Date   

Re: Yocto [thud], [zeus] do_fetch and do_unpack failures with offline/online svn build! #yocto #python

Georgi Georgiev
 

Sorry Khem,
With esc character '\' before & it can't take the full path. So briefly:

Yocto build with char '\' before & in SRC_URI:
online - OK
offline - Error - svn: E170013: Unable to connect to a repository...
Yocto build without '\' in SRC_URI:
online and offline same error - do_unpack: Unpack failure for URL - "package".tar.gz is present in DL_DIR

Thanks

-----Original Message-----
From: yocto@lists.yoctoproject.org [mailto:yocto@lists.yoctoproject.org] On Behalf Of Georgi Georgiev via Lists.Yoctoproject.Org
Sent: Wednesday, February 05, 2020 5:26 PM
To: Khem Raj <raj.khem@gmail.com>; yocto@lists.yoctoproject.org
Cc: yocto@lists.yoctoproject.org
Subject: Re: [EXTERNAL] Re: [yocto] Yocto [thud], [zeus] do_fetch and do_unpack failures with offline/online svn build! #yocto #python

Hi Khem,

Yes, no issues with that. When I am connected to network it fetches the code with and without escape character before & in the path.
svn co " https://urldefense.proofpoint.com/v2/url?u=http-3A__cocosubversion_svn_Embedded_Valve-26Actuator_DVPII_trunk_SOCPACManEnvEngKeys&d=DwIGaQ&c=y6L7g950KfMp92YmLM0QlMdXcRn6b-Cq4AApnSJOenE&r=kHtJrQGzfH1ZmfsNkJpYuH-jtNpv8yMDkqAmsRP99mc&m=vXia-xlcxuLr2fSmBogZpMIaOntXLKWv3mkxcRgtUc4&s=UYH9lueDIeMGZndMxgbZgb76A5EfqN-MR58DBv1iriI&e= " works fine and fetches the package.

Georgi

-----Original Message-----
From: Khem Raj [mailto:raj.khem@gmail.com]
Sent: Friday, January 31, 2020 8:33 PM
To: Georgi Georgiev <Georgi.Georgiev@woodward.com>; yocto@lists.yoctoproject.org
Subject: [EXTERNAL] Re: [yocto] Yocto [thud], [zeus] do_fetch and do_unpack failures with offline/online svn build! #yocto #python

On 1/31/20 4:02 AM, georgi.georgiev via Lists.Yoctoproject.Org wrote:
Hello Community,
This is the third time I am asking for support on this issue. This
time I decided to use the web form.
In our project we have a requirement to be able to build the project
offline. E.g. on the field without any network connection. When we are
connected with the recipe mentioned below we don't have issues:

svn: E170013: Unable to connect to a repository at URL 'https://urldefense.proofpoint.com/v2/url?u=http-3A__cocosubversion_svn_Embedded_Valve-26Actuator_DVPII_trunk_SOCPACManEnvEngKeys_trunk&d=DwIGaQ&c=y6L7g950KfMp92YmLM0QlMdXcRn6b-Cq4AApnSJOenE&r=kHtJrQGzfH1ZmfsNkJpYuH-jtNpv8yMDkqAmsRP99mc&m=ocx6BoVSrDQGWUvUUVvgKtuJbT7eH7jFSjCy1Ys73Vw&s=OTunnmsi_tCUt3JytmU2Hs7i7Xnqhl8-2CMsqmzIl90&e= '
svn: E670003: Temporary failure in name resolution

can you try checking out the repo out side of fetcher and see if the machine can fetch the url you are using in SRC_URI ?

*SUMMARY = "PACMan - Parameter And Configuration MANager"* *LICENSE =
"CLOSED"* *inherit systemd useradd* *REQUIRED_DISTRO_FEATURES =
"systemd"* *# SVN revision* *PV = "121026"* *# Name of SVN project*
*PACMAN_PROJ_NAME="SOCPACManEnvEngKeys"*
*SRC_URI =
"svn://cocosubversion/svn/Embedded/Valve%5C&Actuator/DVPII/trunk/$%7BP
ACMAN_PROJ_NAME%7D;module=trunk;protocol=http;externals=allowed;rev=$%
7BPV%7D%22* *SRC_URI += "file://ww-authpacman.service%22* *SRC_URI +=
"file://ww-pacman.service%22* *S = "${WORKDIR}/trunk/"* *# ${PN}-sys:
content related to system, which goes to base rootfs (only .service
file and symlinks)* *# ${PN}:      real content which may go to
separate partition* *PACKAGES =+ " ${PN}-sys"* .........
When disconnect the network, erase sstate-cache, cache and tmp I see
log file attached (log.do_fetch.32757) and the following output:

