Unihash warning building kernels in Yocto 3.1.7

Unihash warning building kernels in Yocto 3.1.7

[Diego Santa Cruz]

[-- Attachment #1: Type: text/plain, Size: 1039 bytes --]

Hello,

I just updated to Yocto 3.1.7 (from 3.1.5) and I got the following warning while building a kernel recipe.

  do_install: Task do_install requested do_package unihash, but it was not available

I tracked this down to OE-Core commit 0c28edf4bf0d2f92bf3a47406041c63acd90bacf ("package.bbclass: hash equivalency and pr service") and it makes sense since do_package runs after do_install and in dunfell we can find the following in kernel.bbclass

  do_install[prefuncs] += "package_get_auto_pr"

Merging OE-Core commit 36fe4db8ae827a93abe9fce6740459d215411965 ("kernel.bbclass: Remove do_install[prefunc] no longer needed") should normally solve this warning by removing the above line, but I do not know if that change is safe in dunfell.

Could that be merged in? Or is there something else that could be done to avoid the warning?

Our build system flags a production build as bad if there is a warning, so this is kind of annoying.

Best,

Diego
--
Diego Santa Cruz, PhD
Technology Architect
spinetix.com


[-- Attachment #2: Type: text/html, Size: 3598 bytes --]

Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7

[Steve Sakoman]

On Fri, Apr 30, 2021 at 7:00 AM Diego Santa Cruz via
lists.openembedded.org
<diego.santacruz=spinetix.com@lists.openembedded.org> wrote:
>
> Hello,
>
>
>
> I just updated to Yocto 3.1.7 (from 3.1.5) and I got the following warning while building a kernel recipe.
>
>
>
>   do_install: Task do_install requested do_package unihash, but it was not available
>
>
>
> I tracked this down to OE-Core commit 0c28edf4bf0d2f92bf3a47406041c63acd90bacf (“package.bbclass: hash equivalency and pr service”) and it makes sense since do_package runs after do_install and in dunfell we can find the following in kernel.bbclass
>
>
>
>   do_install[prefuncs] += "package_get_auto_pr"
>
>
>
> Merging OE-Core commit 36fe4db8ae827a93abe9fce6740459d215411965 (“kernel.bbclass: Remove do_install[prefunc] no longer needed”) should normally solve this warning by removing the above line, but I do not know if that change is safe in dunfell.
>
>
>
> Could that be merged in? Or is there something else that could be done to avoid the warning?

I can do some testing with that patch.  cc-ing Mark to see if he has
any comments since he is the patch author.

Steve


> Our build system flags a production build as bad if there is a warning, so this is kind of annoying.
>
>
>
> Best,
>
>
>
> Diego
>
> --
> Diego Santa Cruz, PhD
> Technology Architect
> spinetix.com
>
>
>
>
> 
>

Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7

[Steve Sakoman]

Trying again with Mark's correct email address!

On Fri, Apr 30, 2021 at 8:43 AM Steve Sakoman <sakoman@gmail.com> wrote:
>
> On Fri, Apr 30, 2021 at 7:00 AM Diego Santa Cruz via
> lists.openembedded.org
> <diego.santacruz=spinetix.com@lists.openembedded.org> wrote:
> >
> > Hello,
> >
> >
> >
> > I just updated to Yocto 3.1.7 (from 3.1.5) and I got the following warning while building a kernel recipe.
> >
> >
> >
> >   do_install: Task do_install requested do_package unihash, but it was not available
> >
> >
> >
> > I tracked this down to OE-Core commit 0c28edf4bf0d2f92bf3a47406041c63acd90bacf (“package.bbclass: hash equivalency and pr service”) and it makes sense since do_package runs after do_install and in dunfell we can find the following in kernel.bbclass
> >
> >
> >
> >   do_install[prefuncs] += "package_get_auto_pr"
> >
> >
> >
> > Merging OE-Core commit 36fe4db8ae827a93abe9fce6740459d215411965 (“kernel.bbclass: Remove do_install[prefunc] no longer needed”) should normally solve this warning by removing the above line, but I do not know if that change is safe in dunfell.
> >
> >
> >
> > Could that be merged in? Or is there something else that could be done to avoid the warning?
>
> I can do some testing with that patch.  cc-ing Mark to see if he has
> any comments since he is the patch author.
>
> Steve
>
>
> > Our build system flags a production build as bad if there is a warning, so this is kind of annoying.
> >
> >
> >
> > Best,
> >
> >
> >
> > Diego
> >
> > --
> > Diego Santa Cruz, PhD
> > Technology Architect
> > spinetix.com
> >
> >
> >
> >
> > 
> >

Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7

[Diego Santa Cruz]

Hi Steve, Mark,

Any news on this one? I have been doing our own builds with the equivalent of that patch for our BSPs with a linux-intel derived kernel recipe plus another internal kernel recipe (both using kernel.bbclass) and I have not found any issues with it.

