Doubt regarding build warnings

Doubt regarding build warnings

[Bhumika Goyal]

[-- Attachment #1.1: Type: text/plain, Size: 974 bytes --]

I got these warning today while building the kernel. I already got some 
patches accepted but was not not getting any such warnings before.
What can be the reason and how to fix? I used make M=drivers/staging for 
the build.
The warnings are like:

WARNING: "__cpu_online_mask" 
[drivers/staging/lustre/lustre/ptlrpc/ptlrpc.ko] undefined!
WARNING: "__put_page" [drivers/staging/lustre/lustre/ptlrpc/ptlrpc.ko] 
undefined!
WARNING: "__put_page" [drivers/staging/lustre/lustre/obdecho/obdecho.ko] 
undefined!
WARNING: "__cpu_possible_mask" 
[drivers/staging/lustre/lustre/obdclass/obdclass.ko] undefined!
WARNING: "__page_mapcount" [drivers/staging/lustre/lustre/llite/lustre.ko] 
undefined!
WARNING: "__put_page" [drivers/staging/lustre/lustre/llite/lustre.ko] 
undefined!
WARNING: "__cpu_possible_mask" 
[drivers/staging/lustre/lustre/libcfs/libcfs.ko] undefined!
WARNING: "__cpu_online_mask" 
[drivers/staging/lustre/lustre/libcfs/libcfs.ko] undefined!

Thank you,
Bhumika

[-- Attachment #1.2: Type: text/html, Size: 1110 bytes --]

Re: [Outreachy kernel] Doubt regarding build warnings

[Arnd Bergmann]

On Monday 08 February 2016 20:06:15 Bhumika Goyal wrote:
> I got these warning today while building the kernel. I already got some 
> patches accepted but was not not getting any such warnings before.
> What can be the reason and how to fix? I used make M=drivers/staging for 
> the build.
> The warnings are like:
> 
> WARNING: "__cpu_online_mask" 
> [drivers/staging/lustre/lustre/ptlrpc/ptlrpc.ko] undefined!
> WARNING: "__put_page" [drivers/staging/lustre/lustre/ptlrpc/ptlrpc.ko] 
> undefined!
> WARNING: "__put_page" [drivers/staging/lustre/lustre/obdecho/obdecho.ko] 
> undefined!
> WARNING: "__cpu_possible_mask" 
> [drivers/staging/lustre/lustre/obdclass/obdclass.ko] undefined!
> WARNING: "__page_mapcount" [drivers/staging/lustre/lustre/llite/lustre.ko] 
> undefined!
> WARNING: "__put_page" [drivers/staging/lustre/lustre/llite/lustre.ko] 
> undefined!
> WARNING: "__cpu_possible_mask" 
> [drivers/staging/lustre/lustre/libcfs/libcfs.ko] undefined!
> WARNING: "__cpu_online_mask" 
> [drivers/staging/lustre/lustre/libcfs/libcfs.ko] undefined!
> 

See if you get the same warnings when you do a whole kernel build, without the
"M=drivers/staging". I personally prefer "make drivers/staging/" instead of
"make M=drivers/staging", which has slightly different behavior.

	Arnd

Re: [Outreachy kernel] Doubt regarding build warnings

[Bhumika Goyal]

[-- Attachment #1.1: Type: text/plain, Size: 166 bytes --]

Thank you Arnd .I am not getting any warnings with "make drivers/staging". 
Can I know the reason for previous warnings when I used "make M" ?

Thanks,
Bhumika Goyal

[-- Attachment #1.2: Type: text/html, Size: 218 bytes --]

Re: [Outreachy kernel] Doubt regarding build warnings

[Arnd Bergmann]

On Tuesday 09 February 2016 01:01:37 Bhumika Goyal wrote:
> Thank you Arnd .I am not getting any warnings with "make drivers/staging". 
> Can I know the reason for previous warnings when I used "make M" ?
> 
> 

The differences are very subtle:

make drivers/staging

   always succeeds, does nothing at all because the directory exists

make drivers/staging/

   descends into that directory to build all the files in it, but does
   not try to link the modules

make M=drivers/staging

   From the output you get, I think this must try to link the modules
   against the previously-build vmlinux file, which for some reason
   fails here. I suggested doing 'make' to rebuild everything including
   vmlinux, so it is up-to-date with your current configuration.