*ERROR: ww-pacman-121026-r0 do_fetch: Fetcher failure: Fetch command
export PSEUDO_DISABLED=1; unset _PYTHON_SYSCONFIGDATA_NAME; export
DBUS_SESSION_BUS_ADDRESS="unix:path=/run/user/1000/bus"; export
SSH_AGENT_PID="11412"; export
SSH_AUTH_SOCK="/run/user/1000/keyring/ssh"; export
PATH="/home/w23698/projects/proj_dvp2/build_dvp2/tmp/sysroots-uninativ
e/x86_64-linux/usr/bin:/home/w23698/projects/proj_dvp2/sources/poky/sc
ripts:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-n
eon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/b
in/arm-poky-linux-gnueabi:/home/w23698/projects/proj_dvp2/build_dvp2/t
mp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-
sysroot/usr/bin/crossscripts:/home/w23698/projects/proj_dvp2/build_dvp
2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/reci
pe-sysroot-native/usr/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/
tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe
-sysroot-native/usr/bin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp
/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sy
sroot-native/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/
cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-
native/bin:/home/w23698/projects/proj_dvp2/sources/poky/bitbake/bin:/h
ome/w23698/projects/proj_dvp2/build_dvp2/tmp/hosttools";
export HOME="/home/w23698"; /usr/bin/env svn --non-interactive
--trust-server-cert update --no-auth-cache -r 121026 failed with exit
code 1, output:* *Updating '.':*
*svn: E170013: Unable to connect to a repository at URL
'https://urldefense.proofpoint.com/v2/url?u=http-3A__cocosubversion_sv
n_Embedded_Valve-26Actuator_DVPII_trunk_SOCPACManEnvEngKeys_trunk-27-2
A&d=DwIGaQ&c=y6L7g950KfMp92YmLM0QlMdXcRn6b-Cq4AApnSJOenE&r=kHtJrQGzfH1
ZmfsNkJpYuH-jtNpv8yMDkqAmsRP99mc&m=ocx6BoVSrDQGWUvUUVvgKtuJbT7eH7jFSjC
y1Ys73Vw&s=Ffv1JU1QYBh4g49fmoLnsDSFgMMBc_5MbOpy59QUS18&e=
*svn: E670003: Temporary failure in name resolution*
*ERROR: ww-pacman-121026-r0 do_fetch: Fetcher failure for URL:
'svn://cocosubversion/svn/Embedded/Valve%5C&Actuator/DVPII/trunk/SOCPACManEnvEngKeys;module=trunk;protocol=http;externals=allowed;rev=121026'.
Unable to fetch URL from any source.*
*ERROR: Logfile of failure stored in:
/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-po
ky-linux-gnueabi/ww-pacman/121026-r0/temp/log.do_fetch.32757*
*ERROR: Task
(/home/w23698/projects/proj_dvp2/build_dvp2/../sources/meta-ww-dvp2/re
cipes-ww/ww-pacman/ww-pacman.bb:do_fetch)
failed with exit code '1'*

When remove the '\' character in SRC_URI, e.g. to become:

*SRC_URI =
"svn://cocosubversion/svn/Embedded/Valve&Actuator/DVPII/trunk/$%7BPACM
AN_PROJ_NAME%7D;module=trunk;protocol=http;externals=allowed;rev=$%7BP
V%7D%22*

In connected and not connected to network do_fetch() passes
successfully but I see one and same error (log.do_unpack.25226) output:

*ERROR: ww-pacman-121026-r0 do_unpack: Unpack failure for URL:
'svn://cocosubversion/svn/Embedded/Valve&Actuator/DVPII/trunk/SOCPACManEnvEngKeys;module=trunk;protocol=http;externals=allowed;rev=121026'.
Unpack command
PATH="/home/w23698/projects/proj_dvp2/build_dvp2/tmp/sysroots-uninative/x86_64-linux/usr/bin:/home/w23698/projects/proj_dvp2/sources/poky/scripts:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/bin/arm-poky-linux-gnueabi:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot/usr/bin/crossscripts:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/bin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/bin:/home/w23698/projects/proj_dvp2/sources/poky/bitbake/bin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/hosttools"
tar xz --no-same-owner -f
/home/w23698/projects/proj_dvp2/build_dvp2/downloads/trunk_cocosubvers
ion_.svn.Embedded.Valve&Actuator.DVPII.trunk.SOCPACManEnvEngKeys_12102
6_.tar.gz
failed with return value 127*
*ERROR: Logfile of failure stored in:
/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-po
ky-linux-gnueabi/ww-pacman/121026-r0/temp/log.do_unpack.25226*
*ERROR: Task
(/home/w23698/projects/proj_dvp2/build_dvp2/../sources/meta-ww-dvp2/re
cipes-ww/ww-pacman/ww-pacman.bb:do_unpack)
failed with exit code '1'
*
I don't it this matters but, the build machine is baremetal Ubuntu
18.04.3 LTS. In all cases the packed tar.gz remain in downloads
directory with one name!

Cordially,
Georgi




***
The information in this email is confidential and intended solely for the individual or entity to whom it is addressed.  If you have received this email in error please notify the sender by return e-mail, delete this email, and refrain from any disclosure or action based on the information.
***

***
The information in this email is confidential and intended solely for the individual or entity to whom it is addressed.  If you have received this email in error please notify the sender by return e-mail, delete this email, and refrain from any disclosure or action based on the information.
***


Re: [EXTERNAL] Re: [yocto] Yocto [thud], [zeus] do_fetch and do_unpack failures with offline/online svn build! #python #yocto

Georgi Georgiev
 

Hi Khem,

Yes, no issues with that. When I am connected to network it fetches the code with and without escape character before & in the path.
svn co " http://cocosubversion/svn/Embedded/Valve&Actuator/DVPII/trunk/SOCPACManEnvEngKeys" works fine and fetches the package.

Georgi

-----Original Message-----
From: Khem Raj [mailto:raj.khem@gmail.com]
Sent: Friday, January 31, 2020 8:33 PM
To: Georgi Georgiev <Georgi.Georgiev@woodward.com>; yocto@lists.yoctoproject.org
Subject: [EXTERNAL] Re: [yocto] Yocto [thud], [zeus] do_fetch and do_unpack failures with offline/online svn build! #yocto #python

On 1/31/20 4:02 AM, georgi.georgiev via Lists.Yoctoproject.Org wrote:
Hello Community,
This is the third time I am asking for support on this issue. This
time I decided to use the web form.
In our project we have a requirement to be able to build the project
offline. E.g. on the field without any network connection. When we are
connected with the recipe mentioned below we don't have issues:

svn: E170013: Unable to connect to a repository at URL 'https://urldefense.proofpoint.com/v2/url?u=http-3A__cocosubversion_svn_Embedded_Valve-26Actuator_DVPII_trunk_SOCPACManEnvEngKeys_trunk&d=DwIGaQ&c=y6L7g950KfMp92YmLM0QlMdXcRn6b-Cq4AApnSJOenE&r=kHtJrQGzfH1ZmfsNkJpYuH-jtNpv8yMDkqAmsRP99mc&m=ocx6BoVSrDQGWUvUUVvgKtuJbT7eH7jFSjCy1Ys73Vw&s=OTunnmsi_tCUt3JytmU2Hs7i7Xnqhl8-2CMsqmzIl90&e= '
svn: E670003: Temporary failure in name resolution

can you try checking out the repo out side of fetcher and see if the machine can fetch the url you are using in SRC_URI ?

