Tobias Powalowski schrieb:

Hi guys, kernel26-2.6.30-3 please signoff, both arches.

Double-signoff x86_64: - Core 2 Duo, nvidia (latest, proprietary) graphics - Core 2 Duo, laptop, intel graphics with KMS

signoff x86_64. desktop with nouveau running and laptop with ati card. all dmesg looks ok. -Andy

On Sat, Jun 13, 2009 at 11:54 AM, Tobias Powalowski wrote:

Hi guys, kernel26-2.6.30-3 please signoff, both arches.

signoff x86_64 (nvidia proprietary, desktop) Ronald

On Sat, Jun 13, 2009 at 5:54 AM, Tobias Powalowski wrote:

...

Hi guys, kernel26-2.6.30-3 please signoff, both arches.

I'm having network problems (x86_64, forcedeth driver) with the kernel in testing. Upon boot up, netcfg fails to start the network. I just get a "- No connection" error message. After the system boot is completed, restarting net-profiles still fails. I have noticed that if I disconnect and reconnect the ethernet cables, restarting the daemon succeeds and network is working fine. This happened with both -2 and -3 kernels. Does anyone knows what might be causing this? Does anyone else is experiencing this?

BTW, the kernel works fine on my i686.

Eric Could it be that the new netcfg scripts are a bit buggy here, my custom bridge

Am Montag 15 Juni 2009 schrieb Eric Bélanger: script fails too with the same error. Digging a bit in netcfg code shows it uses ip link | grep UP which is not working here it says UNKNOWN, maybe James can enlighten us more. greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org

Thomas Bächler schrieb:

My guess as well, the link detection might be broken. netcfg should really have an option to force enabling the interface even if link detection fails.

However, when no link is detected, some ethernet drivers disable TX completely. When it was first introduced for my old dmfe card, it was broken and I could receive packets, but nothing was ever transmitted - after my bug report they reverted the changes and reintroduced a working link detection in the next version. My point is, even if you fix netcfg, Eric's driver might be just as broken so he still has to reconnect the cable to make it work.

Might be nothing, but you could play with the module parameter mentioned here: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=...

On Tue, Jun 16, 2009 at 12:54 AM, Eric Bélanger wrote:

On Mon, Jun 15, 2009 at 10:19 AM, Thomas Bächler wrote:

...

Tobias Powalowski schrieb:

...

Could it be that the new netcfg scripts are a bit buggy here, my custom bridge script fails too with the same error. Digging a bit in netcfg code shows it uses ip link | grep UP which is not working here it says UNKNOWN, maybe James can enlighten us more.

My guess as well, the link detection might be broken. netcfg should really have an option to force enabling the interface even if link detection fails.

However, when no link is detected, some ethernet drivers disable TX completely. When it was first introduced for my old dmfe card, it was broken and I could receive packets, but nothing was ever transmitted - after my bug report they reverted the changes and reintroduced a working link detection in the next version. My point is, even if you fix netcfg, Eric's driver might be just as broken so he still has to reconnect the cable to make it work.

I tested with the netcfg in core and it works without any problems. So it seems to be an issue with the netcfg in testing.

Update: I tried netcfg 2.2.0b3-1 and it works. My previous failiures were with 2.2.0b2.