Skip to contentSkip to conversation

yocto Digest, Vol 38, Issue 92

DISCUSSION rajan pathak 1 participant 1 message Nov 25, 2013 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

I am not sure but any change to Linux kernel repo can be reflected by modifying the SRCREV variable
in Kernel recipe file.

Thanks
Rajan

Quoted 591 lines · On Mon, Nov 25, 2013 at 10:37 PM, <[email protected]> wrote:

Send yocto mailing list submissions to
[email protected]

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.yoctoproject.org/listinfo/yocto
or, via email, send a message with subject or body 'help' to
[email protected]

You can reach the person managing the list at
[email protected]

When replying, please edit your Subject line so it is more specific than "Re: Contents of yocto digest..."

Today's Topics:

1. Unable to upgrade kernel; package for kernel name includes version (Bryan Evenson)
2. Re: ksize.py and dirsize.py scripts not in dora tarball, is that deliberate? (Rifenbark, Scott M)
3. Re: RFC Autobuilder Naming (Paul Eggleton)
4. Re: unbootable image produced with PACKAGE_CLASSES ?= "package_ipk", /etc missing (Todd Stellanova)
5. FW: stupid question about post-installation scripts (Rifenbark, Scott M)
6. Re: stupid question about post-installation scripts (Bryan Evenson)
7. Re: Unable to upgrade kernel; package for kernel name includes version (Bryan Evenson)
8. Re: FW: stupid question about post-installation scripts (Paul Eggleton)
9. Re: FW: stupid question about post-installation scripts (Paul Eggleton)

----------------------------------------------------------------------

Message: 1
Date: Mon, 25 Nov 2013 09:45:51 -0500
From: Bryan Evenson <[email protected]>
To: "[email protected]" <[email protected]>
Subject: [yocto] Unable to upgrade kernel; package for kernel name
includes version
Message-ID:

<91586D499ADFD74FBCFB8425266A5DE40153C1156010@pluto.melinkcorp.local>
Content-Type: text/plain; charset="us-ascii"

I'm on poky/dylan-9.0.1 I am upgrading the Linux kernel for my device's image from a custom 3.6.9 kernel to a custom 3.10 kernel. I'm using this
layer: https://github.com/evensonbryan/meta-atmel as a basis for the custom kernel recipe (there is a recipe for the 3.10 kernel that I haven't yet deployed to Github in case anyone is wondering). I'm attempting to do a firmware upgrade with my built packages (using opkg for package management) and I've discovered an issue with the kernel naming. On my pre-upgraded device, a list of installed packages includes:

kernel-devicetree
kernel-image-3.6.9-yocto-standard

So when I build my updated Linux kernel, the kernel image is now named "kernel-image-3.10.0-yocto-standard". So when I attempt to do an upgrade, the kernel-devicetree is upgraded just fine but the kernel image is not; kernel-image-3.10.0-yocto-standard is not the same package as kernel-image-3.6.9-yocto-standard, so opkg does not see this as an upgrade.

How do I force the packaging system to recognize that the new kernel is an upgrade of the old kernel, even though the packages have different names?
And for the future, how do I change the package name so that it doesn't
include the version?

Thanks,
Bryan

------------------------------

Message: 2
Date: Mon, 25 Nov 2013 14:52:00 +0000
From: "Rifenbark, Scott M" <[email protected]>
To: "Robert P. J. Day" <[email protected]>, Yocto discussion list
<[email protected]>
Subject: Re: [yocto] ksize.py and dirsize.py scripts not in dora
tarball, is that deliberate?
Message-ID:

        <

41DEA4B02DBDEF40A0F3B6D0DDB1237983F51186@ORSMSX101.amr.corp.intel.com>

Content-Type: text/plain; charset="us-ascii"

Hmm... did these not get included in poky during the release?

Scott

Quoted 6 lines

-----Original Message-----
From: [email protected] [mailto:yocto- [email protected]] On Behalf Of Robert P. J. Day
Sent: Saturday, November 23, 2013 2:59 AM
To: Yocto discussion list
Subject: [yocto] ksize.py and dirsize.py scripts not in dora tarball, is

that

Quoted 14 lines

deliberate?

current dev manual:

http://www.yoctoproject.org/docs/latest/dev-manual/dev-
manual.html#understand-what-gives-your-image-size

claims:

"To help you see where you currently are with kernel and root filesystem sizes, you can use two tools found in the Source Directory in the scripts directory:

Quoted 5 lines

ksize.py: Reports component sizes for the kernel build objects.

dirsize.py: Reports component sizes for the root filesystem."

however, the current dora tarball doesn't contain these scripts, while

oe-

Quoted 12 lines

core contains them more specifically under the scripts/tiny/ subdirectory. thoughts?

