Patchwork connman: Ignore eth0 cleanup when booting over the network

login
register
mail settings
Submitter Saul Wold
Date Aug. 30, 2012, 11:34 p.m.
Message ID <1346369662-17265-1-git-send-email-sgw@linux.intel.com>
Download mbox | patch
Permalink /patch/35571/
State New
Headers show

Comments

Saul Wold - Aug. 30, 2012, 11:34 p.m.
There was a change to connman, such that it cleans up the
route table for devices at startup, this was causing the
network to get lost and NFS to loose it connection.

The -I option will cause the device to be ignored.

[YOCTO #3008]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
---
 meta/recipes-connectivity/connman/connman/connman |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Koen Kooi - Aug. 31, 2012, 7:28 a.m.
Op 31 aug. 2012, om 01:34 heeft Saul Wold <sgw@linux.intel.com> het volgende geschreven:

> There was a change to connman, such that it cleans up the
> route table for devices at startup, this was causing the
> network to get lost and NFS to loose it connection.
> 
> The -I option will cause the device to be ignored.

Apart from the missing PR bump, what happens when you don't boot of the network? Will connman keep ignoring eth0?
FWIW I asked about nfsroot and connman and Marcel said: don't do that! So I'm inclined to say that if you're using nfsroot you shouldn't be using connman.
Simon Busch - Aug. 31, 2012, 7:42 a.m.
Am 31.08.2012 01:34, schrieb Saul Wold:
> There was a change to connman, such that it cleans up the
> route table for devices at startup, this was causing the
> network to get lost and NFS to loose it connection.
> 
> The -I option will cause the device to be ignored.
> 
> [YOCTO #3008]
> 
> Signed-off-by: Saul Wold <sgw@linux.intel.com>
> ---
>  meta/recipes-connectivity/connman/connman/connman |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/meta/recipes-connectivity/connman/connman/connman b/meta/recipes-connectivity/connman/connman/connman
> index f01bf37..2c8354a 100644
> --- a/meta/recipes-connectivity/connman/connman/connman
> +++ b/meta/recipes-connectivity/connman/connman/connman
> @@ -29,7 +29,7 @@ done
>  do_start() {
>  	EXTRA_PARAM=""
>  	if test $nfsroot -eq 1 ; then
> -		EXTRA_PARAM="-P ethernet"
> +		EXTRA_PARAM="-P ethernet -I eth0"
>  	fi
>  	$DAEMON $EXTRA_PARAM
>  }
> 

What about other devices where eht0 is not available but rndis0 for
example which has to be ignored?

As of right we have this fix for several machines in meta-smartphone
already ignoring rndis0 and usb0.

regards,
Simon
Saul Wold - Aug. 31, 2012, 4:48 p.m.
On 08/31/2012 12:28 AM, Koen Kooi wrote:
>
> Op 31 aug. 2012, om 01:34 heeft Saul Wold <sgw@linux.intel.com> het volgende geschreven:
>
>> There was a change to connman, such that it cleans up the
>> route table for devices at startup, this was causing the
>> network to get lost and NFS to loose it connection.
>>
>> The -I option will cause the device to be ignored.
>
> Apart from the missing PR bump, what happens when you don't boot of the network? Will connman keep ignoring eth0?
> FWIW I asked about nfsroot and connman and Marcel said: don't do that! So I'm inclined to say that if you're using nfsroot you shouldn't be using connman.
>
What if there are multiple network interfaces, but just one (eth0, usb0, 
...) is the nfsroot provider?

Is there a way to determine which interface nfs is active on and only 
ignore that one?

I am going to disable connmand for now in a v2 patch, but I think we 
need to address this better either via connman or here in oe-core.

Sau!

Patch

diff --git a/meta/recipes-connectivity/connman/connman/connman b/meta/recipes-connectivity/connman/connman/connman
index f01bf37..2c8354a 100644
--- a/meta/recipes-connectivity/connman/connman/connman
+++ b/meta/recipes-connectivity/connman/connman/connman
@@ -29,7 +29,7 @@  done
 do_start() {
 	EXTRA_PARAM=""
 	if test $nfsroot -eq 1 ; then
-		EXTRA_PARAM="-P ethernet"
+		EXTRA_PARAM="-P ethernet -I eth0"
 	fi
 	$DAEMON $EXTRA_PARAM
 }