[PATCH v1 04/19] intel_iommu: Fill the PASID field when creating an IOMMUTLBEntry

CLEMENT MATHIEU--DRIF posted 19 patches 3 weeks, 6 days ago
[PATCH v1 04/19] intel_iommu: Fill the PASID field when creating an IOMMUTLBEntry
Posted by CLEMENT MATHIEU--DRIF 3 weeks, 6 days ago
From: Clément Mathieu--Drif <clement.mathieu--drif@eviden.com>

PASID value must be used by devices as a key (or part of a key)
when populating their ATC with the IOTLB entries returned by the IOMMU.

Signed-off-by: Clément Mathieu--Drif <clement.mathieu--drif@eviden.com>
---
 hw/i386/intel_iommu.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
index aad132e367..a92ef9fe74 100644
--- a/hw/i386/intel_iommu.c
+++ b/hw/i386/intel_iommu.c
@@ -2119,6 +2119,9 @@ static bool vtd_do_iommu_translate(VTDAddressSpace *vtd_as, PCIBus *bus,
 
     vtd_iommu_lock(s);
 
+    /* fill the pasid before getting rid2pasid */
+    entry->pasid = pasid;
+
     cc_entry = &vtd_as->context_cache_entry;
 
     /* Try to fetch pte from IOTLB, we don't need RID2PASID logic */
@@ -2260,6 +2263,7 @@ out:
     entry->translated_addr = vtd_get_pte_addr(pte, s->aw_bits) & page_mask;
     entry->addr_mask = ~page_mask;
     entry->perm = access_flags;
+    /* pasid already set */
     return true;
 
 error:
@@ -2268,6 +2272,7 @@ error:
     entry->translated_addr = 0;
     entry->addr_mask = 0;
     entry->perm = IOMMU_NONE;
+    entry->pasid = PCI_NO_PASID;
     return false;
 }
 
@@ -2511,6 +2516,7 @@ static void vtd_iotlb_page_invalidate_notify(IntelIOMMUState *s,
                         .translated_addr = 0,
                         .addr_mask = size - 1,
                         .perm = IOMMU_NONE,
+                        .pasid = pasid,
                     },
                 };
                 memory_region_notify_iommu(&vtd_as->iommu, 0, event);
@@ -3098,6 +3104,7 @@ static void do_invalidate_device_tlb(VTDAddressSpace *vtd_dev_as,
     event.entry.iova = addr;
     event.entry.perm = IOMMU_NONE;
     event.entry.translated_addr = 0;
+    event.entry.pasid = vtd_dev_as->pasid;
     memory_region_notify_iommu(&vtd_dev_as->iommu, 0, event);
 }
 
@@ -3680,6 +3687,7 @@ static IOMMUTLBEntry vtd_iommu_translate(IOMMUMemoryRegion *iommu, hwaddr addr,
     IOMMUTLBEntry iotlb = {
         /* We'll fill in the rest later. */
         .target_as = &address_space_memory,
+        .pasid = vtd_as->pasid,
     };
     bool success;
 
@@ -3692,6 +3700,7 @@ static IOMMUTLBEntry vtd_iommu_translate(IOMMUMemoryRegion *iommu, hwaddr addr,
         iotlb.translated_addr = addr & VTD_PAGE_MASK_4K;
         iotlb.addr_mask = ~VTD_PAGE_MASK_4K;
         iotlb.perm = IOMMU_RW;
+        iotlb.pasid = PCI_NO_PASID;
         success = true;
     }
 
-- 
2.47.0
RE: [PATCH v1 04/19] intel_iommu: Fill the PASID field when creating an IOMMUTLBEntry
Posted by Duan, Zhenzhong 3 days, 11 hours ago
Hi Clement,