rday

--

===========================================================
=============
Robert P. J. Day Ottawa, Ontario, CANADA
http://crashcourse.ca

Quoted 8 lines
Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday

===========================================================
=============
_______________________________________________
yocto mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/yocto

------------------------------

Message: 3
Date: Mon, 25 Nov 2013 15:01:29 +0000
From: Paul Eggleton <[email protected]>
To: Michael Halstead <[email protected]>
Cc: [email protected]
Subject: Re: [yocto] RFC Autobuilder Naming
Message-ID: <6760837.T0hJxS54r8@helios>
Content-Type: text/plain; charset="us-ascii"

Hi Michael,

Quoted 11 lines · On Friday 22 November 2013 15:43:58 Michael Halstead wrote:

Currently all builders are simply named ab00-ab13 based on when we acquired them. The name doesn't change except for ab01 which has been reused on new hardware. As we increase the number of builders and make the build process more flexible this becomes inconvenient. I'd like to switch to embedding useful information in the name.

Going forward as I refresh builders I'd like to name them with their distro and location.

For example at OSL:

Quoted 7 lines

fedora19.osl.yoctoproject.org
fedora20.osl.yoctoproject.org
opensuse131.osl.yoctoproject.org
ubuntu1310.osl.yoctoproject.org
etc.

And if we added a build clusters in the EasyStreet datacenter in

Portland:

Quoted 5 lines

centos64.pdx.yoctoproject.org
debian7.pdx.yoctoproject.org
etc.

Quoted 7 lines

This allows for additional build clusters to be added and makes the distro clear to anyone viewing the logs. The current available build slaves can be found on the relevant cluster's buildbot interface
(http://autobuilder.yoctoproject.org/main/buildslaves) so as names change there is an easy place to look them up.

Comments?

I guess this works fine until you want to have two builders with the same distro. Maybe we plan not to do that anymore...?

Cheers,
Paul

--

Paul Eggleton
Intel Open Source Technology Centre

------------------------------

Message: 4
Date: Mon, 25 Nov 2013 07:03:34 -0800
From: Todd Stellanova <[email protected]>
To: Paul Eggleton <[email protected]>
Cc: "[email protected]" <[email protected]>
Subject: Re: [yocto] unbootable image produced with PACKAGE_CLASSES ?=
"package_ipk", /etc missing
Message-ID: <[email protected]>
Content-Type: text/plain; charset=us-ascii

Thanks for the ideas. I'll try creating a new build folder. If that still shows the problem, I'm thinking this has something to do with the fact that I'm running the build inside a vm (inside an Ubuntu vm running on a Mac). It looks like the build is using debugfs...maybe it's running out of ram at some point and not obtaining more in the vm properly?

Quoted 1 line

On Nov 25, 2013, at 5:21 AM, Paul Eggleton <

Quoted 7 lines · [email protected]> wrote:

Hi Nicolas / Todd,

Quoted 4 lines · On Monday 25 November 2013 11:31:42 Nicolas Dechesne wrote:
Quoted 1 line · Todd Stellanova

It appears that copying the files to the ext3 / sdcard image is

failing in

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines

*populate-extfs.sh*
I see a series of these errors:

*copy_file: Could not allocate block in ext2 filesystem*

Any idea what might cause this? I've verified that the initial .tar archive and the bz2 contain the right files.

can you try to create a new <build> folder (do not remove the current

one

Quoted 1 line
Quoted 1 line

for now) and reuse the downloads and sstate folder? i am wondering if

there

Quoted 4 lines
Quoted 2 lines

is a bug when trying to change PACKAGE_CLASSES in an existing <build> folder.

I do this not infrequently and never hit a problem like this, so I doubt

this

Quoted 3 lines

is the case.

Either there is a problem in how the filesystem is being set up (block

sizes,

Quoted 9 lines

etc.) or there is some kind of corruption occurring.

Cheers,
Paul

--

Paul Eggleton
Intel Open Source Technology Centre

------------------------------

Message: 5
Date: Mon, 25 Nov 2013 15:21:41 +0000
From: "Rifenbark, Scott M" <[email protected]>
To: "Robert P. J. Day" <[email protected]>
Cc: Yocto discussion list <[email protected]>
Subject: [yocto] FW: stupid question about post-installation scripts Message-ID:

        <

41DEA4B02DBDEF40A0F3B6D0DDB1237983F511F7@ORSMSX101.amr.corp.intel.com>

Content-Type: text/plain; charset="us-ascii"

Robert,

I don't know... I am forwarding to the discussion list.

