appears to be inconsistency re: base-files between oe-core and meta-angstrom

appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Robert P. J. Day]

  trying to build a "hardware-bringup-image" defined in angstrom layer
and running into:

ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/log -> /var/volatile/log
ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/run -> /run
ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/tmp -> /var/volatile/tmp

this would *appear* to be because oe-core defines, in
base-files_3.0.14.bb:

volatiles = "log tmp"
... snip ...
        for d in ${volatiles}; do
                ln -sf volatile/$d ${D}${localstatedir}/$d
        done
        ln -snf ../run ${D}${localstatedir}/run
        ln -snf ../run/lock ${D}${localstatedir}/lock

which seems to want to create /var/{run,log,tmp} as symlinks, but over
in meta-angstrom, we have base-files_3.0.14.bbappend which seems to
want to create them as actual directories instead:

dirs755 += "${localstatedir}/cache \
            ${localstatedir}/run \
            ${localstatedir}/log \
            ${localstatedir}/lock \
            ${localstatedir}/lock/subsys \
            ${localstatedir}/tmp \
            ${localstatedir}/volatile/tmp \
            /run \
           "

or am i misreading something?

rday

-- 

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

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 2467 bytes --]

On Thu, Aug 29, 2013 at 06:29:17AM -0400, Robert P. J. Day wrote:
> 
>   trying to build a "hardware-bringup-image" defined in angstrom layer
> and running into:
> 
> ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/log -> /var/volatile/log
> ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/run -> /run
> ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/tmp -> /var/volatile/tmp
> 
> this would *appear* to be because oe-core defines, in
> base-files_3.0.14.bb:
> 
> volatiles = "log tmp"
> ... snip ...
>         for d in ${volatiles}; do
>                 ln -sf volatile/$d ${D}${localstatedir}/$d
>         done
>         ln -snf ../run ${D}${localstatedir}/run
>         ln -snf ../run/lock ${D}${localstatedir}/lock
> 
> which seems to want to create /var/{run,log,tmp} as symlinks, but over
> in meta-angstrom, we have base-files_3.0.14.bbappend which seems to
> want to create them as actual directories instead:

Expectations from base-files .bbappend and FILESYSTEM_PERMS_TABLES need
to be consistent, is it possible that you're using .bbappend from
angstrom and not FILESYSTEM_PERMS_TABLES set by angstrom config?

conf/distro/include/angstrom-core-tweaks.inc:FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"

> 
> dirs755 += "${localstatedir}/cache \
>             ${localstatedir}/run \
>             ${localstatedir}/log \
>             ${localstatedir}/lock \
>             ${localstatedir}/lock/subsys \
>             ${localstatedir}/tmp \
>             ${localstatedir}/volatile/tmp \
>             /run \
>            "
> 
> or am i misreading something?
> 
> rday
> 
> -- 
> 
> ========================================================================
> Robert P. J. Day                                 Ottawa, Ontario, CANADA
>                         http://crashcourse.ca
> 
> Twitter:                                       http://twitter.com/rpjday
> LinkedIn:                               http://ca.linkedin.com/in/rpjday
> ========================================================================
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Robert P. J. Day]

On Thu, 29 Aug 2013, Martin Jansa wrote:

> On Thu, Aug 29, 2013 at 06:29:17AM -0400, Robert P. J. Day wrote:
> >
> >   trying to build a "hardware-bringup-image" defined in angstrom layer
> > and running into:
> >
> > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/log -> /var/volatile/log
> > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/run -> /run
> > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/tmp -> /var/volatile/tmp
> >
> > this would *appear* to be because oe-core defines, in
> > base-files_3.0.14.bb:
> >
> > volatiles = "log tmp"
> > ... snip ...
> >         for d in ${volatiles}; do
> >                 ln -sf volatile/$d ${D}${localstatedir}/$d
> >         done
> >         ln -snf ../run ${D}${localstatedir}/run
> >         ln -snf ../run/lock ${D}${localstatedir}/lock
> >
> > which seems to want to create /var/{run,log,tmp} as symlinks, but over
> > in meta-angstrom, we have base-files_3.0.14.bbappend which seems to
> > want to create them as actual directories instead:
>
> Expectations from base-files .bbappend and FILESYSTEM_PERMS_TABLES need
> to be consistent, is it possible that you're using .bbappend from
> angstrom and not FILESYSTEM_PERMS_TABLES set by angstrom config?
>
> conf/distro/include/angstrom-core-tweaks.inc:FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"

