linux-next: i2c tree build failure

linux-next: i2c tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1055 bytes --]

Hi Jean,

Today's linux-next build (powerpc ppc64_defconfig) failed like this:

drivers/i2c/i2c-core.c:234: warning: initialization from incompatible pointer type
drivers/i2c/i2c-core.c:537: warning: initialization from incompatible pointer type
drivers/i2c/i2c-core.c: In function 'i2c_register_adapter':
drivers/i2c/i2c-core.c:603: error: implicit declaration of function 'class_compat_create_link'
drivers/i2c/i2c-core.c: In function 'i2c_del_adapter':
drivers/i2c/i2c-core.c:789: error: implicit declaration of function 'class_compat_remove_link'
drivers/i2c/i2c-core.c: In function 'i2c_init':
drivers/i2c/i2c-core.c:999: error: implicit declaration of function 'class_compat_register'
drivers/i2c/i2c-core.c:999: warning: assignment makes pointer from integer without a cast
drivers/i2c/i2c-core.c:1012: error: implicit declaration of function 'class_compat_unregister'

I have used the i2c tree from next-20090828 for today.
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: linux-next: i2c tree build failure

[Jean Delvare]

Hi Stephen,

On Mon, 31 Aug 2009 11:19:35 +1000, Stephen Rothwell wrote:
> Hi Jean,
> 
> Today's linux-next build (powerpc ppc64_defconfig) failed like this:
> 
> drivers/i2c/i2c-core.c:234: warning: initialization from incompatible pointer type
> drivers/i2c/i2c-core.c:537: warning: initialization from incompatible pointer type
> drivers/i2c/i2c-core.c: In function 'i2c_register_adapter':
> drivers/i2c/i2c-core.c:603: error: implicit declaration of function 'class_compat_create_link'
> drivers/i2c/i2c-core.c: In function 'i2c_del_adapter':
> drivers/i2c/i2c-core.c:789: error: implicit declaration of function 'class_compat_remove_link'
> drivers/i2c/i2c-core.c: In function 'i2c_init':
> drivers/i2c/i2c-core.c:999: error: implicit declaration of function 'class_compat_register'
> drivers/i2c/i2c-core.c:999: warning: assignment makes pointer from integer without a cast
> drivers/i2c/i2c-core.c:1012: error: implicit declaration of function 'class_compat_unregister'
> 
> I have used the i2c tree from next-20090828 for today.

These are caused by the fact that I moved the patch adding
class_compat_create_link() and friends out of the NEXT_PATCHES section.
I did so because Greg KH (Cc'd) added this patch to his driver-core
patch set, which I thought was part of linux-next. I simply didn't want
to apply the same patch twice.

To clarify the dependency, I added the following to my series.conf:

# NEXT_BASE driver-core.current

but apparently it didn't work as intended.

Greg, please clarify whether the patch in question
(driver-core-add-support-for-compatibility-classes.patch) is pushed to
linux-next or not.

Stephen, if my dependency tag is not correct, please tell me how to fix
it.

Thanks, and sorry for the trouble.

-- 
Jean Delvare

Re: linux-next: i2c tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1346 bytes --]

Hi Jean,

On Mon, 31 Aug 2009 12:04:49 +0200 Jean Delvare <khali@linux-fr.org> wrote:
> 
> These are caused by the fact that I moved the patch adding
> class_compat_create_link() and friends out of the NEXT_PATCHES section.
> I did so because Greg KH (Cc'd) added this patch to his driver-core
> patch set, which I thought was part of linux-next. I simply didn't want
> to apply the same patch twice.

It doesn't really matter if the same patch appears twice - git usually
copes quite well.  Also I can tell and I (usually) don't bother
commenting on it.

> To clarify the dependency, I added the following to my series.conf:
> 
> # NEXT_BASE driver-core.current
> 
> but apparently it didn't work as intended.
> 
> Greg, please clarify whether the patch in question
> (driver-core-add-support-for-compatibility-classes.patch) is pushed to
> linux-next or not.
> 
> Stephen, if my dependency tag is not correct, please tell me how to fix
> it.

It was fine, except that Greg has that patch in the driver-core series
(not driver-core.current).  It you decide to leave that patch out and
change the dependency, please let me know as I will then have to merge
your tree after Greg's (it is currently earlier.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: linux-next: i2c tree build failure

[Jean Delvare]