Scott

Quoted 9 lines

-----Original Message-----
From: Robert P. J. Day [mailto:[email protected]]
Sent: Sunday, November 24, 2013 2:37 AM
To: Rifenbark, Scott M
Subject: stupid question about post-installation scripts

when one defines pkg_postinst scripts, are those scripts invoked at
*both* root filesystem creation time and first boot time? so that one

needs to

Quoted 1 line

manually check the value of ${D} to decide whether to run them, say, at

first

Quoted 8 lines

boot time?

i'm reading the section here:

http://www.yoctoproject.org/docs/latest/dev-manual/dev-
manual.html#post-installation-scripts

and i know i've seen elsewhere scripts explicitly checking the value of

the ${D}

Quoted 1 line

prefix to determine when they're being invoked. it *seems* like that's

what's

Quoted 1 line

happening, but if that's the case, it can probably be said much more

clearly.

Quoted 2 lines

rday

------------------------------

Message: 6
Date: Mon, 25 Nov 2013 10:56:17 -0500
From: Bryan Evenson <[email protected]>
To: "Rifenbark, Scott M" <[email protected]>, "Robert P. J.
Day" <[email protected]>
Cc: Yocto discussion list <[email protected]>
Subject: Re: [yocto] stupid question about post-installation scripts Message-ID:

<91586D499ADFD74FBCFB8425266A5DE40153C1156096@pluto.melinkcorp.local>
Content-Type: text/plain; charset="us-ascii"

Robert,

That's how it works in my experience. I have some packages for my system that have a postinst piece that needs to run during image creation, and other pieces that need to run only on a package upgrade. By checking whether "x${D}" = "${D}", I am filtering out whether the postinst script is running during image creation or on the actual hardware. Been working great so far.

-Bryan

Quoted 35 lines

-----Original Message-----
From: [email protected] [mailto:yocto- [email protected]] On Behalf Of Rifenbark, Scott M
Sent: Monday, November 25, 2013 10:22 AM
To: Robert P. J. Day
Cc: Yocto discussion list
Subject: [yocto] FW: stupid question about post-installation scripts

Robert,

I don't know... I am forwarding to the discussion list.

Scott

Quoted 12 lines

-----Original Message-----
From: Robert P. J. Day [mailto:[email protected]]
Sent: Sunday, November 24, 2013 2:37 AM
To: Rifenbark, Scott M
Subject: stupid question about post-installation scripts

when one defines pkg_postinst scripts, are those scripts invoked at
*both* root filesystem creation time and first boot time? so that one needs to manually check the value of ${D} to decide whether to run them, say, at first boot time?

Quoted 6 lines

i'm reading the section here:

http://www.yoctoproject.org/docs/latest/dev-manual/dev-
manual.html#post-installation-scripts

and i know i've seen elsewhere scripts explicitly checking the value

of

Quoted 2 lines

the ${D} prefix to determine when they're being invoked. it *seems* like that's what's happening, but if that's the case, it can probably

Quoted 7 lines

be said much more clearly.

Quoted 2 lines

rday

_______________________________________________
yocto mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/yocto

------------------------------

Message: 7
Date: Mon, 25 Nov 2013 11:35:47 -0500
From: Bryan Evenson <[email protected]>
To: Bryan Evenson <[email protected]>, "[email protected]"
<[email protected]>
Subject: Re: [yocto] Unable to upgrade kernel; package for kernel name
includes version
Message-ID:

<91586D499ADFD74FBCFB8425266A5DE40153C11560E0@pluto.melinkcorp.local>
Content-Type: text/plain; charset="us-ascii"

All,

Quoted 28 lines

-----Original Message-----
From: [email protected] [mailto:yocto- [email protected]] On Behalf Of Bryan Evenson
Sent: Monday, November 25, 2013 9:46 AM
To: [email protected]
Subject: [yocto] Unable to upgrade kernel; package for kernel name includes version

I'm on poky/dylan-9.0.1 I am upgrading the Linux kernel for my device's image from a custom 3.6.9 kernel to a custom 3.10 kernel. I'm using this layer: https://github.com/evensonbryan/meta-atmel as a basis for the custom kernel recipe (there is a recipe for the 3.10 kernel that I haven't yet deployed to Github in case anyone is wondering). I'm attempting to do a firmware upgrade with my built packages (using opkg for package management) and I've discovered an issue with the kernel naming. On my pre-upgraded device, a list of installed packages includes:

kernel-devicetree
kernel-image-3.6.9-yocto-standard

So when I build my updated Linux kernel, the kernel image is now named "kernel-image-3.10.0-yocto-standard". So when I attempt to do an upgrade, the kernel-devicetree is upgraded just fine but the kernel image is not; kernel-image-3.10.0-yocto-standard is not the same package as kernel-image-3.6.9-yocto-standard, so opkg does not see this as an upgrade.

Quoted 4 lines

How do I force the packaging system to recognize that the new kernel is an upgrade of the old kernel, even though the packages have different names? And for the future, how do I change the package name so that it doesn't include the version?

