[RFC] fix for MACHINE_KERNEL_PR stuff

[RFC] fix for MACHINE_KERNEL_PR stuff

[Marcin Juszkiewicz]

[-- Attachment #1: Type: text/plain, Size: 2335 bytes --]


This patch unbreaks current behaviour which was introduced by 
MACHINE_KERNEL_PR variable.

As most of target machines do not use it they have PR with broken value
(set to "r0" instead of value in recipe). I took other way which makes
both types of users happy -- those with MACHINE_KERNEL_PR in use and 
those without it.

By default we set M_K_PR to empty string instead of "r0" - this allows
to check is it set at all or not. If it is set then we set PR to this value.
Otherwise we ignore existance of that variable and use PR from recipe.

Please test and review it. If this will work I will request it for stable/2009.

Signed-off-by: Marcin Juszkiewicz <marcin@juszkiewicz.com.pl>

diff --git a/classes/kernel.bbclass b/classes/kernel.bbclass
index 1323228..757b704 100644
--- a/classes/kernel.bbclass
+++ b/classes/kernel.bbclass
@@ -20,6 +20,12 @@ python __anonymous () {
     image = bb.data.getVar('INITRAMFS_IMAGE', d, True)
     if image != '' and image is not None:
         bb.data.setVar('INITRAMFS_TASK', '${INITRAMFS_IMAGE}:do_rootfs', d)
+
+    pr = bb.data.getVar('PR', d, True)
+    machine_kernel_pr = bb.data.getVar('MACHINE_KERNEL_PR', d, True)
+
+    if machine_kernel_pr:
+       bb.data.setVar('PR', machine_kernel_pr, d)
 }

 INITRAMFS_IMAGE ?= ""
@@ -64,10 +70,6 @@ export CMDLINE_CONSOLE = "console=${@bb.data.getVar("KERNEL_CONSOLE",d,1) or "tt
 KERNEL_VERSION = "${@get_kernelversion('${S}')}"
 KERNEL_MAJOR_VERSION = "${@get_kernelmajorversion('${KERNEL_VERSION}')}"

-# A machine.conf or local.conf can increase MACHINE_KERNEL_PR to force
-# rebuilds for kernel and external modules
-PR = "${MACHINE_KERNEL_PR}"
-
 KERNEL_LOCALVERSION ?= ""

 # kernels are generally machine specific
diff --git a/conf/bitbake.conf b/conf/bitbake.conf
index e711a1f..7bd548d 100644
--- a/conf/bitbake.conf
+++ b/conf/bitbake.conf
@@ -164,7 +164,7 @@ P = "${PN}-${PV}"

 # Define a PR for kernels that machines can override so things like
 # modules get rebuilt
-MACHINE_KERNEL_PR ?= "r0"
+MACHINE_KERNEL_PR = ""

 # Base package name
 # Automatically derives "foo" from "foo-native", "foo-cross" or "foo-initial"


Regards, 
-- 
JID:      hrw@jabber.org
Website:  http://marcin.juszkiewicz.com.pl/
LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 204 bytes --]

Re: [RFC] fix for MACHINE_KERNEL_PR stuff

[Koen Kooi]

On 03-06-09 13:25, Marcin Juszkiewicz wrote:
>
> This patch unbreaks current behaviour which was introduced by
> MACHINE_KERNEL_PR variable.
>
> As most of target machines do not use it they have PR with broken value
> (set to "r0" instead of value in recipe). I took other way which makes
> both types of users happy -- those with MACHINE_KERNEL_PR in use and
> those without it.
>
> By default we set M_K_PR to empty string instead of "r0" - this allows
> to check is it set at all or not. If it is set then we set PR to this value.
> Otherwise we ignore existance of that variable and use PR from recipe.
>
> Please test and review it. If this will work I will request it for stable/2009.
>
> Signed-off-by: Marcin Juszkiewicz<marcin@juszkiewicz.com.pl>
>
> diff --git a/classes/kernel.bbclass b/classes/kernel.bbclass
> index 1323228..757b704 100644
> --- a/classes/kernel.bbclass
> +++ b/classes/kernel.bbclass
> @@ -20,6 +20,12 @@ python __anonymous () {
>       image = bb.data.getVar('INITRAMFS_IMAGE', d, True)
>       if image != '' and image is not None:
>           bb.data.setVar('INITRAMFS_TASK', '${INITRAMFS_IMAGE}:do_rootfs', d)
> +
> +    pr = bb.data.getVar('PR', d, True)

