add viomb driver for virtio balloon device

This allows the host to request some memory back from the openbsd
guest in case of a shortage.

Patch by Dinar Talypov
OK jasper@
OK jmc@ (man page)

3 files changed, 469 insertions, 2 deletions

diff --git a/sys/arch/i386/conf/GENERIC b/sys/arch/i386/conf/GENERIC
index 546407f4017..190677ec1c7 100644
--- a/sys/arch/i386/conf/GENERIC
+++ b/sys/arch/i386/conf/GENERIC
@@ -1,4 +1,4 @@
-#	$OpenBSD: GENERIC,v 1.742 2012/11/29 23:38:25 stsp Exp $
+#	$OpenBSD: GENERIC,v 1.743 2013/01/12 13:02:22 sf Exp $
 #
 # For further information on compiling OpenBSD kernels, see the config(8)
 # man page.
@@ -801,3 +801,4 @@ pseudo-device	crypto		1
 virtio*		at pci? 	# Virtio PCI device
 vioblk*		at virtio?	# Virtio block device
 vio*		at virtio?	# Virtio network device
+viomb*		at virtio?	# Virtio memory ballooning device
diff --git a/sys/dev/pci/files.pci b/sys/dev/pci/files.pci
index 4f8ad7545f9..b4af38167a9 100644
--- a/sys/dev/pci/files.pci
+++ b/sys/dev/pci/files.pci
@@ -1,4 +1,4 @@
-#	$OpenBSD: files.pci,v 1.290 2012/11/29 23:36:34 stsp Exp $
+#	$OpenBSD: files.pci,v 1.291 2013/01/12 13:02:22 sf Exp $
 #	$NetBSD: files.pci,v 1.20 1996/09/24 17:47:15 christos Exp $
 #
 # Config file and device description for machine-independent PCI code.
@@ -841,3 +841,7 @@ file	dev/pci/if_vio.c		vio
 device	vioblk: scsi
 attach	vioblk at virtio
 file	dev/pci/vioblk.c		vioblk
+
+device	viomb
+attach	viomb at virtio
+file	dev/pci/viomb.c			viomb
diff --git a/sys/dev/pci/viomb.c b/sys/dev/pci/viomb.c
new file mode 100644
index 00000000000..4ad2b1f0322
--- /dev/null
+++ b/sys/dev/pci/viomb.c
@@ -0,0 +1,462 @@
+/* $OpenBSD: viomb.c,v 1.1 2013/01/12 13:02:22 sf Exp $	 */
+/* $NetBSD: viomb.c,v 1.1 2011/10/30 12:12:21 hannken Exp $	 */
+/*
+ * Copyright (c) 2012 Talypov Dinar <dinar@i-nk.ru>
+ * Copyright (c) 2010 Minoura Makoto.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
+ * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+ * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/malloc.h>
+#include <sys/device.h>
+#include <sys/workq.h>
+#include <sys/stdint.h>
+#include <uvm/uvm.h>
+#include <dev/pci/pcidevs.h>
+#include <dev/pci/pcivar.h>
+
+#include <dev/pci/virtioreg.h>
+#include <dev/pci/virtiovar.h>
+
+#if VIRTIO_PAGE_SIZE!=PAGE_SIZE
+#error non-4K page sizes are not supported yet
+#endif
+
+#define	DEVNAME(sc)	sc->sc_dev.dv_xname
+#if VIRTIO_DEBUG
+#define VIOMBDEBUG(sc, format, args...)	\
+		do { printf("%s: " format, sc->sc_dev.dv_xname, ##args);} \
+		while (0)
+#else
+#define VIOMBDEBUG(...)
