* Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
@ 2011-10-27 20:30 Andreas Müller
2011-10-27 20:43 ` Scott Garman
` (3 more replies)
0 siblings, 4 replies; 22+ messages in thread
From: Andreas Müller @ 2011-10-27 20:30 UTC (permalink / raw)
To: openembedded-core
Since the last updates of all layers (angstrom based) I have issues with
* networkmanager/nm-applet (meta-oe): I am not allowed to change the settings
any more
* xfce-session (meta-oe): Logging out takes very long and Restart/Shutdown is
disabled
I remember that I had similar issues a while ago. That could be worked around
by starting some dbus stuff ( do not exactly remember what it was ).
With this vague idea I checked and found in oe-core:
| dbus: use useradd class to allow use in read-only filesystems
| author Otavio Salvador <otavio@ossystems.com.br>
| commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
Not understanding totally what is all about I see in /etc/passw
| messagebus:x:999:998::/var/lib/dbus:/bin/sh
but I don't have /var/lib/dbus on my sytem!!
I did not check yet, but maybe this is an incompatibilty with angstrom custom
| FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
somebody around to enlight the darkness?
Andreas
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-27 20:30 Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)? Andreas Müller
@ 2011-10-27 20:43 ` Scott Garman
2011-10-27 21:42 ` Andreas Müller
2011-10-28 7:30 ` Henning Heinold
` (2 subsequent siblings)
3 siblings, 1 reply; 22+ messages in thread
From: Scott Garman @ 2011-10-27 20:43 UTC (permalink / raw)
To: openembedded-core
On 10/27/2011 01:30 PM, Andreas Müller wrote:
> Since the last updates of all layers (angstrom based) I have issues with
>
> * networkmanager/nm-applet (meta-oe): I am not allowed to change the settings
> any more
> * xfce-session (meta-oe): Logging out takes very long and Restart/Shutdown is
> disabled
>
> I remember that I had similar issues a while ago. That could be worked around
> by starting some dbus stuff ( do not exactly remember what it was ).
>
> With this vague idea I checked and found in oe-core:
>
> | dbus: use useradd class to allow use in read-only filesystems
> | author Otavio Salvador<otavio@ossystems.com.br>
> | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
>
> Not understanding totally what is all about I see in /etc/passw
>
> | messagebus:x:999:998::/var/lib/dbus:/bin/sh
>
> but I don't have /var/lib/dbus on my sytem!!
>
> I did not check yet, but maybe this is an incompatibilty with angstrom custom
>
> | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
>
> somebody around to enlight the darkness?
useradd should be creating the home directory automatically, but I'd
guess it's not being picked up during packaging. Try adding:
${localstatedir}/lib/dbus
to FILES_${PN}
This is a reasonable guess off the top of my head.
Scott
--
Scott Garman
Embedded Linux Engineer - Yocto Project
Intel Open Source Technology Center
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-27 20:43 ` Scott Garman
@ 2011-10-27 21:42 ` Andreas Müller
2011-10-27 22:30 ` Scott Garman
0 siblings, 1 reply; 22+ messages in thread
From: Andreas Müller @ 2011-10-27 21:42 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Thursday, October 27, 2011 10:43:17 PM Scott Garman wrote:
> On 10/27/2011 01:30 PM, Andreas Müller wrote:
> > Since the last updates of all layers (angstrom based) I have issues with
> >
> > * networkmanager/nm-applet (meta-oe): I am not allowed to change the settings
> > any more
> > * xfce-session (meta-oe): Logging out takes very long and Restart/Shutdown is
> > disabled
> >
> > I remember that I had similar issues a while ago. That could be worked around
> > by starting some dbus stuff ( do not exactly remember what it was ).
> >
> > With this vague idea I checked and found in oe-core:
> >
> > | dbus: use useradd class to allow use in read-only filesystems
> > | author Otavio Salvador<otavio@ossystems.com.br>
> > | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
> >
> > Not understanding totally what is all about I see in /etc/passw
> >
> > | messagebus:x:999:998::/var/lib/dbus:/bin/sh
> >
> > but I don't have /var/lib/dbus on my sytem!!
> >
> > I did not check yet, but maybe this is an incompatibilty with angstrom custom
> >
> > | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> >
> > somebody around to enlight the darkness?
>
> useradd should be creating the home directory automatically, but I'd
> guess it's not being picked up during packaging. Try adding:
>
> ${localstatedir}/lib/dbus
>
> to FILES_${PN}
>
> This is a reasonable guess off the top of my head.
>
> Scott
>
>
On the machine I created /var/lib/dbus and made messagebus owner but no change
in behaviour..
Andreas
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-27 21:42 ` Andreas Müller
@ 2011-10-27 22:30 ` Scott Garman
0 siblings, 0 replies; 22+ messages in thread
From: Scott Garman @ 2011-10-27 22:30 UTC (permalink / raw)
To: openembedded-core
On 10/27/2011 02:42 PM, Andreas Müller wrote:
> On Thursday, October 27, 2011 10:43:17 PM Scott Garman wrote:
>> On 10/27/2011 01:30 PM, Andreas Müller wrote:
>>> Since the last updates of all layers (angstrom based) I have issues with
>>>
>>> * networkmanager/nm-applet (meta-oe): I am not allowed to change the settings
>>> any more
>>> * xfce-session (meta-oe): Logging out takes very long and Restart/Shutdown is
>>> disabled
>>>
>>> I remember that I had similar issues a while ago. That could be worked around
>>> by starting some dbus stuff ( do not exactly remember what it was ).