>-----Original Message-----
>From: CLEMENT MATHIEU--DRIF <clement.mathieu--drif@eviden.com>
>Subject: [PATCH v1 04/19] intel_iommu: Fill the PASID field when creating an
>IOMMUTLBEntry
>
>From: Clément Mathieu--Drif <clement.mathieu--drif@eviden.com>
>
>PASID value must be used by devices as a key (or part of a key)
>when populating their ATC with the IOTLB entries returned by the IOMMU.
>
>Signed-off-by: Clément Mathieu--Drif <clement.mathieu--drif@eviden.com>
>---
> hw/i386/intel_iommu.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
>diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
>index aad132e367..a92ef9fe74 100644
>--- a/hw/i386/intel_iommu.c
>+++ b/hw/i386/intel_iommu.c
>@@ -2119,6 +2119,9 @@ static bool vtd_do_iommu_translate(VTDAddressSpace
>*vtd_as, PCIBus *bus,
>
>     vtd_iommu_lock(s);
>
>+    /* fill the pasid before getting rid2pasid */
>+    entry->pasid = pasid;

Seems unnecessary because vtd_iommu_translate() already assigned it.

>+
>     cc_entry = &vtd_as->context_cache_entry;
>
>     /* Try to fetch pte from IOTLB, we don't need RID2PASID logic */
>@@ -2260,6 +2263,7 @@ out:
>     entry->translated_addr = vtd_get_pte_addr(pte, s->aw_bits) & page_mask;
>     entry->addr_mask = ~page_mask;
>     entry->perm = access_flags;
>+    /* pasid already set */
>     return true;
>
> error:
>@@ -2268,6 +2272,7 @@ error:
>     entry->translated_addr = 0;
>     entry->addr_mask = 0;
>     entry->perm = IOMMU_NONE;
>+    entry->pasid = PCI_NO_PASID;

Shouldn't we keep original pasid value?

>     return false;
> }
>
>@@ -2511,6 +2516,7 @@ static void
>vtd_iotlb_page_invalidate_notify(IntelIOMMUState *s,
>                         .translated_addr = 0,
>                         .addr_mask = size - 1,
>                         .perm = IOMMU_NONE,
>+                        .pasid = pasid,

Shouldn't we assign vtd_dev_as->pasid?

>                     },
>                 };
>                 memory_region_notify_iommu(&vtd_as->iommu, 0, event);
>@@ -3098,6 +3104,7 @@ static void do_invalidate_device_tlb(VTDAddressSpace
>*vtd_dev_as,
>     event.entry.iova = addr;
>     event.entry.perm = IOMMU_NONE;
>     event.entry.translated_addr = 0;
>+    event.entry.pasid = vtd_dev_as->pasid;
>     memory_region_notify_iommu(&vtd_dev_as->iommu, 0, event);
> }
>
>@@ -3680,6 +3687,7 @@ static IOMMUTLBEntry
>vtd_iommu_translate(IOMMUMemoryRegion *iommu, hwaddr addr,
>     IOMMUTLBEntry iotlb = {
>         /* We'll fill in the rest later. */
>         .target_as = &address_space_memory,
>+        .pasid = vtd_as->pasid,
>     };
>     bool success;
>
>@@ -3692,6 +3700,7 @@ static IOMMUTLBEntry
>vtd_iommu_translate(IOMMUMemoryRegion *iommu, hwaddr addr,
>         iotlb.translated_addr = addr & VTD_PAGE_MASK_4K;
>         iotlb.addr_mask = ~VTD_PAGE_MASK_4K;
>         iotlb.perm = IOMMU_RW;
>+        iotlb.pasid = PCI_NO_PASID;

Shouldn't we keep earlier assigned pasid value as above?