+#endif
+
+/* flags used to specify kind of operation,
+ * actually should be moved to virtiovar.h
+ */
+#define VRING_READ		0
+#define VRING_WRITE		1
+
+/* notify or don't notify */
+#define VRING_NO_NOTIFY		0
+#define VRING_NOTIFY		1
+
+/* Configuration registers */
+#define VIRTIO_BALLOON_CONFIG_NUM_PAGES	0	/* 32bit */
+#define VIRTIO_BALLOON_CONFIG_ACTUAL	4	/* 32bit */
+
+/* Feature bits */
+#define VIRTIO_BALLOON_F_MUST_TELL_HOST (1<<0)
+#define VIRTIO_BALLOON_F_STATS_VQ	(1<<1)
+
+static const struct virtio_feature_name viomb_feature_names[] = {
+	{VIRTIO_BALLOON_F_MUST_TELL_HOST, "TellHost"},
+	{VIRTIO_BALLOON_F_STATS_VQ, "StatVQ"},
+	{0, NULL}
+};
+#define PGS_PER_REQ		256	/* 1MB, 4KB/page */
+#define VQ_INFLATE	0
+#define VQ_DEFLATE	1
+
+struct balloon_req {
+	bus_dmamap_t	bl_dmamap;
+	struct pglist	bl_pglist;
+	int		bl_nentries;
+	u_int32_t	bl_pages[PGS_PER_REQ];
+};
+
+struct viomb_softc {
+	struct device		sc_dev;
+	struct virtio_softc	*sc_virtio;
+	struct virtqueue	sc_vq[2];
+	u_int32_t		sc_npages;
+	u_int32_t		sc_actual;
+	struct balloon_req	sc_req;
+	struct workq_task	sc_task;
+	int			sc_workq_queued;
+	struct pglist		sc_balloon_pages;
+};
+
+int	viomb_match(struct device *, void *, void *);
+void	viomb_attach(struct device *, struct device *, void *);
+void	viomb_worker(void *, void *);
+void	viomb_inflate(struct viomb_softc *);
+void	viomb_deflate(struct viomb_softc *);
+int	viomb_config_change(struct virtio_softc *);
+void	viomb_read_config(struct viomb_softc *);
+int	viomb_vq_dequeue(struct virtqueue *);
+int	viomb_inflate_intr(struct virtqueue *);
+int	viomb_deflate_intr(struct virtqueue *);
+
+struct cfattach viomb_ca = {
+	sizeof(struct viomb_softc), viomb_match, viomb_attach
+};
+
+struct cfdriver viomb_cd = {
+	NULL, "viomb", DV_DULL
+};
+
+struct workq *viomb_workq;
+
+int
+viomb_match(struct device *parent, void *match, void *aux)
+{
+	struct virtio_softc *va = aux;
+	if (va->sc_childdevid == PCI_PRODUCT_VIRTIO_BALLOON)
+		return (1);
+	return (0);
+}
+
+void
+viomb_attach(struct device *parent, struct device *self, void *aux)
+{
+	struct viomb_softc *sc = (struct viomb_softc *)self;
+	struct virtio_softc *vsc = (struct virtio_softc *)parent;
+	u_int32_t features;
+	int i;
+
+	if (vsc->sc_child != NULL) {
+		printf("child already attached for %s; something wrong...\n",
+		    parent->dv_xname);
+		return;
+	}
+
+	/* fail on non-4K page size archs */
+	if (VIRTIO_PAGE_SIZE != PAGE_SIZE){
+		printf("non-4K page size arch found, needs %d, got %d\n",
+		    VIRTIO_PAGE_SIZE, PAGE_SIZE);
+		return;
+	}
+
+	sc->sc_virtio = vsc;
+	vsc->sc_vqs = &sc->sc_vq[VQ_INFLATE];
+	vsc->sc_nvqs = 0;
+	vsc->sc_child = self;
+	vsc->sc_ipl = IPL_VM;
+	vsc->sc_config_change = viomb_config_change;
+	vsc->sc_intrhand = virtio_vq_intr;