>>>
>>> With this vague idea I checked and found in oe-core:
>>>
>>> | dbus: use useradd class to allow use in read-only filesystems
>>> | author Otavio Salvador<otavio@ossystems.com.br>
>>> | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
>>>
>>> Not understanding totally what is all about I see in /etc/passw
>>>
>>> | messagebus:x:999:998::/var/lib/dbus:/bin/sh
>>>
>>> but I don't have /var/lib/dbus on my sytem!!
>>>
>>> I did not check yet, but maybe this is an incompatibilty with angstrom custom
>>>
>>> | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
>>>
>>> somebody around to enlight the darkness?
>>
>> useradd should be creating the home directory automatically, but I'd
>> guess it's not being picked up during packaging. Try adding:
>>
>> ${localstatedir}/lib/dbus
>>
>> to FILES_${PN}
>>
>> This is a reasonable guess off the top of my head.
>>
>> Scott
>>
>>
> On the machine I created /var/lib/dbus and made messagebus owner but no change
> in behaviour..
And in retrospect, I now realize the useradd utility should create this
home directory at rootfs creation time, so there's no chance it could be
a packaging problem. So forget my initial suggestion.
Scott
--
Scott Garman
Embedded Linux Engineer - Yocto Project
Intel Open Source Technology Center
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-27 20:30 Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)? Andreas Müller
2011-10-27 20:43 ` Scott Garman
@ 2011-10-28 7:30 ` Henning Heinold
2011-10-28 10:22 ` Richard Purdie
2011-10-28 20:58 ` Klaus 'mrmoku' Kurzmann
3 siblings, 0 replies; 22+ messages in thread
From: Henning Heinold @ 2011-10-28 7:30 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Thu, Oct 27, 2011 at 10:30:27PM +0200, Andreas Müller wrote:
> I remember that I had similar issues a while ago. That could be worked around
> by starting some dbus stuff ( do not exactly remember what it was ).
You probably mean policykit, which has a ruleset for actions a user can do or not on dbus.
Bye Henning
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-27 20:30 Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)? Andreas Müller
2011-10-27 20:43 ` Scott Garman
2011-10-28 7:30 ` Henning Heinold
@ 2011-10-28 10:22 ` Richard Purdie
2011-10-28 22:33 ` Andreas Müller
2011-10-28 20:58 ` Klaus 'mrmoku' Kurzmann
3 siblings, 1 reply; 22+ messages in thread
From: Richard Purdie @ 2011-10-28 10:22 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Thu, 2011-10-27 at 22:30 +0200, Andreas Müller wrote:
> Since the last updates of all layers (angstrom based) I have issues with
>
> * networkmanager/nm-applet (meta-oe): I am not allowed to change the settings
> any more
> * xfce-session (meta-oe): Logging out takes very long and Restart/Shutdown is
> disabled
>
> I remember that I had similar issues a while ago. That could be worked around
> by starting some dbus stuff ( do not exactly remember what it was ).
>
> With this vague idea I checked and found in oe-core:
>
> | dbus: use useradd class to allow use in read-only filesystems
> | author Otavio Salvador <otavio@ossystems.com.br>
> | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
>
> Not understanding totally what is all about I see in /etc/passw
>
> | messagebus:x:999:998::/var/lib/dbus:/bin/sh
>
> but I don't have /var/lib/dbus on my sytem!!
>
> I did not check yet, but maybe this is an incompatibilty with angstrom custom
>
> | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
>
> somebody around to enlight the darkness?
There are some issues due to the orcering of when we now start some of
the services. After the above change the dbus service starts much
earlier on first boot, before some of the other postinstalls have run.
This can mean you see errors due to users which have not yet been
created by postinstalls (e.g. avahi or xuser) yet the dbus files
reference these and its trying to start those services.
We're looking at fixing the remaining adduser type postinstalls to
address this. I'm not sure if this is related to the problem you were
seeing above or not.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-27 20:30 Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)? Andreas Müller
` (2 preceding siblings ...)
2011-10-28 10:22 ` Richard Purdie
@ 2011-10-28 20:58 ` Klaus 'mrmoku' Kurzmann
2011-10-29 10:53 ` [PATCH] dbus: add work around to set right owner in postinst Martin Jansa
3 siblings, 1 reply; 22+ messages in thread
From: Klaus 'mrmoku' Kurzmann @ 2011-10-28 20:58 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Thu, 27 Oct 2011, Andreas Müller wrote:
> Since the last updates of all layers (angstrom based) I have issues with
> * networkmanager/nm-applet (meta-oe): I am not allowed to change the settings
> any more
> * xfce-session (meta-oe): Logging out takes very long and Restart/Shutdown is
> disabled
> I remember that I had similar issues a while ago. That could be worked around
> by starting some dbus stuff ( do not exactly remember what it was ).
> With this vague idea I checked and found in oe-core:
> | dbus: use useradd class to allow use in read-only filesystems
> | author Otavio Salvador <otavio@ossystems.com.br>
> | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
> Not understanding totally what is all about I see in /etc/passw
> | messagebus:x:999:998::/var/lib/dbus:/bin/sh
> but I don't have /var/lib/dbus on my sytem!!
> I did not check yet, but maybe this is an incompatibilty with angstrom custom
> | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> somebody around to enlight the darkness?
one problem seems to be /usr/lib/dbus-1.0/dbus-daemon-launch-helper
having wrong ownership. It is root:root instead of root:messagebus how
it should be. Dunno if that is the only problem but fixing that makes
dbus-activation work again.
