linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 464 bytes --]

Hi Vinod,

In commit 548860697046 ("dmaengine: dw_dmac: Enhance device tree
support") (which was changed since yesterday :-(), an instance of
__devinit has been added.  We are in the process of making CONFIG_HOTPLUG
always true, and since commit 78d86c213f28 ("init.h: Remove __dev*
sections from the kernel") (included in v3.7), __devinit does nothing.
Please do not add any more.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Greg Kroah-Hartman]

On Tue, Jan 08, 2013 at 09:35:57AM +1100, Stephen Rothwell wrote:
> Hi Vinod,
> 
> In commit 548860697046 ("dmaengine: dw_dmac: Enhance device tree
> support") (which was changed since yesterday :-(), an instance of
> __devinit has been added.  We are in the process of making CONFIG_HOTPLUG
> always true, and since commit 78d86c213f28 ("init.h: Remove __dev*
> sections from the kernel") (included in v3.7), __devinit does nothing.
> Please do not add any more.

That brings up something I wanted to ask you.

I'd like to send a patch to Linus to just delete __devinit and friends
entirely from init.h, but that will break the linux-next tree from
building in places (like this one, and in a few others.)

Is that ok?  That should finally get everyone's attention and keep them
from the cargo-cult programming mode of using these attributes :)

thanks,

greg k-h

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 2655 bytes --]

Hi Greg,

On Mon, 7 Jan 2013 14:43:08 -0800 Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
>
> That brings up something I wanted to ask you.
> 
> I'd like to send a patch to Linus to just delete __devinit and friends
> entirely from init.h, but that will break the linux-next tree from
> building in places (like this one, and in a few others.)
> 
> Is that ok?  That should finally get everyone's attention and keep them
> from the cargo-cult programming mode of using these attributes :)

I guess I can apply fixup patches that just remove any instances that
turn up ... and you and the tree owner will get copies of those.

So, if Linus takes it, I will cope.

This is what "git grep -E '__dev(init|exit)'" returns for next-20130107:

arch/ia64/kernel/iosapic.c: *				__init/__devinit attributes.
drivers/dma/dw_dmac.c:__devinit dw_dma_parse_dt(struct platform_device *pdev)
drivers/media/platform/sh_veu.c:static int __devinit sh_veu_probe(struct platform_device *pdev)
drivers/media/platform/sh_veu.c:static int __devexit sh_veu_remove(struct platform_device *pdev)
drivers/media/platform/sh_veu.c:	.remove		= __devexit_p(sh_veu_remove),
drivers/net/ethernet/qlogic/qlcnic/qlcnic_main.c:static int __devinit
drivers/net/ethernet/qlogic/qlcnic/qlcnic_main.c:static void __devexit qlcnic_remove(struct pci_dev *pdev)
drivers/power/reset/restart-poweroff.c:static int __devinit restart_poweroff_probe(struct platform_device *pdev)
drivers/power/reset/restart-poweroff.c:static int __devexit restart_poweroff_remove(struct platform_device *pdev)
drivers/power/reset/restart-poweroff.c:	.remove = __devexit_p(restart_poweroff_remove),
drivers/regulator/lp8755.c:	.remove = __devexit_p(lp8755_remove),
drivers/scsi/ipr.c:static int __devinit ipr_enable_msix(struct ipr_ioa_cfg *ioa_cfg)
drivers/scsi/ipr.c:static int __devinit ipr_enable_msi(struct ipr_ioa_cfg *ioa_cfg)
drivers/scsi/ipr.c:static void __devinit name_msi_vectors(struct ipr_ioa_cfg *ioa_cfg)
drivers/scsi/ipr.c:static int __devinit ipr_request_other_msi_irqs(struct ipr_ioa_cfg *ioa_cfg)
drivers/scsi/ips.c:/* 5.30.00  - use __devexit_p()                                              */
scripts/tags.sh:	-I __initdata,__exitdata,__initconst,__devinitdata	\
scripts/tags.sh:	-I __devinitconst,__cpuinitdata,__initdata_memblock	\
sound/oss/swarm_cs4297a.c:* 11/10/00 bkz - added __devinit to cs4297a_hw_init()

so there are just a few left.

BTW, I am *not* reporting merge fixes I am doing due to your tree wide
removal of __dev*, but they are still turning up :-(

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Vinod Koul]

On Mon, Jan 07, 2013 at 02:43:08PM -0800, Greg Kroah-Hartman wrote:
> On Tue, Jan 08, 2013 at 09:35:57AM +1100, Stephen Rothwell wrote:
> > Hi Vinod,
> > 
> > In commit 548860697046 ("dmaengine: dw_dmac: Enhance device tree
> > support") (which was changed since yesterday :-(), an instance of
> > __devinit has been added.  We are in the process of making CONFIG_HOTPLUG
> > always true, and since commit 78d86c213f28 ("init.h: Remove __dev*
> > sections from the kernel") (included in v3.7), __devinit does nothing.
> > Please do not add any more.
> 
> That brings up something I wanted to ask you.
> 
> I'd like to send a patch to Linus to just delete __devinit and friends
> entirely from init.h, but that will break the linux-next tree from
> building in places (like this one, and in a few others.)
> 
> Is that ok?  That should finally get everyone's attention and keep them
> from the cargo-cult programming mode of using these attributes :)
I think if you remove this from linus's tree and since we are in rc2 we can get
it done and remove all other intances from other trees.

--
~Vinod

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Viresh Kumar]

On Tue, Jan 8, 2013 at 4:05 AM, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> In commit 548860697046 ("dmaengine: dw_dmac: Enhance device tree
> support") (which was changed since yesterday :-(), an instance of
> __devinit has been added.  We are in the process of making CONFIG_HOTPLUG
> always true, and since commit 78d86c213f28 ("init.h: Remove __dev*
> sections from the kernel") (included in v3.7), __devinit does nothing.
> Please do not add any more.

When this patch got added, __devinit discussion just started. Because Vinod
missed the merge window, we still have this patch in linux-next.

@Vinod: I hope you can edit this patch yourself?

--
viresh

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Vinod Koul]

On Tue, Jan 08, 2013 at 10:19:12AM +0530, Viresh Kumar wrote:
> @Vinod: I hope you can edit this patch yourself?
Done, I have removed this and another instance

--
~Vinod

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Vinod Koul]

On Tue, Jan 08, 2013 at 09:35:57AM +1100, Stephen Rothwell wrote:
> Hi Vinod,
> 
> In commit 548860697046 ("dmaengine: dw_dmac: Enhance device tree
> support") (which was changed since yesterday :-(),
Yes due to build conflict I redid the rebase of my tree.

> an instance of
> __devinit has been added.  We are in the process of making CONFIG_HOTPLUG
> always true, and since commit 78d86c213f28 ("init.h: Remove __dev*
> sections from the kernel") (included in v3.7), __devinit does nothing.
> Please do not add any more.
I think I was trying to remove it. Anyway I have changed the commit to remove
this and also another instance. So you should be happy about it :)
> 
> -- 
> Cheers,
> Stephen Rothwell                    sfr@canb.auug.org.au

Re: linux-next: comment in commit in the slave-dma tree (dmaengine: dw_dmac: Enhance device tree support)

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 447 bytes --]

Hi Vinod,

On Tue, 8 Jan 2013 00:17:35 -0800 Vinod Koul <vinod.koul@intel.com> wrote:
>
> I think I was trying to remove it. Anyway I have changed the commit to remove
> this and also another instance. So you should be happy about it :)

Thanks.  I would have been even happier if you had just added another
commit that removed them :-)  But I shan't be churlish.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]