+
+	/* negotiate features */
+	features = VIRTIO_F_RING_INDIRECT_DESC;
+	features = virtio_negotiate_features(vsc, features,
+					     viomb_feature_names);
+
+	if ((virtio_alloc_vq(vsc, &sc->sc_vq[VQ_INFLATE], VQ_INFLATE,
+	     sizeof(u_int32_t) * PGS_PER_REQ, 1, "inflate") != 0))
+		goto err;
+	vsc->sc_nvqs++;
+	if ((virtio_alloc_vq(vsc, &sc->sc_vq[VQ_DEFLATE], VQ_DEFLATE,
+	     sizeof(u_int32_t) * PGS_PER_REQ, 1, "deflate") != 0))
+		goto err;
+	vsc->sc_nvqs++;
+
+	sc->sc_vq[VQ_INFLATE].vq_done = viomb_inflate_intr;
+	sc->sc_vq[VQ_DEFLATE].vq_done = viomb_deflate_intr;
+	virtio_start_vq_intr(vsc, &sc->sc_vq[VQ_INFLATE]);
+	virtio_start_vq_intr(vsc, &sc->sc_vq[VQ_DEFLATE]);
+
+	viomb_read_config(sc);
+	TAILQ_INIT(&sc->sc_balloon_pages);
+
+	if (bus_dmamap_create(vsc->sc_dmat, sizeof(u_int32_t) * PGS_PER_REQ,
+			      1, sizeof(u_int32_t) * PGS_PER_REQ, 0,
+			      BUS_DMA_NOWAIT, &sc->sc_req.bl_dmamap)) {
+		printf("%s: dmamap creation failed.\n", DEVNAME(sc));
+		goto err;
+	}
+	if (bus_dmamap_load(vsc->sc_dmat, sc->sc_req.bl_dmamap,
+			    &sc->sc_req.bl_pages[0],
+			    sizeof(uint32_t) * PGS_PER_REQ,
+			    NULL, BUS_DMA_NOWAIT)) {
+		printf("%s: dmamap load failed.\n", DEVNAME(sc));
+		goto err_dmamap;
+	}
+
+	viomb_workq = workq_create("viomb", 1, IPL_VM);
+	if (viomb_workq == NULL)
+		goto err_dmamap;
+	printf("\n");
+	return;
+err_dmamap:
+	bus_dmamap_destroy(vsc->sc_dmat, sc->sc_req.bl_dmamap);
+err:
+	for (i = 0; i < vsc->sc_nvqs; i++)
+		virtio_free_vq(vsc, &sc->sc_vq[i]);
+	vsc->sc_nvqs = 0;
+	vsc->sc_child = VIRTIO_CHILD_ERROR;
+	return;
+}
+
+/*
+ * Config change
+ */
+int
+viomb_config_change(struct virtio_softc *vsc)
+{
+	struct viomb_softc *sc = (struct viomb_softc *)vsc->sc_child;
+
+	if (sc->sc_workq_queued == 0){
+		workq_queue_task(viomb_workq, &sc->sc_task, 0,
+				 viomb_worker, sc, NULL);
+		sc->sc_workq_queued = 1;
+	}
+	return (1);
+}
+
+void
+viomb_worker(void *arg1, void *arg2)
+{
+	struct viomb_softc *sc = (struct viomb_softc *)arg1;
+	int s;
+
+	s = splvm();
+	sc->sc_workq_queued = 0;
+	viomb_read_config(sc);
+	if (sc->sc_npages > sc->sc_actual){
+		VIOMBDEBUG(sc, "inflating balloon from %u to %u.\n",
+			   sc->sc_actual, sc->sc_npages);
+		viomb_inflate(sc);
+		}
+	else if (sc->sc_npages < sc->sc_actual){
+		viomb_deflate(sc);
+		VIOMBDEBUG(sc, "deflating balloon from %u to %u.\n",
+			   sc->sc_actual, sc->sc_npages);
+	}
+	splx(s);
+}
+
+void
+viomb_inflate(struct viomb_softc *sc)
+{
+	struct virtio_softc *vsc = (struct virtio_softc *)sc->sc_virtio;
+	struct balloon_req *b;
+	struct vm_page *p;
+	struct virtqueue *vq = &sc->sc_vq[VQ_INFLATE];
+	u_int32_t nvpages;
+	int slot, error, i = 0;
+
+	nvpages = sc->sc_npages - sc->sc_actual;
+	if (nvpages > PGS_PER_REQ)
+		nvpages = PGS_PER_REQ;
