[PATCH 1/2] udev: fixed obsolete udevadm syntax for udev init.d script.

[PATCH 1/2] udev: fixed obsolete udevadm syntax for udev init.d script.

[Ihar Hrachyshka]

Fixed the following warning message when executing udev init script:
"udevadm control commands requires the --<command> format, this will
stop working in a future release."

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
---
 recipes/udev/udev-124/init |    4 ++--
 recipes/udev/udev-141/init |    4 ++--
 recipes/udev/udev_124.bb   |    2 +-
 recipes/udev/udev_141.bb   |    2 +-
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/recipes/udev/udev-124/init b/recipes/udev/udev-124/init
index efe4470..7188b5a 100644
--- a/recipes/udev/udev-124/init
+++ b/recipes/udev/udev-124/init
@@ -56,10 +56,10 @@ kill_udevd > "/dev/null" 2>&1
         echo -e '\000\000\000\000' > /proc/sys/kernel/hotplug
         /sbin/udevd -d
 
-        /sbin/udevadm control env STARTUP=1
+        /sbin/udevadm control --env STARTUP=1
 		if [ "$not_first_boot" != "" ];then
 			/sbin/udevadm trigger --subsystem-nomatch=tty --subsystem-nomatch=mem --subsystem-nomatch=vc --subsystem-nomatch=vtconsole --subsystem-nomatch=misc --subsystem-nomatch=dcon --subsystem-nomatch=pci_bus  --subsystem-nomatch=graphics  --subsystem-nomatch=backlight --subsystem-nomatch=video4linux  --subsystem-nomatch=platform
-			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control env STARTUP=)&
+			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control --env STARTUP=)&
 			if [ "$DEVCACHE" != "" ]; then
 				rm -f /tmp/uname
 				rm -f /tmp/cmdline
diff --git a/recipes/udev/udev-141/init b/recipes/udev/udev-141/init
index efe4470..7188b5a 100644
--- a/recipes/udev/udev-141/init
+++ b/recipes/udev/udev-141/init
@@ -56,10 +56,10 @@ kill_udevd > "/dev/null" 2>&1
         echo -e '\000\000\000\000' > /proc/sys/kernel/hotplug
         /sbin/udevd -d
 
-        /sbin/udevadm control env STARTUP=1
+        /sbin/udevadm control --env STARTUP=1
 		if [ "$not_first_boot" != "" ];then
 			/sbin/udevadm trigger --subsystem-nomatch=tty --subsystem-nomatch=mem --subsystem-nomatch=vc --subsystem-nomatch=vtconsole --subsystem-nomatch=misc --subsystem-nomatch=dcon --subsystem-nomatch=pci_bus  --subsystem-nomatch=graphics  --subsystem-nomatch=backlight --subsystem-nomatch=video4linux  --subsystem-nomatch=platform
-			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control env STARTUP=)&
+			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control --env STARTUP=)&
 			if [ "$DEVCACHE" != "" ]; then
 				rm -f /tmp/uname
 				rm -f /tmp/cmdline
diff --git a/recipes/udev/udev_124.bb b/recipes/udev/udev_124.bb
index 834a70a..be545f7 100644
--- a/recipes/udev/udev_124.bb
+++ b/recipes/udev/udev_124.bb
@@ -2,7 +2,7 @@ DESCRIPTION = "udev is a daemon which dynamically creates and removes device nod
 /dev/, handles hotplug events and loads drivers at boot time. It replaces \
 the hotplug package and requires a kernel not older than 2.6.12."
 RPROVIDES_${PN} = "hotplug"
-PR = "r17"
+PR = "r18"
 
 SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
 	   file://noasmlinkage.patch;patch=1 \
diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
index e7438fb..7e26798 100644
--- a/recipes/udev/udev_141.bb
+++ b/recipes/udev/udev_141.bb
@@ -6,7 +6,7 @@ LICENSE = "GPL"
 # Untested
 DEFAULT_PREFERENCE = "-1"
 
-PR = "r6"
+PR = "r7"
 
 SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
 	   file://mount.blacklist \
-- 
1.6.3.1

[PATCH 2/2] udev: moved udevadm to the main udev package

[Ihar Hrachyshka]

The reason for the move is that udevadm binary is executed from udev
init.d script.

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
---
 recipes/udev/udev_141.bb |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
index 7e26798..42b96e4 100644
--- a/recipes/udev/udev_141.bb
+++ b/recipes/udev/udev_141.bb
@@ -6,7 +6,7 @@ LICENSE = "GPL"
 # Untested
 DEFAULT_PREFERENCE = "-1"
 
-PR = "r7"
+PR = "r8"
 
 SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
 	   file://mount.blacklist \
@@ -41,11 +41,11 @@ INITSCRIPT_PARAMS = "start 03 S ."
 PACKAGES =+ "udev-utils libvolume-id libvolume-id-dev"
 