  ah, i suspect that's what's happening. so what's the correct way to
set up this OE build? i'm trying to build the angstrom
"hardware-bringup-image" using a current oe-core, and my bblayers.conf
file looks like:

BBLAYERS ?= " \
  /home/rpjday/oe/dist/layers/oe-core/meta \
  /home/rpjday/oe/dist/layers/meta-openembedded/meta-oe \
  /home/rpjday/oe/dist/layers/meta-angstrom \
  /home/rpjday/oe/dist/layers/meta-ti \
  "

what should i have done differently?

rday

-- 

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

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 2928 bytes --]

On Thu, Aug 29, 2013 at 07:00:45AM -0400, Robert P. J. Day wrote:
> On Thu, 29 Aug 2013, Martin Jansa wrote:
> 
> > On Thu, Aug 29, 2013 at 06:29:17AM -0400, Robert P. J. Day wrote:
> > >
> > >   trying to build a "hardware-bringup-image" defined in angstrom layer
> > > and running into:
> > >
> > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/log -> /var/volatile/log
> > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/run -> /run
> > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/tmp -> /var/volatile/tmp
> > >
> > > this would *appear* to be because oe-core defines, in
> > > base-files_3.0.14.bb:
> > >
> > > volatiles = "log tmp"
> > > ... snip ...
> > >         for d in ${volatiles}; do
> > >                 ln -sf volatile/$d ${D}${localstatedir}/$d
> > >         done
> > >         ln -snf ../run ${D}${localstatedir}/run
> > >         ln -snf ../run/lock ${D}${localstatedir}/lock
> > >
> > > which seems to want to create /var/{run,log,tmp} as symlinks, but over
> > > in meta-angstrom, we have base-files_3.0.14.bbappend which seems to
> > > want to create them as actual directories instead:
> >
> > Expectations from base-files .bbappend and FILESYSTEM_PERMS_TABLES need
> > to be consistent, is it possible that you're using .bbappend from
> > angstrom and not FILESYSTEM_PERMS_TABLES set by angstrom config?
> >
> > conf/distro/include/angstrom-core-tweaks.inc:FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> 
>   ah, i suspect that's what's happening. so what's the correct way to
> set up this OE build? i'm trying to build the angstrom
> "hardware-bringup-image" using a current oe-core, and my bblayers.conf
> file looks like:
> 
> BBLAYERS ?= " \
>   /home/rpjday/oe/dist/layers/oe-core/meta \
>   /home/rpjday/oe/dist/layers/meta-openembedded/meta-oe \
>   /home/rpjday/oe/dist/layers/meta-angstrom \
>   /home/rpjday/oe/dist/layers/meta-ti \
>   "

What's your DISTRO value? distroless oe-core?

FWIW I had similar problem with meta-shr (distro layer) in distroless
world builds, in the end I've splitted meta-shr and meta-shr-distro.

https://github.com/shr-distribution/meta-smartphone/commit/a3e2b26f8bf13a47c63419623959e6dbe5bc1c15

> 
> what should i have done differently?
> 
> rday
> 
> -- 
> 
> ========================================================================
> Robert P. J. Day                                 Ottawa, Ontario, CANADA
>                         http://crashcourse.ca
> 
> Twitter:                                       http://twitter.com/rpjday
> LinkedIn:                               http://ca.linkedin.com/in/rpjday
> ========================================================================

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Robert P. J. Day]

On Thu, 29 Aug 2013, Martin Jansa wrote:

