Merge pull request #35 from xez/vmnet-tap

Fix and reenable optional TAP device backend for virtio-net

Makefile

@@ -50,7 +50,8 @@ XHYVE_SRC := \
 	src/pci_lpc.c \
 	src/pci_uart.c \
 	src/pci_virtio_block.c \
-	src/pci_virtio_vmnet.c \
+	src/pci_virtio_net_tap.c \
+	src/pci_virtio_net_vmnet.c \
 	src/pci_virtio_rnd.c \
 	src/pm.c \
 	src/post.c \

README.md

@@ -156,6 +156,18 @@ If you want the same IP address across VM reboots, assign a UUID to a particular
 
     $ xhyve [-U uuid]
 
+**Optional:**
+
+If you need more advanced networking and already have a configured [TAP](http://tuntaposx.sourceforge.net) device you can use it with:
+
+	virtio-tap,tapX
+
+instead of:
+
+    virtio-net
+
+Where *X* is your tap device, i.e. */dev/tapX*.
+
 Issues
 ------
 If you are, or were, running any version of VirtualBox, prior to 5.0, and
@@ -174,25 +186,26 @@ TODO
 ----
 
 - vmm:
-  - enable APIC access page to speed up APIC emulation
-  - enable x2APIC MSRs (even faster)
+  - enable APIC access page to speed up APIC emulation (**performance**)
+  - enable x2APIC MSRs (even faster) (**performance**)
   - vmm_callout:
       - is a quick'n'dirty implementation of the FreeBSD kernel callout mechanism
       - seems to be racy
       - fix races or perhaps replace with something better
-      - use per vCPU timer event thread (performance)?
+      - use per vCPU timer event thread (**performance**)?
+      - use hardware VMX preemption timer instead of `pthread_cond_wait` (**performance**)
   - some 32-bit guests are broken (support PAE paging in VMCS)
-  - PCID guest support (performance)
+  - PCID guest support (**performance**)
 - block_if:
-  - OS X does not support preadv/pwritev, we need to serialize reads and writes for the time being until we find a better solution.
+  - OS X does not support `preadv`/`pwritev`, we need to serialize reads and writes for the time being until we find a better solution. (**performance**)
   - support block devices other than plain files
 - virtio_net:
-  - make it not require root
   - unify TAP and vmnet backends
-  - performance: send/receive more than a single packet at a time
+  - vmnet: make it not require root
+  - vmnet: send/receive more than a single packet at a time (**performance**)
 - virtio_rnd:
   - is untested
 - remove explicit state transitions:
-  - since only the owning task/thread can modify the VM/vCPUs a lot of the synchronization might be unnecessary
+  - since only the owning task/thread can modify the VM/vCPUs a lot of the synchronization might be unnecessary (**performance**)
 - performance, performance and performance
 - remove vestigial code, cleanup

src/pci_virtio_net_tap.c

@@ -51,6 +51,8 @@
 #include <xhyve/mevent.h>
 #include <xhyve/virtio.h>
 
+#define USE_MEVENT 0
+
 #define VTNET_RINGSZ 1024
 #define VTNET_MAXSEGS 32
 
@@ -370,6 +372,7 @@ pci_vtnet_tap_rx(struct pci_vtnet_softc *sc)
 	vq_endchains(vq, 1);
 }
 
+#if USE_MEVENT
 static void
 pci_vtnet_tap_callback(UNUSED int fd, UNUSED enum ev_type type, void *param)
 {
@@ -383,6 +386,37 @@ pci_vtnet_tap_callback(UNUSED int fd, UNUSED enum ev_type type, void *param)
 
 }
 
+#else /* !USE_MEVENT */
+
+static void *
+pci_vtnet_tap_select_func(void *vsc) {
+	struct pci_vtnet_softc *sc;
+	fd_set rfd;
+
+	sc = vsc;
+
+	assert(sc);
+	assert(sc->vsc_tapfd != -1);
+
+	FD_ZERO(&rfd);
+	FD_SET(sc->vsc_tapfd, &rfd);
+
+	while (1) {
+		if (select((sc->vsc_tapfd + 1), &rfd, NULL, NULL, NULL) == -1) {
+			abort();
+		}
+
+		pthread_mutex_lock(&sc->rx_mtx);
+		sc->rx_in_progress = 1;
+		pci_vtnet_tap_rx(sc);
+		sc->rx_in_progress = 0;
+		pthread_mutex_unlock(&sc->rx_mtx);
+	}
+
+	return (NULL);
+}
+#endif
+
 static void
 pci_vtnet_ping_rxq(void *vsc, struct vqueue_info *vq)
 {
@@ -542,6 +576,9 @@ pci_vtnet_init(struct pci_devinst *pi, char *opts)
 	char *devname;
 	char *vtopts;
 	int mac_provided;
+#if !USE_MEVENT
+	pthread_t sthrd;
+#endif
 
 	sc = calloc(1, sizeof(struct pci_vtnet_softc));
 
@@ -601,6 +638,7 @@ pci_vtnet_init(struct pci_devinst *pi, char *opts)
 				sc->vsc_tapfd = -1;
 			}
 
+#if USE_MEVENT
 			sc->vsc_mevp = mevent_add(sc->vsc_tapfd,
 						  EVF_READ,
 						  pci_vtnet_tap_callback,
@@ -610,7 +648,15 @@ pci_vtnet_init(struct pci_devinst *pi, char *opts)
 				close(sc->vsc_tapfd);
 				sc->vsc_tapfd = -1;
 			}
-		}		
+
+#else /* !USE_MEVENT */
+			if (pthread_create(&sthrd, NULL, pci_vtnet_tap_select_func, sc)) {
+				WPRINTF(("Could not create tap receive thread\n"));
+				close(sc->vsc_tapfd);
+				sc->vsc_tapfd = -1;
+			}
+#endif
+		}
 	}
 
 	/*
@@ -715,10 +761,10 @@ pci_vtnet_neg_features(void *vsc, uint64_t negotiated_features)
 	}
 }
 
-static struct pci_devemu pci_de_vnet = {
-	.pe_emu = 	"virtio-net",
+static struct pci_devemu pci_de_vnet_tap = {
+	.pe_emu = 	"virtio-tap",
 	.pe_init =	pci_vtnet_init,
 	.pe_barwrite =	vi_pci_write,
 	.pe_barread =	vi_pci_read
 };
-PCI_EMUL_SET(pci_de_vnet);
+PCI_EMUL_SET(pci_de_vnet_tap);

src/pci_virtio_net_vmnet.c

@@ -811,10 +811,10 @@ pci_vtnet_neg_features(void *vsc, uint64_t negotiated_features)
 	}
 }
 
-static struct pci_devemu pci_de_vnet = {
+static struct pci_devemu pci_de_vnet_vmnet = {
 	.pe_emu = 	"virtio-net",
 	.pe_init =	pci_vtnet_init,
 	.pe_barwrite =	vi_pci_write,
 	.pe_barread =	vi_pci_read
 };
-PCI_EMUL_SET(pci_de_vnet);
+PCI_EMUL_SET(pci_de_vnet_vmnet);