 FILES_libvolume-id-dev = "${includedir}/libvolume_id.h ${libdir}/libvolume_id.a ${libdir}/libvolume_id.so ${libdir}/pkgconfig/libvolume_id.pc"
-FILES_udev-utils = "${usrbindir}/udevinfo ${sbindir}/udevadm ${usrbindir}/udevtest"
+FILES_udev-utils = "${usrbindir}/udevinfo ${usrbindir}/udevtest"
 FILES_libvolume-id = "${base_libdir}/libvolume_id.so.*"
 
 RPROVIDES_${PN} = "hotplug"
-FILES_${PN} += "${usrbindir}/* ${usrsbindir}/udevd"
+FILES_${PN} += "${usrbindir}/* ${sbindir}/udevadm ${usrsbindir}/udevd"
 FILES_${PN}-dbg += "${usrbindir}/.debug ${usrsbindir}/.debug"
 
 FILES_${PN} += "${libdir}/udev/* ${base_libdir}/udev/*"
-- 
1.6.3.1

Re: [PATCH 1/2] udev: fixed obsolete udevadm syntax for udev init.d script.

[Phil Blundell]

On Sun, 2009-05-24 at 21:33 +0300, Ihar Hrachyshka wrote:
> Fixed the following warning message when executing udev init script:
> "udevadm control commands requires the --<command> format, this will
> stop working in a future release."
> 
> Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
> ---
>  recipes/udev/udev-124/init |    4 ++--
>  recipes/udev/udev-141/init |    4 ++--
>  recipes/udev/udev_124.bb   |    2 +-
>  recipes/udev/udev_141.bb   |    2 +-
>  4 files changed, 6 insertions(+), 6 deletions(-)

There doesn't seem much point in patching udev 124 for this: that
version doesn't actually emit the warning as far as I can tell.  If the
initscript is maintained separately for different versions then you
might as well just patch the -141 version and save a rebuild/reinstall
for people using 124.  Or, alternatively, if the two initscripts are
actually the same, maybe you could unify them and save on future
maintenance.

p.

[PATCH 1/4] udev: fixed obsolete udevadm syntax for udev init.d script.

[Ihar Hrachyshka]

Fixed the following warning message when executing udev init script:
"udevadm control commands requires the --<command> format, this will
stop working in a future release."

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
---
 recipes/udev/udev-141/init |    4 ++--
 recipes/udev/udev_141.bb   |    2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/recipes/udev/udev-141/init b/recipes/udev/udev-141/init
index efe4470..7188b5a 100644
--- a/recipes/udev/udev-141/init
+++ b/recipes/udev/udev-141/init
@@ -56,10 +56,10 @@ kill_udevd > "/dev/null" 2>&1
         echo -e '\000\000\000\000' > /proc/sys/kernel/hotplug
         /sbin/udevd -d
 
-        /sbin/udevadm control env STARTUP=1
+        /sbin/udevadm control --env STARTUP=1
 		if [ "$not_first_boot" != "" ];then
 			/sbin/udevadm trigger --subsystem-nomatch=tty --subsystem-nomatch=mem --subsystem-nomatch=vc --subsystem-nomatch=vtconsole --subsystem-nomatch=misc --subsystem-nomatch=dcon --subsystem-nomatch=pci_bus  --subsystem-nomatch=graphics  --subsystem-nomatch=backlight --subsystem-nomatch=video4linux  --subsystem-nomatch=platform
-			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control env STARTUP=)&
+			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control --env STARTUP=)&
 			if [ "$DEVCACHE" != "" ]; then
 				rm -f /tmp/uname
 				rm -f /tmp/cmdline
diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
index e7438fb..7e26798 100644
--- a/recipes/udev/udev_141.bb
+++ b/recipes/udev/udev_141.bb
@@ -6,7 +6,7 @@ LICENSE = "GPL"
 # Untested
 DEFAULT_PREFERENCE = "-1"
 
-PR = "r6"
+PR = "r7"
 
 SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
 	   file://mount.blacklist \
-- 
1.6.3.1

Re: [PATCH 1/2] udev: fixed obsolete udevadm syntax for udev init.d script.

[Tom Rini]

On Sun, May 24, 2009 at 08:18:10PM +0100, Phil Blundell wrote:
> On Sun, 2009-05-24 at 21:33 +0300, Ihar Hrachyshka wrote:
> > Fixed the following warning message when executing udev init script:
> > "udevadm control commands requires the --<command> format, this will
> > stop working in a future release."
> > 
> > Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
> > ---
> >  recipes/udev/udev-124/init |    4 ++--
> >  recipes/udev/udev-141/init |    4 ++--
> >  recipes/udev/udev_124.bb   |    2 +-
> >  recipes/udev/udev_141.bb   |    2 +-
> >  4 files changed, 6 insertions(+), 6 deletions(-)
> 
> There doesn't seem much point in patching udev 124 for this: that
> version doesn't actually emit the warning as far as I can tell.  If the
> initscript is maintained separately for different versions then you
> might as well just patch the -141 version and save a rebuild/reinstall
> for people using 124.  Or, alternatively, if the two initscripts are
> actually the same, maybe you could unify them and save on future
> maintenance.

I disagree.  There are, I assume, things that could be done on top of
the -141 script as further enchancements/cleanups that would not apply
to -124.  But keeping older files in sync, so long as (a) it works and
(b) the older version is cared about makes future buging easier, I've
found.

