Message-ID: <20081120030012.GC26448@xw6200.broadcom.net>
Date:	Wed, 19 Nov 2008 19:00:12 -0800
From:	"Matt Carlson" <mcarlson@...adcom.com>
To:	"Roger Heflin" <rogerheflin@...il.com>
cc:	"Peter Zijlstra" <peterz@...radead.org>,
	LKML <linux-kernel@...r.kernel.org>,
	netdev <netdev@...r.kernel.org>
Subject: Re: WARNING: at net/sched/sch_generic.c:219
 dev_watchdog+0xfe/0x17e() with tg3 network

On Fri, Nov 14, 2008 at 08:01:46PM -0800, Roger Heflin wrote:
> Peter Zijlstra wrote:
> > (netdev CC'ed)
> > 
> > On Tue, 2008-11-11 at 03:48 -0600, Roger Heflin wrote:
> >> I have duplicate this with kernel 2.6.27.2 and 2.6.27.5, no
> >> extra modules, tg3 Gbit networking.   I have not yet tested
> >> earlier kernels to see if this has been around for a while.
> > 
> > How do more recent kernels do?
> 
> I did not try more recent kernels, more testing seems to indicate
> that at the very least it the bug depends on a certain version of
> either tg3 and/or firmware to happen, as my second tg3 port does not
> have it happen.    More about this below.
> 
> > 
> >> So far I have had this error happen 5 times (MTBF is maybe
> >> 12 hours), 4 of the 5 times resulted in the networking being
> >> broken, one time things came back by itself without a reboot,
> >> I believe in this case the hang was traffic coming into the
> >> machine vs the other times going out of the machine.
> >>
> >> Unloading all of the network modules and reloading them did
> >> not correct the problem.
> >>
> >> Searching google finds a couple of other people getting the
> >> same error but they have a different network chipset (e1000
> >> and a rt811C chipset), which makes me thing that there is
> >> something interacting bad with the network.    Or does this
> >> error truly mean that the network chipset for some unknown reason
> >> locked itself up?
> >>
> >> http://www.google.com/url?sa=U&start=4&q=http://kerneltrap.org/mailarchive/linux-netdev/2008/8/6/2838184&ei=rU8ZScysAon8edz5xKgO&sig2=Wxp7IkUtdgORGZiflxvppg&usg=AFQjCNHzPwsCOmLGKmtX4q_FEpk6oubxxg
> >> http://article.gmane.org/gmane.linux.network/110238
> >>
> >> The changes I made recently were to upgrade my MB (old
> >> was E100 on a 100Mbit network,new is tg3 on a Gbit network,
> >> cpu and memory are the same, MB chipset is a intel 955
> >> chipset vs the old being a intel 915 chipset).
> >>
> >> Autoneg is turned on all around, the GBit switch is a
> >> 8-port Dlink switch.   The network seems to otherwise be working
> >> correctly.
> >>
> >> I did test the network under decent load and the error did not
> >> appear to be any more likely under load, and typically the network
> >> is under very light load 2-3MB/second.
> >>
> >> The machine originally had 2 HT CPU's showing up, I turned off HT
> >> so that only one cpu was showing, but this did not change the error.
> >>
> >> I am first turning off all offload capabilities on tg3 and going
> >> to see if that changes anything.
> 
> This made no difference in the error.
> 
> >>
> >> The next thing I am going to be doing is to turn of GB capability
> >> on the networking and see if that does anything.
> 
> Did not try.
> 
> >>
> >> I also have a second tg3 port that is slightly different, so I may
> >> try that eventually.
> 
> I tried this, and with the second port I don't appear to be getting
> the error.   The first port is a 5789-v3.29a and the second port is a
> 5788-v3.04, I know the first port is faster (pcie-x1) than the second
> port (pci bus-built-in, unknown exact connection).   The second port
> will sustain about 50MB/second, were as the first port will get
>  >90MB/second.

The 5789 is a PCIe device, the 5788 is a PCI device.  There's a lot of
room for differences between the two devices.