> Andreas
Klaus 'mrmoku' Kurzmann
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-28 10:22 ` Richard Purdie
@ 2011-10-28 22:33 ` Andreas Müller
2011-11-09 20:38 ` Andreas Müller
0 siblings, 1 reply; 22+ messages in thread
From: Andreas Müller @ 2011-10-28 22:33 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Friday, October 28, 2011 12:22:11 PM Richard Purdie wrote:
> On Thu, 2011-10-27 at 22:30 +0200, Andreas Müller wrote:
> > Since the last updates of all layers (angstrom based) I have issues with
> >
> > * networkmanager/nm-applet (meta-oe): I am not allowed to change the
> > settings
> >
> > any more
> >
> > * xfce-session (meta-oe): Logging out takes very long and
> > Restart/Shutdown is
> >
> > disabled
> >
> > I remember that I had similar issues a while ago. That could be worked
> > around by starting some dbus stuff ( do not exactly remember what it was
> > ).
> >
> > With this vague idea I checked and found in oe-core:
> > | dbus: use useradd class to allow use in read-only filesystems
> > | author Otavio Salvador <otavio@ossystems.com.br>
> > | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
> >
> > Not understanding totally what is all about I see in /etc/passw
> >
> > | messagebus:x:999:998::/var/lib/dbus:/bin/sh
> >
> > but I don't have /var/lib/dbus on my sytem!!
> >
> > I did not check yet, but maybe this is an incompatibilty with angstrom
> > custom
> >
> > | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> >
> > somebody around to enlight the darkness?
>
> There are some issues due to the orcering of when we now start some of
> the services. After the above change the dbus service starts much
> earlier on first boot, before some of the other postinstalls have run.
> This can mean you see errors due to users which have not yet been
> created by postinstalls (e.g. avahi or xuser) yet the dbus files
> reference these and its trying to start those services.
>
> We're looking at fixing the remaining adduser type postinstalls to
> address this. I'm not sure if this is related to the problem you were
> seeing above or not.
>
FYI: reverting the patch fixes mentioned issues here.
Andreas
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH] dbus: add work around to set right owner in postinst
2011-10-28 20:58 ` Klaus 'mrmoku' Kurzmann
@ 2011-10-29 10:53 ` Martin Jansa
2011-10-30 4:13 ` Otavio Salvador
` (2 more replies)
0 siblings, 3 replies; 22+ messages in thread
From: Martin Jansa @ 2011-10-29 10:53 UTC (permalink / raw)
To: openembedded-core
* even when I see right chown/chmod in log.do_install, files are still
owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
$ tail -n 4 temp/log.do_install.27768
+ chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
+ chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
+ chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
+ ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
$ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
-rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
lper*
-rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
lper*
drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
* and we _cannot_ use ${D} in postinst script because it get's expanded
before including postinst file to ipk's control.tar.gz
and on target I see something like this:
if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
which is wrong $D without {} works
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
---
meta/recipes-core/dbus/dbus.inc | 9 +++++++--
1 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc
index 2a97c02..b862bb4 100644
--- a/meta/recipes-core/dbus/dbus.inc
+++ b/meta/recipes-core/dbus/dbus.inc
@@ -10,7 +10,7 @@ DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X
DEPENDS_virtclass-native = "expat-native virtual/libintl-native"
DEPENDS_virtclass-nativesdk = "expat-nativesdk virtual/libintl-nativesdk virtual/libx11"
-PR = "r1"
+PR = "r2"
SRC_URI = "http://dbus.freedesktop.org/releases/dbus/dbus-${PV}.tar.gz \
file://tmpdir.patch; \
@@ -51,9 +51,14 @@ RRECOMMENDS_${PN}-lib = "${PN}"
FILES_${PN}-dev += "${libdir}/dbus-1.0/include ${bindir}/dbus-glib-tool"
pkg_postinst_dbus() {
- if [ -z "${D}" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
+ if [ -z "$D" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
/etc/init.d/populate-volatile.sh update
fi
+ if [ -z "$D" ] ; then
+ chown messagebus:messagebus ${localstatedir}/run/dbus ${localstatedir}/lib/dbus
+ chown root:messagebus ${libexecdir}/dbus-daemon-launch-helper
+ chmod 4754 ${libexecdir}/dbus-daemon-launch-helper
+ fi
}
EXTRA_OECONF_X = "${@base_contains('DISTRO_FEATURES', 'x11', '--with-x', '--without-x', d)}"
--
1.7.7.1
^ permalink raw reply related [flat|nested] 22+ messages in thread
* Re: [PATCH] dbus: add work around to set right owner in postinst
2011-10-29 10:53 ` [PATCH] dbus: add work around to set right owner in postinst Martin Jansa
@ 2011-10-30 4:13 ` Otavio Salvador
2011-10-30 4:17 ` Otavio Salvador
2011-10-31 8:54 ` Koen Kooi
2011-10-31 17:43 ` [PATCHv2] dbus: fix for " Martin Jansa
2 siblings, 1 reply; 22+ messages in thread
From: Otavio Salvador @ 2011-10-30 4:13 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Sat, Oct 29, 2011 at 08:53, Martin Jansa <martin.jansa@gmail.com> wrote:
...
