Re: [PATCH v2 01/20] memory: Adjust event ranges to fit within notifier boundaries

[Alejandro Jimenez]

On 6/12/25 2:54 AM, Vasant Hegde wrote:
Alejandro,


On 5/2/2025 7:45 AM, Alejandro Jimenez wrote:
Invalidating the entire address space (i.e. range of [0, ~0ULL]) is a
valid and required operation by vIOMMU implementations. However, such
invalidations currently trigger an assertion unless they originate from
device IOTLB invalidations.

Although in recent Linux guests this case is not exercised by the VTD
implementation due to various optimizations, the assertion will be hit
by upcoming AMD vIOMMU changes to support DMA address translation. More
specifically, when running a Linux guest with VFIO passthrough device,
and a kernel that does not contain commmit 3f2571fed2fa ("iommu/amd:
Remove redundant domain flush from attach_device()").

FYI. Its easy to send invalidate all without above commit (as it does it in
every attach), there are other paths where kernel will still send invalidate
all.. Like detaching/attaching device, etc.


ACK, thank you for the comment/confirmation. I just mentioned an easily 
reproducible case where it causes a crash during guest boot with just 
passing a VF to the guest in the QEMU cmdline, not hoptplug or other 
more complex scenarios required. Your argument above is part of why I 
chose to remove the assertion completely instead of special casing and 
breaking down this range on the amd_iommu.c code handling the invalidation.

Thank you,
Alejandro


-Vasant




Remove the assertion altogether and adjust the range to ensure it does
not cross notifier boundaries.

Signed-off-by: Alejandro Jimenez <alejandro.j.jime...@oracle.com>
---
  system/memory.c | 10 +++-------
  1 file changed, 3 insertions(+), 7 deletions(-)

diff --git a/system/memory.c b/system/memory.c
index 71434e7ad02c..7ad2fc098341 100644
--- a/system/memory.c
+++ b/system/memory.c
@@ -2021,13 +2021,9 @@ void memory_region_notify_iommu_one(IOMMUNotifier 
*notifier,
          return;
      }
  
-    if (notifier->notifier_flags & IOMMU_NOTIFIER_DEVIOTLB_UNMAP) {
-        /* Crop (iova, addr_mask) to range */
-        tmp.iova = MAX(tmp.iova, notifier->start);
-        tmp.addr_mask = MIN(entry_end, notifier->end) - tmp.iova;
-    } else {
-        assert(entry->iova >= notifier->start && entry_end <= notifier->end);
-    }
+    /* Crop (iova, addr_mask) to range */
+    tmp.iova = MAX(tmp.iova, notifier->start);
+    tmp.addr_mask = MIN(entry_end, notifier->end) - tmp.iova;
  
      if (event->type & notifier->notifier_flags) {
          notifier->notify(notifier, &tmp);