*SUMMARY = "PACMan - Parameter And Configuration MANager"* *LICENSE =
"CLOSED"* *inherit systemd useradd* *REQUIRED_DISTRO_FEATURES =
"systemd"* *# SVN revision* *PV = "121026"* *# Name of SVN project*
*PACMAN_PROJ_NAME="SOCPACManEnvEngKeys"*
*SRC_URI =
"svn://cocosubversion/svn/Embedded/Valve%5C&Actuator/DVPII/trunk/$%7BP
ACMAN_PROJ_NAME%7D;module=trunk;protocol=http;externals=allowed;rev=$%
7BPV%7D%22* *SRC_URI += "file://ww-authpacman.service%22* *SRC_URI +=
"file://ww-pacman.service%22* *S = "${WORKDIR}/trunk/"* *# ${PN}-sys:
content related to system, which goes to base rootfs (only .service
file and symlinks)* *# ${PN}:      real content which may go to
separate partition* *PACKAGES =+ " ${PN}-sys"* .........
When disconnect the network, erase sstate-cache, cache and tmp I see
log file attached (log.do_fetch.32757) and the following output:

*ERROR: ww-pacman-121026-r0 do_fetch: Fetcher failure: Fetch command
export PSEUDO_DISABLED=1; unset _PYTHON_SYSCONFIGDATA_NAME; export
DBUS_SESSION_BUS_ADDRESS="unix:path=/run/user/1000/bus"; export
SSH_AGENT_PID="11412"; export
SSH_AUTH_SOCK="/run/user/1000/keyring/ssh"; export
PATH="/home/w23698/projects/proj_dvp2/build_dvp2/tmp/sysroots-uninativ
e/x86_64-linux/usr/bin:/home/w23698/projects/proj_dvp2/sources/poky/sc
ripts:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-n
eon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/b
in/arm-poky-linux-gnueabi:/home/w23698/projects/proj_dvp2/build_dvp2/t
mp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-
sysroot/usr/bin/crossscripts:/home/w23698/projects/proj_dvp2/build_dvp
2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/reci
pe-sysroot-native/usr/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/
tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe
-sysroot-native/usr/bin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp
/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sy
sroot-native/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/
cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-
native/bin:/home/w23698/projects/proj_dvp2/sources/poky/bitbake/bin:/h
ome/w23698/projects/proj_dvp2/build_dvp2/tmp/hosttools";
export HOME="/home/w23698"; /usr/bin/env svn --non-interactive
--trust-server-cert update --no-auth-cache -r 121026 failed with exit
code 1, output:* *Updating '.':*
*svn: E170013: Unable to connect to a repository at URL
'https://urldefense.proofpoint.com/v2/url?u=http-3A__cocosubversion_sv
n_Embedded_Valve-26Actuator_DVPII_trunk_SOCPACManEnvEngKeys_trunk-27-2
A&d=DwIGaQ&c=y6L7g950KfMp92YmLM0QlMdXcRn6b-Cq4AApnSJOenE&r=kHtJrQGzfH1
ZmfsNkJpYuH-jtNpv8yMDkqAmsRP99mc&m=ocx6BoVSrDQGWUvUUVvgKtuJbT7eH7jFSjC
y1Ys73Vw&s=Ffv1JU1QYBh4g49fmoLnsDSFgMMBc_5MbOpy59QUS18&e=
*svn: E670003: Temporary failure in name resolution*
*ERROR: ww-pacman-121026-r0 do_fetch: Fetcher failure for URL:
'svn://cocosubversion/svn/Embedded/Valve%5C&Actuator/DVPII/trunk/SOCPACManEnvEngKeys;module=trunk;protocol=http;externals=allowed;rev=121026'.
Unable to fetch URL from any source.*
*ERROR: Logfile of failure stored in:
/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-po
ky-linux-gnueabi/ww-pacman/121026-r0/temp/log.do_fetch.32757*
*ERROR: Task
(/home/w23698/projects/proj_dvp2/build_dvp2/../sources/meta-ww-dvp2/re
cipes-ww/ww-pacman/ww-pacman.bb:do_fetch)
failed with exit code '1'*

When remove the '\' character in SRC_URI, e.g. to become:

*SRC_URI =
"svn://cocosubversion/svn/Embedded/Valve&Actuator/DVPII/trunk/$%7BPACM
AN_PROJ_NAME%7D;module=trunk;protocol=http;externals=allowed;rev=$%7BP
V%7D%22*

In connected and not connected to network do_fetch() passes
successfully but I see one and same error (log.do_unpack.25226) output:

*ERROR: ww-pacman-121026-r0 do_unpack: Unpack failure for URL:
'svn://cocosubversion/svn/Embedded/Valve&Actuator/DVPII/trunk/SOCPACManEnvEngKeys;module=trunk;protocol=http;externals=allowed;rev=121026'.
Unpack command
PATH="/home/w23698/projects/proj_dvp2/build_dvp2/tmp/sysroots-uninative/x86_64-linux/usr/bin:/home/w23698/projects/proj_dvp2/sources/poky/scripts:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/bin/arm-poky-linux-gnueabi:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot/usr/bin/crossscripts:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/usr/bin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/sbin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-poky-linux-gnueabi/ww-pacman/121026-r0/recipe-sysroot-native/bin:/home/w23698/projects/proj_dvp2/sources/poky/bitbake/bin:/home/w23698/projects/proj_dvp2/build_dvp2/tmp/hosttools"
tar xz --no-same-owner -f
/home/w23698/projects/proj_dvp2/build_dvp2/downloads/trunk_cocosubvers
ion_.svn.Embedded.Valve&Actuator.DVPII.trunk.SOCPACManEnvEngKeys_12102
6_.tar.gz
failed with return value 127*
*ERROR: Logfile of failure stored in:
/home/w23698/projects/proj_dvp2/build_dvp2/tmp/work/cortexa9hf-neon-po
ky-linux-gnueabi/ww-pacman/121026-r0/temp/log.do_unpack.25226*
*ERROR: Task
(/home/w23698/projects/proj_dvp2/build_dvp2/../sources/meta-ww-dvp2/re
cipes-ww/ww-pacman/ww-pacman.bb:do_unpack)
failed with exit code '1'
*
I don't it this matters but, the build machine is baremetal Ubuntu
18.04.3 LTS. In all cases the packed tar.gz remain in downloads
directory with one name!

Cordially,
Georgi




***
The information in this email is confidential and intended solely for the individual or entity to whom it is addressed.  If you have received this email in error please notify the sender by return e-mail, delete this email, and refrain from any disclosure or action based on the information.
***


Re: Busybox and scripts licensing

Ross Burton
 

On Wed, 5 Feb 2020 at 12:16, <lukasz.odw39@gmail.com> wrote:
Let's assume we have busybox on our device (that's populated by default by Yocto). What license shall be chosen for package containing custom shell scripts?
I heard opinion, that all shell scripts running on Busybox shall be GPLv2(?).
That opinion is very wrong indeed.

If you write the scripts then you get to pick the license.

Ross


Busybox and scripts licensing

Lukasz Domowy
 

Hello,

Let's assume we have busybox on our device (that's populated by default by Yocto). What license shall be chosen for package containing custom shell scripts?
I heard opinion, that all shell scripts running on Busybox shall be GPLv2(?).

Best regards,
Lukasz


Re: [meta-qt5, zeus-next, python3-pyqt5] QtChart can't be imported

Adrian
 

Hi,

I turn out PyQtChart was simply missing in the meta-qt5 layer.
I have created a recipe which adds it and hopefully will be merged to the upstream.

Regards,

Adrian

On 31.01.2020 17:52, Adrian Fiergolski wrote:

Hi,

I use python3-pyqt5 from meta-qt5 (in zeus-next). I get an error in python3 when I try to import QtChart module (from PyQt5 import QtChart). Has anybody experience the same issue and knows a fix? The deployed image comes with qtchart packages.

I have created an issue on the layer's issue tracker (link), but so far no reply.

Regards,

Adrian


Re: Building and selecting multiple kernel versions

Quentin Schulz
 

On Tue, Feb 04, 2020 at 04:32:34PM +0000, Ryan Harkin wrote:
[...]

Actually, it does work like this:

export MACHINE=xyz
bitbake image

But not like this:

MACHINE=xyz bitbake image

No big deal, but I expected the last one to work.
I use it every day. I don't think we're doing something special (well,
for that particular thing :D), so there's at least a way to make it work
for sure :)

Quentin


Re: Building and selecting multiple kernel versions

Ryan Harkin <ryan.harkin@...>
 



On Tue, 4 Feb 2020 at 16:15, Quentin Schulz <quentin.schulz@...> wrote:
Hi Ryan,

On Tue, Feb 04, 2020 at 04:00:01PM +0000, Ryan Harkin wrote:
> Hi Quentin,
>
> Thanks for your reply.
>
> On Tue, 4 Feb 2020 at 08:51, Quentin Schulz <
> quentin.schulz@...> wrote:
>
> > Hi Ryan,
> >
> > On Mon, Feb 03, 2020 at 05:34:23PM +0000, Ryan Harkin wrote:
> > > Hello all,
> > >
> > > I'm looking for advice on how to support multiple kernel versions in my
> > > distro with minimal changes, and minimal disruption to my users.
> > >
> > > Some background:
> > >
> > > I have a legacy Sumo based distro with an image config, and a machine
> > > config, with the machine using a 4.9 kernel. Last year, I upgraded
> > > everything to Warrior, and moved to a 4.19 kernel.
> > >
> > > Some of my users who are migrating from Sumo, wish to keep their 4.9
> > > kernel. So I'm trying to work out how to handle this in the simplest way.
> > >
> > > I know that I can add a 4.9 recipes to my Warrior branches, set
> > > PREFERRED_VERSION in my distro.conf. But I don't want to change the
> > default
> > > preferred version globally. And I don't really want users to update the
> > > distro.conf. Ideally, they should be able to take what I give them and
> > > "just" build it to get a 4.9 or 4.19 variant.
> > >
> >
> > You can make two machine configuration files. One with
> > PREFERRED_VERSION_virtual/kernel = "4.9%" and the other with 4.19.
> >
> > They pick the correct machine when building, they should expect the
> > correct kernel in output. Only applies to images built for this machine
> > so I guess that's what you're looking for?
> >
>
> Yes, this works.
>
> Trying it showed a few small problems. Eg. I have a package that only builds
> for the 4.19 kernel, and needs to be excluded for 4.9. That's something I
> can
> handle in the recipes using the machine type, of course.
>

It depends on what exactly does this recipe represent? A kernel module?
In which case, you can put it in your machine configuration file under
MACHINE_ESSENTIAL_EXTRA_RDEPENDS or RRECOMMENDS and omit it for 4.9.