-- 
Tom Rini

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Marcin Juszkiewicz]

Dnia niedziela, 24 maja 2009 o 20:33:05 Ihar Hrachyshka napisał(a):
> The reason for the move is that udevadm binary is executed from udev
> init.d script.

NACK, NACK, NACK - udevadm is also udevinfo and is used by HAL. Your 
patch will make my mdev images not working properly because I would need 
to install mdev and udev to get HAL working and then find a way to not 
run udev at boot time (even after package upgrades).

Regards, 
-- 
JID:      hrw@jabber.org
Website:  http://marcin.juszkiewicz.com.pl/
LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Ihar Hrachyshka]

On Mon, May 25, 2009 at 12:25 PM, Marcin Juszkiewicz
<marcin@juszkiewicz.com.pl> wrote:
> Dnia niedziela, 24 maja 2009 o 20:33:05 Ihar Hrachyshka napisał(a):
>> The reason for the move is that udevadm binary is executed from udev
>> init.d script.
>
> NACK, NACK, NACK - udevadm is also udevinfo and is used by HAL. Your
> patch will make my mdev images not working properly because I would need
> to install mdev and udev to get HAL working and then find a way to not
> run udev at boot time (even after package upgrades).

I tried to understand your complain (really) but don't get it. What do
you mean when saying "udevadm is also udevinfo and is used by HAL"?
And also please provide your solution for the problem of broken images
with unfuctional udev subsystem.

>
> Regards,
> --
> JID:      hrw@jabber.org
> Website:  http://marcin.juszkiewicz.com.pl/
> LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz
>
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Koen Kooi]

On 25-05-09 11:54, Ihar Hrachyshka wrote:
> On Mon, May 25, 2009 at 12:25 PM, Marcin Juszkiewicz
> <marcin@juszkiewicz.com.pl>  wrote:
>> Dnia niedziela, 24 maja 2009 o 20:33:05 Ihar Hrachyshka napisał(a):
>>> The reason for the move is that udevadm binary is executed from udev
>>> init.d script.
>>
>> NACK, NACK, NACK - udevadm is also udevinfo and is used by HAL. Your
>> patch will make my mdev images not working properly because I would need
>> to install mdev and udev to get HAL working and then find a way to not
>> run udev at boot time (even after package upgrades).
>
> I tried to understand your complain (really) but don't get it. What do
> you mean when saying "udevadm is also udevinfo and is used by HAL"?
> And also please provide your solution for the problem of broken images
> with unfuctional udev subsystem.

If you want to have a system that uses mdev as /dev manager and hal, you 
need to install udevadm to make hal happy. I think the solution is to do 
something like this:

RDEPENDS_udev += "udev-utils" in udev_141.bb

regards,

Koen

PS: mdev users need to start thinking about how devicekit and 
util-linux-ng will integrate into their systems, since the hal world 
seems to be moving away from hal to libudev + util-linux-ng

regards,

Koen

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Phil Blundell]

On Mon, 2009-05-25 at 12:54 +0300, Ihar Hrachyshka wrote:
> On Mon, May 25, 2009 at 12:25 PM, Marcin Juszkiewicz
> <marcin@juszkiewicz.com.pl> wrote:
> > Dnia niedziela, 24 maja 2009 o 20:33:05 Ihar Hrachyshka napisał(a):
> >> The reason for the move is that udevadm binary is executed from udev
> >> init.d script.
> >
> > NACK, NACK, NACK - udevadm is also udevinfo and is used by HAL. Your
> > patch will make my mdev images not working properly because I would need
> > to install mdev and udev to get HAL working and then find a way to not
> > run udev at boot time (even after package upgrades).
> 
> I tried to understand your complain (really) but don't get it. What do
> you mean when saying "udevadm is also udevinfo and is used by HAL"?
> And also please provide your solution for the problem of broken images
> with unfuctional udev subsystem.

I think he's saying that udevadm is needed by HAL for correct operation,
even if you're using mdev instead of udev itself.

It looks like udev used to Depend on udev-utils due to the existence
of /sbin/udevsettle in version 124.  But, in version 141, that symlink
has gone away and hence the dependency is no longer created.  The right
fix is probably to add an explicit RDEPENDS_${PN} += "udev-utils" or
some such.

p.

[PATCH] udev: include udev-utils into udev rootfs

[Ihar Hrachyshka]

The reason for the move is that udevadm binary is executed from udev
init.d script.

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
---
 recipes/udev/udev_141.bb |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
index 7e26798..068d23c 100644
--- a/recipes/udev/udev_141.bb
+++ b/recipes/udev/udev_141.bb
@@ -6,7 +6,10 @@ LICENSE = "GPL"
 # Untested
 DEFAULT_PREFERENCE = "-1"
 
-PR = "r7"
+PR = "r8"
+
+# needed for init.d script
+RDEPENDS += "udev-utils"
 
 SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
 	   file://mount.blacklist \
-- 
1.6.3.1

[PATCH] udev: include udev-utils into udev rootfs

[Ihar Hrachyshka]

The reason for the move is that udevadm binary is executed from udev
init.d script.