> + if [ -z "$D" ] ; then
> + chown messagebus:messagebus ${localstatedir}/run/dbus ${localstatedir}/lib/dbus
> + chown root:messagebus ${libexecdir}/dbus-daemon-launch-helper
> + chmod 4754 ${libexecdir}/dbus-daemon-launch-helper
> + fi
If you use $D in directories it will work for in-target and rootfs
installs allowing use of it in read-only-fs.
--
Otavio Salvador O.S. Systems
E-mail: otavio@ossystems.com.br http://www.ossystems.com.br
Mobile: +55 53 9981-7854 http://projetos.ossystems.com.br
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH] dbus: add work around to set right owner in postinst
2011-10-30 4:13 ` Otavio Salvador
@ 2011-10-30 4:17 ` Otavio Salvador
0 siblings, 0 replies; 22+ messages in thread
From: Otavio Salvador @ 2011-10-30 4:17 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Sun, Oct 30, 2011 at 02:13, Otavio Salvador <otavio@ossystems.com.br> wrote:
> On Sat, Oct 29, 2011 at 08:53, Martin Jansa <martin.jansa@gmail.com> wrote:
> ...
>> + if [ -z "$D" ] ; then
>> + chown messagebus:messagebus ${localstatedir}/run/dbus ${localstatedir}/lib/dbus
>> + chown root:messagebus ${libexecdir}/dbus-daemon-launch-helper
>> + chmod 4754 ${libexecdir}/dbus-daemon-launch-helper
>> + fi
>
> If you use $D in directories it will work for in-target and rootfs
> installs allowing use of it in read-only-fs.
As in http://paste.debian.net/141111/ (untested)
--
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
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH] dbus: add work around to set right owner in postinst
2011-10-29 10:53 ` [PATCH] dbus: add work around to set right owner in postinst Martin Jansa
2011-10-30 4:13 ` Otavio Salvador
@ 2011-10-31 8:54 ` Koen Kooi
2011-10-31 16:49 ` Saul Wold
2011-10-31 17:43 ` [PATCHv2] dbus: fix for " Martin Jansa
2 siblings, 1 reply; 22+ messages in thread
From: Koen Kooi @ 2011-10-31 8:54 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
[-- Attachment #1: Type: text/plain, Size: 331 bytes --]
Op 29 okt. 2011, om 12:53 heeft Martin Jansa het volgende geschreven:
> * even when I see right chown/chmod in log.do_install, files are still
> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
I suspect fsperms.txt is breaking things, but I haven't tested it yet.
regards,
Koen
[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 169 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH] dbus: add work around to set right owner in postinst
2011-10-31 8:54 ` Koen Kooi
@ 2011-10-31 16:49 ` Saul Wold
0 siblings, 0 replies; 22+ messages in thread
From: Saul Wold @ 2011-10-31 16:49 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer; +Cc: Koen Kooi
On 10/31/2011 01:54 AM, Koen Kooi wrote:
>
> Op 29 okt. 2011, om 12:53 heeft Martin Jansa het volgende geschreven:
>
>> * even when I see right chown/chmod in log.do_install, files are still
>> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
>
> I suspect fsperms.txt is breaking things, but I haven't tested it yet.
>
I did try to reset FILESYSTEM_PERMS_TABLE to "" and the issue still existed.
Further testing and debugging showed that both IPK and RPM packaging
seems to kill off the ownership, but DEB packaging retained it correctly!
More research and digging is required.
Sau!
> regards,
>
> Koen
>
>
>
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCHv2] dbus: fix for right owner in postinst
2011-10-29 10:53 ` [PATCH] dbus: add work around to set right owner in postinst Martin Jansa
2011-10-30 4:13 ` Otavio Salvador
2011-10-31 8:54 ` Koen Kooi
@ 2011-10-31 17:43 ` Martin Jansa
2011-10-31 22:40 ` Richard Purdie
2 siblings, 1 reply; 22+ messages in thread
From: Martin Jansa @ 2011-10-31 17:43 UTC (permalink / raw)
To: openembedded-core
* even when I see right chown/chmod in log.do_install, files are still
owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
$ tail -n 4 temp/log.do_install.27768
+ chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
+ chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
+ chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
+ ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
$ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
-rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
lper*
-rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
lper*
drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
* and we _cannot_ use ${D} in postinst script because it get's expanded
before including postinst file to ipk's control.tar.gz
and on target I see something like this:
if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
which is wrong $D without {} works
* V2: updated with Otavio version to fix it also for do_rootfs
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
---
meta/recipes-core/dbus/dbus.inc | 12 +++++-------
1 files changed, 5 insertions(+), 7 deletions(-)
diff --git a/meta/recipes-core/dbus/dbus.inc b/meta/recipes-core/dbus/dbus.inc
index 2a97c02..a72407c 100644
--- a/meta/recipes-core/dbus/dbus.inc
+++ b/meta/recipes-core/dbus/dbus.inc
@@ -10,7 +10,7 @@ DEPENDS = "expat virtual/libintl ${@base_contains('DISTRO_FEATURES', 'x11', '${X
DEPENDS_virtclass-native = "expat-native virtual/libintl-native"
DEPENDS_virtclass-nativesdk = "expat-nativesdk virtual/libintl-nativesdk virtual/libx11"
-PR = "r1"
+PR = "r3"
SRC_URI = "http://dbus.freedesktop.org/releases/dbus/dbus-${PV}.tar.gz \
file://tmpdir.patch; \
@@ -51,9 +51,12 @@ RRECOMMENDS_${PN}-lib = "${PN}"
FILES_${PN}-dev += "${libdir}/dbus-1.0/include ${bindir}/dbus-glib-tool"
pkg_postinst_dbus() {
- if [ -z "${D}" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
+ if [ -z "$D" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
/etc/init.d/populate-volatile.sh update
fi
+ chown messagebus:messagebus $D${localstatedir}/run/dbus $D${localstatedir}/lib/dbus
+ chown root:messagebus $D${libexecdir}/dbus-daemon-launch-helper
+ chmod 4754 $D${libexecdir}/dbus-daemon-launch-helper
}
EXTRA_OECONF_X = "${@base_contains('DISTRO_FEATURES', 'x11', '--with-x', '--without-x', d)}"
@@ -81,11 +84,6 @@ do_install() {
mkdir -p ${D}${localstatedir}/run/dbus ${D}${localstatedir}/lib/dbus
- chown messagebus:messagebus ${D}${localstatedir}/run/dbus ${D}${localstatedir}/lib/dbus
-
- chown root:messagebus ${D}${libexecdir}/dbus-daemon-launch-helper
- chmod 4754 ${D}${libexecdir}/dbus-daemon-launch-helper