We have a way to specify runtime dependencies on specific versions:
https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#var-RDEPENDS

I unfortunately have no idea if Yocto supports such a thing in DEPENDS.


Good tips, thanks again! I'll have a play with it.
 
>
> >
> > > Ideally, I don't want to *have* to build both kernels and then create two
> > > images. I expect that will cause confusion and lead to people flashing
> > the
> > > wrong images. So I'd prefer it to be either/or. Of course, I have to test
> > > all of this, so I want to be able to build both variants in CI, which
> > makes
> > > editing distro.conf even more  unattractive.
> > >
> >
> > Same image (POV of bitbake <image>) but different machines, does that
> > match your requirements?
> >
> > FWIW, you can pass MACHINE= to the command line just before bitbake
> > <image> making it rather obvious which machine they pick.
> >
>
> Incidentally, that didn't work for me, but that's a symptom of how we setup
> the environment, where local.conf sets MACHINE. I changed it to "MACHINE ?=
> ",
> thinking it would let me override it via the shell. But it didn't. Strange,
> but not a
> big problem.
>

To check who's setting it, run `bitbake <somesimplerecipe> -e | less` and look
for the line starting with MACHINE=. Then you have the whole bitbake
logic to set it above that line. That could help you find out what's
happening.

Actually, it does work like this:

export MACHINE=xyz
bitbake image

But not like this:

MACHINE=xyz bitbake image

No big deal, but I expected the last one to work.


Quentin


Re: Building and selecting multiple kernel versions

Quentin Schulz
 

Hi Ryan,

On Tue, Feb 04, 2020 at 04:00:01PM +0000, Ryan Harkin wrote:
Hi Quentin,

Thanks for your reply.

On Tue, 4 Feb 2020 at 08:51, Quentin Schulz <
quentin.schulz@streamunlimited.com> wrote:

Hi Ryan,

On Mon, Feb 03, 2020 at 05:34:23PM +0000, Ryan Harkin wrote:
Hello all,

I'm looking for advice on how to support multiple kernel versions in my
distro with minimal changes, and minimal disruption to my users.

Some background:

I have a legacy Sumo based distro with an image config, and a machine
config, with the machine using a 4.9 kernel. Last year, I upgraded
everything to Warrior, and moved to a 4.19 kernel.

Some of my users who are migrating from Sumo, wish to keep their 4.9
kernel. So I'm trying to work out how to handle this in the simplest way.

I know that I can add a 4.9 recipes to my Warrior branches, set
PREFERRED_VERSION in my distro.conf. But I don't want to change the
default
preferred version globally. And I don't really want users to update the
distro.conf. Ideally, they should be able to take what I give them and
"just" build it to get a 4.9 or 4.19 variant.
You can make two machine configuration files. One with
PREFERRED_VERSION_virtual/kernel = "4.9%" and the other with 4.19.

They pick the correct machine when building, they should expect the
correct kernel in output. Only applies to images built for this machine
so I guess that's what you're looking for?
Yes, this works.

Trying it showed a few small problems. Eg. I have a package that only builds
for the 4.19 kernel, and needs to be excluded for 4.9. That's something I
can
handle in the recipes using the machine type, of course.
It depends on what exactly does this recipe represent? A kernel module?
In which case, you can put it in your machine configuration file under
MACHINE_ESSENTIAL_EXTRA_RDEPENDS or RRECOMMENDS and omit it for 4.9.

We have a way to specify runtime dependencies on specific versions:
https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#var-RDEPENDS

I unfortunately have no idea if Yocto supports such a thing in DEPENDS.



Ideally, I don't want to *have* to build both kernels and then create two
images. I expect that will cause confusion and lead to people flashing
the
wrong images. So I'd prefer it to be either/or. Of course, I have to test
all of this, so I want to be able to build both variants in CI, which
makes
editing distro.conf even more unattractive.
Same image (POV of bitbake <image>) but different machines, does that
match your requirements?

FWIW, you can pass MACHINE= to the command line just before bitbake
<image> making it rather obvious which machine they pick.
Incidentally, that didn't work for me, but that's a symptom of how we setup
the environment, where local.conf sets MACHINE. I changed it to "MACHINE ?=
",
thinking it would let me override it via the shell. But it didn't. Strange,
but not a
big problem.
To check who's setting it, run `bitbake <somesimplerecipe> -e | less` and look
for the line starting with MACHINE=. Then you have the whole bitbake
logic to set it above that line. That could help you find out what's
happening.

Quentin


Re: Building and selecting multiple kernel versions

Ryan Harkin <ryan.harkin@...>
 

Hi Quentin,

Thanks for your reply.

On Tue, 4 Feb 2020 at 08:51, Quentin Schulz <quentin.schulz@...> wrote:
Hi Ryan,

On Mon, Feb 03, 2020 at 05:34:23PM +0000, Ryan Harkin wrote:
> Hello all,
>
> I'm looking for advice on how to support multiple kernel versions in my
> distro with minimal changes, and minimal disruption to my users.
>
> Some background:
>
> I have a legacy Sumo based distro with an image config, and a machine
> config, with the machine using a 4.9 kernel. Last year, I upgraded
> everything to Warrior, and moved to a 4.19 kernel.
>
> Some of my users who are migrating from Sumo, wish to keep their 4.9
> kernel. So I'm trying to work out how to handle this in the simplest way.
>
> I know that I can add a 4.9 recipes to my Warrior branches, set
> PREFERRED_VERSION in my distro.conf. But I don't want to change the default
> preferred version globally. And I don't really want users to update the
> distro.conf. Ideally, they should be able to take what I give them and
> "just" build it to get a 4.9 or 4.19 variant.
>

You can make two machine configuration files. One with
PREFERRED_VERSION_virtual/kernel = "4.9%" and the other with 4.19.