	Arnd

Re: [Outreachy kernel] Doubt regarding build warnings

[Julia Lawall]

On Tue, 9 Feb 2016, Arnd Bergmann wrote:

> On Tuesday 09 February 2016 01:01:37 Bhumika Goyal wrote:
> > Thank you Arnd .I am not getting any warnings with "make drivers/staging".
> > Can I know the reason for previous warnings when I used "make M" ?
> >
> >
>
> The differences are very subtle:
>
> make drivers/staging
>
>    always succeeds, does nothing at all because the directory exists
>
> make drivers/staging/
>
>    descends into that directory to build all the files in it, but does
>    not try to link the modules
>
> make M=drivers/staging
>
>    From the output you get, I think this must try to link the modules
>    against the previously-build vmlinux file, which for some reason
>    fails here. I suggested doing 'make' to rebuild everything including
>    vmlinux, so it is up-to-date with your current configuration.

CouldI add this to the wiki?

julia

>
> 	Arnd
>
> --
> You received this message because you are subscribed to the Google Groups "outreachy-kernel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to outreachy-kernel+unsubscribe@googlegroups.com.
> To post to this group, send email to outreachy-kernel@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/outreachy-kernel/5477208.ntDhW1kj3a%40wuerfel.
> For more options, visit https://groups.google.com/d/optout.
>

Re: [Outreachy kernel] Doubt regarding build warnings

[Arnd Bergmann]

On Tuesday 09 February 2016 12:09:38 Julia Lawall wrote:
> On Tue, 9 Feb 2016, Arnd Bergmann wrote:
> >
> > The differences are very subtle:
> >
> > make drivers/staging
> >
> >    always succeeds, does nothing at all because the directory exists
> >
> > make drivers/staging/
> >
> >    descends into that directory to build all the files in it, but does
> >    not try to link the modules
> >
> > make M=drivers/staging
> >
> >    From the output you get, I think this must try to link the modules
> >    against the previously-build vmlinux file, which for some reason
> >    fails here. I suggested doing 'make' to rebuild everything including
> >    vmlinux, so it is up-to-date with your current configuration.
> 
> Could I add this to the wiki?
> 

Sure.

	Arnd

Re: [Outreachy kernel] Doubt regarding build warnings

[Julia Lawall]

On Tue, 9 Feb 2016, Arnd Bergmann wrote:

> On Tuesday 09 February 2016 12:09:38 Julia Lawall wrote:
> > On Tue, 9 Feb 2016, Arnd Bergmann wrote:
> > >
> > > The differences are very subtle:
> > >
> > > make drivers/staging
> > >
> > >    always succeeds, does nothing at all because the directory exists
> > >
> > > make drivers/staging/
> > >
> > >    descends into that directory to build all the files in it, but does
> > >    not try to link the modules
> > >
> > > make M=drivers/staging
> > >
> > >    From the output you get, I think this must try to link the modules
> > >    against the previously-build vmlinux file, which for some reason
> > >    fails here. I suggested doing 'make' to rebuild everything including
> > >    vmlinux, so it is up-to-date with your current configuration.
> >
> > Could I add this to the wiki?
> >
>
> Sure.

I put the following:

Compiling only part of the kernel

There are several ways to compile only part of the kernel:

    make path/file.o: This compiles only a single file. It may not be
sufficient to check changes that affect interactions with other files.

    make path, e.g. make drivers/staging: This always succeeds. It does
nothing at all because the directory exists.

    make path/, e.g. make drivers/staging/: This descends into that
directory to build all the files in it, but does not try to link the
modules.

    make M=drivers/staging: This seems to try to link the modules in a
previously build vmlinux file.

Make path/ could be a reasonable choice for a localized change within a
single driver, or for a change localized to drivers staging. In any case,
as a minimum always check that compilation produces a .o file for every
file that your patch changes. If there is no .o file, you have not
compiled the file, and you need to look for other compiler options.

julia

Re: [Outreachy kernel] Doubt regarding build warnings

[Arnd Bergmann]

On Tuesday 09 February 2016 16:18:55 Julia Lawall wrote:
>     make M=drivers/staging: This seems to try to link the modules in a
> previously build vmlinux file.
> 
> Make path/ could be a reasonable choice for a localized change within a
> single driver, or for a change localized to drivers staging. In any case,
> as a minimum always check that compilation produces a .o file for every
> file that your patch changes. If there is no .o file, you have not
> compiled the file, and you need to look for other compiler options.

