[PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Benard]

Signed-off-by: Eric Benard <eric@eukrea.com>
---
 classes/module-base.bbclass |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/classes/module-base.bbclass b/classes/module-base.bbclass
index bc53e1b..9aaaa4e 100644
--- a/classes/module-base.bbclass
+++ b/classes/module-base.bbclass
@@ -7,7 +7,12 @@ export CROSS_COMPILE = "${TARGET_PREFIX}"
 
 # A machine.conf or local.conf can increase MACHINE_KERNEL_PR to force
 # rebuilds for kernel and external modules
-PR = "${MACHINE_KERNEL_PR}"
+python __anonymous () {
+    machine_kernel_pr = bb.data.getVar('MACHINE_KERNEL_PR', d, True)
+
+    if machine_kernel_pr:
+       bb.data.setVar('PR', machine_kernel_pr, d)
+}
 
 export KERNEL_VERSION = "${@base_read_file('${STAGING_KERNEL_DIR}/kernel-abiversion')}"
 export KERNEL_SOURCE = "${@base_read_file('${STAGING_KERNEL_DIR}/kernel-source')}"
-- 
1.6.3.3

Re: [PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Bénard]

Le 03/05/2010 23:21, Eric Benard a écrit :
> Signed-off-by: Eric Benard<eric@eukrea.com>
> ---
>   classes/module-base.bbclass |    7 ++++++-
>   1 files changed, 6 insertions(+), 1 deletions(-)
>
> diff --git a/classes/module-base.bbclass b/classes/module-base.bbclass
> index bc53e1b..9aaaa4e 100644
> --- a/classes/module-base.bbclass
> +++ b/classes/module-base.bbclass
> @@ -7,7 +7,12 @@ export CROSS_COMPILE = "${TARGET_PREFIX}"
>
>   # A machine.conf or local.conf can increase MACHINE_KERNEL_PR to force
>   # rebuilds for kernel and external modules
> -PR = "${MACHINE_KERNEL_PR}"
> +python __anonymous () {
> +    machine_kernel_pr = bb.data.getVar('MACHINE_KERNEL_PR', d, True)
> +
> +    if machine_kernel_pr:
> +       bb.data.setVar('PR', machine_kernel_pr, d)
> +}
>
>   export KERNEL_VERSION = "${@base_read_file('${STAGING_KERNEL_DIR}/kernel-abiversion')}"
>   export KERNEL_SOURCE = "${@base_read_file('${STAGING_KERNEL_DIR}/kernel-source')}"

Actually, we loose PR if MACHINE_KERNEL_PR is not set.

Is this patch fine to solve this problem or is there another prefered way ?

Thanks,
Eric

[RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Bénard]

Hi Koen,

I was told you are the person to contact about MACHINE_KERNEL_PR so may 
you please have a look to the patch below - also available here : 
http://patchwork.openembedded.org/patch/2012/

Actually, when compiling modules (out of tree), we loose PR if 
MACHINE_KERNEL_PR is not set.

Is this patch fine to solve this problem or is there another prefered way ?

The same problem was already solved in kernel.bbclass by the following 
comit
http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=35e7b34d3b5b96ed11bc7ed6b2a5d08e6183d8a3

Thanks,
Eric

Signed-off-by: Eric Benard<eric@eukrea.com>
---
   classes/module-base.bbclass |    7 ++++++-
   1 files changed, 6 insertions(+), 1 deletions(-)

diff --git a/classes/module-base.bbclass b/classes/module-base.bbclass
index bc53e1b..9aaaa4e 100644
--- a/classes/module-base.bbclass
+++ b/classes/module-base.bbclass
@@ -7,7 +7,12 @@ export CROSS_COMPILE = "${TARGET_PREFIX}"