They pick the correct machine when building, they should expect the
correct kernel in output. Only applies to images built for this machine
so I guess that's what you're looking for?

Yes, this works.

Trying it showed a few small problems. Eg. I have a package that only builds
for the 4.19 kernel, and needs to be excluded for 4.9. That's something I can
handle in the recipes using the machine type, of course.
 

> Ideally, I don't want to *have* to build both kernels and then create two
> images. I expect that will cause confusion and lead to people flashing the
> wrong images. So I'd prefer it to be either/or. Of course, I have to test
> all of this, so I want to be able to build both variants in CI, which makes
> editing distro.conf even more  unattractive.
>

Same image (POV of bitbake <image>) but different machines, does that
match your requirements?

FWIW, you can pass MACHINE= to the command line just before bitbake
<image> making it rather obvious which machine they pick.

Incidentally, that didn't work for me, but that's a symptom of how we setup
the environment, where local.conf sets MACHINE. I changed it to "MACHINE ?= ",
thinking it would let me override it via the shell. But it didn't. Strange, but not a
big problem.

I'd be interested to hear if anyone else has a different approach I could try for
comparison.

Thanks,
Ryan.


Yocto Project Status WW05'20

Stephen Jolley
 

Current Dev Position: YP 3.1 M3

Next Deadline: YP 3.1 M3 build date 2/24/2020

 

Next Team Meetings:

 

Key Status/Updates:

  • YP 3.1 M2 is due to be released with reservations.
  • There are significant concerns with M2 due to the build reproducibility failures which continue to cause major problems with master too. These will have to block the M3 build/release.
  • The NPM fetcher rework patches have now merged, we’d like to thank Jean-Marie Lemetayer for his patience whilst they were made into good shape with regression tests, thanks, its a good end result.
  • The host dependencies on makeinfo and pod2man have now been dropped to simplify build dependencies and improve reproducibility.
  • The mpc8315e-rdb machine/BSP has been dropped from meta-yocto-bsp.
  • Build reproducibility issues have been tracked down to a number of causes but many more remain, some with open bugs, some with bugs yet to be filed. Any assistance with reproducibility issues is appreciated as is the major factor blocking patch merging at present.
  • Reproducibility failure debugging has been improved with better integration into the autobuilder to assist with this. Saved problematic output is now saved at: https://autobuilder.yocto.io/pub/repro-fail/
  • Ongoing kernel issues still prevent kernel upgrades merging at present but work on debugging the issues is moving forward (thanks Victor/Cisco in particular)
  • We’re collecting a list of companies, products and projects which use the Yocto Project on the wiki: https://wiki.yoctoproject.org/wiki/Project_Users Please add any you know are missing (or email Richard/Stephen who can add).
  • We have created a “documentation” list and a “licensing” list on the yocto project mailing list infrastructure. We believe these topics would benefit from dedicated discussions, the people involved usually aren’t interested in bulk email as would be found on other lists. The new lists can be found at: https://lists.yoctoproject.org/g/licensing/join and https://lists.yoctoproject.org/g/docs/join 
  • The triage team is worried about attendance at triage meetings and the project is finding it hard to find people to help fix bugs. If anyone is willing to work on bugs, assistance would be greatly appreciated.

 

YP 3.1 Milestone Dates:

  • YP 3.1 M2 should release shortly
  • YP 3.1 M3 build date 2/24/2020
  • YP 3.1 M3 release date 3/6/2020
  • YP 3.1 M4 build date  3/30/2020
  • YP 3.1 M4 release date  4/24/2020

 

Planned upcoming dot releases:

  • YP 2.7.3 build date  2/10/2020
  • YP 2.7.3 release date 2/21/2020
  • YP 3.0.2 build date  2/3/2020
  • YP 3.0.2 release date 2/14/2020

 

Tracking Metrics:

 

The Yocto Project’s technical governance is through its Technical Steering Committee, more information is available at:

https://wiki.yoctoproject.org/wiki/TSC

 

The Status reports are now stored on the wiki at: https://wiki.yoctoproject.org/wiki/Weekly_Status

 

[If anyone has suggestions for other information you’d like to see on this weekly status update, let us know!]

 

Thanks,

 

Stephen K. Jolley

Yocto Project Program Manager

(    Cell:                (208) 244-4460

* Email:              sjolley.yp.pm@...

 


CONFFILES for RPM: %config or %config(noreplace) #yocto

Gary Huband
 

I'm using Morty.  When I add CONFFILES_${PN} to a recipe the recipe RPM spec file contains %config lines not %config(noreplace).  Is there a way to get %config(noreplace)?

Thanks

Gary


Re: Which package for nandbiterrs?

Marek Belisko
 

On Tue, Feb 4, 2020 at 1:22 PM JH <jupiter.hce@gmail.com> wrote:

On 2/4/20, Belisko Marek <marek.belisko@gmail.com> wrote:
Well did you check build directory if it was really build (check
.config)? Or better try to build as module and install kernel-modules
and then you can modprobe it.
Looking at code did you supply kernel arguments for test module:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c#L42
Hmm, tried:

