Re: tg3 adapter losing link - PM related?

[Nikola Ciprich <nikola.ciprich@linuxbox.cz>]

[-- Attachment #1: Type: text/plain, Size: 5109 bytes --]

Hello Nithin,

On Fri, May 03, 2013 at 08:13:15AM -0700, Nithin Nayak Sujir wrote:
> Hi Nikola,
> 1. What device is present on this server? Can you give the tg3 messages in
> /var/log/messages? Can you also give the output of "ethtool -i <iface>"?

sure, here it goes:

[  +0.273255] tg3.c:v3.119 (May 18, 2011)
[  +0.000184] tg3 0000:03:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32
[  +0.000175] tg3 0000:03:00.0: setting latency timer to 64
[  +0.034616] tg3 0000:03:00.0: eth0: Tigon3 [partno(none) rev 5719001] (PCI Express) MAC address 2c:76:8a:52:a5:1c
[  +0.000255] tg3 0000:03:00.0: eth0: attached PHY is 5719C (10/100/1000Base-T Ethernet) (WireSpeed[1], EEE[1])
[  +0.000251] tg3 0000:03:00.0: eth0: RXcsums[1] LinkChgREG[0] MIirq[0] ASF[1] TSOcap[0]
[  +0.000251] tg3 0000:03:00.0: eth0: dma_rwctrl[00000001] dma_mask[64-bit]


[root@atlovav1a ~]# ethtool -i eth0
driver: tg3
version: 3.119
firmware-version: 5719-v1.29 NCSI v1.0.88.0
bus-info: 0000:03:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: no

the box is HP ProLiant DL360p, device identifies itself in lspci as Broadcom Corporation NetXtreme BCM5719


> 
> 2. Is it possible for you to try the latest 3.9 upstream kernel?

of course, It's compiling now. I'll report how it behaves.

> 
> 3. Any reason why the link is at 100Mb? The switch does not support gig?
> What switch is it?

it's some lowcost cisco, not sure whether it's gigabit capable.. I'll find out on monday
and report if it's important.

> 
> 4. I don't think you can turn off EEE in the bios, but you can try turning
> autoneg off. Try "ethtool -s <iface> speed 100 duplex full autoneg off".
> It's not exactly the same thing since the device still has EEE enabled but
> not negotiated.

forcing speed helped, I no longer see link going down! 

of course I'll report about 3.9 kernel too..

thanks for Your reply!

nik

> 
> Nithin.
> 
> 
> On 5/3/2013 2:28 AM, Nikola Ciprich wrote:
> >Hello,
> >
> >I'd like to ask about trouble I've got with new HP server: tg3 adapter keeps losing
> >link every few minutes:
> >
> >Aug  3 03:58:02 atlovav1a kernel: [616741.147598] tg3 0000:03:00.0: eth0: Link is down
> >Aug  3 03:58:04 atlovav1a kernel: [616743.943456] tg3 0000:03:00.0: eth0: Link is up at 100 Mbps, full duplex
> >Aug  3 03:58:04 atlovav1a kernel: [616743.943598] tg3 0000:03:00.0: eth0: Flow control is on for TX and on for RX
> >Aug  3 03:58:04 atlovav1a kernel: [616743.943736] tg3 0000:03:00.0: eth0: EEE is enabled
> >Aug  3 04:14:29 atlovav1a kernel: [617727.980487] tg3 0000:03:00.0: eth0: Link is down
> >Aug  3 04:14:32 atlovav1a kernel: [617730.847245] tg3 0000:03:00.0: eth0: Link is up at 100 Mbps, full duplex
> >Aug  3 04:14:32 atlovav1a kernel: [617730.847387] tg3 0000:03:00.0: eth0: Flow control is on for TX and on for RX
> >Aug  3 04:14:32 atlovav1a kernel: [617730.847525] tg3 0000:03:00.0: eth0: EEE is enabled
> >Aug  3 06:47:13 atlovav1a kernel: [626885.452974] tg3 0000:03:00.0: eth0: Link is down
> >Aug  3 06:47:15 atlovav1a kernel: [626888.218702] tg3 0000:03:00.0: eth0: Link is up at 100 Mbps, full duplex
> >Aug  3 06:47:15 atlovav1a kernel: [626888.218844] tg3 0000:03:00.0: eth0: Flow control is on for TX and on for RX
> >Aug  3 06:47:15 atlovav1a kernel: [626888.218982] tg3 0000:03:00.0: eth0: EEE is enabled
> >Aug  3 06:51:44 atlovav1a kernel: [627156.293386] tg3 0000:03:00.0: eth0: Link is down
> >Aug  3 06:51:46 atlovav1a kernel: [627159.123347] tg3 0000:03:00.0: eth0: Link is up at 100 Mbps, full duplex
> >Aug  3 06:51:46 atlovav1a kernel: [627159.123491] tg3 0000:03:00.0: eth0: Flow control is on for TX and on for RX
> >Aug  3 06:51:46 atlovav1a kernel: [627159.123629] tg3 0000:03:00.0: eth0: EEE is enabled
> >Aug  3 07:13:10 atlovav1a kernel: [628441.722197] tg3 0000:03:00.0: eth0: Link is down
> >Aug  3 07:13:13 atlovav1a kernel: [628444.615548] tg3 0000:03:00.0: eth0: Link is up at 100 Mbps, full duplex
> >Aug  3 07:13:13 atlovav1a kernel: [628444.615690] tg3 0000:03:00.0: eth0: Flow control is on for TX and on for RX
> >Aug  3 07:13:13 atlovav1a kernel: [628444.615828] tg3 0000:03:00.0: eth0: EEE is enabled
> >
> >I can't exclude the possibility it's switch problem, but I don't have access to box now and the
> >switch is not managed, so I'd like to try other ways first..
> >
> >Could this somehow be power management related? I don't see what else PM-related could I disable
> >in the bios to disable this EEE stuff.. Didn't anyone met similar problem?
> >
> >the system is running Centos6 with x86_64 3.0.76 kernel
> >
> >thanks in advance for any reply
> >
> >BR
> >
> >nik
> >
> >
> 

-- 
-------------------------------------
Ing. Nikola CIPRICH
LinuxBox.cz, s.r.o.
28. rijna 168, 709 00 Ostrava

tel.:   +420 591 166 214
fax:    +420 596 621 273
mobil:  +420 777 093 799

www.linuxbox.cz

mobil servis: +420 737 238 656
email servis: servis@linuxbox.cz
-------------------------------------

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]