You're not using 'pr' anywhere in the code you add. Looks fine 
otherwise, although I haven't tested it.

regards,

Koen

Re: [RFC] fix for MACHINE_KERNEL_PR stuff

[Marcin Juszkiewicz]

Dnia środa, 3 czerwca 2009 o 13:45:24 Koen Kooi napisał(a):
> On 03-06-09 13:25, Marcin Juszkiewicz wrote:

> > +    pr = bb.data.getVar('PR', d, True)
>
> You're not using 'pr' anywhere in the code you add. Looks fine
> otherwise, although I haven't tested it.

Thats leftover from older version of patch.

Regards, 
-- 
JID:      hrw@jabber.org
Website:  http://marcin.juszkiewicz.com.pl/
LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz

Re: [RFC] fix for MACHINE_KERNEL_PR stuff

[Koen Kooi]

On 03-06-09 13:59, Marcin Juszkiewicz wrote:
> Dnia środa, 3 czerwca 2009 o 13:45:24 Koen Kooi napisał(a):
>> On 03-06-09 13:25, Marcin Juszkiewicz wrote:
>
>>> +    pr = bb.data.getVar('PR', d, True)
>>
>> You're not using 'pr' anywhere in the code you add. Looks fine
>> otherwise, although I haven't tested it.
>
> Thats leftover from older version of patch.

So where's the new version of this patch without cruft?

regards,

koen

Re: [RFC] fix for MACHINE_KERNEL_PR stuff

[Phil Blundell]

On Wed, 2009-06-03 at 13:25 +0200, Marcin Juszkiewicz wrote:
> This patch unbreaks current behaviour which was introduced by 
> MACHINE_KERNEL_PR variable.
> 
> As most of target machines do not use it they have PR with broken value
> (set to "r0" instead of value in recipe). I took other way which makes
> both types of users happy -- those with MACHINE_KERNEL_PR in use and 
> those without it.
> 
> By default we set M_K_PR to empty string instead of "r0" - this allows
> to check is it set at all or not. If it is set then we set PR to this value.
> Otherwise we ignore existance of that variable and use PR from recipe.

Thanks for the patch.  This does look like a reasonable compromise: as
you say, it should prevent MACHINE_KERNEL_PR causing trouble for those
people who don't want it.  So, +1 from me.

p.

Re: [RFC] fix for MACHINE_KERNEL_PR stuff

[Marcin Juszkiewicz]

Dnia środa, 3 czerwca 2009 o 13:25:43 Marcin Juszkiewicz napisał(a):
> This patch unbreaks current behaviour which was introduced by
> MACHINE_KERNEL_PR variable.
>
> As most of target machines do not use it they have PR with broken
> value (set to "r0" instead of value in recipe). I took other way
> which makes both types of users happy -- those with MACHINE_KERNEL_PR
> in use and those without it.
>
> By default we set M_K_PR to empty string instead of "r0" - this
> allows to check is it set at all or not. If it is set then we set PR
> to this value. Otherwise we ignore existance of that variable and use
> PR from recipe.
>
> Please test and review it. If this will work I will request it for
> stable/2009.

Someone tested that? I would like to get it in .dev and stable/2009 but 
not without testing.

Regards, 
-- 
JID:      hrw@jabber.org
Website:  http://marcin.juszkiewicz.com.pl/
LinkedIn: http://www.linkedin.com/in/marcinjuszkiewicz

Re: [RFC] fix for MACHINE_KERNEL_PR stuff

[Rolf Leggewie]

Marcin Juszkiewicz wrote:
> Someone tested that?

Looks like its doing the right thing.

Tested with minimal distro and machine wrt54.  bb file specifies r2. 
without patch, it's r0 in bitbake.  With the patch, it's r2 as it should be.

Thank you for fixing this, especially since it wasn't you who broke things.