Bug#704304: e1000e driver hangs and system becomes unstable

Hi,

Narrowed down the issue to laptop-mode-tools. If ethernet control is disabled the issue is solved.

Tried with kernel 3.2.41-2, now available in testing, same results.

Apr 5 18:51:02 laptop kernel: [ 7.928034] ------------[ cut here ]------------

Apr 5 18:51:02 laptop kernel: [ 7.928063] WARNING: at /build/buildd-linux_3.2.41-2-amd64-Wvc92F/linux-3.2.41/drivers/net/ethernet/intel/e1000e/netdev.c:5333 __e1000_shutdown+0x6f/0x3ea [e1000e]()

Apr 5 18:51:02 laptop kernel: [ 7.928067] Hardware name: 7465CTO

Apr 5 18:51:02 laptop kernel: [ 7.928069] Modules linked in: ip6table_filter ip6_tables ebtable_nat ebtables ipt_MASQUERADE iptable_nat nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_state nf_conntrack ipt_REJECT xt_CHECKSUM iptable_mangle xt_tcpudp iptable_filter ip_tables x_tables tun bridge stp parport_pc ppdev lp parport bnep rfcomm uinput ext2 tp_smapi(O) thinkpad_ec(O) loop kvm_intel kvm snd_hda_codec_conexant arc4 thinkpad_acpi nvram snd_hda_intel snd_hda_codec snd_hwdep snd_pcm snd_page_alloc snd_seq snd_seq_device snd_timer snd iwlwifi mac80211 psmouse iTCO_wdt btusb serio_raw bluetooth pcspkr cfg80211 coretemp i2c_i801 iTCO_vendor_support evdev soundcore rfkill i915 battery ac power_supply drm_kms_helper drm acpi_cpufreq wmi mperf video i2c_algo_bit i2c_core button processor ext4 crc16 jbd2 mbcache dm_mod microcode sg sd_mod crc_t10dif ahci libahci thermal thermal_sys uhci_hcd ata_generic libata scsi_mod ehci_hcd e1000e usbcore usb_common [last unloaded: scsi_wait_scan]

Apr 5 18:51:02 laptop kernel: [ 7.928140] Pid: 11, comm: kworker/0:1 Tainted: G O 3.2.0-4-amd64 #1 Debian 3.2.41-2

Apr 5 18:51:02 laptop kernel: [ 7.928142] Call Trace:

Apr 5 18:51:02 laptop kernel: [ 7.928149] [<ffffffff81046a55>] ? warn_slowpath_common+0x78/0x8c

Apr 5 18:51:02 laptop kernel: [ 7.928160] [<ffffffffa007cfc8>] ? __e1000_shutdown+0x6f/0x3ea [e1000e]

Apr 5 18:51:02 laptop kernel: [ 7.928171] [<ffffffffa007d374>] ? e1000_runtime_suspend+0x31/0x39 [e1000e]

Apr 5 18:51:02 laptop kernel: [ 7.928177] [<ffffffff811c5ce8>] ? pci_pm_runtime_suspend+0x43/0xd6

Apr 5 18:51:02 laptop kernel: [ 7.928181] [<ffffffff811c5ca5>] ? pci_legacy_suspend_late+0xbd/0xbd

Apr 5 18:51:02 laptop kernel: [ 7.928186] [<ffffffff81257662>] ? __rpm_callback+0x2c/0x50

Apr 5 18:51:02 laptop kernel: [ 7.928189] [<ffffffff81257bc1>] ? rpm_suspend+0x271/0x401

Apr 5 18:51:02 laptop kernel: [ 7.928194] [<ffffffff8103981a>] ? finish_task_switch+0x4e/0xb9

Apr 5 18:51:02 laptop kernel: [ 7.928197] [<ffffffff812588d7>] ? pm_runtime_work+0x65/0x7e

Apr 5 18:51:02 laptop kernel: [ 7.928202] [<ffffffff8105b225>] ? process_one_work+0x161/0x264

Apr 5 18:51:02 laptop kernel: [ 7.928205] [<ffffffff8105c1e6>] ? worker_thread+0xc2/0x145

Apr 5 18:51:02 laptop kernel: [ 7.928209] [<ffffffff8105c124>] ? manage_workers.isra.25+0x15b/0x15b

Apr 5 18:51:02 laptop kernel: [ 7.928213] [<ffffffff8105f321>] ? kthread+0x76/0x7e

Apr 5 18:51:02 laptop kernel: [ 7.928217] [<ffffffff81354ab4>] ? kernel_thread_helper+0x4/0x10

Apr 5 18:51:02 laptop kernel: [ 7.928222] [<ffffffff8105f2ab>] ? kthread_worker_fn+0x139/0x139

Apr 5 18:51:02 laptop kernel: [ 7.928225] [<ffffffff81354ab0>] ? gs_change+0x13/0x13

Apr 5 18:51:02 laptop kernel: [ 7.928227] ---[ end trace 2ee513d3cae1946f ]---

Will let you know what happens with kernel 3.8.y once i upgrade.

Ilan

On Mon, Apr 1, 2013 at 5:50 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

Ilan Cohen wrote:

> have done some more research, disabled the wol in pm-utils and disabled
> ethernet control in laptop-mode-tools
> and the problem has gone away.

There was an e1000e fix in __e1000e_shutdown() in 3.2.41. Can you
reproduce this with 3.2.41-2 from unstable? (If not, the fix will be
coming to wheezy soon and we can close the bug.)

If 3.2.41-2 produces the same symptoms, can you reproduce them with
3.8.y from experimental? (If not, we can try to find the fix and
apply it to wheezy.)

If 3.8.y produces the same symptoms, please send a summary of what
you have tried and what happened to the upstream developers at
e1000-devel@lists.sourceforge.net, cc-ing either me or this bug log so
we can track it. Hopefully they will have ideas for tracking it down
further.

Thanks and hope that helps,
Jonathan

Ilan Cohen