Hi Zhenzhong,  

I traced the calls we have and I fear the flow is becoming a bit spaghetti.

```
vtd_do_iommu_translate
        |
         --> if (vtd_dev_pt_enabled) vtd_pt_enable_fast_path
                                                |
                                                v
                                    vtd_switch_address_space
                                        (updates the pasid
                                        entry based on the
                                        parameter it has
                                        been given)
                                                |
                                                v
                                        vtd_as_pt_enabled
                                          (drops the
                                          entry returned
                                          by vtd_dev_pt_enabled)
                                                |
                                                v
                                        vtd_dev_pt_enabled
```

It looks like we are checking things twice, ignoring the result on the second 
call and re-using the first one.

I think we are trying to gate the call to vtd_pt_enable_fast_path from 
vtd_do_iommu_translate whereas the callee could do the thing on its own.

What is your opinion about this?



On Wed, 2026-11-04 at 15:07 +0800, Zhenzhong Duan wrote:
> Caution: External email. Do not open attachments or click links, unless this 
> email comes from a known sender and you know the content is safe.
> 
> 
> With caching mode disabled, a guest can install a scalable-mode  
> pass-through PASID entry without issuing a PASID-cache invalidation. The  
> first DMA enables the PT fast path, but leaves QEMU's PASID cache invalid.
> 
> When the guest tears down that entry to switch back to a DMA domain, the  
> invalidation handler sees both a non-present guest entry and an invalid  
> cached entry and returns without re-enabling the IOMMU memory region.  
> Subsequent DMA therefore continues to bypass translation.
> 
> The issue can be reproduced with an emulated device in scalable mode  
> with caching-mode=off:
> 
> 1. Unbind the guest device driver and change the group's type to  
>    identity through /sys/kernel/iommu_groups/<group>/type.  
> 2. Rebind the driver and issue a DMA write.  
> 3. Unbind the driver, change the group's type back to DMA-FQ, and  
>    rebind the driver.  
> 4. Issue another DMA write using an IOVA mapped to a different GPA.
> 
> The second write incorrectly bypasses translation instead of reaching  
> the mapped GPA.
> 
> Save the PASID entry when enabling the scalable-mode PT fast path so  
> teardown invalidation can restore the IOMMU address space. Return after  
> handling a non-present entry, and refresh the cached entry when a present  
> entry changes.
> 
> Opportunistically clarify the invalidation modes in the fast-path comment.
> 
> Fixes: c9cf33075a4c ("intel_iommu: Fix DMA failure when guest switches IOMMU 
> domain")  
> Reported-by: Clement Mathieu--Drif 
> <[[email protected]](mailto:[email protected])>  
> Signed-off-by: Zhenzhong Duan 
> <[[email protected]](mailto:[email protected])>  
> Tested-by: Clement Mathieu--Drif 
> <[[email protected]](mailto:[email protected])>  
> ---  
>  hw/i386/intel_iommu.c | 32 ++++++++++++++++++++++++--------  
>  1 file changed, 24 insertions(+), 8 deletions(-)
> 
> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c  
> index b7da5d086d..2c5aa04f69 100644  
> --- a/hw/i386/intel_iommu.c  
> +++ b/hw/i386/intel_iommu.c  
> @@ -1649,7 +1649,7 @@ static int vtd_address_space_sync(VTDAddressSpace 
> *vtd_as)  
>   * or second stage translation, it depends on PGTT setting.  
>   */  
>  static bool vtd_dev_pt_enabled(IntelIOMMUState *s, VTDContextEntry *ce,  
> -                               uint32_t pasid)  
> +                               uint32_t pasid, VTDPASIDEntry *pasid_entry)  
>  {  
>      VTDPASIDEntry pe;  