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
---
 recipes/udev/udev_141.bb |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
index 7e26798..068d23c 100644
--- a/recipes/udev/udev_141.bb
+++ b/recipes/udev/udev_141.bb
@@ -6,7 +6,10 @@ LICENSE = "GPL"
 # Untested
 DEFAULT_PREFERENCE = "-1"
 
-PR = "r7"
+PR = "r8"
+
+# needed for init.d script
+RDEPENDS_udev += "udev-utils"
 
 SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
 	   file://mount.blacklist \
-- 
1.6.3.1

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Graeme Gregory]

On Mon, 2009-05-25 at 12:24 +0200, Koen Kooi wrote:
> PS: mdev users need to start thinking about how devicekit and 
> util-linux-ng will integrate into their systems, since the hal world 
> seems to be moving away from hal to libudev + util-linux-ng
> 

I think the mdev market is for devices not using hal/devicekit to boot.
Personally I think if your trying to replace udev with mdev on a device
that still need hal/devicekit for dynamic devices your probably peeing
into the wind!

Graeme

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Ihar Hrachyshka]

On Mon, May 25, 2009 at 1:30 PM, Phil Blundell <philb@gnu.org> wrote:
> On Mon, 2009-05-25 at 12:54 +0300, Ihar Hrachyshka wrote:
>> On Mon, May 25, 2009 at 12:25 PM, Marcin Juszkiewicz
>> <marcin@juszkiewicz.com.pl> wrote:
>> > Dnia niedziela, 24 maja 2009 o 20:33:05 Ihar Hrachyshka napisał(a):
>> >> The reason for the move is that udevadm binary is executed from udev
>> >> init.d script.
>> >
>> > NACK, NACK, NACK - udevadm is also udevinfo and is used by HAL. Your
>> > patch will make my mdev images not working properly because I would need
>> > to install mdev and udev to get HAL working and then find a way to not
>> > run udev at boot time (even after package upgrades).
>>
>> I tried to understand your complain (really) but don't get it. What do
>> you mean when saying "udevadm is also udevinfo and is used by HAL"?
>> And also please provide your solution for the problem of broken images
>> with unfuctional udev subsystem.
>
> I think he's saying that udevadm is needed by HAL for correct operation,
> even if you're using mdev instead of udev itself.
>
> It looks like udev used to Depend on udev-utils due to the existence
> of /sbin/udevsettle in version 124.  But, in version 141, that symlink
> has gone away and hence the dependency is no longer created.  The right
> fix is probably to add an explicit RDEPENDS_${PN} += "udev-utils" or
> some such.

Just sent the fix to ML.

>
> p.
>
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH] udev: include udev-utils into udev rootfs

[Koen Kooi]

On 25-05-09 12:51, Ihar Hrachyshka wrote:
> The reason for the move is that udevadm binary is executed from udev
> init.d script.

Acked-by: Koen Kooi <koen@openembedded.org>


>
> Signed-off-by: Ihar Hrachyshka<ihar.hrachyshka@gmail.com>
> ---
>   recipes/udev/udev_141.bb |    5 ++++-
>   1 files changed, 4 insertions(+), 1 deletions(-)
>
> diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
> index 7e26798..068d23c 100644
> --- a/recipes/udev/udev_141.bb
> +++ b/recipes/udev/udev_141.bb
> @@ -6,7 +6,10 @@ LICENSE = "GPL"
>   # Untested
>   DEFAULT_PREFERENCE = "-1"
>
> -PR = "r7"
> +PR = "r8"
> +
> +# needed for init.d script
> +RDEPENDS_udev += "udev-utils"
>
>   SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
>   	   file://mount.blacklist \

Re: [PATCH] udev: include udev-utils into udev rootfs

[Otavio Salvador]

Hello,

NACK!

On Mon, May 25, 2009 at 8:14 AM, Koen Kooi <k.kooi@student.utwente.nl> wrote:
> On 25-05-09 12:51, Ihar Hrachyshka wrote:
>>
>> The reason for the move is that udevadm binary is executed from udev
>> init.d script.

Wouldn't be better to move udevadm to udev binary?

It makes non-sense to have another package installed by default and
bring more things then required. If it is used by init.d, and this is
ok, so please move the required binary for the udev package and that's
solved.

My 2c.

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Otavio Salvador]

Hello,

Having it in mind

On Mon, May 25, 2009 at 7:55 AM, Ihar Hrachyshka
<ihar.hrachyshka@gmail.com> wrote:
[...]
>> It looks like udev used to Depend on udev-utils due to the existence
>> of /sbin/udevsettle in version 124.  But, in version 141, that symlink
>> has gone away and hence the dependency is no longer created.  The right
>> fix is probably to add an explicit RDEPENDS_${PN} += "udev-utils" or
>> some such.
>
> Just sent the fix to ML.

I belive that the right way to fix it is add a binary package only for
udevadm and make udev depends on it and mdev suggests it.

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 1/2] udev: fixed obsolete udevadm syntax for udev init.d script.