Best,

Diego

--
Diego Santa Cruz, PhD
Technology Architect
T +41 21 341 15 50
diego.santacruz@spinetix.com | Subscribe to our Newlsetter
spinetix.com




> -----Original Message-----
> From: Steve Sakoman <sakoman@gmail.com>
> Sent: 30 April 2021 20:46
> To: Diego Santa Cruz <Diego.SantaCruz@spinetix.com>
> Cc: openembedded-core@lists.openembedded.org; Mark Hatle
> <mark.hatle@kernel.crashing.org>
> Subject: Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7
> 
> Trying again with Mark's correct email address!
> 
> On Fri, Apr 30, 2021 at 8:43 AM Steve Sakoman <sakoman@gmail.com>
> wrote:
> >
> > On Fri, Apr 30, 2021 at 7:00 AM Diego Santa Cruz via
> > lists.openembedded.org
> > <diego.santacruz=spinetix.com@lists.openembedded.org> wrote:
> > >
> > > Hello,
> > >
> > >
> > >
> > > I just updated to Yocto 3.1.7 (from 3.1.5) and I got the following warning
> while building a kernel recipe.
> > >
> > >
> > >
> > >   do_install: Task do_install requested do_package unihash, but it was not
> available
> > >
> > >
> > >
> > > I tracked this down to OE-Core commit
> 0c28edf4bf0d2f92bf3a47406041c63acd90bacf ("package.bbclass: hash
> equivalency and pr service") and it makes sense since do_package runs after
> do_install and in dunfell we can find the following in kernel.bbclass
> > >
> > >
> > >
> > >   do_install[prefuncs] += "package_get_auto_pr"
> > >
> > >
> > >
> > > Merging OE-Core commit 36fe4db8ae827a93abe9fce6740459d215411965
> ("kernel.bbclass: Remove do_install[prefunc] no longer needed") should
> normally solve this warning by removing the above line, but I do not know if
> that change is safe in dunfell.
> > >
> > >
> > >
> > > Could that be merged in? Or is there something else that could be done
> to avoid the warning?
> >
> > I can do some testing with that patch.  cc-ing Mark to see if he has
> > any comments since he is the patch author.
> >
> > Steve
> >
> >
> > > Our build system flags a production build as bad if there is a warning, so
> this is kind of annoying.
> > >
> > >
> > >
> > > Best,
> > >
> > >
> > >
> > > Diego
> > >
> > > --
> > > Diego Santa Cruz, PhD
> > > Technology Architect
> > > spinetix.com
> > >
> > >
> > >
> > >
> > > 
> > >

Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7

[Steve Sakoman]

On Tue, May 4, 2021 at 10:17 PM Diego Santa Cruz via
lists.openembedded.org
<diego.santacruz=spinetix.com@lists.openembedded.org> wrote:

> Any news on this one? I have been doing our own builds with the equivalent of that patch for our BSPs with a linux-intel derived kernel recipe plus another internal kernel recipe (both using kernel.bbclass) and I have not found any issues with it.

This is in the group of patches I am currently testing.  So far it looks good.

Steve

> > -----Original Message-----
> > From: Steve Sakoman <sakoman@gmail.com>
> > Sent: 30 April 2021 20:46
> > To: Diego Santa Cruz <Diego.SantaCruz@spinetix.com>
> > Cc: openembedded-core@lists.openembedded.org; Mark Hatle
> > <mark.hatle@kernel.crashing.org>
> > Subject: Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7
> >
> > Trying again with Mark's correct email address!
> >
> > On Fri, Apr 30, 2021 at 8:43 AM Steve Sakoman <sakoman@gmail.com>
> > wrote:
> > >
> > > On Fri, Apr 30, 2021 at 7:00 AM Diego Santa Cruz via
> > > lists.openembedded.org
> > > <diego.santacruz=spinetix.com@lists.openembedded.org> wrote:
> > > >
> > > > Hello,
> > > >
> > > >
> > > >
> > > > I just updated to Yocto 3.1.7 (from 3.1.5) and I got the following warning
> > while building a kernel recipe.
> > > >
> > > >
> > > >
> > > >   do_install: Task do_install requested do_package unihash, but it was not
> > available
> > > >
> > > >
> > > >
> > > > I tracked this down to OE-Core commit
> > 0c28edf4bf0d2f92bf3a47406041c63acd90bacf ("package.bbclass: hash
> > equivalency and pr service") and it makes sense since do_package runs after
> > do_install and in dunfell we can find the following in kernel.bbclass
> > > >
> > > >
> > > >
> > > >   do_install[prefuncs] += "package_get_auto_pr"
> > > >
> > > >
> > > >
> > > > Merging OE-Core commit 36fe4db8ae827a93abe9fce6740459d215411965
> > ("kernel.bbclass: Remove do_install[prefunc] no longer needed") should
> > normally solve this warning by removing the above line, but I do not know if
> > that change is safe in dunfell.
> > > >
> > > >
> > > >
> > > > Could that be merged in? Or is there something else that could be done
> > to avoid the warning?
> > >
> > > I can do some testing with that patch.  cc-ing Mark to see if he has
> > > any comments since he is the patch author.
> > >
> > > Steve
> > >
> > >
> > > > Our build system flags a production build as bad if there is a warning, so
> > this is kind of annoying.
> > > >
> > > >
> > > >
> > > > Best,
> > > >
> > > >
> > > >
> > > > Diego
> > > >
> > > > --
> > > > Diego Santa Cruz, PhD
> > > > Technology Architect
> > > > spinetix.com
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
>
> 
>

Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7

[Diego Santa Cruz]

> -----Original Message-----
> From: Steve Sakoman <steve@sakoman.com>
> Sent: 06 May 2021 16:14
> To: Diego Santa Cruz <Diego.SantaCruz@spinetix.com>
> Cc: Steve Sakoman <sakoman@gmail.com>; openembedded-
> core@lists.openembedded.org; Mark Hatle
> <mark.hatle@kernel.crashing.org>
> Subject: Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7
> 
> On Tue, May 4, 2021 at 10:17 PM Diego Santa Cruz via
> lists.openembedded.org
> <diego.santacruz=spinetix.com@lists.openembedded.org> wrote:
> 
> > Any news on this one? I have been doing our own builds with the
> equivalent of that patch for our BSPs with a linux-intel derived kernel recipe
> plus another internal kernel recipe (both using kernel.bbclass) and I have not
> found any issues with it.
> 
> This is in the group of patches I am currently testing.  So far it looks good.
> 

Great, thanks!

> Steve
> 
> > > -----Original Message-----
> > > From: Steve Sakoman <sakoman@gmail.com>
> > > Sent: 30 April 2021 20:46
> > > To: Diego Santa Cruz <Diego.SantaCruz@spinetix.com>
> > > Cc: openembedded-core@lists.openembedded.org; Mark Hatle
> > > <mark.hatle@kernel.crashing.org>
> > > Subject: Re: [OE-core] Unihash warning building kernels in Yocto 3.1.7
> > >
> > > Trying again with Mark's correct email address!
> > >
> > > On Fri, Apr 30, 2021 at 8:43 AM Steve Sakoman <sakoman@gmail.com>
> > > wrote:
> > > >
> > > > On Fri, Apr 30, 2021 at 7:00 AM Diego Santa Cruz via
> > > > lists.openembedded.org
> > > > <diego.santacruz=spinetix.com@lists.openembedded.org> wrote:
> > > > >
> > > > > Hello,
> > > > >
> > > > >
> > > > >
> > > > > I just updated to Yocto 3.1.7 (from 3.1.5) and I got the following
> warning
> > > while building a kernel recipe.
> > > > >
> > > > >
> > > > >
> > > > >   do_install: Task do_install requested do_package unihash, but it was
> not
> > > available
> > > > >
> > > > >
> > > > >
> > > > > I tracked this down to OE-Core commit
> > > 0c28edf4bf0d2f92bf3a47406041c63acd90bacf ("package.bbclass: hash
> > > equivalency and pr service") and it makes sense since do_package runs
> after
> > > do_install and in dunfell we can find the following in kernel.bbclass
> > > > >
> > > > >
> > > > >
> > > > >   do_install[prefuncs] += "package_get_auto_pr"
> > > > >
> > > > >
> > > > >
> > > > > Merging OE-Core commit
> 36fe4db8ae827a93abe9fce6740459d215411965
> > > ("kernel.bbclass: Remove do_install[prefunc] no longer needed") should
> > > normally solve this warning by removing the above line, but I do not
> know if
> > > that change is safe in dunfell.
> > > > >
> > > > >
> > > > >
> > > > > Could that be merged in? Or is there something else that could be
> done
> > > to avoid the warning?
> > > >
> > > > I can do some testing with that patch.  cc-ing Mark to see if he has
> > > > any comments since he is the patch author.
> > > >
> > > > Steve
> > > >
> > > >
> > > > > Our build system flags a production build as bad if there is a warning,
> so
> > > this is kind of annoying.
> > > > >
> > > > >
> > > > >
> > > > > Best,
> > > > >
> > > > >
> > > > >
> > > > > Diego
> > > > >
> > > > > --
> > > > > Diego Santa Cruz, PhD
> > > > > Technology Architect
> > > > > spinetix.com
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> >
> > 
> >
-- 
Diego Santa Cruz, PhD
Technology Architect
spinetix.com