# modprobe mtd_nandbiterrs
[ 695.090585]
[ 695.092143] ==================================================
[ 695.098317] mtd_nandbiterrs: MTD device: 0
[ 695.114256] mtd_nandbiterrs: MTD device size 1048576, eraseblock=131072, pag4
[ 695.122867] mtd_nandbiterrs: Device uses 1 subpages of 2048 bytes
[ 695.129138] mtd_nandbiterrs: Using page=0, offset=0, eraseblock=0
[ 695.144888] mtd_nandbiterrs: incremental biterrors test
[ 695.150594] mtd_nandbiterrs: write_page
[ 695.158629] mtd_nandbiterrs: rewrite page
[ 695.163488] mtd_nandbiterrs: read_page
[ 695.170790] mtd_nandbiterrs: verify_page
[ 695.174887] mtd_nandbiterrs: Successfully corrected 0 bit errors per subpage
[ 695.182279] mtd_nandbiterrs: Inserted biterror @ 0/5
[ 695.187387] mtd_nandbiterrs: rewrite page
[ 695.196243] mtd_nandbiterrs: read_page
[ 695.202608] mtd_nandbiterrs: Read reported 1 corrected bit errors
[ 695.209115] mtd_nandbiterrs: verify_page
[ 695.213192] mtd_nandbiterrs: Successfully corrected 1 bit errors per subpage
[ 695.220361] mtd_nandbiterrs: Inserted biterror @ 0/2
[ 695.225361] mtd_nandbiterrs: rewrite page
[ 695.235261] mtd_nandbiterrs: read_page
[ 695.240237] mtd_nandbiterrs: Read reported 2 corrected bit errors
[ 695.246384] mtd_nandbiterrs: verify_page
[ 695.250771] mtd_nandbiterrs: Successfully corrected 2 bit errors per subpage
[ 695.257984] mtd_nandbiterrs: Inserted biterror @ 0/0
[ 695.262984] mtd_nandbiterrs: rewrite page
[ 695.273646] mtd_nandbiterrs: read_page
[ 695.280000] mtd_nandbiterrs: Read reported 2 corrected bit errors
[ 695.286230] mtd_nandbiterrs: verify_page
[ 695.290489] mtd_nandbiterrs: Error: page offset 0, expected 25, got 00
[ 695.297155] mtd_nandbiterrs: Error: page offset 282, expected 29, got 28
[ 695.303897] mtd_nandbiterrs: Error: page offset 359, expected a7, got 27
[ 695.310834] mtd_nandbiterrs: ECC failure, read data is incorrect despite reas
modprobe: ERROR: could not insert 'mtd_nandbiterrs': Input/output error
Well looks like some ecc issue (just blind guess). This is
unfortunately out of scope of this forum.
Pls ask on kernel mailing list. Thanks.

Now I am in trouble, what is that error means?

Thank you marek,

- jh
BR,

marek


Re: Which package for nandbiterrs?

JH
 

On 2/4/20, Belisko Marek <marek.belisko@gmail.com> wrote:
Well did you check build directory if it was really build (check
.config)? Or better try to build as module and install kernel-modules
and then you can modprobe it.
Looking at code did you supply kernel arguments for test module:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c#L42
Hmm, tried:

# modprobe mtd_nandbiterrs
[ 695.090585]
[ 695.092143] ==================================================
[ 695.098317] mtd_nandbiterrs: MTD device: 0
[ 695.114256] mtd_nandbiterrs: MTD device size 1048576, eraseblock=131072, pag4
[ 695.122867] mtd_nandbiterrs: Device uses 1 subpages of 2048 bytes
[ 695.129138] mtd_nandbiterrs: Using page=0, offset=0, eraseblock=0
[ 695.144888] mtd_nandbiterrs: incremental biterrors test
[ 695.150594] mtd_nandbiterrs: write_page
[ 695.158629] mtd_nandbiterrs: rewrite page
[ 695.163488] mtd_nandbiterrs: read_page
[ 695.170790] mtd_nandbiterrs: verify_page
[ 695.174887] mtd_nandbiterrs: Successfully corrected 0 bit errors per subpage
[ 695.182279] mtd_nandbiterrs: Inserted biterror @ 0/5
[ 695.187387] mtd_nandbiterrs: rewrite page
[ 695.196243] mtd_nandbiterrs: read_page
[ 695.202608] mtd_nandbiterrs: Read reported 1 corrected bit errors
[ 695.209115] mtd_nandbiterrs: verify_page
[ 695.213192] mtd_nandbiterrs: Successfully corrected 1 bit errors per subpage
[ 695.220361] mtd_nandbiterrs: Inserted biterror @ 0/2
[ 695.225361] mtd_nandbiterrs: rewrite page
[ 695.235261] mtd_nandbiterrs: read_page
[ 695.240237] mtd_nandbiterrs: Read reported 2 corrected bit errors
[ 695.246384] mtd_nandbiterrs: verify_page
[ 695.250771] mtd_nandbiterrs: Successfully corrected 2 bit errors per subpage
[ 695.257984] mtd_nandbiterrs: Inserted biterror @ 0/0
[ 695.262984] mtd_nandbiterrs: rewrite page
[ 695.273646] mtd_nandbiterrs: read_page
[ 695.280000] mtd_nandbiterrs: Read reported 2 corrected bit errors
[ 695.286230] mtd_nandbiterrs: verify_page
[ 695.290489] mtd_nandbiterrs: Error: page offset 0, expected 25, got 00
[ 695.297155] mtd_nandbiterrs: Error: page offset 282, expected 29, got 28
[ 695.303897] mtd_nandbiterrs: Error: page offset 359, expected a7, got 27
[ 695.310834] mtd_nandbiterrs: ECC failure, read data is incorrect despite reas
modprobe: ERROR: could not insert 'mtd_nandbiterrs': Input/output error

Now I am in trouble, what is that error means?

Thank you marek,

- jh


Re: Which package for nandbiterrs?

