vmd(8): only add vionet tap read events after a notify event.

My recent refactor introduced a bug where the vionet device will
add the tap(4) read event on an unpause regardless of driver and
device state, causing the cpu to peg as the read event fires
indefinitely because the guest's virtqueue isn't ready to receive
the data.

Add in a global flag that tracks if the rx-side is enabled or not
to coordinate adding the tap read event.

ok mlarkin@

diff --git a/usr.sbin/vmd/vionet.c b/usr.sbin/vmd/vionet.c
index f9c0c13..f4bb1a9 100644 (file)
--- a/usr.sbin/vmd/vionet.c
+++ b/usr.sbin/vmd/vionet.c
@@ -1,4 +1,4 @@
-/*     $OpenBSD: vionet.c,v 1.13 2024/02/20 21:40:37 dv Exp $  */
+/*     $OpenBSD: vionet.c,v 1.14 2024/02/22 02:38:53 dv Exp $  */
 
 /*
  * Copyright (c) 2023 Dave Voutila <dv@openbsd.org>
@@ -89,9 +89,8 @@ int pipe_inject[2];
 struct iovec iov_rx[VIONET_QUEUE_SIZE];
 struct iovec iov_tx[VIONET_QUEUE_SIZE];
 pthread_rwlock_t lock = NULL;          /* Guards device config state. */
-
-/* Transient reset state used by the main thread to coordinate device reset. */
-int resetting = 0;
+int resetting = 0;     /* Transient reset state used to coordinate reset. */
+int rx_enabled = 0;    /* 1: we expect to read the tap, 0: wait for notify. */
 
 __dead void
 vionet_main(int fd, int fd_vmm)
@@ -693,6 +692,7 @@ vionet_notifyq(struct virtio_dev *dev)
 
        switch (vionet->cfg.queue_notify) {
        case RXQ:
+               rx_enabled = 1;
                vm_pipe_send(&pipe_rx, VIRTIO_NOTIFY);
                break;
        case TXQ:
@@ -900,7 +900,6 @@ dev_dispatch_vm(int fd, short event, void *arg)
        struct imsg              imsg;
        ssize_t                  n = 0;
        int                      verbose;
-       uint8_t                  status = 0;
 
        if (dev == NULL)
                fatalx("%s: missing vionet pointer", __func__);
@@ -948,11 +947,7 @@ dev_dispatch_vm(int fd, short event, void *arg)
                        break;
                case IMSG_VMDOP_UNPAUSE_VM:
                        log_debug("%s: unpausing", __func__);
-                       pthread_rwlock_rdlock(&lock);
-                       status = vionet->cfg.device_status &
-                           VIRTIO_CONFIG_DEVICE_STATUS_DRIVER_OK;
-                       pthread_rwlock_unlock(&lock);
-                       if (status)
+                       if (rx_enabled)
                                vm_pipe_send(&pipe_rx, VIRTIO_THREAD_START);
                        break;
                case IMSG_CTL_VERBOSE:
@@ -1092,6 +1087,7 @@ handle_io_write(struct viodev_msg *msg, struct virtio_dev *dev)
 
        pthread_rwlock_unlock(&lock);
        if (pause_devices) {
+               rx_enabled = 0;
                vionet_deassert_pic_irq(dev);
                vm_pipe_send(&pipe_rx, VIRTIO_THREAD_PAUSE);
                vm_pipe_send(&pipe_tx, VIRTIO_THREAD_PAUSE);