-
# disable dbus-1 sysv script on systemd installs
# nearly all distros call the initscript plain 'dbus', but OE-core is different
ln -sf /dev/null ${D}/${base_libdir}/systemd/system/dbus-1.service
--
1.7.7.1
^ permalink raw reply related [flat|nested] 22+ messages in thread
* Re: [PATCHv2] dbus: fix for right owner in postinst
2011-10-31 17:43 ` [PATCHv2] dbus: fix for " Martin Jansa
@ 2011-10-31 22:40 ` Richard Purdie
2011-11-01 11:26 ` Koen Kooi
2011-11-01 13:36 ` Martin Jansa
0 siblings, 2 replies; 22+ messages in thread
From: Richard Purdie @ 2011-10-31 22:40 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Mon, 2011-10-31 at 18:43 +0100, Martin Jansa wrote:
> * even when I see right chown/chmod in log.do_install, files are still
> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
> $ tail -n 4 temp/log.do_install.27768
> + chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
> e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
> + chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> + chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> + ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
>
> $ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
> lper*
> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
> lper*
> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
>
> * and we _cannot_ use ${D} in postinst script because it get's expanded
> before including postinst file to ipk's control.tar.gz
> and on target I see something like this:
> if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
> which is wrong $D without {} works
>
> * V2: updated with Otavio version to fix it also for do_rootfs
>
> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
> ---
> meta/recipes-core/dbus/dbus.inc | 12 +++++-------
> 1 files changed, 5 insertions(+), 7 deletions(-)
This is just wallpapering over the cracks I'm afraid. There is something
more fundamental going wrong here and we need to track it down...
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCHv2] dbus: fix for right owner in postinst
2011-10-31 22:40 ` Richard Purdie
@ 2011-11-01 11:26 ` Koen Kooi
2011-11-01 14:16 ` Richard Purdie
2011-11-01 13:36 ` Martin Jansa
1 sibling, 1 reply; 22+ messages in thread
From: Koen Kooi @ 2011-11-01 11:26 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
[-- Attachment #1: Type: text/plain, Size: 2831 bytes --]
Op 31 okt. 2011, om 23:40 heeft Richard Purdie het volgende geschreven:
> On Mon, 2011-10-31 at 18:43 +0100, Martin Jansa wrote:
>> * even when I see right chown/chmod in log.do_install, files are still
>> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
>> $ tail -n 4 temp/log.do_install.27768
>> + chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
>> e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
>> + chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
>> + chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
>> + ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
>>
>> $ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
>> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
>> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
>> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
>> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
>> lper*
>> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
>> lper*
>> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
>> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
>>
>> * and we _cannot_ use ${D} in postinst script because it get's expanded
>> before including postinst file to ipk's control.tar.gz
>> and on target I see something like this:
>> if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
>> which is wrong $D without {} works
>>
>> * V2: updated with Otavio version to fix it also for do_rootfs
>>
>> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
>> ---
>> meta/recipes-core/dbus/dbus.inc | 12 +++++-------
>> 1 files changed, 5 insertions(+), 7 deletions(-)
>
> This is just wallpapering over the cracks I'm afraid. There is something
> more fundamental going wrong here and we need to track it down...
Any news on this?
[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 169 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCHv2] dbus: fix for right owner in postinst
2011-10-31 22:40 ` Richard Purdie
2011-11-01 11:26 ` Koen Kooi
@ 2011-11-01 13:36 ` Martin Jansa
2011-11-01 14:15 ` Richard Purdie
1 sibling, 1 reply; 22+ messages in thread
From: Martin Jansa @ 2011-11-01 13:36 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Mon, Oct 31, 2011 at 11:40 PM, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> On Mon, 2011-10-31 at 18:43 +0100, Martin Jansa wrote:
>> * even when I see right chown/chmod in log.do_install, files are still
>> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
>> $ tail -n 4 temp/log.do_install.27768
>> + chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
>> e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
>> + chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
>> + chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
>> + ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
>>
>> $ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
>> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
>> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
>> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
>> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
>> lper*
>> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
>> lper*
>> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
>> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
>>
>> * and we _cannot_ use ${D} in postinst script because it get's expanded
>> before including postinst file to ipk's control.tar.gz
>> and on target I see something like this:
>> if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
>> which is wrong $D without {} works
>>
>> * V2: updated with Otavio version to fix it also for do_rootfs
>>
>> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
>> ---
>> meta/recipes-core/dbus/dbus.inc | 12 +++++-------
>> 1 files changed, 5 insertions(+), 7 deletions(-)
>
> This is just wallpapering over the cracks I'm afraid. There is something
> more fundamental going wrong here and we need to track it down...