   # A machine.conf or local.conf can increase MACHINE_KERNEL_PR to force
   # rebuilds for kernel and external modules
-PR = "${MACHINE_KERNEL_PR}"
+python __anonymous () {
+    machine_kernel_pr = bb.data.getVar('MACHINE_KERNEL_PR', d, True)
+
+    if machine_kernel_pr:
+       bb.data.setVar('PR', machine_kernel_pr, d)
+}

   export KERNEL_VERSION = 
"${@base_read_file('${STAGING_KERNEL_DIR}/kernel-abiversion')}"
   export KERNEL_SOURCE = 
"${@base_read_file('${STAGING_KERNEL_DIR}/kernel-source')}"

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Koen Kooi]

Op 14 mei 2010, om 17:10 heeft Eric Bénard het volgende geschreven:

> Hi Koen,
> 
> I was told you are the person to contact about MACHINE_KERNEL_PR so may you please have a look to the patch below - also available here : http://patchwork.openembedded.org/patch/2012/
> 
> Actually, when compiling modules (out of tree), we loose PR if MACHINE_KERNEL_PR is not set.

I don't believe that when MACHINE_KERNEL_PR isn't set people care about PR.

regards,

Koen

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Phil Blundell]

On Fri, 2010-05-14 at 17:10 +0200, Eric Bénard wrote:
> -PR = "${MACHINE_KERNEL_PR}"
> +python __anonymous () {
> +    machine_kernel_pr = bb.data.getVar('MACHINE_KERNEL_PR', d, True)
> +
> +    if machine_kernel_pr:
> +       bb.data.setVar('PR', machine_kernel_pr, d)
> +}

Looks reasonable to me.  It seems a little bit sad to have this code
duplicated in both kernel.bbclass and module-base.bbclass but I guess
that is not a very big deal.

p.

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Bénard]

Le 14/05/2010 17:52, Koen Kooi a écrit :
> Op 14 mei 2010, om 17:10 heeft Eric Bénard het volgende geschreven:
>> I was told you are the person to contact about MACHINE_KERNEL_PR so may you please have a look to the patch below - also available here : http://patchwork.openembedded.org/patch/2012/
>>
>> Actually, when compiling modules (out of tree), we loose PR if MACHINE_KERNEL_PR is not set.
>
> I don't believe that when MACHINE_KERNEL_PR isn't set people care about PR.
>
Does that mean that policy is that MACHINE_KERNEL_PR should be set for 
every machine ?
If yes, maybe this should be added in the doc.

Actually only 7 files (4 machines + 3 include) in conf/machine have this 
set and conf/bitbake set it to "" as a default so what aboutdoing ?= 
"r0" in bitbake.conf in order to have a default value as for PR ?

Eric

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 14-05-10 19:17, Eric Bénard wrote:
> Le 14/05/2010 17:52, Koen Kooi a écrit :
>> Op 14 mei 2010, om 17:10 heeft Eric Bénard het volgende geschreven:
>>> I was told you are the person to contact about MACHINE_KERNEL_PR so
>>> may you please have a look to the patch below - also available here :
>>> http://patchwork.openembedded.org/patch/2012/
>>>
>>> Actually, when compiling modules (out of tree), we loose PR if
>>> MACHINE_KERNEL_PR is not set.
>>
>> I don't believe that when MACHINE_KERNEL_PR isn't set people care
>> about PR.
>>
> Does that mean that policy is that MACHINE_KERNEL_PR should be set for
> every machine ?
> If yes, maybe this should be added in the doc.

