(No subject)

(No subject)

[Denis 'GNUtoo' Carikli]

Hi,

Is it possible to move /etc/rcS.d/S98configure before S40networking, 
so usb0 comes up at first boot, or is there some problems that would 
be introduced from such move?

Here's my problem:
on the bug machine,with Angstrom, at first boot usb0 doesn't come up 
in a console-image.
It is very annoying because the only way to get control of the machine 
is trough ssh on usbnet.

At second boot it works fine.

If I move /etc/rcS.d/S98configure before S40networking it 
works fine at first boot.

I will send a patch that changes the order of S40networking  
and S98configure to S40configure and S41networking.

Denis.

[PATCH] netbase and opkg: reorder init scripts to make networking come up at first boot

[Denis 'GNUtoo' Carikli]

Without that fix, usb0 doesn't come up at first boot
  (tested on the bug machine, with Angstrom on a console-image).

Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@no-log.org>
---
 recipes/netbase/netbase_4.21.bb |    2 +-
 recipes/opkg/files/configure    |    2 +-
 recipes/opkg/opkg.inc           |    4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/recipes/netbase/netbase_4.21.bb b/recipes/netbase/netbase_4.21.bb
index f3ae6e9..075c9db 100644
--- a/recipes/netbase/netbase_4.21.bb
+++ b/recipes/netbase/netbase_4.21.bb
@@ -9,7 +9,7 @@ inherit update-rc.d
 SRC_URI_OVERRIDES_PACKAGE_ARCH = "1"
 
 INITSCRIPT_NAME = "networking"
-INITSCRIPT_PARAMS = "start 40 S . stop 40 0 6 1 ."
+INITSCRIPT_PARAMS = "start 41 S . stop 41 0 6 1 ."
 # On MNCI etc, start very late so that our own apps come up faster
 INITSCRIPT_PARAMS_openmn = "start 85 1 2 3 4 5 . stop 85 0 6 1 ."
 # On SlugOS (NSLU2) delay the stop until after network apps have exited
diff --git a/recipes/opkg/files/configure b/recipes/opkg/files/configure
index d5dd74d..633d8d3 100644
--- a/recipes/opkg/files/configure
+++ b/recipes/opkg/files/configure
@@ -22,4 +22,4 @@ else
 fi
 
 # delete myself
-rm -f /etc/rcS.d/S98configure
+rm -f /etc/rcS.d/S40configure
diff --git a/recipes/opkg/opkg.inc b/recipes/opkg/opkg.inc
index 6970f2f..8980c14 100644
--- a/recipes/opkg/opkg.inc
+++ b/recipes/opkg/opkg.inc
@@ -5,7 +5,7 @@ LICENSE = "GPLv2"
 SRCREV = "609"
 PV = "0.1.8+svnr${SRCPV}"
 CONFLICTS = "ipkg"
-INC_PR = "r4"
+INC_PR = "r5"
 
 SRC_URI = "svn://opkg.googlecode.com/svn;module=trunk;proto=http \
 	   file://configure \
@@ -17,7 +17,7 @@ inherit autotools pkgconfig
 
 do_install_prepend() {
   install -d ${D}${sysconfdir}/rcS.d
-  install -m 0755 ${WORKDIR}/configure ${D}${sysconfdir}/rcS.d/S98configure
+  install -m 0755 ${WORKDIR}/configure ${D}${sysconfdir}/rcS.d/S40configure
 }
 
 EXTRA_OECONF += " --disable-gpg \ 
-- 
1.7.0.4

Re: [PATCH] netbase and opkg: reorder init scripts to make networking come up at first boot

[Phil Blundell]

On Tue, 2011-03-01 at 18:45 +0100, Denis 'GNUtoo' Carikli wrote:
> Without that fix, usb0 doesn't come up at first boot
>   (tested on the bug machine, with Angstrom on a console-image).
> 
> Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@no-log.org>

That sounds like it's just papering over some kind of problem in
netbase.  What exactly stops usb0 from working with the previous
arrangement?

p.

Re: [PATCH] netbase and opkg: reorder init scripts to make networking come up at first boot

[Denis 'GNUtoo' Carikli]

On Tue, 2011-03-01 at 17:58 +0000, Phil Blundell wrote:
> On Tue, 2011-03-01 at 18:45 +0100, Denis 'GNUtoo' Carikli wrote:
> > Without that fix, usb0 doesn't come up at first boot
> >   (tested on the bug machine, with Angstrom on a console-image).
> > 
> > Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@no-log.org>
> 
> That sounds like it's just papering over some kind of problem in
> netbase.  What exactly stops usb0 from working with the previous
> arrangement?
> 
> p.
The usb0 interface doesn't exist when it's configured by netbase in the
first boot.

Altough it's there in the second boot.

Denis.

(No subject)

[Denis 'GNUtoo' Carikli]

Sorry for the lack of Subject, 

I tried to send a first mail and had an error while sending it.

Then I pasted the content of the saved mail(git-send-email often errors
for me so I saved it as a precaution),forgetting to paste the Subject.

Denis.

Re: [PATCH] netbase and opkg: reorder init scripts to make networking come up at first boot

[Tom Rini]

On 03/01/2011 10:58 AM, Phil Blundell wrote:
> On Tue, 2011-03-01 at 18:45 +0100, Denis 'GNUtoo' Carikli wrote:
>> Without that fix, usb0 doesn't come up at first boot
>>    (tested on the bug machine, with Angstrom on a console-image).
>>
>> Signed-off-by: Denis 'GNUtoo' Carikli<GNUtoo@no-log.org>
>
> That sounds like it's just papering over some kind of problem in
> netbase.  What exactly stops usb0 from working with the previous
> arrangement?

Presumably having not configured the system.  We should try and look at 
moving opkg's configure up to be as early as possible

-- 
Tom Rini
Mentor Graphics Corporation

Re: [PATCH] netbase and opkg: reorder init scripts to make networking come up at first boot

[Phil Blundell]

On Tue, 2011-03-01 at 11:27 -0700, Tom Rini wrote:
> On 03/01/2011 10:58 AM, Phil Blundell wrote:
> > On Tue, 2011-03-01 at 18:45 +0100, Denis 'GNUtoo' Carikli wrote:
> >> Without that fix, usb0 doesn't come up at first boot
> >>    (tested on the bug machine, with Angstrom on a console-image).
> >>
> >> Signed-off-by: Denis 'GNUtoo' Carikli<GNUtoo@no-log.org>
> >
> > That sounds like it's just papering over some kind of problem in
> > netbase.  What exactly stops usb0 from working with the previous
> > arrangement?
> 
> Presumably having not configured the system.  We should try and look at 
> moving opkg's configure up to be as early as possible

It oughtn't really to matter.  Generally, if a package depends on having
been configured before its initscript runs, that means it will also fail
to work in the case where it's subsequently installed (or upgraded) on a
running system.  (Obviously that's a nonissue for distros that don't use
O_P_M but it would still be nice to get it right in the metadata.)

Also, packages which depend on having "opkg configure" run on the target
before they will work are a bit of a pain for folks with readonly
rootfses.  In some cases that's just unavoidable but I would have
sort-of hoped that a base package like netbase would manage this a bit
better.

p.