> On Thu, Aug 29, 2013 at 07:00:45AM -0400, Robert P. J. Day wrote:
> > On Thu, 29 Aug 2013, Martin Jansa wrote:
> >
> > > On Thu, Aug 29, 2013 at 06:29:17AM -0400, Robert P. J. Day wrote:
> > > >
> > > >   trying to build a "hardware-bringup-image" defined in angstrom layer
> > > > and running into:
> > > >
> > > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/log -> /var/volatile/log
> > > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/run -> /run
> > > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/tmp -> /var/volatile/tmp
> > > >
> > > > this would *appear* to be because oe-core defines, in
> > > > base-files_3.0.14.bb:
> > > >
> > > > volatiles = "log tmp"
> > > > ... snip ...
> > > >         for d in ${volatiles}; do
> > > >                 ln -sf volatile/$d ${D}${localstatedir}/$d
> > > >         done
> > > >         ln -snf ../run ${D}${localstatedir}/run
> > > >         ln -snf ../run/lock ${D}${localstatedir}/lock
> > > >
> > > > which seems to want to create /var/{run,log,tmp} as symlinks, but over
> > > > in meta-angstrom, we have base-files_3.0.14.bbappend which seems to
> > > > want to create them as actual directories instead:
> > >
> > > Expectations from base-files .bbappend and FILESYSTEM_PERMS_TABLES need
> > > to be consistent, is it possible that you're using .bbappend from
> > > angstrom and not FILESYSTEM_PERMS_TABLES set by angstrom config?
> > >
> > > conf/distro/include/angstrom-core-tweaks.inc:FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> >
> >   ah, i suspect that's what's happening. so what's the correct way to
> > set up this OE build? i'm trying to build the angstrom
> > "hardware-bringup-image" using a current oe-core, and my bblayers.conf
> > file looks like:
> >
> > BBLAYERS ?= " \
> >   /home/rpjday/oe/dist/layers/oe-core/meta \
> >   /home/rpjday/oe/dist/layers/meta-openembedded/meta-oe \
> >   /home/rpjday/oe/dist/layers/meta-angstrom \
> >   /home/rpjday/oe/dist/layers/meta-ti \
> >   "
>
> What's your DISTRO value? distroless oe-core?

  distroless. i realize i *could* set a value for DISTRO, but it would
seem that i shouldn't have to in order to avoid a build error.

  in any event, you clearly see exactly what's causing the error --
that the base-files bbappend file from meta-angstrom is clashing with
the base-files recipe from oe-core. if you want me to try anything
different, i have lots of time today to test. if you want me to select
a DISTRO, i can certainly do that.

rday

-- 

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

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Phil Blundell]

On Thu, 2013-08-29 at 07:37 -0400, Robert P. J. Day wrote:
>   distroless. i realize i *could* set a value for DISTRO, but it would
> seem that i shouldn't have to in order to avoid a build error.

Well, either you're using angstrom or you aren't.  If you do want to use
it then you ought to set DISTRO appropriately, and if you don't want to
use it then you probably oughtn't to have the angstrom layer included at
all.  

Trying to use the angstrom .bbappends and files with a different (or no)
distro configuration will give you a set of metadata that's neither one
thing nor the other and, even if you didn't get a build error, there's
no guarantee that the resulting image would actually work.

p.

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 3352 bytes --]