Ok, sounds good. I've added a little bit more detail now:

http://kernelnewbies.org/FirstKernelPatch?action=diff&rev2=25&rev1=24



	Arnd 

Re: [Outreachy kernel] Doubt regarding build warnings

[Julia Lawall]

On Tue, 9 Feb 2016, Arnd Bergmann wrote:

> On Tuesday 09 February 2016 16:18:55 Julia Lawall wrote:
> >     make M=drivers/staging: This seems to try to link the modules in a
> > previously build vmlinux file.
> >
> > Make path/ could be a reasonable choice for a localized change within a
> > single driver, or for a change localized to drivers staging. In any case,
> > as a minimum always check that compilation produces a .o file for every
> > file that your patch changes. If there is no .o file, you have not
> > compiled the file, and you need to look for other compiler options.
>
> Ok, sounds good. I've added a little bit more detail now:
>
> http://kernelnewbies.org/FirstKernelPatch?action=diff&rev2=25&rev1=24

Thanks.  What is -sj2?

julia

Re: [Outreachy kernel] Doubt regarding build warnings

[Arnd Bergmann]

On Tuesday 09 February 2016 16:49:31 Julia Lawall wrote:
> On Tue, 9 Feb 2016, Arnd Bergmann wrote:
> 
> > On Tuesday 09 February 2016 16:18:55 Julia Lawall wrote:
> > >     make M=drivers/staging: This seems to try to link the modules in a
> > > previously build vmlinux file.
> > >
> > > Make path/ could be a reasonable choice for a localized change within a
> > > single driver, or for a change localized to drivers staging. In any case,
> > > as a minimum always check that compilation produces a .o file for every
> > > file that your patch changes. If there is no .o file, you have not
> > > compiled the file, and you need to look for other compiler options.
> >
> > Ok, sounds good. I've added a little bit more detail now:
> >
> > http://kernelnewbies.org/FirstKernelPatch?action=diff&rev2=25&rev1=24
> 
> Thanks.  What is -sj2?
> 
> 

-s is 'silent', -j2 is used to build on two threads in parallel rather than
one. On second thought, I'll remove those again, as this prevents you
from seeing which files get built.

	Arnd

Re: [Outreachy kernel] Doubt regarding build warnings

[Julia Lawall]

On Tue, 9 Feb 2016, Arnd Bergmann wrote:

> On Tuesday 09 February 2016 16:49:31 Julia Lawall wrote:
> > On Tue, 9 Feb 2016, Arnd Bergmann wrote:
> >
> > > On Tuesday 09 February 2016 16:18:55 Julia Lawall wrote:
> > > >     make M=drivers/staging: This seems to try to link the modules in a
> > > > previously build vmlinux file.
> > > >
> > > > Make path/ could be a reasonable choice for a localized change within a
> > > > single driver, or for a change localized to drivers staging. In any case,
> > > > as a minimum always check that compilation produces a .o file for every
> > > > file that your patch changes. If there is no .o file, you have not
> > > > compiled the file, and you need to look for other compiler options.
> > >
> > > Ok, sounds good. I've added a little bit more detail now:
> > >
> > > http://kernelnewbies.org/FirstKernelPatch?action=diff&rev2=25&rev1=24
> >
> > Thanks.  What is -sj2?
> >
> >
>
> -s is 'silent', -j2 is used to build on two threads in parallel rather than
> one. On second thought, I'll remove those again, as this prevents you
> from seeing which files get built.

-j2 could be left, but one could also infer it from the rest.

thanks,
julia

Re: [Outreachy kernel] Doubt regarding build warnings

[Daniel Baluta]

On Tue, Feb 9, 2016 at 5:55 PM, Julia Lawall <julia.lawall@lip6.fr> wrote:
> On Tue, 9 Feb 2016, Arnd Bergmann wrote:
>
>> On Tuesday 09 February 2016 16:49:31 Julia Lawall wrote:
>> > On Tue, 9 Feb 2016, Arnd Bergmann wrote:
>> >
>> > > On Tuesday 09 February 2016 16:18:55 Julia Lawall wrote:
>> > > >     make M=drivers/staging: This seems to try to link the modules in a
>> > > > previously build vmlinux file.
>> > > >
>> > > > Make path/ could be a reasonable choice for a localized change within a
>> > > > single driver, or for a change localized to drivers staging. In any case,
>> > > > as a minimum always check that compilation produces a .o file for every
>> > > > file that your patch changes. If there is no .o file, you have not
>> > > > compiled the file, and you need to look for other compiler options.
>> > >
>> > > Ok, sounds good. I've added a little bit more detail now:
>> > >
>> > > http://kernelnewbies.org/FirstKernelPatch?action=diff&rev2=25&rev1=24
>> >
>> > Thanks.  What is -sj2?
>> >
>> >
>>
>> -s is 'silent', -j2 is used to build on two threads in parallel rather than
>> one. On second thought, I'll remove those again, as this prevents you
>> from seeing which files get built.
>
> -j2 could be left, but one could also infer it from the rest.

I would also remove -j2, doesn't make any difference for building few modules.
It can only confuse things.

Daniel.