Patchwork sstate skips STAGING_DIR_KERNEL?

login
register
mail settings
Submitter Richard Purdie
Date June 23, 2011, 3:29 p.m.
Message ID <1308842954.20015.126.camel@rex>
Download mbox | patch
Permalink /patch/6425/
State New, archived
Headers show

Comments

Richard Purdie - June 23, 2011, 3:29 p.m.
On Wed, 2011-06-22 at 17:20 -0700, Tom Rini wrote:
> As part of looking into the problem I had that I just posted a patch for
> I was going "why isn't -c clean (or cleanall or cleansstate) knocking
> out STAGING_DIR_KERNEL contents?  Answer, sstate is totally skipping
> that out.  Is there some variable we need to be setting somewhere to add
> this in?  Or are we going to have to take in the various bits of OE.dev
> kernel.bbclass instead here?

I think there is a bug in kernel.bbclass. This should fix it:


I'll merge something like this is nobody objects.

Cheers,

Richard
Tom Rini - June 23, 2011, 4:15 p.m.
On 06/23/2011 08:29 AM, Richard Purdie wrote:
> On Wed, 2011-06-22 at 17:20 -0700, Tom Rini wrote:
>> As part of looking into the problem I had that I just posted a patch for
>> I was going "why isn't -c clean (or cleanall or cleansstate) knocking
>> out STAGING_DIR_KERNEL contents?  Answer, sstate is totally skipping
>> that out.  Is there some variable we need to be setting somewhere to add
>> this in?  Or are we going to have to take in the various bits of OE.dev
>> kernel.bbclass instead here?
> 
> I think there is a bug in kernel.bbclass. This should fix it:
> 
> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
> index aaf341b..9f09107 100644
> --- a/meta/classes/kernel.bbclass
> +++ b/meta/classes/kernel.bbclass
> @@ -119,7 +119,7 @@ kernel_do_install() {
>         # Support for external module building - create a minimal copy of the
>         # kernel source tree.
>         #
> -       kerneldir=${STAGING_KERNEL_DIR}
> +       kerneldir=${D}/kernel
>         install -d $kerneldir
>  
>         #
> 
> I'll merge something like this is nobody objects.

That makes sense, testing this now.
Tom Rini - June 23, 2011, 4:51 p.m.
On 06/23/2011 09:15 AM, Tom Rini wrote:
> On 06/23/2011 08:29 AM, Richard Purdie wrote:
>> On Wed, 2011-06-22 at 17:20 -0700, Tom Rini wrote:
>>> As part of looking into the problem I had that I just posted a patch for
>>> I was going "why isn't -c clean (or cleanall or cleansstate) knocking
>>> out STAGING_DIR_KERNEL contents?  Answer, sstate is totally skipping
>>> that out.  Is there some variable we need to be setting somewhere to add
>>> this in?  Or are we going to have to take in the various bits of OE.dev
>>> kernel.bbclass instead here?
>>
>> I think there is a bug in kernel.bbclass. This should fix it:
>>
>> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
>> index aaf341b..9f09107 100644
>> --- a/meta/classes/kernel.bbclass
>> +++ b/meta/classes/kernel.bbclass
>> @@ -119,7 +119,7 @@ kernel_do_install() {
>>         # Support for external module building - create a minimal copy of the
>>         # kernel source tree.
>>         #
>> -       kerneldir=${STAGING_KERNEL_DIR}
>> +       kerneldir=${D}/kernel
>>         install -d $kerneldir
>>  
>>         #
>>
>> I'll merge something like this is nobody objects.
> 
> That makes sense, testing this now.

Yeah, that does it.

