On Sat, 2012-06-16 at 14:52 +0530, Ritesh Raj Sarraf wrote:
> Added Ben and the Debian Kernel Team
>
>
> On Saturday 16 June 2012 07:29 AM, Guillem Jover wrote:
> >> On Friday 15 June 2012 09:27 PM, Guillem Jover wrote:
> >>> > > Since some time now (I only sat down to track it down pretty recently,
> >>> > > but this has going on for probably a year or more), laptop-mode-tools
> >>> > > has broken the wired network (e1000e) whenever I unplug the laptop from
> >>> > > the wall power.
> >>> > >
> >>> > > The culprit is BATT_THROTTLE_ETHERNET=1 in
> >>> > > «/etc/laptop-mode/conf.d/ethernet.conf».
> >>> > >
> >>> > > This turns the working ethernet device from this state:
> >>> > >
> >>> > > ,---
> >>> > > # ethtool eth0
> >>> > > Settings for eth0:
> >>> > > Supported ports: [ TP ]
> >>> > > Supported link modes: 10baseT/Half 10ba
> >>> > > 100baseT/Half 100baseT/Full
> >>> > > 1000baseT/Full
> >>> > > Supported pause frame use: No
> >>> > > Supports auto-negotiation: Yes
> >>> > > Advertised link modes: 10baseT/Half 10ba
> >>> > > 100baseT/Half 100baseT/Full
> >>> > > 1000baseT/Full
> >>> > > Advertised pause frame use: No
> >>> > > Advertised auto-negotiation: Yes
> >>> > > Speed: 100Mb/s
> >>> > > Duplex: Full
> >>> > > Port: Twisted Pair
> >>> > > PHYAD: 2
> >>> > > Transceiver: internal
> >>> > > Auto-negotiation: on
> >>> > > MDI-X: on
> >>> > > Supports Wake-on: pumbg
> >>> > > Wake-on: g
> >>> > > Current message level: 0x00000001 (1)
> >>> > > drv
> >>> > > Link detected: yes
> >>> > > `---
> >>> > >
> >>> > > To this non-working state:
> >>> > >
> >>> > > ,---
> >>> > > # ethtool eth0
> >>> > > Settings for eth0:
> >>> > > Supported ports: [ TP ]
> >>> > > Supported link modes: 10baseT/Half 10ba
> >>> > > 100baseT/Half 100baseT/Full
> >>> > > 1000baseT/Full
> >>> > > Supported pause frame use: No
> >>> > > Supports auto-negotiation: Yes
> >>> > > Advertised link modes: 10baseT/Half 10ba
> >>> > > 100baseT/Half 100baseT/Full
> >>> > > 1000baseT/Full
> >>> > > Advertised pause frame use: No
> >>> > > Advertised auto-negotiation: Yes
> >>> > > Speed: Unknown!
> >>> > > Duplex: Unknown! (255)
> >>> > > Port: Twisted Pair
> >>> > > PHYAD: 2
> >>> > > Transceiver: internal
> >>> > > Auto-negotiation: on
> >>> > > MDI-X: Unknown
> >>> > > Supports Wake-on: pumbg
> >>> > > Wake-on: d
> >>> > > Current message level: 0x00000001 (1)
> >>> > > drv
> >>> > > Link detected: no
> >>> > > `---
> >>> > >
> >>> > > If I set BATT_THROTTLE_ETHERNET to 0, then everything works as before.
> >>> > >
> >>> > > I don't think this setting should be enabled by default if it might
> >>> > > cause this type of breakage, even if it ends up being a driver
> >>> > > problem.
> >> >
> >> > This is the first report on this behavior.
> >> > From your logs, it shows that when switched to battery, nothing is set
> >> > correct. Neither the speed, nor the Link.
> >> >
> >> > Could you please set DEBUG=1 in the ethernet module only, and then see
> >> > if it can provide more information on what is failing?
> >> >
> >> > e1000e must be a common driver supporting multiple Intel chipsets. I
> >> > have no problem setting the default to 0, but like I said, this is the
> >> > first time I've heard of this behavior.
> > I've tracked it further down to «ethtool -s eth0 autoneg off», which
> > totally messes up the network device. I've tested this with 3.2, 3.3
> > and 3.4 kernels, all are affected. I'll try to check later on with
> > earlier kernels to see if something changes.
Turning off autonegotiation is a totally fucking stupid thing to do.
But, that said, both the above reports have it turned on.
(The right thing to do is leave autonegotiation on but only advertise
the speed you want. Also, coming soon to ethtool: Energy Efficient
Ethernet settings.)
> > It might be related to the PCIe controller, ACPI setup, the PM settings,
> > or the driver itself, I've got some error messages on dmesg about the
> > driver being unable to read from PHY registers and also about Hardware
> > Errors, but then I've seen several instances of that on google, so I
> > don't think it's really a hardware problem, as those have changed over
> > time with different kernels.
>
> Since this happens with 3.2 also, I have pulled in the kernel team.
> Perhaps they have some information.
I rarely use wired networking on my laptop but sometimes find that
e1000e fails to link up. Re-loading the module fixes it. Never got
round to reporting this as a bug, though. I don't have
laptop-mode-tools installed.
Ben.
--
Ben Hutchings
We get into the habit of living before acquiring the habit of thinking.
- Albert Camus
Attachment:
signature.asc
Description: This is a digitally signed message part