[Otavio Salvador]

Hello Tom,

On Sun, May 24, 2009 at 11:06 PM, Tom Rini <trini@embeddedalley.com> wrote:
> I disagree.  There are, I assume, things that could be done on top of
> the -141 script as further enchancements/cleanups that would not apply
> to -124.  But keeping older files in sync, so long as (a) it works and
> (b) the older version is cared about makes future buging easier, I've
> found.

I also disagree but due different reasons. I believe we ought to drop
the old version except if there're a real reason to keep it around.
For udev 124 I think there's no reason for it to be kept (maybe I'm
missing something here).

Obviously this ought to be done in dev, not stable, but it can make
our life easier and avoid us to need to care about so many recipes.
The final result is that old versions usually has ugly and unknown
bugs that noone cares about.

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Phil Blundell]

On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
> Hello,
> 
> Having it in mind
> 
> On Mon, May 25, 2009 at 7:55 AM, Ihar Hrachyshka
> <ihar.hrachyshka@gmail.com> wrote:
> [...]
> >> It looks like udev used to Depend on udev-utils due to the existence
> >> of /sbin/udevsettle in version 124.  But, in version 141, that symlink
> >> has gone away and hence the dependency is no longer created.  The right
> >> fix is probably to add an explicit RDEPENDS_${PN} += "udev-utils" or
> >> some such.
> >
> > Just sent the fix to ML.
> 
> I belive that the right way to fix it is add a binary package only for
> udevadm 

That's pretty much what udev-utils is: it doesn't contain any binaries
apart from udevadm.  

> and make udev depends on it and mdev suggests it.

Mdev itself doesn't have any use for udevadm, so having mdev Recommend
udev-utils wouldn't be the right thing; this would just be a waste of
space for people who are using mdev without HAL (which, I suspect, is
actually the majority of mdev users).

p.

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Otavio Salvador]

Hello,

On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
> On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
[...]
>> I belive that the right way to fix it is add a binary package only for
>> udevadm
>
> That's pretty much what udev-utils is: it doesn't contain any binaries
> apart from udevadm.

In this case the binary package is wrong.

>> and make udev depends on it and mdev suggests it.
>
> Mdev itself doesn't have any use for udevadm, so having mdev Recommend
> udev-utils wouldn't be the right thing; this would just be a waste of
> space for people who are using mdev without HAL (which, I suspect, is
> actually the majority of mdev users).

Agreed. But AFAIK suggested packages do not get installed by bitbake by default.

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Phil Blundell]

On Mon, 2009-05-25 at 13:59 -0300, Otavio Salvador wrote:
> Hello,
> 
> On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
> > On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
> [...]
> >> I belive that the right way to fix it is add a binary package only for
> >> udevadm
> >
> > That's pretty much what udev-utils is: it doesn't contain any binaries
> > apart from udevadm.
> 
> In this case the binary package is wrong.

Why so?

> >> and make udev depends on it and mdev suggests it.
> >
> > Mdev itself doesn't have any use for udevadm, so having mdev Recommend
> > udev-utils wouldn't be the right thing; this would just be a waste of
> > space for people who are using mdev without HAL (which, I suspect, is
> > actually the majority of mdev users).
> 
> Agreed. But AFAIK suggested packages do not get installed by bitbake by default.

Ah yes, sorry, I misread your first mail.  Having mdev Suggest
udev-utils would indeed be harmless since you're right, ipkg doesn't
install suggestions automatically.  It would still be fairly pointless
though since nothing in mdev makes any use of udev-utils.

p.

Re: [PATCH 1/2] udev: fixed obsolete udevadm syntax for udev init.d script.

[Ihar Hrachyshka]

On Mon, May 25, 2009 at 7:38 PM, Otavio Salvador
<otavio@ossystems.com.br> wrote:
> Hello Tom,
>
> On Sun, May 24, 2009 at 11:06 PM, Tom Rini <trini@embeddedalley.com> wrote:
>> I disagree.  There are, I assume, things that could be done on top of
>> the -141 script as further enchancements/cleanups that would not apply
>> to -124.  But keeping older files in sync, so long as (a) it works and
>> (b) the older version is cared about makes future buging easier, I've
>> found.
>
> I also disagree but due different reasons. I believe we ought to drop
> the old version except if there're a real reason to keep it around.
> For udev 124 I think there's no reason for it to be kept (maybe I'm
> missing something here).

Some distros still prefer this version. I think that's the reason you missed.

>
> Obviously this ought to be done in dev, not stable, but it can make
> our life easier and avoid us to need to care about so many recipes.
> The final result is that old versions usually has ugly and unknown
> bugs that noone cares about.

If someone prefer the version then we should consider it as needed. Or
drop the obsolete distro (btw I think it's a good thing to drop
obsolete infrastructure things for which maintainers are too "busy" to
sync their third-party gits with upstream).

