| Message ID | 20250317235546.4546-2-dongli.zhang@oracle.com (mailing list archive) |
|---|---|
| State | Not Applicable |
| Headers | show |
| Series | vhost-scsi: log write descriptors for live migration (and three bugfix) | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Not a local patch |
On Tue, Mar 18, 2025 at 7:51 AM Dongli Zhang <dongli.zhang@oracle.com> wrote: > > The vhost-scsi completion path may access vq->log_base when vq->log_used is > already set to false. > > vhost-thread QEMU-thread > > vhost_scsi_complete_cmd_work() > -> vhost_add_used() > -> vhost_add_used_n() > if (unlikely(vq->log_used)) > QEMU disables vq->log_used > via VHOST_SET_VRING_ADDR. > mutex_lock(&vq->mutex); > vq->log_used = false now! > mutex_unlock(&vq->mutex); > > QEMU gfree(vq->log_base) > log_used() > -> log_write(vq->log_base) > > Assuming the VMM is QEMU. The vq->log_base is from QEMU userpace and can be > reclaimed via gfree(). As a result, this causes invalid memory writes to > QEMU userspace. > > The control queue path has the same issue. > > Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com> > --- Acked-by: Jason Wang <jasowang@redhat.com> Thanks
diff --git a/drivers/vhost/scsi.c b/drivers/vhost/scsi.c index f6f5a7ac7894..f846f2aa7c87 100644 --- a/drivers/vhost/scsi.c +++ b/drivers/vhost/scsi.c @@ -627,6 +627,9 @@ static void vhost_scsi_complete_cmd_work(struct vhost_work *work) int ret; llnode = llist_del_all(&svq->completion_list); + + mutex_lock(&svq->vq.mutex); + llist_for_each_entry_safe(cmd, t, llnode, tvc_completion_list) { se_cmd = &cmd->tvc_se_cmd; @@ -660,6 +663,8 @@ static void vhost_scsi_complete_cmd_work(struct vhost_work *work) vhost_scsi_release_cmd_res(se_cmd); } + mutex_unlock(&svq->vq.mutex); + if (signal) vhost_signal(&svq->vs->dev, &svq->vq); } @@ -1432,8 +1437,11 @@ static void vhost_scsi_tmf_resp_work(struct vhost_work *work) else resp_code = VIRTIO_SCSI_S_FUNCTION_REJECTED; + mutex_lock(&tmf->svq->vq.mutex); vhost_scsi_send_tmf_resp(tmf->vhost, &tmf->svq->vq, tmf->in_iovs, tmf->vq_desc, &tmf->resp_iov, resp_code); + mutex_unlock(&tmf->svq->vq.mutex); + vhost_scsi_release_tmf_res(tmf); }
The vhost-scsi completion path may access vq->log_base when vq->log_used is already set to false. vhost-thread QEMU-thread vhost_scsi_complete_cmd_work() -> vhost_add_used() -> vhost_add_used_n() if (unlikely(vq->log_used)) QEMU disables vq->log_used via VHOST_SET_VRING_ADDR. mutex_lock(&vq->mutex); vq->log_used = false now! mutex_unlock(&vq->mutex); QEMU gfree(vq->log_base) log_used() -> log_write(vq->log_base) Assuming the VMM is QEMU. The vq->log_base is from QEMU userpace and can be reclaimed via gfree(). As a result, this causes invalid memory writes to QEMU userspace. The control queue path has the same issue. Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com> --- Changed since v1: - Move lock to the begin and end of vhost_scsi_complete_cmd_work() as it is per-vq now. This reduces the number of mutex_lock(). - Move this bugfix patch to before dirty log tracking patches. drivers/vhost/scsi.c | 8 ++++++++ 1 file changed, 8 insertions(+)