Re: [PATCH v5 4/7] iommu/sva: Stop using ioasid_set for SVA

[Jacob Pan]

Hi Baolu,

On Fri, 10 Mar 2023 13:07:51 +0800, Baolu Lu <baolu.lu@xxxxxxxxxxxxxxx>
wrote:

> On 3/10/23 6:21 AM, Jacob Pan wrote:
> > From: Jason Gunthorpe <jgg@xxxxxxxxxx>
> > 
> > Instead SVA drivers can use a simple global IDA to allocate PASIDs for
> > each mm_struct.
> > 
> > Future work would be to allow drivers using the SVA APIs to reserve
> > global PASIDs from this IDA for their internal use, eg with the DMA API
> > PASID support.
> > 
> > Signed-off-by: Jason Gunthorpe <jgg@xxxxxxxxxx>
> > Signed-off-by: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx>
> > ---
> > v5:
> > 	- Put removing iommu_sva_find() to a separate patch (Kevin)
> > 	- Make pasid allocation range to be inclusive (Tina)
> > 	- Simplified return code handling (Baolu)
> > v4:
> > 	- Keep GFP_ATOMIC flag for PASID allocation, will changed to
> > 	GFP_KERNEL in a separate patch.
> > ---
> >   drivers/iommu/iommu-sva.c | 42 +++++++++++++--------------------------
> >   drivers/iommu/iommu-sva.h |  2 --
> >   2 files changed, 14 insertions(+), 30 deletions(-)
> > 
> > diff --git a/drivers/iommu/iommu-sva.c b/drivers/iommu/iommu-sva.c
> > index 4f357ef14f04..b75711bdbe97 100644
> > --- a/drivers/iommu/iommu-sva.c
> > +++ b/drivers/iommu/iommu-sva.c
> > @@ -9,47 +9,33 @@
> >   #include "iommu-sva.h"
> >   
> >   static DEFINE_MUTEX(iommu_sva_lock);
> > -static DECLARE_IOASID_SET(iommu_sva_pasid);
> > +static DEFINE_IDA(iommu_global_pasid_ida);
> >   
> > -/**
> > - * iommu_sva_alloc_pasid - Allocate a PASID for the mm
> > - * @mm: the mm
> > - * @min: minimum PASID value (inclusive)
> > - * @max: maximum PASID value (inclusive)
> > - *
> > - * Try to allocate a PASID for this mm, or take a reference to the
> > existing one
> > - * provided it fits within the [@min, @max] range. On success the
> > PASID is
> > - * available in mm->pasid and will be available for the lifetime of
> > the mm.
> > - *
> > - * Returns 0 on success and < 0 on error.
> > - */
> > -int iommu_sva_alloc_pasid(struct mm_struct *mm, ioasid_t min, ioasid_t
> > max) +static int iommu_sva_alloc_pasid(struct mm_struct *mm, ioasid_t
> > min, ioasid_t max) {
> >   	int ret = 0;
> > -	ioasid_t pasid;
> >   
> > -	if (min == INVALID_IOASID || max == INVALID_IOASID ||
> > -	    min == 0 || max < min)
> > +	if (!pasid_valid(min) || !pasid_valid(max) ||
> > +	     min == 0 || max < min)  
> 
> No need to change above line.
> 
> >   		return -EINVAL;
> >   
> >   	mutex_lock(&iommu_sva_lock);
> >   	/* Is a PASID already associated with this mm? */
> >   	if (pasid_valid(mm->pasid)) {
> > -		if (mm->pasid < min || mm->pasid >= max)
> > +		if (mm->pasid < min || mm->pasid > max)  
> 
> I forgot why do we need to change above line. But it's better to put
> some comments there so that people don't need to dive into
> ioasid_alloc() to know the inclusion or exclusion of @min or @max.
> 
just to be consistent for both limits. sure, we can add the comment.

> >   			ret = -EOVERFLOW;
> >   		goto out;
> >   	}
> >   
> > -	pasid = ioasid_alloc(&iommu_sva_pasid, min, max, mm);
> > -	if (!pasid_valid(pasid))
> > -		ret = -ENOMEM;
> > -	else
> > -		mm->pasid = pasid;
> > +	ret = ida_alloc_range(&iommu_global_pasid_ida, min, max,
> > GFP_ATOMIC);
> > +	if (ret < 0)
> > +		goto out;
> > +	mm->pasid = ret;
> > +	ret = 0;
> >   out:
> >   	mutex_unlock(&iommu_sva_lock);
> >   	return ret;
> >   }
> > -EXPORT_SYMBOL_GPL(iommu_sva_alloc_pasid);
> >   
> >   /**
> >    * iommu_sva_bind_device() - Bind a process address space to a device
> > @@ -221,8 +207,8 @@ iommu_sva_handle_iopf(struct iommu_fault *fault,
> > void *data) 
> >   void mm_pasid_drop(struct mm_struct *mm)
> >   {
> > -	if (pasid_valid(mm->pasid)) {
> > -		ioasid_free(mm->pasid);
> > -		mm->pasid = INVALID_IOASID;
> > -	}
> > +	if (likely(!pasid_valid(mm->pasid)))
> > +		return;
> > +
> > +	ida_free(&iommu_global_pasid_ida, mm->pasid);  
> 
> Any reason why do you drop "mm->pasid = INVALID_IOASID;" here?
> 
mm_drop is called when mm gets released because "mm_count" becomes zero. so
there is no more mm_users. No one will see mm->pasid after this.

On the other hand, iommu_sva_bind_device() needs to hold mm_users refcount
so it won;t happen after mm_drop.

> >   }
> > diff --git a/drivers/iommu/iommu-sva.h b/drivers/iommu/iommu-sva.h
> > index 102eae1817a2..c22d0174ad61 100644
> > --- a/drivers/iommu/iommu-sva.h
> > +++ b/drivers/iommu/iommu-sva.h
> > @@ -8,8 +8,6 @@
> >   #include <linux/ioasid.h>
> >   #include <linux/mm_types.h>
> >   
> > -int iommu_sva_alloc_pasid(struct mm_struct *mm, ioasid_t min, ioasid_t
> > max); -
> >   /* I/O Page fault */
> >   struct device;
> >   struct iommu_fault;  
> 
> Best regards,
> baolu


Thanks,

Jacob