Marek Belisko
 

On Tue, Feb 4, 2020 at 12:43 PM Marek Belisko via
Lists.Yoctoproject.Org
<marek.belisko=gmail.com@lists.yoctoproject.org> wrote:

On Tue, Feb 4, 2020 at 12:22 PM JH <jupiter.hce@gmail.com> wrote:

On 2/4/20, Belisko Marek <marek.belisko@gmail.com> wrote:
On Tue, Feb 4, 2020 at 12:09 PM JH <jupiter.hce@gmail.com> wrote:

Hi,

I built the image in thud to include following MTD receipts:

mtd-utils
mtd-utils-ubifs
mtd-utils-tests

But the nandbiterrs is still missing, which package for nandbiterrs?
Are you referring to this one:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c
Then it's in kernel (not sure if tests are build though).
I think so, I have enabled CONFIG_MTD_TESTS=y, it still does not have
nandbiterrs. I built it with kernel 4.19.75, did you mean the that
build could not include that test file?
Well did you check build directory if it was really build (check
.config)? Or better try to build as module and install kernel-modules
and then you can modprobe it.
Looking at code did you supply kernel arguments for test module:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c#L42

Thanks marek,

- jh
BR,

marek
marek


Re: Which package for nandbiterrs?

Marek Belisko
 

On Tue, Feb 4, 2020 at 12:22 PM JH <jupiter.hce@gmail.com> wrote:

On 2/4/20, Belisko Marek <marek.belisko@gmail.com> wrote:
On Tue, Feb 4, 2020 at 12:09 PM JH <jupiter.hce@gmail.com> wrote:

Hi,

I built the image in thud to include following MTD receipts:

mtd-utils
mtd-utils-ubifs
mtd-utils-tests

But the nandbiterrs is still missing, which package for nandbiterrs?
Are you referring to this one:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c
Then it's in kernel (not sure if tests are build though).
I think so, I have enabled CONFIG_MTD_TESTS=y, it still does not have
nandbiterrs. I built it with kernel 4.19.75, did you mean the that
build could not include that test file?
Well did you check build directory if it was really build (check
.config)? Or better try to build as module and install kernel-modules
and then you can modprobe it.

Thanks marek,

- jh
BR,

marek


Re: Which package for nandbiterrs?

JH
 

On 2/4/20, Belisko Marek <marek.belisko@gmail.com> wrote:
On Tue, Feb 4, 2020 at 12:09 PM JH <jupiter.hce@gmail.com> wrote:

Hi,

I built the image in thud to include following MTD receipts:

mtd-utils
mtd-utils-ubifs
mtd-utils-tests

But the nandbiterrs is still missing, which package for nandbiterrs?
Are you referring to this one:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c
Then it's in kernel (not sure if tests are build though).
I think so, I have enabled CONFIG_MTD_TESTS=y, it still does not have
nandbiterrs. I built it with kernel 4.19.75, did you mean the that
build could not include that test file?

Thanks marek,

- jh


Re: Which package for nandbiterrs?

Marek Belisko
 

On Tue, Feb 4, 2020 at 12:09 PM JH <jupiter.hce@gmail.com> wrote:

Hi,

I built the image in thud to include following MTD receipts:

mtd-utils
mtd-utils-ubifs
mtd-utils-tests

But the nandbiterrs is still missing, which package for nandbiterrs?
Are you referring to this one:
https://elixir.bootlin.com/linux/latest/source/drivers/mtd/tests/nandbiterrs.c
Then it's in kernel (not sure if tests are build though).

Thank you.

Kind regards,

- jh
BR,

marek


Which package for nandbiterrs?

JH
 

Hi,

I built the image in thud to include following MTD receipts:

mtd-utils
mtd-utils-ubifs
mtd-utils-tests

But the nandbiterrs is still missing, which package for nandbiterrs?

Thank you.

Kind regards,

- jh


Re: Sad news about Scott Rifenbark

Robert P. J. Day
 

On Fri, 31 Jan 2020, Richard Purdie wrote:

I'm sorry to have to pass on the sad news that Scott Rifenbark, our
tech writer for the project passed away on Wednesday after a battle
with cancer.
... snip ...

Well, this is horrible news ... I got to know Scott from earlier days
of poring over the docs and submitting patches. I'm not sure how
documentation maintenance is now going to work but, for what it's
worth, I'm willing to step up and get involved again. Let me know what
I can do to help.

Rob Day


Re: Building and selecting multiple kernel versions

Quentin Schulz
 

Hi Ryan,

On Mon, Feb 03, 2020 at 05:34:23PM +0000, Ryan Harkin wrote:
Hello all,

I'm looking for advice on how to support multiple kernel versions in my
distro with minimal changes, and minimal disruption to my users.

Some background:

I have a legacy Sumo based distro with an image config, and a machine
config, with the machine using a 4.9 kernel. Last year, I upgraded
everything to Warrior, and moved to a 4.19 kernel.

Some of my users who are migrating from Sumo, wish to keep their 4.9
kernel. So I'm trying to work out how to handle this in the simplest way.

I know that I can add a 4.9 recipes to my Warrior branches, set
PREFERRED_VERSION in my distro.conf. But I don't want to change the default
preferred version globally. And I don't really want users to update the
distro.conf. Ideally, they should be able to take what I give them and
"just" build it to get a 4.9 or 4.19 variant.
You can make two machine configuration files. One with
PREFERRED_VERSION_virtual/kernel = "4.9%" and the other with 4.19.

They pick the correct machine when building, they should expect the
correct kernel in output. Only applies to images built for this machine
so I guess that's what you're looking for?

Ideally, I don't want to *have* to build both kernels and then create two
images. I expect that will cause confusion and lead to people flashing the
wrong images. So I'd prefer it to be either/or. Of course, I have to test
all of this, so I want to be able to build both variants in CI, which makes
editing distro.conf even more unattractive.
Same image (POV of bitbake <image>) but different machines, does that
match your requirements?

FWIW, you can pass MACHINE= to the command line just before bitbake
<image> making it rather obvious which machine they pick.

Quentin

5621 - 5640 of 53901