>
> --
> Otavio Salvador                  O.S. Systems
> E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
> Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH] udev: include udev-utils into udev rootfs

[Ihar Hrachyshka]

On Mon, May 25, 2009 at 7:28 PM, Otavio Salvador
<otavio@ossystems.com.br> wrote:
> Hello,
>
> NACK!
>
> On Mon, May 25, 2009 at 8:14 AM, Koen Kooi <k.kooi@student.utwente.nl> wrote:
>> On 25-05-09 12:51, Ihar Hrachyshka wrote:
>>>
>>> The reason for the move is that udevadm binary is executed from udev
>>> init.d script.
>
> Wouldn't be better to move udevadm to udev binary?
>
> It makes non-sense to have another package installed by default and
> bring more things then required. If it is used by init.d, and this is
> ok, so please move the required binary for the udev package and that's
> solved.

We *already* clarified the reasons to do the exact thing (in another
thread): some people (f.e. with mdev rather udev) wants to be able to
install udev stuff needed by HAL but with no udev as a default device
manager.

>
> My 2c.
>
> --
> Otavio Salvador                  O.S. Systems
> E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
> Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Otavio Salvador]

Hello,

On Mon, May 25, 2009 at 2:32 PM, Phil Blundell <philb@gnu.org> wrote:
> On Mon, 2009-05-25 at 13:59 -0300, Otavio Salvador wrote:
>> Hello,
>>
>> On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
>> > On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
>> [...]
>> >> I belive that the right way to fix it is add a binary package only for
>> >> udevadm
>> >
>> > That's pretty much what udev-utils is: it doesn't contain any binaries
>> > apart from udevadm.
>>
>> In this case the binary package is wrong.
>
> Why so?

I think udev-udevadm is more sound for me.

>> >> and make udev depends on it and mdev suggests it.
>> >
>> > Mdev itself doesn't have any use for udevadm, so having mdev Recommend
>> > udev-utils wouldn't be the right thing; this would just be a waste of
>> > space for people who are using mdev without HAL (which, I suspect, is
>> > actually the majority of mdev users).
>>
>> Agreed. But AFAIK suggested packages do not get installed by bitbake by default.
>
> Ah yes, sorry, I misread your first mail.  Having mdev Suggest
> udev-utils would indeed be harmless since you're right, ipkg doesn't
> install suggestions automatically.  It would still be fairly pointless
> though since nothing in mdev makes any use of udev-utils.

Or maybe hal ought to depends on udev-(utils|udevadm) instead ;-)

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Ihar Hrachyshka]

On Mon, May 25, 2009 at 9:57 PM, Otavio Salvador
<otavio@ossystems.com.br> wrote:
> Hello,
>
> On Mon, May 25, 2009 at 2:32 PM, Phil Blundell <philb@gnu.org> wrote:
>> On Mon, 2009-05-25 at 13:59 -0300, Otavio Salvador wrote:
>>> Hello,
>>>
>>> On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
>>> > On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
>>> [...]
>>> >> I belive that the right way to fix it is add a binary package only for
>>> >> udevadm
>>> >
>>> > That's pretty much what udev-utils is: it doesn't contain any binaries
>>> > apart from udevadm.
>>>
>>> In this case the binary package is wrong.
>>
>> Why so?
>
> I think udev-udevadm is more sound for me.

It's not only udevadm:
FILES_udev-utils = "${usrbindir}/udevinfo ${sbindir}/udevadm
${usrbindir}/udevtest"

>
>>> >> and make udev depends on it and mdev suggests it.
>>> >
>>> > Mdev itself doesn't have any use for udevadm, so having mdev Recommend
>>> > udev-utils wouldn't be the right thing; this would just be a waste of
>>> > space for people who are using mdev without HAL (which, I suspect, is
>>> > actually the majority of mdev users).
>>>
>>> Agreed. But AFAIK suggested packages do not get installed by bitbake by default.
>>
>> Ah yes, sorry, I misread your first mail.  Having mdev Suggest
>> udev-utils would indeed be harmless since you're right, ipkg doesn't
>> install suggestions automatically.  It would still be fairly pointless
>> though since nothing in mdev makes any use of udev-utils.
>
> Or maybe hal ought to depends on udev-(utils|udevadm) instead ;-)

Not all hal users need these binaries, don't they?

>
> --
> Otavio Salvador                  O.S. Systems
> E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
> Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Phil Blundell]

On Mon, 2009-05-25 at 15:57 -0300, Otavio Salvador wrote:
> Hello,
> 
> On Mon, May 25, 2009 at 2:32 PM, Phil Blundell <philb@gnu.org> wrote:
> > On Mon, 2009-05-25 at 13:59 -0300, Otavio Salvador wrote:
> >> Hello,
> >>
> >> On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
> >> > On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
> >> [...]
> >> >> I belive that the right way to fix it is add a binary package only for
> >> >> udevadm
> >> >
> >> > That's pretty much what udev-utils is: it doesn't contain any binaries
> >> > apart from udevadm.
> >>
> >> In this case the binary package is wrong.
> >
> > Why so?
> 
> I think udev-udevadm is more sound for me.

Ah, you mean the package name?  Yes, udev-udevadm would probably be
better if we were doing this again from scratch, but renaming an
existing package is a fairly disruptive thing to do and I don't think
the existing one is wrong enough to make it worthwhile.