I have a partial solution, but I'm still curious as to if there is a cleaner method than what I've found so far. I have a package that is an image version so that way I can easily tell the overall distribution version. In that package, I added the line:

RDEPENDS_${PN} = "kernel-image (>= 3.10)"

When I did this, upgrading my image version package caused kernel-image-3.10.0-yocto-standard to be installed. However, opkg still claims that kernel-image-3.6.9-yocto-standard is still installed; it'd be nice to remove this package also. Would adding:

RCONFLICTS_${PN} = "kernel-image-3.6.9-yocto-standard" RREPLACES_${PN} = "kernel-image-3.6.9-yocto-standard"

to the 3.10 kernel recipe fix the issue? And if so, moving forward would I then need to append the list with each kernel version that is created, in case I have an old system that has never been upgraded? I think this will work, but it seems messy.

Thanks,
Bryan

Quoted 10 lines

Thanks,
Bryan

_______________________________________________
yocto mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/yocto

------------------------------

Message: 8
Date: Mon, 25 Nov 2013 17:01:10 +0000
From: Paul Eggleton <[email protected]>
To: "Rifenbark, Scott M" <[email protected]>, "Robert P. J.
Day" <[email protected]>
Cc: [email protected]
Subject: Re: [yocto] FW: stupid question about post-installation
scripts
Message-ID: <2877100.Goa8NZvDMD@helios>
Content-Type: text/plain; charset="us-ascii"

Quoted 8 lines · Scott M
Quoted 8 lines

From: Robert P. J. Day [mailto:[email protected]]
Sent: Sunday, November 24, 2013 2:37 AM
To: Rifenbark, Scott M
Subject: stupid question about post-installation scripts

when one defines pkg_postinst scripts, are those scripts invoked at

*both* root filesystem creation time and first boot time? so that one

needs

Quoted 1 line
Quoted 1 line

to manually check the value of ${D} to decide whether to run them, say,

at

Quoted 1 line
Quoted 1 line

first boot time?

Not both - it's either-or; i.e. it will be run at rootfs creation time and if
it fails then, it will be run on first boot. Yes you can use the value of $D
(note: *not* ${D}!) to find out where the script is being called from, if necessary.

Quoted 5 lines
Quoted 2 lines

i'm reading the section here:
http://www.yoctoproject.org/docs/latest/dev-manual/dev->

manual.html#post-installation-scripts

Quoted 2 lines

and i know i've seen elsewhere scripts explicitly checking the value of

the

Quoted 3 lines
Quoted 3 lines

${D} prefix to determine when they're being invoked. it *seems* like that's what's happening, but if that's the case, it can probably be said much more clearly.

The answer is, you can do this if you have to, but ideally you shouldn't need
to. In the ideal situation the script should be written such that it functions
equally well no matter where it executes; that avoids the need to run anything
on first boot, *and* (assuming you have package management enabled for the target) it will work if the package is installed on the target some time afterwards.

Cheers,
Paul

--

Paul Eggleton
Intel Open Source Technology Centre

------------------------------

Message: 9
Date: Mon, 25 Nov 2013 17:06:51 +0000
From: Paul Eggleton <[email protected]>
To: "Rifenbark, Scott M" <[email protected]>, "Robert P. J.
Day" <[email protected]>
Cc: [email protected]
Subject: Re: [yocto] FW: stupid question about post-installation
scripts
Message-ID: <197917002.9dER9YZC1c@helios>
Content-Type: text/plain; charset="us-ascii"

Quoted 11 lines · On Monday 25 November 2013 17:01:10 Paul Eggleton wrote:
Quoted 10 lines · Scott M
Quoted 10 lines

From: Robert P. J. Day [mailto:[email protected]]
Sent: Sunday, November 24, 2013 2:37 AM
To: Rifenbark, Scott M
Subject: stupid question about post-installation scripts

when one defines pkg_postinst scripts, are those scripts invoked at

*both* root filesystem creation time and first boot time? so that one needs
to manually check the value of ${D} to decide whether to run them,

say,

Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

at
first boot time?

Not both - it's either-or; i.e. it will be run at rootfs creation time

and

Quoted 3 lines

if it fails then, it will be run on first boot. Yes you can use the value of $D (note: *not* ${D}!) to find out where the script is being called from, if necessary.

PS, if you really do want to force the script to be postponed until first boot,
you *must* make the postinst script fail if $D has a value - if you just let
it pass the system will assume it doesn't need to be called again on first boot.

Cheers,
Paul

--

Paul Eggleton
Intel Open Source Technology Centre

------------------------------

_______________________________________________
yocto mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/yocto

End of yocto Digest, Vol 38, Issue 92
*************************************