Is pseudo really supposed to allow chown in do_install()? And then ie
keep changed owners in sysroot or even sstate to recreate packages
later with right file owners?
Because normally all produced files are owned by UID who is building
them (in my example bitbake) and only when creating packages/rootfs
it's packaged with expected UID for target (ie root).
So this wallpapering makes current dbus at least usable on target.
Cheers,
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCHv2] dbus: fix for right owner in postinst
2011-11-01 13:36 ` Martin Jansa
@ 2011-11-01 14:15 ` Richard Purdie
0 siblings, 0 replies; 22+ messages in thread
From: Richard Purdie @ 2011-11-01 14:15 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Tue, 2011-11-01 at 14:36 +0100, Martin Jansa wrote:
> On Mon, Oct 31, 2011 at 11:40 PM, Richard Purdie
> <richard.purdie@linuxfoundation.org> wrote:
> > On Mon, 2011-10-31 at 18:43 +0100, Martin Jansa wrote:
> >> * even when I see right chown/chmod in log.do_install, files are still
> >> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
> >> $ tail -n 4 temp/log.do_install.27768
> >> + chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
> >> e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
> >> + chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> >> + chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> >> + ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
> >>
> >> $ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
> >> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
> >> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
> >> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> >> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
> >> lper*
> >> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
> >> lper*
> >> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
> >> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
> >>
> >> * and we _cannot_ use ${D} in postinst script because it get's expanded
> >> before including postinst file to ipk's control.tar.gz
> >> and on target I see something like this:
> >> if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
> >> which is wrong $D without {} works
> >>
> >> * V2: updated with Otavio version to fix it also for do_rootfs
> >>
> >> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
> >> ---
> >> meta/recipes-core/dbus/dbus.inc | 12 +++++-------
> >> 1 files changed, 5 insertions(+), 7 deletions(-)
> >
> > This is just wallpapering over the cracks I'm afraid. There is something
> > more fundamental going wrong here and we need to track it down...
>
> Is pseudo really supposed to allow chown in do_install()? And then ie
> keep changed owners in sysroot or even sstate to recreate packages
> later with right file owners?
Yes, it is.
> Because normally all produced files are owned by UID who is building
> them (in my example bitbake) and only when creating packages/rootfs
> it's packaged with expected UID for target (ie root).
I appreciate that.
> So this wallpapering makes current dbus at least usable on target.
This doesn't mean the fix is correct though. I have a patch series to
fix the real issues (there are several) that will be published shortly.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCHv2] dbus: fix for right owner in postinst
2011-11-01 11:26 ` Koen Kooi
@ 2011-11-01 14:16 ` Richard Purdie
0 siblings, 0 replies; 22+ messages in thread
From: Richard Purdie @ 2011-11-01 14:16 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Tue, 2011-11-01 at 12:26 +0100, Koen Kooi wrote:
> Op 31 okt. 2011, om 23:40 heeft Richard Purdie het volgende geschreven:
>
> > On Mon, 2011-10-31 at 18:43 +0100, Martin Jansa wrote:
> >> * even when I see right chown/chmod in log.do_install, files are still
> >> owned by bitbake:bitbake (user running bitbake) and we need to fix it on target
> >> $ tail -n 4 temp/log.do_install.27768
> >> + chown messagebus:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus /OE/shr-core/tmp/work/armv7a-vfp-neon-o
> >> e-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus
> >> + chown root:messagebus /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> >> + chmod 4754 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> >> + ln -sf /dev/null /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image//lib/systemd/system/dbus-1.service
> >>
> >> $ ll -d /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus \
> >> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus \
> >> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper \
> >> /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-helper
> >> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
> >> lper*
> >> -rwsr-xr-- 1 bitbake bitbake 819869 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/usr/libexec/dbus-daemon-launch-he
> >> lper*
> >> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/lib/dbus/
> >> drwxr-xr-x 2 bitbake bitbake 4096 Oct 29 12:35 /OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image/var/run/dbus/
> >>
> >> * and we _cannot_ use ${D} in postinst script because it get's expanded
> >> before including postinst file to ipk's control.tar.gz
> >> and on target I see something like this:
> >> if [ -z "/OE/shr-core/tmp/work/armv7a-vfp-neon-oe-linux-gnueabi/dbus-1.4.12-r2/image" ] && [ -e /etc/init.d/populate-volatile.sh ] ; then
> >> which is wrong $D without {} works
> >>
> >> * V2: updated with Otavio version to fix it also for do_rootfs
> >>
> >> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
> >> ---
> >> meta/recipes-core/dbus/dbus.inc | 12 +++++-------
> >> 1 files changed, 5 insertions(+), 7 deletions(-)
> >
> > This is just wallpapering over the cracks I'm afraid. There is something
> > more fundamental going wrong here and we need to track it down...
>
> Any news on this?