> Or maybe hal ought to depends on udev-(utils|udevadm) instead ;-)

It already does.  See hal_0.5.9.1.bb for example, which has:

RDEPENDS_${PN} += "udev-utils hal-info"

p.

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Otavio Salvador]

Hello,

On Mon, May 25, 2009 at 4:52 PM, Phil Blundell <philb@gnu.org> wrote:
> On Mon, 2009-05-25 at 15:57 -0300, Otavio Salvador wrote:
>> Hello,
>>
>> On Mon, May 25, 2009 at 2:32 PM, Phil Blundell <philb@gnu.org> wrote:
>> > On Mon, 2009-05-25 at 13:59 -0300, Otavio Salvador wrote:
>> >> Hello,
>> >>
>> >> On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
>> >> > On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
>> >> [...]
>> >> >> I belive that the right way to fix it is add a binary package only for
>> >> >> udevadm
>> >> >
>> >> > That's pretty much what udev-utils is: it doesn't contain any binaries
>> >> > apart from udevadm.
>> >>
>> >> In this case the binary package is wrong.
>> >
>> > Why so?
>>
>> I think udev-udevadm is more sound for me.
>
> Ah, you mean the package name?  Yes, udev-udevadm would probably be
> better if we were doing this again from scratch, but renaming an
> existing package is a fairly disruptive thing to do and I don't think
> the existing one is wrong enough to make it worthwhile.

Well, I think it makes sense. What does not make sense to me is to
install udevtest and udevinfo in a system that doesn't use them.

FILES_udev-utils = "${usrbindir}/udevinfo ${sbindir}/udevadm
${usrbindir}/udevtest"

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Ihar Hrachyshka]

On Mon, May 25, 2009 at 11:22 PM, Otavio Salvador
<otavio@ossystems.com.br> wrote:
> Hello,
>
> On Mon, May 25, 2009 at 4:52 PM, Phil Blundell <philb@gnu.org> wrote:
>> On Mon, 2009-05-25 at 15:57 -0300, Otavio Salvador wrote:
>>> Hello,
>>>
>>> On Mon, May 25, 2009 at 2:32 PM, Phil Blundell <philb@gnu.org> wrote:
>>> > On Mon, 2009-05-25 at 13:59 -0300, Otavio Salvador wrote:
>>> >> Hello,
>>> >>
>>> >> On Mon, May 25, 2009 at 1:53 PM, Phil Blundell <philb@gnu.org> wrote:
>>> >> > On Mon, 2009-05-25 at 13:31 -0300, Otavio Salvador wrote:
>>> >> [...]
>>> >> >> I belive that the right way to fix it is add a binary package only for
>>> >> >> udevadm
>>> >> >
>>> >> > That's pretty much what udev-utils is: it doesn't contain any binaries
>>> >> > apart from udevadm.
>>> >>
>>> >> In this case the binary package is wrong.
>>> >
>>> > Why so?
>>>
>>> I think udev-udevadm is more sound for me.
>>
>> Ah, you mean the package name?  Yes, udev-udevadm would probably be
>> better if we were doing this again from scratch, but renaming an
>> existing package is a fairly disruptive thing to do and I don't think
>> the existing one is wrong enough to make it worthwhile.
>
> Well, I think it makes sense. What does not make sense to me is to
> install udevtest and udevinfo in a system that doesn't use them.

Do you suggest splitting uboot-utils like "uboot-udevadm +
uboot-utils" where uboot-utils contains udevinfo and udevtest? Or do
you suggest creating separate packages for every udev* binary? What's
your point?

>
> FILES_udev-utils = "${usrbindir}/udevinfo ${sbindir}/udevadm
> ${usrbindir}/udevtest"
>
> --
> Otavio Salvador                  O.S. Systems
> E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
> Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Phil Blundell]

On Mon, 2009-05-25 at 17:22 -0300, Otavio Salvador wrote:
> Well, I think it makes sense. What does not make sense to me is to
> install udevtest and udevinfo in a system that doesn't use them.
> 
> FILES_udev-utils = "${usrbindir}/udevinfo ${sbindir}/udevadm
> ${usrbindir}/udevtest"

I don't know what udevtest is for but, notwithstanding its presence in
FILES, no recent version of udev (at least, not 124 or 141) seems to
ship it.  udevinfo is just a symlink to udevadm in -124, and seems to be
missing entirely in -141, so again that doesn't seem like a big deal.

p.

Re: [PATCH 2/2] udev: moved udevadm to the main udev package

[Otavio Salvador]

Hello,

On Mon, May 25, 2009 at 5:29 PM, Ihar Hrachyshka
<ihar.hrachyshka@gmail.com> wrote:
> Do you suggest splitting uboot-utils like "uboot-udevadm +
> uboot-utils" where uboot-utils contains udevinfo and udevtest? Or do
> you suggest creating separate packages for every udev* binary? What's
> your point?

uboot-udevadm + uboot-utils

-- 
Otavio Salvador                  O.S. Systems
E-mail: otavio@ossystems.com.br  http://www.ossystems.com.br
Mobile: +55 53 9981-7854         http://projetos.ossystems.com.br

Re: [PATCH 1/4] udev: fixed obsolete udevadm syntax for udev init.d script.