>      int ret;  
> @@ -1663,6 +1663,11 @@ static bool vtd_dev_pt_enabled(IntelIOMMUState *s, 
> VTDContextEntry *ce,  
>               */  
>              return false;  
>          }  
> +  
> +        if (pasid_entry) {  
> +            *pasid_entry = pe;  
> +        }  
> +  
>          return vtd_pe_pgtt_is_pt(&pe);  
>      }
> 
> @@ -1689,7 +1694,7 @@ static bool vtd_as_pt_enabled(VTDAddressSpace *as)  
>          return false;  
>      }
> 
> -    return vtd_dev_pt_enabled(s, &ce, as->pasid);  
> +    return vtd_dev_pt_enabled(s, &ce, as->pasid, NULL);  
>  }
> 
>  /* Return whether the device is using IOMMU translation. */  
> @@ -1887,7 +1892,8 @@ VTDAddressSpace *vtd_get_as_by_sid(IntelIOMMUState *s, 
> uint16_t sid)  
>      return vtd_get_as_by_sid_and_pasid(s, sid, IOMMU_NO_PASID);  
>  }
> 
> -static void vtd_pt_enable_fast_path(IntelIOMMUState *s, uint16_t source_id)  
> +static void vtd_pt_enable_fast_path(IntelIOMMUState *s, uint16_t source_id,  
> +                                    VTDPASIDEntry *pe)  
>  {  
>      VTDAddressSpace *vtd_as;  
>      bool success = false;  
> @@ -1900,6 +1906,13 @@ static void vtd_pt_enable_fast_path(IntelIOMMUState 
> *s, uint16_t source_id)  
>      if (vtd_switch_address_space(vtd_as) == false) {  
>          /* We switched off IOMMU region successfully. */  
>          success = true;  
> +  
> +        if (s->root_scalable) {  
> +            VTDPASIDCacheEntry *pc_entry = &vtd_as->pasid_cache_entry;  
> +  
> +            pc_entry->pasid_entry = *pe;  
> +            pc_entry->valid = true;  
> +        }  
>      }
> 
>  out:  
> @@ -2091,6 +2104,7 @@ static bool vtd_do_iommu_translate(VTDAddressSpace 
> *vtd_as, PCIBus *bus,  
>  {  
>      IntelIOMMUState *s = vtd_as->iommu_state;  
>      VTDContextEntry ce;  
> +    VTDPASIDEntry pe;  
>      uint8_t bus_num = pci_bus_num(bus);  
>      VTDContextCacheEntry *cc_entry;  
>      uint64_t pte, page_mask;  
> @@ -2167,7 +2181,7 @@ static bool vtd_do_iommu_translate(VTDAddressSpace 
> *vtd_as, PCIBus *bus,  
>       * We don't need to translate for pass-through context entries.  
>       * Also, let's ignore IOTLB caching as well for PT devices.  
>       */  
> -    if (vtd_dev_pt_enabled(s, &ce, pasid)) {  
> +    if (vtd_dev_pt_enabled(s, &ce, pasid, &pe)) {  
>          entry->iova = addr & VTD_PAGE_MASK_4K;  
>          entry->translated_addr = entry->iova;  
>          entry->addr_mask = ~VTD_PAGE_MASK_4K;  
> @@ -2180,10 +2194,11 @@ static bool vtd_do_iommu_translate(VTDAddressSpace 
> *vtd_as, PCIBus *bus,  
>           * the device. Let's enable the fast path for passthrough.  
>           *  
>           * When passthrough is disabled again for the device, we can  
> -         * capture it via the context entry invalidation, then the  
> -         * IOMMU region can be swapped back.  
> +         * capture it via context-cache invalidation in legacy mode or  
> +         * PASID-cache invalidation in scalable mode, then the IOMMU  
> +         * region can be swapped back.  
>           */  
> -        vtd_pt_enable_fast_path(s, source_id);  
> +        vtd_pt_enable_fast_path(s, source_id, &pe);  
>          vtd_iommu_unlock(s);  
>          return true;  
>      }  
> @@ -3133,6 +3148,7 @@ static void vtd_pasid_cache_sync_locked(gpointer key, 
> gpointer value,  
>              vtd_address_space_unmap(vtd_as, n);  
>          }  
>          vtd_switch_address_space(vtd_as);  
> +        return;  
>      }
> 
>      /*  
> @@ -3158,11 +3174,11 @@ static void vtd_pasid_cache_sync_locked(gpointer key, 
> gpointer value,  
>      }
> 
>      if (!pc_entry->valid) {  
> -        pc_entry->pasid_entry = pe;  
>          pc_entry->valid = true;  
>      } else if (!vtd_pasid_entry_compare(&pe, &pc_entry->pasid_entry)) {  
>          return;  
>      }  
> +    pc_entry->pasid_entry = pe;
> 
>      vtd_switch_address_space(vtd_as);  
>      vtd_address_space_sync(vtd_as);  
> --  
> 2.52.0
> 

Reply via email to