Re: [PATCH 7/7] iommu/arm-smmu: Allow building as a module

[Will Deacon]

Hi Isaac,

On Mon, Nov 04, 2019 at 11:34:00AM -0800, Isaac J. Manjarres wrote:
> On Wed, Oct 30, 2019 at 02:51:12PM +0000, Will Deacon wrote:
> > diff --git a/drivers/iommu/arm-smmu-impl.c b/drivers/iommu/arm-smmu-impl.c
> > index 5c87a38620c4..2f82d40317d6 100644
> > --- a/drivers/iommu/arm-smmu-impl.c
> > +++ b/drivers/iommu/arm-smmu-impl.c
> > @@ -5,6 +5,7 @@
> >  #define pr_fmt(fmt) "arm-smmu: " fmt
> >  
> >  #include <linux/bitfield.h>
> > +#include <linux/module.h>
> >  #include <linux/of.h>
> >  
> >  #include "arm-smmu.h"
> > @@ -172,3 +173,8 @@ struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
> >  
> >  	return smmu;
> >  }
> > +EXPORT_SYMBOL_GPL(arm_smmu_impl_init);
> > +
> > +MODULE_DESCRIPTION("IOMMU quirks for ARM architected SMMU implementations");
> > +MODULE_AUTHOR("Robin Murphy <robin.murphy@xxxxxxx>");
> > +MODULE_LICENSE("GPL v2");
> 
> A minor comment: I was curious about why arm-smmu.c and arm-smmu-impl.c
> were being compiled as two separate modules, as opposed to combining
> them into one module? The latter approach seemed more appropriate, given
> that arm-smmu-impl doesn't offer much as a module on its own. Thoughts?

Yes, you're right. The simple answer is that I couldn't come up with a good
name for the combined module, since "arm-smmu" is already taken by the core
part of the driver and I don't want to rename that file. Looking at what a
few other drivers do, it seems that "arm-smmu-mod" might be the best bet
so I'll incorporate that change for v2 and put you on cc.

Thanks!

Will