On Thu, Aug 29, 2013 at 07:37:17AM -0400, Robert P. J. Day wrote:
> On Thu, 29 Aug 2013, Martin Jansa wrote:
> 
> > On Thu, Aug 29, 2013 at 07:00:45AM -0400, Robert P. J. Day wrote:
> > > On Thu, 29 Aug 2013, Martin Jansa wrote:
> > >
> > > > On Thu, Aug 29, 2013 at 06:29:17AM -0400, Robert P. J. Day wrote:
> > > > >
> > > > >   trying to build a "hardware-bringup-image" defined in angstrom layer
> > > > > and running into:
> > > > >
> > > > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/log -> /var/volatile/log
> > > > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/run -> /run
> > > > > ERROR: QA Issue: Fixup Perms: Unable to correct directory link, target already exists: /var/tmp -> /var/volatile/tmp
> > > > >
> > > > > this would *appear* to be because oe-core defines, in
> > > > > base-files_3.0.14.bb:
> > > > >
> > > > > volatiles = "log tmp"
> > > > > ... snip ...
> > > > >         for d in ${volatiles}; do
> > > > >                 ln -sf volatile/$d ${D}${localstatedir}/$d
> > > > >         done
> > > > >         ln -snf ../run ${D}${localstatedir}/run
> > > > >         ln -snf ../run/lock ${D}${localstatedir}/lock
> > > > >
> > > > > which seems to want to create /var/{run,log,tmp} as symlinks, but over
> > > > > in meta-angstrom, we have base-files_3.0.14.bbappend which seems to
> > > > > want to create them as actual directories instead:
> > > >
> > > > Expectations from base-files .bbappend and FILESYSTEM_PERMS_TABLES need
> > > > to be consistent, is it possible that you're using .bbappend from
> > > > angstrom and not FILESYSTEM_PERMS_TABLES set by angstrom config?
> > > >
> > > > conf/distro/include/angstrom-core-tweaks.inc:FILESYSTEM_PERMS_TABLES = "fs-perms-angstrom.txt"
> > >
> > >   ah, i suspect that's what's happening. so what's the correct way to
> > > set up this OE build? i'm trying to build the angstrom
> > > "hardware-bringup-image" using a current oe-core, and my bblayers.conf
> > > file looks like:
> > >
> > > BBLAYERS ?= " \
> > >   /home/rpjday/oe/dist/layers/oe-core/meta \
> > >   /home/rpjday/oe/dist/layers/meta-openembedded/meta-oe \
> > >   /home/rpjday/oe/dist/layers/meta-angstrom \
> > >   /home/rpjday/oe/dist/layers/meta-ti \
> > >   "
> >
> > What's your DISTRO value? distroless oe-core?
> 
>   distroless. i realize i *could* set a value for DISTRO, but it would
> seem that i shouldn't have to in order to avoid a build error.
> 
>   in any event, you clearly see exactly what's causing the error --
> that the base-files bbappend file from meta-angstrom is clashing with
> the base-files recipe from oe-core. if you want me to try anything
> different, i have lots of time today to test. if you want me to select
> a DISTRO, i can certainly do that.

You can add that line from angstrom-core-tweaks into your local.conf,
but it depends on distro maintainer if he wants to support different
DISTROs in distro layer.

In other words, everything in distro layer can usually assume correct
distro configuration, when you're using some recipes from distro layer
with your own distro or distroless config then you're on your own
unsupported combination.

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Robert P. J. Day]

On Thu, 29 Aug 2013, Phil Blundell wrote:

> On Thu, 2013-08-29 at 07:37 -0400, Robert P. J. Day wrote:
> >   distroless. i realize i *could* set a value for DISTRO, but it would
> > seem that i shouldn't have to in order to avoid a build error.
>
> Well, either you're using angstrom or you aren't.  If you do want to
> use it then you ought to set DISTRO appropriately, and if you don't
> want to use it then you probably oughtn't to have the angstrom layer
> included at all.
>
> Trying to use the angstrom .bbappends and files with a different (or
> no) distro configuration will give you a set of metadata that's
> neither one thing nor the other and, even if you didn't get a build
> error, there's no guarantee that the resulting image would actually
> work.

  ah, gotcha. so if i want to live dangerously, i can select
"angstrom-next"?

rday

-- 

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

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Richard Purdie]

On Thu, 2013-08-29 at 12:49 +0100, Phil Blundell wrote:
> On Thu, 2013-08-29 at 07:37 -0400, Robert P. J. Day wrote:
> >   distroless. i realize i *could* set a value for DISTRO, but it would
> > seem that i shouldn't have to in order to avoid a build error.
> 
> Well, either you're using angstrom or you aren't.  If you do want to use
> it then you ought to set DISTRO appropriately, and if you don't want to
> use it then you probably oughtn't to have the angstrom layer included at
> all.  
> 
> Trying to use the angstrom .bbappends and files with a different (or no)
> distro configuration will give you a set of metadata that's neither one
> thing nor the other and, even if you didn't get a build error, there's
> no guarantee that the resulting image would actually work.

