Re: [PATCH v2 00/17] Add Nested Translation Support for SMMUv3

From: zhangfei gao
Date: Thu May 25 2023 - 21:59:19 EST

在 2023/5/26 07:42, Nicolin Chen 写道:
On Tue, May 16, 2023 at 11:12:44AM +0800, Zhangfei Gao wrote:

However when debugging hotplug PCI device, it still does not work,
Segmentation fault same as 6.2.

guest kernel

boot guest (this info does not appear in 6.2)
qemu-system-aarch64: -device
Failed to set data -1
qemu-system-aarch64: -device
failed to set device data
Hmm.. I wonder what fails the set_dev_data ioctl...
Simply debug, it is because dev_data.sid=0, causing
arm_smmu_set_dev_user_data fail
I found that too. The input pci bus number is 1, yet the in
the context of set_dev_data, the pci bus number is 0, which
resulted in a 0-valued sid. I will take another look to get

$ sudo nc -U /tmp/qmpm_1.socket
(qemu) info pci
(qemu) device_del acc1

qemu-system-aarch64: IOMMU_IOAS_UNMAP failed: No such file or directory
qemu-system-aarch64: vfio_container_dma_unmap(0xaaaae1fc0380,
0x8000000000, 0x10000) = -2 (No such file or directory)
From ex-email reply
(Eric) In qemu arm virt machine 0x8000000000 matches the PCI MMIO region.
(Yi) Currently, iommufd kernel part doesn't support mapping device BAR MMIO.
This is a known gap.

qemu-system-aarch64: Failed to unset data -1
Segmentation fault (core dumped). // also happened in 6.2
Hmm, would it be possible for you to run the test again by
adding the following tracers to your QEMU command?
--trace "iommufd*" \
--trace "smmu*" \
--trace "vfio_*" \
--trace "pci_*"

Have sent you the log directly, since it is too big.
I have found two missing pieces in the device detach routine.
Applying the following should fix the crash at hotplug path.

diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 89a256efa999..2344307523cb 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -151,8 +151,10 @@ void vfio_container_destroy(VFIOContainer *container)

QLIST_FOREACH_SAFE(giommu, &container->giommu_list, giommu_next, tmp) {
- memory_region_unregister_iommu_notifier(
- MEMORY_REGION(giommu->iommu_mr), &giommu->n);
+ if (giommu->n.notifier_flags) {
+ memory_region_unregister_iommu_notifier(
+ MEMORY_REGION(giommu->iommu_mr), &giommu->n);
+ }
QLIST_REMOVE(giommu, giommu_next);
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index 844c60892db2..35d31480390d 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -652,6 +652,9 @@ found:
if (QLIST_EMPTY(&container->hwpt_list)) {
vfio_as_del_container(space, bcontainer);
+ if (bcontainer->nested) {
+ memory_listener_unregister(& bcontainer->prereg_listener);
+ }
__vfio_device_detach_container(vbasedev, container, &err);
if (err) {

Would you please try your case with it?

Yes, this solve the hotplug segmentation fault

Still report

qemu-system-aarch64: IOMMU_IOAS_UNMAP failed: No such file or directory
qemu-system-aarch64: vfio_container_dma_unmap(0xaaaae622e300, 0x8000000000, 0x10000) = -2 (No such file or directory)
qemu-system-aarch64: Failed to unset data -1 (only the first time of device_del)

Test with device_del and device_add