> It seems to me to likely be the firmware on the tg3, and it would seem
> unlikely that the driver could do anything more than work around the
> issue that is in the firmware, and currently my system works on the
> second port, and the second port is fast enough for my needs.

I don't think the 5789 should have any firmware running at the time of
the failure.  There's always the possibility that there is a problem
with the bootcode, but I'm guessing the problem is elsewhere.

> If someone else runs into this issue, since I have 2 ports I would be
> able to do some testing on it, right now my first port is locked up, and
> the machine is running fine on the second port.
> 
> lspci -vvv for the first (bad) port:

Ah.  There it is.

> 02:00.0 Ethernet controller: Broadcom Corporation NetLink BCM5789 Gigabit 
> Ethernet PCI Express (rev 11)
>          Subsystem: Foxconn International, Inc. Unknown device 0cc1
>          Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- 
> Stepping- SERR- FastB2B-
>          Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- 
> <MAbort- >SERR- <PERR-
>          Latency: 0, Cache Line Size: 32 bytes
>          Interrupt: pin A routed to IRQ 19
>          Region 0: Memory at fd8f0000 (64-bit, non-prefetchable) [size=64K]
>          Expansion ROM at <ignored> [disabled]
>          Capabilities: [48] Power Management version 2
>                  Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA 
> PME(D0-,D1-,D2-,D3hot+,D3cold+)
>                  Status: D3 PME-Enable- DSel=0 DScale=1 PME-
>          Capabilities: [50] Vital Product Data
>          Capabilities: [58] Message Signalled Interrupts: Mask- 64bit+ Queue=0/3 
> Enable-
>                  Address: 0101b8102a0f7b0c  Data: f21e
>          Capabilities: [d0] Express Endpoint IRQ 0
>                  Device: Supported: MaxPayload 128 bytes, PhantFunc 0, ExtTag+
>                  Device: Latency L0s <4us, L1 unlimited
>                  Device: AtnBtn- AtnInd- PwrInd-
>                  Device: Errors: Correctable- Non-Fatal- Fatal- Unsupported-
>                  Device: RlxdOrd- ExtTag- PhantFunc- AuxPwr- NoSnoop-
>                  Device: MaxPayload 128 bytes, MaxReadReq 4096 bytes
>                  Link: Supported Speed 2.5Gb/s, Width x1, ASPM L0s, Port 0
>                  Link: Latency L0s <2us, L1 <64us
>                  Link: ASPM Disabled RCB 64 bytes CommClk+ ExtSynch-
>                  Link: Speed 2.5Gb/s, Width x1
>          Capabilities: [100] Advanced Error Reporting
>          Capabilities: [13c] Virtual Channel

Hmmm.  No smoking gun.  Perhaps the register dump will help.


> >> Nov 11 00:44:39 computer kernel: ------------[ cut here ]------------
> >> Nov 11 00:44:39 computer kernel: WARNING: at net/sched/sch_generic.c:219 
> >> dev_watchdog+0xfe/0x17e()
> >> Nov 11 00:44:39 computer kernel: NETDEV WATCHDOG: eth0 (tg3): transmit timed out

Usually the tg3_tx_timeout function dumps a few registers before
resetting the chip, but I don't see that here.  Have you seen any dumps
since then?