Yes, expect to see some patches for review shortly. There are several
interwoven issues.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-10-28 22:33 ` Andreas Müller
@ 2011-11-09 20:38 ` Andreas Müller
2011-11-09 20:39 ` Martin Jansa
0 siblings, 1 reply; 22+ messages in thread
From: Andreas Müller @ 2011-11-09 20:38 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Saturday, October 29, 2011 12:33:46 AM Andreas Müller wrote:
> On Friday, October 28, 2011 12:22:11 PM Richard Purdie wrote:
> > On Thu, 2011-10-27 at 22:30 +0200, Andreas Müller wrote:
> > > Since the last updates of all layers (angstrom based) I have issues
> > > with
> > >
> > > * networkmanager/nm-applet (meta-oe): I am not allowed to change the
> > > settings
> > >
> > > any more
> > >
> > > * xfce-session (meta-oe): Logging out takes very long and
> > > Restart/Shutdown is
> > >
> > > disabled
> > >
> > > I remember that I had similar issues a while ago. That could be worked
> > > around by starting some dbus stuff ( do not exactly remember what it
> > > was ).
> > >
> > > With this vague idea I checked and found in oe-core:
> > > | dbus: use useradd class to allow use in read-only filesystems
> > > | author Otavio Salvador <otavio@ossystems.com.br>
> > > | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
> > >
> > > Not understanding totally what is all about I see in /etc/passw
> > >
> > > | messagebus:x:999:998::/var/lib/dbus:/bin/sh
> > >
> > > but I don't have /var/lib/dbus on my sytem!!
> > >
> > > I did not check yet, but maybe this is an incompatibilty with angstrom
> > > custom
> > >
> > > | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> > >
> > > somebody around to enlight the darkness?
> >
> > There are some issues due to the orcering of when we now start some of
> > the services. After the above change the dbus service starts much
> > earlier on first boot, before some of the other postinstalls have run.
> > This can mean you see errors due to users which have not yet been
> > created by postinstalls (e.g. avahi or xuser) yet the dbus files
> > reference these and its trying to start those services.
> >
> > We're looking at fixing the remaining adduser type postinstalls to
> > address this. I'm not sure if this is related to the problem you were
> > seeing above or not.
>
> FYI: reverting the patch fixes mentioned issues here.
>
Maybe I missed something but after starting a new image - to get the errors
mentioned above worked around - I still have to do manually on the machine :
chown messagebus:messagebus /var/run/dbus ${D}${localstatedir}/lib/dbus
chown root:messagebus /usr/libexec/dbus-daemon-launch-helper
chmod 4754 /usr/libexec/dbus-daemon-launch-helper
As long as this is not fixed, can we have this additionally in pkg_postinst_dbus
temporary?
Andreas
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-11-09 20:38 ` Andreas Müller
@ 2011-11-09 20:39 ` Martin Jansa
2011-11-09 20:56 ` Andreas Müller
0 siblings, 1 reply; 22+ messages in thread
From: Martin Jansa @ 2011-11-09 20:39 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
[-- Attachment #1: Type: text/plain, Size: 2954 bytes --]
On Wed, Nov 09, 2011 at 09:38:57PM +0100, Andreas Müller wrote:
> On Saturday, October 29, 2011 12:33:46 AM Andreas Müller wrote:
> > On Friday, October 28, 2011 12:22:11 PM Richard Purdie wrote:
> > > On Thu, 2011-10-27 at 22:30 +0200, Andreas Müller wrote:
> > > > Since the last updates of all layers (angstrom based) I have issues
> > > > with
> > > >
> > > > * networkmanager/nm-applet (meta-oe): I am not allowed to change the
> > > > settings
> > > >
> > > > any more
> > > >
> > > > * xfce-session (meta-oe): Logging out takes very long and
> > > > Restart/Shutdown is
> > > >
> > > > disabled
> > > >
> > > > I remember that I had similar issues a while ago. That could be worked
> > > > around by starting some dbus stuff ( do not exactly remember what it
> > > > was ).
> > > >
> > > > With this vague idea I checked and found in oe-core:
> > > > | dbus: use useradd class to allow use in read-only filesystems
> > > > | author Otavio Salvador <otavio@ossystems.com.br>
> > > > | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
> > > >
> > > > Not understanding totally what is all about I see in /etc/passw
> > > >
> > > > | messagebus:x:999:998::/var/lib/dbus:/bin/sh
> > > >
> > > > but I don't have /var/lib/dbus on my sytem!!
> > > >
> > > > I did not check yet, but maybe this is an incompatibilty with angstrom
> > > > custom
> > > >
> > > > | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> > > >
> > > > somebody around to enlight the darkness?
> > >
> > > There are some issues due to the orcering of when we now start some of
> > > the services. After the above change the dbus service starts much
> > > earlier on first boot, before some of the other postinstalls have run.
> > > This can mean you see errors due to users which have not yet been
> > > created by postinstalls (e.g. avahi or xuser) yet the dbus files
> > > reference these and its trying to start those services.
> > >
> > > We're looking at fixing the remaining adduser type postinstalls to
> > > address this. I'm not sure if this is related to the problem you were
> > > seeing above or not.
> >
> > FYI: reverting the patch fixes mentioned issues here.
> >
> Maybe I missed something but after starting a new image - to get the errors
> mentioned above worked around - I still have to do manually on the machine :
>
> chown messagebus:messagebus /var/run/dbus ${D}${localstatedir}/lib/dbus
> chown root:messagebus /usr/libexec/dbus-daemon-launch-helper
> chmod 4754 /usr/libexec/dbus-daemon-launch-helper
>
> As long as this is not fixed, can we have this additionally in pkg_postinst_dbus
> temporary?