IMO yes, but some people were against it. Either their kernel never
changes or they don't care about out-of-tree modules breaking on kernel
changes.

regards,

Koen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFL7YmVMkyGM64RGpERAsf2AJ45yUaAIVL0HCm7jCfxwkmq9C5gnACfcV8w
vGcI7BJka3fCXJJ+RL3Y+Yc=
=RLGB
-----END PGP SIGNATURE-----

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Bénard]

Le 14/05/2010 19:34, Koen Kooi a écrit :
> On 14-05-10 19:17, Eric Bénard wrote:
>> Le 14/05/2010 17:52, Koen Kooi a écrit :
>>> Op 14 mei 2010, om 17:10 heeft Eric Bénard het volgende geschreven:
>>>> I was told you are the person to contact about MACHINE_KERNEL_PR so
>>>> may you please have a look to the patch below - also available here :
>>>> http://patchwork.openembedded.org/patch/2012/
>>>>
>>>> Actually, when compiling modules (out of tree), we loose PR if
>>>> MACHINE_KERNEL_PR is not set.
>>>
>>> I don't believe that when MACHINE_KERNEL_PR isn't set people care
>>> about PR.
>>>
>> Does that mean that policy is that MACHINE_KERNEL_PR should be set for
>> every machine ?
>> If yes, maybe this should be added in the doc.
>
> IMO yes, but some people were against it. Either their kernel never
> changes or they don't care about out-of-tree modules breaking on kernel
> changes.
>
so why not setting r0 as a default in bitbake.conf as a compromise ?

Eric

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Phil Blundell]

On Fri, 2010-05-14 at 19:45 +0200, Eric Bénard wrote:
> so why not setting r0 as a default in bitbake.conf as a compromise ?

A default MACHINE_KERNEL_PR, you mean?  That would cause PR to suddenly
go backwards for a whole bunch of machines, which would definitely not
be a good thing.  See the mailing list archive; there was quite a lot of
discussion of these kinds of issues at the time that the M_K_PR change
was originally landed.

p.

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Bénard]

Le 14/05/2010 21:55, Phil Blundell a écrit :
> On Fri, 2010-05-14 at 19:45 +0200, Eric Bénard wrote:
>> so why not setting r0 as a default in bitbake.conf as a compromise ?
>
> A default MACHINE_KERNEL_PR, you mean?  That would cause PR to suddenly
> go backwards for a whole bunch of machines, which would definitely not
> be a good thing.  See the mailing list archive; there was quite a lot of
> discussion of these kinds of issues at the time that the M_K_PR change
> was originally landed.
>
yes, in this case MACHINE_KERNEL_PR would overide kernel's PR which is 
not what we want.

So actually the fix I propose is needed as while MACHINE_KERNEL_PR is 
not required by policy, module-base.bbclass is broken for machines 
without MACHINE_KERNEL_PR.

Eric

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Phil Blundell]

On Fri, 2010-05-14 at 22:15 +0200, Eric Bénard wrote:
> So actually the fix I propose is needed as while MACHINE_KERNEL_PR is 
> not required by policy, module-base.bbclass is broken for machines 
> without MACHINE_KERNEL_PR.

Yes, agreed.  I think your fix is the right thing.

p.

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Eric Bénard]

Le 14/05/2010 23:02, Phil Blundell a écrit :
> On Fri, 2010-05-14 at 22:15 +0200, Eric Bénard wrote:
>> So actually the fix I propose is needed as while MACHINE_KERNEL_PR is
>> not required by policy, module-base.bbclass is broken for machines
>> without MACHINE_KERNEL_PR.
>
> Yes, agreed.  I think your fix is the right thing.
>
Is this an ack so can I push it ?

Eric

Re: [RFC][PATCH] module-base.bbclass: keep PR if MACHINE_KERNEL_PR is not set

[Phil Blundell]

On Mon, 2010-05-17 at 09:12 +0200, Eric Bénard wrote:
> Le 14/05/2010 23:02, Phil Blundell a écrit :
> > On Fri, 2010-05-14 at 22:15 +0200, Eric Bénard wrote:
> >> So actually the fix I propose is needed as while MACHINE_KERNEL_PR is
> >> not required by policy, module-base.bbclass is broken for machines
> >> without MACHINE_KERNEL_PR.
> >
> > Yes, agreed.  I think your fix is the right thing.
> >
> Is this an ack so can I push it ?

Yes, please go ahead.

p.