[Ihar Hrachyshka]

Any change we can get it in tree?

On Sun, May 24, 2009 at 11:54 PM, Ihar Hrachyshka
<ihar.hrachyshka@gmail.com> wrote:
> Fixed the following warning message when executing udev init script:
> "udevadm control commands requires the --<command> format, this will
> stop working in a future release."
>
> Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
> ---
>  recipes/udev/udev-141/init |    4 ++--
>  recipes/udev/udev_141.bb   |    2 +-
>  2 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/recipes/udev/udev-141/init b/recipes/udev/udev-141/init
> index efe4470..7188b5a 100644
> --- a/recipes/udev/udev-141/init
> +++ b/recipes/udev/udev-141/init
> @@ -56,10 +56,10 @@ kill_udevd > "/dev/null" 2>&1
>         echo -e '\000\000\000\000' > /proc/sys/kernel/hotplug
>         /sbin/udevd -d
>
> -        /sbin/udevadm control env STARTUP=1
> +        /sbin/udevadm control --env STARTUP=1
>                if [ "$not_first_boot" != "" ];then
>                        /sbin/udevadm trigger --subsystem-nomatch=tty --subsystem-nomatch=mem --subsystem-nomatch=vc --subsystem-nomatch=vtconsole --subsystem-nomatch=misc --subsystem-nomatch=dcon --subsystem-nomatch=pci_bus  --subsystem-nomatch=graphics  --subsystem-nomatch=backlight --subsystem-nomatch=video4linux  --subsystem-nomatch=platform
> -                       (/sbin/udevadm settle --timeout=3; /sbin/udevadm control env STARTUP=)&
> +                       (/sbin/udevadm settle --timeout=3; /sbin/udevadm control --env STARTUP=)&
>                        if [ "$DEVCACHE" != "" ]; then
>                                rm -f /tmp/uname
>                                rm -f /tmp/cmdline
> diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
> index e7438fb..7e26798 100644
> --- a/recipes/udev/udev_141.bb
> +++ b/recipes/udev/udev_141.bb
> @@ -6,7 +6,7 @@ LICENSE = "GPL"
>  # Untested
>  DEFAULT_PREFERENCE = "-1"
>
> -PR = "r6"
> +PR = "r7"
>
>  SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
>           file://mount.blacklist \
> --
> 1.6.3.1
>
>

Re: [PATCH 1/4] udev: fixed obsolete udevadm syntax for udev init.d script.

[Tom Rini]

On Sun, May 24, 2009 at 11:54:00PM +0300, Ihar Hrachyshka wrote:

> Fixed the following warning message when executing udev init script:
> "udevadm control commands requires the --<command> format, this will
> stop working in a future release."
> 
> Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>

So, while I would have preferred we fixup both, looking more long-term I
think dev is going to move towards dropping some old programs, so..:

Acked-by: Tom Rini <trini@embeddedalley.com>

> ---
>  recipes/udev/udev-141/init |    4 ++--
>  recipes/udev/udev_141.bb   |    2 +-
>  2 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/recipes/udev/udev-141/init b/recipes/udev/udev-141/init
> index efe4470..7188b5a 100644
> --- a/recipes/udev/udev-141/init
> +++ b/recipes/udev/udev-141/init
> @@ -56,10 +56,10 @@ kill_udevd > "/dev/null" 2>&1
>          echo -e '\000\000\000\000' > /proc/sys/kernel/hotplug
>          /sbin/udevd -d
>  
> -        /sbin/udevadm control env STARTUP=1
> +        /sbin/udevadm control --env STARTUP=1
>  		if [ "$not_first_boot" != "" ];then
>  			/sbin/udevadm trigger --subsystem-nomatch=tty --subsystem-nomatch=mem --subsystem-nomatch=vc --subsystem-nomatch=vtconsole --subsystem-nomatch=misc --subsystem-nomatch=dcon --subsystem-nomatch=pci_bus  --subsystem-nomatch=graphics  --subsystem-nomatch=backlight --subsystem-nomatch=video4linux  --subsystem-nomatch=platform
> -			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control env STARTUP=)&
> +			(/sbin/udevadm settle --timeout=3; /sbin/udevadm control --env STARTUP=)&
>  			if [ "$DEVCACHE" != "" ]; then
>  				rm -f /tmp/uname
>  				rm -f /tmp/cmdline
> diff --git a/recipes/udev/udev_141.bb b/recipes/udev/udev_141.bb
> index e7438fb..7e26798 100644
> --- a/recipes/udev/udev_141.bb
> +++ b/recipes/udev/udev_141.bb
> @@ -6,7 +6,7 @@ LICENSE = "GPL"
>  # Untested
>  DEFAULT_PREFERENCE = "-1"
>  
> -PR = "r6"
> +PR = "r7"
>  
>  SRC_URI = "http://kernel.org/pub/linux/utils/kernel/hotplug/udev-${PV}.tar.gz \
>  	   file://mount.blacklist \
> -- 
> 1.6.3.1
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> 

-- 
Tom Rini