> > Hello Jinqian > > On 3/25/25 09:00, yangjinqian via wrote: > > Hi, > > > > I'm observing intermittent failures when enabling the HNS3 network > > port in VM using > > > > QEMU with the mainline kernel. > > > Would you mind opening a gitlab issue with more details on the context, host, > guest, HW ? > > https://gitlab.com/qemu-project/qemu/-/issues/ > > Thanks, > > C. > > Hi, Cédric
I've added the details to the gitlab issue. https://gitlab.com/qemu-project/qemu/-/issues/2884 thanks, Jinqian > > > > HNS3 drive in kernel: > > > > static int hns3_nic_net_up(struct net_device *netdev) > > > > { > > > > ...... > > > > /* enable the vectors */ > > > > for (i = 0; i < vector_num; i++) > > > > { > > > > napi_enable(napi); > > > > enable_irq(vector_irq); > > > > hns3_mask_vector_irq(tqp_vector, 1); > > > > } > > > > ...... > > > > } > > > > When the VM has two vCPUs, the guest HNS3 driver performs two > > sequential > > > > enable_irq operations (vector_num=2). The enable_irq will trap into > > KVM for interrupt > > > > configuration and exit to QEMU for PCI device emulation. When > > emulating interrupt > > > > enabling in QEMU, vfio_[intx/msi/msix]_enable calls > > vfio_disable_interrupts to disable > > > > all interrupts on the vdev. > > > > The questions are: > > > > 1.Why does the vfio-pci disable all interrupts of the device before enabling > irqs? > > > > 2.Is this a normal vfio-pci process or a problem? > > > > Thanks > > > > Jinqian > > > > Would >