> >> Nov 11 00:44:39 computer kernel: Modules linked in: nfsd auth_rpcgss exportfs 
> >> w83627ehf hwmon_vid hwmon nfs lockd nfs_acl sunrpc ipv6 xfs raid456 async_xor 
> >> async_memcpy async_tx xor video output sbs sbshc battery ac lgdt330x cx88_dvb 
> >> wm8775 cx88_vp3054_i2c cx25840 tuner_simple tuner_types tda9887 tda8290 tuner 
> >> mt2131 s5h1409 snd_hda_intel snd_seq_dummy ivtv cx8800 snd_seq_oss cx88_alsa 
> >> cx8802 cx88xx cx23885 snd_seq_midi_event snd_seq ir_common videodev v4l1_compat 
> >> i2c_algo_bit cx2341x firewire_ohci iTCO_wdt snd_seq_device compat_ioctl32 
> >> videobuf_dvb i2c_i801 firewire_core tveeprom floppy iTCO_vendor_support 
> >> v4l2_common snd_pcm_oss dvb_core pcspkr tg3 sata_sil i2c_core btcx_risc 
> >> videobuf_dma_sg crc_itu_t snd_mixer_oss libphy videobuf_core snd_pcm parport_pc 
> >> parport snd_timer snd soundcore button snd_page_alloc sg dm_snapshot dm_zero 
> >> dm_mirror dm_log dm_mod ahci ata_piix ata_generic libata sd_mod scsi_mod ext3 
> >> jbd mbcache ehci_hcd ohci_hcd uhci_hcd [last unloaded: eeprom]
> >> Nov 11 00:44:39 computer kernel: Pid: 0, comm: swapper Not tainted 2.6.27.5 #2
> >> Nov 11 00:44:39 computer kernel:  [<c042524f>] warn_slowpath+0x61/0x83
> >> Nov 11 00:44:39 computer kernel:  [<c05663a4>] usb_hcd_submit_urb+0x75c/0x811
> >> Nov 11 00:44:39 computer kernel:  [<c0594972>] hiddev_hid_event+0x0/0x64
> >> Nov 11 00:44:39 computer kernel:  [<c058ce80>] hid_process_event+0x58/0x5f
> >> Nov 11 00:44:39 computer kernel:  [<c04e13d6>] __next_cpu+0x12/0x21
> >> Nov 11 00:44:39 computer kernel:  [<c041cbe3>] find_busiest_group+0x23e/0x672
> >> Nov 11 00:44:39 computer kernel:  [<c0439d1e>] clocksource_get_next+0x39/0x3f
> >> Nov 11 00:44:39 computer kernel:  [<c0438e51>] update_wall_time+0x567/0x70c
> >> Nov 11 00:44:39 computer kernel:  [<c040783e>] read_tsc+0x6/0x22
> >> Nov 11 00:44:39 computer kernel:  [<c04387e8>] getnstimeofday+0x37/0xc1
> >> Nov 11 00:44:39 computer kernel:  [<f8829a83>] uhci_scan_schedule+0x11b/0x6b0 
> >> [uhci_hcd]
> >> Nov 11 00:44:39 computer kernel:  [<c05b16ba>] dev_watchdog+0xfe/0x17e
> >> Nov 11 00:44:39 computer kernel:  [<c042c66f>] __mod_timer+0x99/0xa3
> >> Nov 11 00:44:39 computer kernel:  [<c05654b6>] rh_timer_func+0x0/0x5
> >> Nov 11 00:44:39 computer kernel:  [<c05654ae>] usb_hcd_poll_rh_status+0x12b/0x133
> >> Nov 11 00:44:39 computer kernel:  [<c043bca8>] tick_dev_program_event+0x1e/0x81
> >> Nov 11 00:44:39 computer kernel:  [<c05b15bc>] dev_watchdog+0x0/0x17e
> >> Nov 11 00:44:39 computer kernel:  [<c042c2b4>] run_timer_softirq+0x10e/0x167
> >> Nov 11 00:44:39 computer kernel:  [<c05b15bc>] dev_watchdog+0x0/0x17e
> >> Nov 11 00:44:39 computer kernel:  [<c0428d3e>] __do_softirq+0x5d/0xc1
> >> Nov 11 00:44:39 computer kernel:  [<c0428dd4>] do_softirq+0x32/0x36
> >> Nov 11 00:44:39 computer kernel:  [<c0412939>] smp_apic_timer_interrupt+0x6e/0x79
> >> Nov 11 00:44:39 computer kernel:  [<c040431c>] apic_timer_interrupt+0x28/0x30
> >> Nov 11 00:44:39 computer kernel:  [<c0408582>] mwait_idle+0x32/0x38
> >> Nov 11 00:44:39 computer kernel:  [<c040255d>] cpu_idle+0xbd/0xd5
> >> --
> >> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> >> the body of a message to majordomo@...r.kernel.org
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >> Please read the FAQ at  http://www.tux.org/lkml/
> > 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@...r.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists