qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Eric Auger <eric.auger@redhat.com>
To: eric.auger.pro@gmail.com, eric.auger@redhat.com,
	qemu-devel@nongnu.org, qemu-arm@nongnu.org, mst@redhat.com,
	jean-philippe@linaro.org, peter.maydell@linaro.org,
	clg@redhat.com, zhenzhong.duan@intel.com,
	alex.williamson@redhat.com, jasowang@redhat.com
Cc: yanghliu@redhat.com
Subject: [PATCH 1/6] Revert "virtio-iommu: Clear IOMMUDevice when VFIO device is unplugged"
Date: Tue, 16 Jul 2024 11:45:03 +0200	[thread overview]
Message-ID: <20240716094619.1713905-2-eric.auger@redhat.com> (raw)
In-Reply-To: <20240716094619.1713905-1-eric.auger@redhat.com>

This reverts commit 1b889d6e39c32d709f1114699a014b381bcf1cb1.
There are different problems with that tentative fix:
- Some resources are left dangling (resv_regions,
  host_resv_ranges) and memory subregions are left attached to
  the root MR although freed as embedded in the sdev IOMMUDevice.
  Finally the sdev->as is not destroyed and associated listeners
  are left.
- Even when fixing the above we observe a memory corruption
  associated with the deallocation of the IOMMUDevice. This can
  be observed when a VFIO device is hotplugged, hot-unplugged
  and a system reset is issued. At this stage we have not been
  able to identify the root cause (IOMMU MR or as structs beeing
  overwritten and used later on?).
- Another issue is HostIOMMUDevice are indexed by non aliased
  BDF whereas the IOMMUDevice is indexed by aliased BDF - yes the
  current naming is really misleading -. Given the state of the
  code I don't think the virtio-iommu device works in non
  singleton group case though.

So let's revert the patch for now. This means the IOMMU MR/as survive
the hotunplug. This is what is done in the intel_iommu for instance.
It does not sound very logical to keep those but currently there is
no symetric function to pci_device_iommu_address_space().

probe_done issue will be handled in a subsequent patch. Also
resv_regions and host_resv_regions will be deallocated separately.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
---
 hw/virtio/virtio-iommu.c | 21 ---------------------
 1 file changed, 21 deletions(-)

diff --git a/hw/virtio/virtio-iommu.c b/hw/virtio/virtio-iommu.c
index 33ae61c4a6..4e34dacd6e 100644
--- a/hw/virtio/virtio-iommu.c
+++ b/hw/virtio/virtio-iommu.c
@@ -467,26 +467,6 @@ static AddressSpace *virtio_iommu_find_add_as(PCIBus *bus, void *opaque,
     return &sdev->as;
 }
 
-static void virtio_iommu_device_clear(VirtIOIOMMU *s, PCIBus *bus, int devfn)
-{
-    IOMMUPciBus *sbus = g_hash_table_lookup(s->as_by_busptr, bus);
-    IOMMUDevice *sdev;
-
-    if (!sbus) {
-        return;
-    }
-
-    sdev = sbus->pbdev[devfn];
-    if (!sdev) {
-        return;
-    }
-
-    g_list_free_full(sdev->resv_regions, g_free);
-    sdev->resv_regions = NULL;
-    g_free(sdev);
-    sbus->pbdev[devfn] = NULL;
-}
-
 static gboolean hiod_equal(gconstpointer v1, gconstpointer v2)
 {
     const struct hiod_key *key1 = v1;
@@ -728,7 +708,6 @@ virtio_iommu_unset_iommu_device(PCIBus *bus, void *opaque, int devfn)
     }
 
     g_hash_table_remove(viommu->host_iommu_devices, &key);
-    virtio_iommu_device_clear(viommu, bus, devfn);
 }
 
 static const PCIIOMMUOps virtio_iommu_ops = {
-- 
2.41.0



  reply	other threads:[~2024-07-16  9:47 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-07-16  9:45 [PATCH 0/6] VIRTIO-IOMMU/VFIO: Revert IOMMUDevice clear and fix hotunplug Eric Auger
2024-07-16  9:45 ` Eric Auger [this message]
2024-07-16 13:05   ` [PATCH 1/6] Revert "virtio-iommu: Clear IOMMUDevice when VFIO device is unplugged" Cédric Le Goater
2024-07-16  9:45 ` [PATCH 2/6] virtio-iommu: Remove probe_done Eric Auger
2024-07-16 13:06   ` Cédric Le Goater
2024-07-16  9:45 ` [PATCH 3/6] virtio-iommu: Free [host_]resv_ranges on unset_iommu_devices Eric Auger
2024-07-16 13:36   ` Cédric Le Goater
2024-07-17  3:06   ` Duan, Zhenzhong
2024-07-17  7:40     ` Eric Auger
2024-07-17  7:56       ` Duan, Zhenzhong
2024-07-16  9:45 ` [PATCH 4/6] virtio-iommu: Remove the end point on detach Eric Auger
2024-07-16 13:36   ` Cédric Le Goater
2024-07-16  9:45 ` [PATCH 5/6] hw/vfio/common: Add vfio_listener_region_del_iommu trace event Eric Auger
2024-07-16 13:36   ` Cédric Le Goater
2024-07-16  9:45 ` [PATCH 6/6] virtio-iommu: Add trace point on virtio_iommu_detach_endpoint_from_domain Eric Auger
2024-07-16 13:36   ` Cédric Le Goater
2024-07-16 14:02 ` [PATCH 0/6] VIRTIO-IOMMU/VFIO: Revert IOMMUDevice clear and fix hotunplug Cédric Le Goater
2024-07-16 14:59   ` Eric Auger

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240716094619.1713905-2-eric.auger@redhat.com \
    --to=eric.auger@redhat.com \
    --cc=alex.williamson@redhat.com \
    --cc=clg@redhat.com \
    --cc=eric.auger.pro@gmail.com \
    --cc=jasowang@redhat.com \
    --cc=jean-philippe@linaro.org \
    --cc=mst@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=yanghliu@redhat.com \
    --cc=zhenzhong.duan@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).