Hi Stephen,

On Mon, 31 Aug 2009 21:03:35 +1000, Stephen Rothwell wrote:
> On Mon, 31 Aug 2009 12:04:49 +0200 Jean Delvare <khali@linux-fr.org> wrote:
> > 
> > These are caused by the fact that I moved the patch adding
> > class_compat_create_link() and friends out of the NEXT_PATCHES section.
> > I did so because Greg KH (Cc'd) added this patch to his driver-core
> > patch set, which I thought was part of linux-next. I simply didn't want
> > to apply the same patch twice.
> 
> It doesn't really matter if the same patch appears twice - git usually
> copes quite well.  Also I can tell and I (usually) don't bother
> commenting on it.
> 
> > To clarify the dependency, I added the following to my series.conf:
> > 
> > # NEXT_BASE driver-core.current
> > 
> > but apparently it didn't work as intended.
> > 
> > Greg, please clarify whether the patch in question
> > (driver-core-add-support-for-compatibility-classes.patch) is pushed to
> > linux-next or not.
> > 
> > Stephen, if my dependency tag is not correct, please tell me how to fix
> > it.
> 
> It was fine, except that Greg has that patch in the driver-core series
> (not driver-core.current).

OK, thanks for the clarification. I did not notice Greg had split some
of his trees that way. Makes sense.

> It you decide to leave that patch out and
> change the dependency, please let me know as I will then have to merge
> your tree after Greg's (it is currently earlier.

Regardless of what I do, I think it would make sense to merge
driver-core before driver subsystem trees. I would insert if before pci.

I am also surprised that I would have to tell you. What is the purpose
of the NEXT_BASE tag if you do not check for dependencies automatically?

Anyway, if you say git can deal with duplicate patches OK, then I think
I'll simply re-add the patch on my end. But I do not have a strong
opinion on this either, so if you prefer clean dependencies and
ordering without duplicate patches, it is just as easy for me to leave
the redundant patch out and restore the NEXT_BASE tag.

Thanks,
-- 
Jean Delvare

Re: linux-next: i2c tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1672 bytes --]

Hi Jean,

On Mon, 31 Aug 2009 13:16:47 +0200 Jean Delvare <khali@linux-fr.org> wrote:
>
> OK, thanks for the clarification. I did not notice Greg had split some
> of his trees that way. Makes sense.
> 
> > It you decide to leave that patch out and
> > change the dependency, please let me know as I will then have to merge
> > your tree after Greg's (it is currently earlier.
> 
> Regardless of what I do, I think it would make sense to merge
> driver-core before driver subsystem trees. I would insert if before pci.

I merge driver-core near the end because it often has API changes in it
and Linus suggested that the pain of API changes should be with the
changer.

> I am also surprised that I would have to tell you. What is the purpose
> of the NEXT_BASE tag if you do not check for dependencies automatically?

I use the tag to choose the base when I import the quilt series into git.
So far I have not needed to automate the ordering of the imports.

> Anyway, if you say git can deal with duplicate patches OK, then I think
> I'll simply re-add the patch on my end. But I do not have a strong
> opinion on this either, so if you prefer clean dependencies and
> ordering without duplicate patches, it is just as easy for me to leave
> the redundant patch out and restore the NEXT_BASE tag.

I prefer less dependencies (so you could put the patch back in), however,
that means if Greg changes the patch, then we will have two slightly
different versions to merge.  I can cope with that as well.

Just readd the patch :-)

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: linux-next: i2c tree build failure

[Jean Delvare]

On Mon, 31 Aug 2009 21:59:05 +1000, Stephen Rothwell wrote:
> On Mon, 31 Aug 2009 13:16:47 +0200 Jean Delvare <khali@linux-fr.org> wrote:
> > I am also surprised that I would have to tell you. What is the purpose
> > of the NEXT_BASE tag if you do not check for dependencies automatically?
> 
> I use the tag to choose the base when I import the quilt series into git.
> So far I have not needed to automate the ordering of the imports.
> 
> > Anyway, if you say git can deal with duplicate patches OK, then I think
> > I'll simply re-add the patch on my end. But I do not have a strong
> > opinion on this either, so if you prefer clean dependencies and
> > ordering without duplicate patches, it is just as easy for me to leave
> > the redundant patch out and restore the NEXT_BASE tag.
> 
> I prefer less dependencies (so you could put the patch back in), however,
> that means if Greg changes the patch, then we will have two slightly
> different versions to merge.  I can cope with that as well.

I've made sure my patch has the same name as Greg's, to make it easier
for you to spot duplicates.

> Just readd the patch :-)

OK, done. Hopefully the next merge window is coming soon now anyway, and
then the trouble with this specific patch will be solved.

-- 
Jean Delvare

Re: linux-next: i2c tree build failure

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 531 bytes --]

Hi Jean,

On Mon, 31 Aug 2009 14:15:46 +0200 Jean Delvare <khali@linux-fr.org> wrote:
>
> I've made sure my patch has the same name as Greg's, to make it easier
> for you to spot duplicates.

Thanks.

> > Just readd the patch :-)
> 
> OK, done. Hopefully the next merge window is coming soon now anyway, and
> then the trouble with this specific patch will be solved.

Indeed - and I can get more sleep :-)

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]