Thanks
Zhenzhong
Re: [PATCH v1 04/19] intel_iommu: Fill the PASID field when creating an IOMMUTLBEntry
Posted by CLEMENT MATHIEU--DRIF 2 days, 11 hours ago
Hi zhenzhong,


On 23/12/2024 06:50, Duan, Zhenzhong 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.
>
>
> Hi Clement,
>
>> -----Original Message-----
>> From: CLEMENT MATHIEU--DRIF <clement.mathieu--drif@eviden.com>
>> Subject: [PATCH v1 04/19] intel_iommu: Fill the PASID field when creating an
>> IOMMUTLBEntry
>>
>> From: Clément Mathieu--Drif <clement.mathieu--drif@eviden.com>
>>
>> PASID value must be used by devices as a key (or part of a key)
>> when populating their ATC with the IOTLB entries returned by the IOMMU.
>>
>> Signed-off-by: Clément Mathieu--Drif <clement.mathieu--drif@eviden.com>
>> ---
>> hw/i386/intel_iommu.c | 9 +++++++++
>> 1 file changed, 9 insertions(+)
>>
>> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
>> index aad132e367..a92ef9fe74 100644
>> --- a/hw/i386/intel_iommu.c
>> +++ b/hw/i386/intel_iommu.c
>> @@ -2119,6 +2119,9 @@ static bool vtd_do_iommu_translate(VTDAddressSpace
>> *vtd_as, PCIBus *bus,
>>
>>      vtd_iommu_lock(s);
>>
>> +    /* fill the pasid before getting rid2pasid */
>> +    entry->pasid = pasid;
> Seems unnecessary because vtd_iommu_translate() already assigned it.

You are right

>
>> +
>>      cc_entry = &vtd_as->context_cache_entry;
>>
>>      /* Try to fetch pte from IOTLB, we don't need RID2PASID logic */
>> @@ -2260,6 +2263,7 @@ out:
>>      entry->translated_addr = vtd_get_pte_addr(pte, s->aw_bits) & page_mask;
>>      entry->addr_mask = ~page_mask;
>>      entry->perm = access_flags;
>> +    /* pasid already set */
>>      return true;
>>
>> error:
>> @@ -2268,6 +2272,7 @@ error:
>>      entry->translated_addr = 0;
>>      entry->addr_mask = 0;
>>      entry->perm = IOMMU_NONE;
>> +    entry->pasid = PCI_NO_PASID;
> Shouldn't we keep original pasid value?

will do

>
>>      return false;
>> }
>>
>> @@ -2511,6 +2516,7 @@ static void
>> vtd_iotlb_page_invalidate_notify(IntelIOMMUState *s,
>>                          .translated_addr = 0,
>>                          .addr_mask = size - 1,
>>                          .perm = IOMMU_NONE,
>> +                        .pasid = pasid,
> Shouldn't we assign vtd_dev_as->pasid?

will do

>
>>                      },
>>                  };
>>                  memory_region_notify_iommu(&vtd_as->iommu, 0, event);
>> @@ -3098,6 +3104,7 @@ static void do_invalidate_device_tlb(VTDAddressSpace
>> *vtd_dev_as,
>>      event.entry.iova = addr;
>>      event.entry.perm = IOMMU_NONE;
>>      event.entry.translated_addr = 0;
>> +    event.entry.pasid = vtd_dev_as->pasid;
>>      memory_region_notify_iommu(&vtd_dev_as->iommu, 0, event);
>> }
>>
>> @@ -3680,6 +3687,7 @@ static IOMMUTLBEntry
>> vtd_iommu_translate(IOMMUMemoryRegion *iommu, hwaddr addr,
>>      IOMMUTLBEntry iotlb = {
>>          /* We'll fill in the rest later. */
>>          .target_as = &address_space_memory,
>> +        .pasid = vtd_as->pasid,
>>      };
>>      bool success;
>>
>> @@ -3692,6 +3700,7 @@ static IOMMUTLBEntry
>> vtd_iommu_translate(IOMMUMemoryRegion *iommu, hwaddr addr,
>>          iotlb.translated_addr = addr & VTD_PAGE_MASK_4K;
>>          iotlb.addr_mask = ~VTD_PAGE_MASK_4K;
>>          iotlb.perm = IOMMU_RW;
>> +        iotlb.pasid = PCI_NO_PASID;
> Shouldn't we keep earlier assigned pasid value as above?

will do


>
> Thanks
> Zhenzhong

I'll wait for more reviews before sending a v2

Thanks
cmd