [ 1694.327924] xhci_giveback_urb_in_irq.isra.55+0x6d/0x90 > [ 1694.333765] xhci_td_cleanup+0xcb/0x150 > [ 1694.338051] xhci_irq+0x66a/0x20b0 > [ 1694.341854] ? rcu_accelerate_cbs+0x3a/0x190 > [ 1694.346628] __handle_irq_event_percpu+0x32/0xd0 > [ 1694.351791] handle_irq_event_percpu+0x2b/0x70 > [ 1694.356757] handle_irq_event+0x2f/0x50 > [ 1694.361044] handle_edge_irq+0x72/0x190 > [ 1694.365331] handle_irq+0x17/0x20 > [ 1694.369038] do_IRQ+0x3c/0xc0 > [ 1694.372355] common_interrupt+0x7a/0x7a > [ 1694.376639] > [ 1694.378981] RIP: 0010:cpuidle_enter_state+0x122/0x200 > [ 1694.384622] RSP: 0018:ffffffff84003ea8 EFLAGS: 00000286 ORIG_RAX: ffffffffffffffbd > [ 1694.393087] RAX: ffff8b2b3fc1fa00 RBX: 0000000000000002 RCX: 0000018a721a5ae8 > [ 1694.401063] RDX: 0000018a721a5ae8 RSI: 0000000050523fbe RDI: 0000000000000000 > [ 1694.409040] RBP: ffff8b2b3aa13600 R08: 0000000000000193 R09: 000000000000012a > [ 1694.417015] R10: ffffffff84003e90 R11: 00000000000000ef R12: 0000018a721a5ae8 > [ 1694.424990] R13: 0000018a72199308 R14: 0000000000000002 R15: 0000000072efca78 > [ 1694.432970] ? cpuidle_enter_state+0x11e/0x200 > [ 1694.437937] do_idle+0x161/0x1a0 > [ 1694.441544] cpu_startup_entry+0x6a/0x70 > [ 1694.445927] start_kernel+0x414/0x434 > [ 1694.450023] secondary_startup_64+0xa5/0xb0 > [ 1694.454697] Code: 88 96 4f 17 00 74 02 5b c3 5b e9 4e 15 bc ff e8 d9 33 04 00 eb dd e8 c2 5a 0d 00 eb 9f 48 83 e7 fe e8 07 d7 01 00 e9 75 ff ff ff <0f> 0b eb ad 0f 1f 40 00 66 2e 0f 1f 84 00 00 00 00 00 53 48 89 > [ 1694.475869] ---[ end trace 063196a1141471b3 ]--- >