Tested-by: Tom Rini <tom_rini@mentor.com>

Koen, once this is in oe-core need a request to bring it over too or
just want to come grab it?
Koen Kooi - June 23, 2011, 5:12 p.m.
Op 23 jun 2011, om 18:51 heeft Tom Rini het volgende geschreven:

> On 06/23/2011 09:15 AM, Tom Rini wrote:
>> On 06/23/2011 08:29 AM, Richard Purdie wrote:
>>> On Wed, 2011-06-22 at 17:20 -0700, Tom Rini wrote:
>>>> As part of looking into the problem I had that I just posted a  
>>>> patch for
>>>> I was going "why isn't -c clean (or cleanall or cleansstate)  
>>>> knocking
>>>> out STAGING_DIR_KERNEL contents?  Answer, sstate is totally  
>>>> skipping
>>>> that out.  Is there some variable we need to be setting somewhere  
>>>> to add
>>>> this in?  Or are we going to have to take in the various bits of  
>>>> OE.dev
>>>> kernel.bbclass instead here?
>>>
>>> I think there is a bug in kernel.bbclass. This should fix it:
>>>
>>> diff --git a/meta/classes/kernel.bbclass b/meta/classes/ 
>>> kernel.bbclass
>>> index aaf341b..9f09107 100644
>>> --- a/meta/classes/kernel.bbclass
>>> +++ b/meta/classes/kernel.bbclass
>>> @@ -119,7 +119,7 @@ kernel_do_install() {
>>>        # Support for external module building - create a minimal  
>>> copy of the
>>>        # kernel source tree.
>>>        #
>>> -       kerneldir=${STAGING_KERNEL_DIR}
>>> +       kerneldir=${D}/kernel
>>>        install -d $kerneldir
>>>
>>>        #
>>>
>>> I'll merge something like this is nobody objects.
>>
>> That makes sense, testing this now.
>
> Yeah, that does it.
>
> Tested-by: Tom Rini <tom_rini@mentor.com>
>
> Koen, once this is in oe-core need a request to bring it over too or
> just want to come grab it?

I have it applied locally, just waiting for it to land in oe-core so I  
can transplant the commit headers :)
Richard Purdie - June 23, 2011, 7:47 p.m.
On Thu, 2011-06-23 at 09:51 -0700, Tom Rini wrote:
> On 06/23/2011 09:15 AM, Tom Rini wrote:
> > On 06/23/2011 08:29 AM, Richard Purdie wrote:
> >> On Wed, 2011-06-22 at 17:20 -0700, Tom Rini wrote:
> >>> As part of looking into the problem I had that I just posted a patch for
> >>> I was going "why isn't -c clean (or cleanall or cleansstate) knocking
> >>> out STAGING_DIR_KERNEL contents?  Answer, sstate is totally skipping
> >>> that out.  Is there some variable we need to be setting somewhere to add
> >>> this in?  Or are we going to have to take in the various bits of OE.dev
> >>> kernel.bbclass instead here?
> >>
> >> I think there is a bug in kernel.bbclass. This should fix it:
> >>
> >> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
> >> index aaf341b..9f09107 100644
> >> --- a/meta/classes/kernel.bbclass
> >> +++ b/meta/classes/kernel.bbclass
> >> @@ -119,7 +119,7 @@ kernel_do_install() {
> >>         # Support for external module building - create a minimal copy of the
> >>         # kernel source tree.
> >>         #
> >> -       kerneldir=${STAGING_KERNEL_DIR}
> >> +       kerneldir=${D}/kernel
> >>         install -d $kerneldir
> >>  
> >>         #
> >>
> >> I'll merge something like this is nobody objects.
> > 
> > That makes sense, testing this now.
> 
> Yeah, that does it.
> 
> Tested-by: Tom Rini <tom_rini@mentor.com>
> 
> Koen, once this is in oe-core need a request to bring it over too or
> just want to come grab it?

I've merged it into OE-Core.

Cheers,

Richard

Patch

diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
index aaf341b..9f09107 100644
--- a/meta/classes/kernel.bbclass
+++ b/meta/classes/kernel.bbclass
@@ -119,7 +119,7 @@  kernel_do_install() {
        # Support for external module building - create a minimal copy of the
        # kernel source tree.
        #
-       kerneldir=${STAGING_KERNEL_DIR}
+       kerneldir=${D}/kernel
        install -d $kerneldir
 
        #