You have wrong ${D} there, but you can apply
http://git.openembedded.org/openembedded-core-contrib/commit/?h=shr&id=283b046e6e38afb63125776afdb5ee5fea0d0b7e
which still works, but it's not proper fix.
--
Martin 'JaMa' Jansa jabber: Martin.Jansa@gmail.com
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)?
2011-11-09 20:39 ` Martin Jansa
@ 2011-11-09 20:56 ` Andreas Müller
0 siblings, 0 replies; 22+ messages in thread
From: Andreas Müller @ 2011-11-09 20:56 UTC (permalink / raw)
To: Patches and discussions about the oe-core layer
On Wednesday, November 09, 2011 09:39:16 PM Martin Jansa wrote:
> On Wed, Nov 09, 2011 at 09:38:57PM +0100, Andreas Müller wrote:
> > On Saturday, October 29, 2011 12:33:46 AM Andreas Müller wrote:
> > > On Friday, October 28, 2011 12:22:11 PM Richard Purdie wrote:
> > > > On Thu, 2011-10-27 at 22:30 +0200, Andreas Müller wrote:
> > > > > Since the last updates of all layers (angstrom based) I have issues
> > > > > with
> > > > >
> > > > > * networkmanager/nm-applet (meta-oe): I am not allowed to change
> > > > > the settings
> > > > >
> > > > > any more
> > > > >
> > > > > * xfce-session (meta-oe): Logging out takes very long and
> > > > > Restart/Shutdown is
> > > > >
> > > > > disabled
> > > > >
> > > > > I remember that I had similar issues a while ago. That could be
> > > > > worked around by starting some dbus stuff ( do not exactly
> > > > > remember what it was ).
> > > > >
> > > > > With this vague idea I checked and found in oe-core:
> > > > > | dbus: use useradd class to allow use in read-only filesystems
> > > > > | author Otavio Salvador <otavio@ossystems.com.br>
> > > > > | commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
> > > > >
> > > > > Not understanding totally what is all about I see in /etc/passw
> > > > >
> > > > > | messagebus:x:999:998::/var/lib/dbus:/bin/sh
> > > > >
> > > > > but I don't have /var/lib/dbus on my sytem!!
> > > > >
> > > > > I did not check yet, but maybe this is an incompatibilty with
> > > > > angstrom custom
> > > > >
> > > > > | FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> > > > >
> > > > > somebody around to enlight the darkness?
> > > >
> > > > There are some issues due to the orcering of when we now start some
> > > > of the services. After the above change the dbus service starts much
> > > > earlier on first boot, before some of the other postinstalls have
> > > > run. This can mean you see errors due to users which have not yet
> > > > been created by postinstalls (e.g. avahi or xuser) yet the dbus
> > > > files reference these and its trying to start those services.
> > > >
> > > > We're looking at fixing the remaining adduser type postinstalls to
> > > > address this. I'm not sure if this is related to the problem you were
> > > > seeing above or not.
> > >
> > > FYI: reverting the patch fixes mentioned issues here.
> >
> > Maybe I missed something but after starting a new image - to get the
> > errors mentioned above worked around - I still have to do manually on
> > the machine :
> >
> > chown messagebus:messagebus /var/run/dbus ${D}${localstatedir}/lib/dbus
> > chown root:messagebus /usr/libexec/dbus-daemon-launch-helper
> > chmod 4754 /usr/libexec/dbus-daemon-launch-helper
> >
> > As long as this is not fixed, can we have this additionally in
> > pkg_postinst_dbus temporary?
>
> You have wrong ${D} there, but you can apply
> http://git.openembedded.org/openembedded-core-contrib/commit/?h=shr&id=283b
> 046e6e38afb63125776afdb5ee5fea0d0b7e which still works, but it's not proper
> fix.
That's exactly the workaround I meant :)
Andreas
^ permalink raw reply [flat|nested] 22+ messages in thread
end of thread, other threads:[~2011-11-09 20:57 UTC | newest]
Thread overview: 22+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-10-27 20:30 Trouble by last dbus patch (46e6c3fa8034b12d178d605f3f5d7efe69671a13)? Andreas Müller
2011-10-27 20:43 ` Scott Garman
2011-10-27 21:42 ` Andreas Müller
2011-10-27 22:30 ` Scott Garman
2011-10-28 7:30 ` Henning Heinold
2011-10-28 10:22 ` Richard Purdie
2011-10-28 22:33 ` Andreas Müller
2011-11-09 20:38 ` Andreas Müller
2011-11-09 20:39 ` Martin Jansa
2011-11-09 20:56 ` Andreas Müller
2011-10-28 20:58 ` Klaus 'mrmoku' Kurzmann
2011-10-29 10:53 ` [PATCH] dbus: add work around to set right owner in postinst Martin Jansa
2011-10-30 4:13 ` Otavio Salvador
2011-10-30 4:17 ` Otavio Salvador
2011-10-31 8:54 ` Koen Kooi
2011-10-31 16:49 ` Saul Wold
2011-10-31 17:43 ` [PATCHv2] dbus: fix for " Martin Jansa
2011-10-31 22:40 ` Richard Purdie
2011-11-01 11:26 ` Koen Kooi
2011-11-01 14:16 ` Richard Purdie
2011-11-01 13:36 ` Martin Jansa
2011-11-01 14:15 ` Richard Purdie
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.