I'm not sure I agree. Ideally layers should be includable without
turning on behaviour unless it is enabled. This is why we have things
like an array of OVERRIDES to chose from.

So I'd actually suggest this is a bug in the angstrom layer. There are
also other layers which probably have issues, maybe including meta-yocto
but we should fix them.

I'll go further and suggest that Yocto Project Compatible status might
depend on this in future (the current questions suggest it should be the
case already).

Cheers,

Richard

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Chris Larson]

[-- Attachment #1: Type: text/plain, Size: 1625 bytes --]

On Thu, Aug 29, 2013 at 4:57 AM, Richard Purdie <
richard.purdie@linuxfoundation.org> wrote:

> On Thu, 2013-08-29 at 12:49 +0100, Phil Blundell wrote:
> > On Thu, 2013-08-29 at 07:37 -0400, Robert P. J. Day wrote:
> > >   distroless. i realize i *could* set a value for DISTRO, but it would
> > > seem that i shouldn't have to in order to avoid a build error.
> >
> > Well, either you're using angstrom or you aren't.  If you do want to use
> > it then you ought to set DISTRO appropriately, and if you don't want to
> > use it then you probably oughtn't to have the angstrom layer included at
> > all.
> >
> > Trying to use the angstrom .bbappends and files with a different (or no)
> > distro configuration will give you a set of metadata that's neither one
> > thing nor the other and, even if you didn't get a build error, there's
> > no guarantee that the resulting image would actually work.
>
> I'm not sure I agree. Ideally layers should be includable without
> turning on behaviour unless it is enabled. This is why we have things
> like an array of OVERRIDES to chose from.
>
> So I'd actually suggest this is a bug in the angstrom layer. There are
> also other layers which probably have issues, maybe including meta-yocto
> but we should fix them.
>
> I'll go further and suggest that Yocto Project Compatible status might
> depend on this in future (the current questions suggest it should be the
> case already).
>

I agree wholeheartedly with this. Distro and BSP layers really need to be
leveraging OVERRIDES for anything distro or machine specific.
-- 
Christopher Larson

[-- Attachment #2: Type: text/html, Size: 2098 bytes --]

Re: appears to be inconsistency re: base-files between oe-core and meta-angstrom

[Robert P. J. Day]

On Thu, 29 Aug 2013, Richard Purdie wrote:

> On Thu, 2013-08-29 at 12:49 +0100, Phil Blundell wrote:
> > On Thu, 2013-08-29 at 07:37 -0400, Robert P. J. Day wrote:
> > >   distroless. i realize i *could* set a value for DISTRO, but it would
> > > seem that i shouldn't have to in order to avoid a build error.
> >
> > Well, either you're using angstrom or you aren't.  If you do want
> > to use it then you ought to set DISTRO appropriately, and if you
> > don't want to use it then you probably oughtn't to have the
> > angstrom layer included at all.
> >
> > Trying to use the angstrom .bbappends and files with a different
> > (or no) distro configuration will give you a set of metadata
> > that's neither one thing nor the other and, even if you didn't get
> > a build error, there's no guarantee that the resulting image would
> > actually work.
>
> I'm not sure I agree. Ideally layers should be includable without
> turning on behaviour unless it is enabled. This is why we have
> things like an array of OVERRIDES to choose from.

  after i thought about it for a while, this is what i started to
believe as well -- i should be able to include various layers and let
them work it out amongst themselves the correct behaviour and bbappend
processing and so on.

  the reason i selected the layers i did in my example was that i
wanted to build cutting-edge components for my beaglebone black and,
as i read it, this required me to use oe-core, meta-angstom and
meta-ti (among other things). and as you saw, oe-core and
meta-angstrom didn't play well together WRT base-files.

  anyway, i'll let folks higher up the food chain ponder on this.

rday

-- 

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

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================