Re: [PATCH v3 2/4] drivers/iommu: Add calls for IOMMU_DEBUG_PAGEALLOC

[Baolu Lu <baolu.lu@linux.intel.com>]

On 12/9/25 04:09, Mostafa Saleh wrote:
> On Tue, Nov 25, 2025 at 10:01 AM Mostafa Saleh<smostafa@google.com> wrote:
>> On Tue, Nov 25, 2025 at 03:35:08PM +0800, Baolu Lu wrote:
>>> On 11/25/25 04:08, Mostafa Saleh wrote:
>>>> Add calls for the new iommu debug config IOMMU_DEBUG_PAGEALLOC:
>>>> - iommu_debug_init: Enable the debug mode if configured by the user.
>>>> - iommu_debug_map: Track iommu pages mapped, using physical address.
>>>> - iommu_debug_unmap_begin: Track start of iommu unmap operation, with
>>>>     IOVA and size.
>>>> - iommu_debug_unmap_end: Track the end of unmap operation, passing the
>>>>     actual unmapped size versus the tracked one at unmap_begin.
>>>>
>>>> We have to do the unmap_begin/end as once pages are unmapped we lose
>>>> the information of the physical address.
>>>> This is racy, but the API is racy by construction as it uses refcounts
>>>> and doesn't attempt to lock/synchronize with the IOMMU API as that will
>>>> be costly, meaning that possibility of false negative exists.
>>>>
>>>> Signed-off-by: Mostafa Saleh<smostafa@google.com>
>>>> ---
>>>>    drivers/iommu/iommu-debug-pagealloc.c | 26 +++++++++++++
>>>>    drivers/iommu/iommu.c                 | 12 +++++-
>>>>    include/linux/iommu-debug-pagealloc.h | 56 +++++++++++++++++++++++++++
>>>>    3 files changed, 92 insertions(+), 2 deletions(-)
>>>>
>>> Remove "drivers/" from the commit title.
>>>
>>> $ git log --oneline drivers/iommu/iommu.c
>> My bad, I will fix it.
>>
>>> [...]
>>>> diff --git a/include/linux/iommu-debug-pagealloc.h b/include/linux/iommu-debug-pagealloc.h
>>>> index 83e64d70bf6c..454303ec09c2 100644
>>>> --- a/include/linux/iommu-debug-pagealloc.h
>>>> +++ b/include/linux/iommu-debug-pagealloc.h
>>>> @@ -8,10 +8,66 @@
>>>>    #ifndef __LINUX_IOMMU_DEBUG_PAGEALLOC_H
>>>>    #define __LINUX_IOMMU_DEBUG_PAGEALLOC_H
>>>> +struct iommu_domain;
>>>> +
>>>>    #ifdef CONFIG_IOMMU_DEBUG_PAGEALLOC
>>>> +DECLARE_STATIC_KEY_FALSE(iommu_debug_initialized);
>>>> +
>>>>    extern struct page_ext_operations page_iommu_debug_ops;
>>>> +void __iommu_debug_map(struct iommu_domain *domain, phys_addr_t phys,
>>>> +                  size_t size);
>>>> +void __iommu_debug_unmap_begin(struct iommu_domain *domain,
>>>> +                          unsigned long iova, size_t size);
>>>> +void __iommu_debug_unmap_end(struct iommu_domain *domain,
>>>> +                        unsigned long iova, size_t size, size_t unmapped);
>>>> +
>>>> +static inline void iommu_debug_map(struct iommu_domain *domain,
>>>> +                              phys_addr_t phys, size_t size)
>>>> +{
>>>> +   if (static_branch_unlikely(&iommu_debug_initialized))
>>>> +           __iommu_debug_map(domain, phys, size);
>>>> +}
>>>> +
>>>> +static inline void iommu_debug_unmap_begin(struct iommu_domain *domain,
>>>> +                                      unsigned long iova, size_t size)
>>>> +{
>>>> +   if (static_branch_unlikely(&iommu_debug_initialized))
>>>> +           __iommu_debug_unmap_begin(domain, iova, size);
>>>> +}
>>>> +
>>>> +static inline void iommu_debug_unmap_end(struct iommu_domain *domain,
>>>> +                                    unsigned long iova, size_t size,
>>>> +                                    size_t unmapped)
>>>> +{
>>>> +   if (static_branch_unlikely(&iommu_debug_initialized))
>>>> +           __iommu_debug_unmap_end(domain, iova, size, unmapped);
>>>> +}
>>>> +
>>>> +void iommu_debug_init(void);
>>>> +
>>>> +#else
>>>> +static inline void iommu_debug_map(struct iommu_domain *domain,
>>>> +                              phys_addr_t phys, size_t size)
>>>> +{
>>>> +}
>>>> +
>>>> +static inline void iommu_debug_unmap_begin(struct iommu_domain *domain,
>>>> +                                      unsigned long iova, size_t size)
>>>> +{
>>>> +}
>>>> +
>>>> +static inline void iommu_debug_unmap_end(struct iommu_domain *domain,
>>>> +                                    unsigned long iova, size_t size,
>>>> +                                    size_t unmapped)
>>>> +{
>>>> +}
>>>> +
>>>> +static inline void iommu_debug_init(void)
>>>> +{
>>>> +}
>>> I suppose that all these should go to drivers/iommu/iommu-priv.h, as
>>> they are for use in other files inside the IOMMU subsystem.
>> It seemed better to have all the feature functions/declarations in one
>> isolated file, as it is included outside of the iommu susbsystem also.
>> I have no strong opinion, I can keep them in drivers/iommu/iommu-priv.h
>> if you think it's better. But then we will have to include also
>> "iommu-debug-pagealloc.h" for the static key to avoid including extra
>> files to linux/mm.h.
>>
>> Thanks,
>> Mostafa
>>
> Hi Baolu,
> 
> I plan to post a new version soon. So far, I only have the commit
> subject rewrites. Do you have any opinion on the header file split or
> the unmap size part?

I have no option regarding the unmap size part, as the comments already
declare the possible cases. For the header file split, I feel it's
better to put staff that is only used in the IOMMU subsystem in drivers/
iommu in iommu-pri.h instead of the global ones under include/linux.

Thanks,
baolu