+	b = &sc->sc_req;
+
+	if ((error = uvm_pglistalloc(nvpages * PAGE_SIZE, 0,
+				     dma_constraint.ucr_high,
+				     0, 0, &b->bl_pglist, nvpages,
+				     UVM_PLA_NOWAIT))) {
+		printf("%s unable to allocate %lu physmem pages,"
+		    "error %d\n", DEVNAME(sc), nvpages, error);
+		return;
+	}
+
+	b->bl_nentries = nvpages;
+	TAILQ_FOREACH(p, &b->bl_pglist, pageq)
+		b->bl_pages[i++] = p->phys_addr / VIRTIO_PAGE_SIZE;
+
+	KASSERT(i == nvpages);
+
+	if ((virtio_enqueue_prep(vq, &slot)) > 0) {
+		printf("%s:virtio_enqueue_prep() vq_num %d\n",
+		       DEVNAME(sc), vq->vq_num);
+		goto err;
+	}
+	if (virtio_enqueue_reserve(vq, slot, 1)) {
+		printf("%s:virtio_enqueue_reserve vq_num %d\n",
+		       DEVNAME(sc), vq->vq_num);
+		goto err;
+	}
+	bus_dmamap_sync(vsc->sc_dmat, b->bl_dmamap, 0,
+			sizeof(u_int32_t) * nvpages, BUS_DMASYNC_PREWRITE);
+	virtio_enqueue_p(vq, slot, b->bl_dmamap, 0,
+			 sizeof(u_int32_t) * nvpages, VRING_READ);
+	virtio_enqueue_commit(vsc, vq, slot, VRING_NOTIFY);
+	return;
+err:
+	uvm_pglistfree(&b->bl_pglist);
+	return;
+}
+
+void
+viomb_deflate(struct viomb_softc *sc)
+{
+	struct virtio_softc *vsc = (struct virtio_softc *)sc->sc_virtio;
+	struct balloon_req *b;
+	struct vm_page *p;
+	struct virtqueue *vq = &sc->sc_vq[VQ_DEFLATE];
+	u_int64_t nvpages;
+	int i, slot;
+
+	nvpages = sc->sc_actual - sc->sc_npages;
+	if (nvpages > PGS_PER_REQ)
+		nvpages = PGS_PER_REQ;
+	b = &sc->sc_req;
+	b->bl_nentries = nvpages;
+
+	TAILQ_INIT(&b->bl_pglist);
+	for (i = 0; i < nvpages; i++) {
+		p = TAILQ_FIRST(&sc->sc_balloon_pages);
+		if (p == NULL){
+		    b->bl_nentries = i - 1;
+		    break;
+		}
+		TAILQ_REMOVE(&sc->sc_balloon_pages, p, pageq);
+		TAILQ_INSERT_TAIL(&b->bl_pglist, p, pageq);
+		b->bl_pages[i] = p->phys_addr / VIRTIO_PAGE_SIZE;
+	}
+
+	if (virtio_enqueue_prep(vq, &slot)) {
+		printf("%s:virtio_get_slot(def) vq_num %d\n",
+		       DEVNAME(sc), vq->vq_num);
+		goto err;
+	}
+	if (virtio_enqueue_reserve(vq, slot, 1)) {
+		printf("%s:virtio_enqueue_reserve() vq_num %d\n",
+		       DEVNAME(sc), vq->vq_num);
+		goto err;
+	}
+	bus_dmamap_sync(vsc->sc_dmat, b->bl_dmamap, 0,
+		    sizeof(u_int32_t) * nvpages,
+		    BUS_DMASYNC_PREWRITE);
+	virtio_enqueue_p(vq, slot, b->bl_dmamap, 0,
+			 sizeof(u_int32_t) * nvpages, VRING_READ);
+
+	if (!(vsc->sc_features & VIRTIO_BALLOON_F_MUST_TELL_HOST))
+		uvm_pglistfree(&b->bl_pglist);
+	virtio_enqueue_commit(vsc, vq, slot, VRING_NOTIFY);
+	return;
+err:
+	while ((p = TAILQ_LAST(&b->bl_pglist, pglist))) {
+		TAILQ_REMOVE(&b->bl_pglist, p, pageq);
+		TAILQ_INSERT_HEAD(&sc->sc_balloon_pages, p, pageq);
+	}
+	return;
+}
+
+void
+viomb_read_config(struct viomb_softc *sc)
+{
+	struct virtio_softc *vsc = (struct virtio_softc *)sc->sc_virtio;
+	u_int32_t reg;
+
+	/* these values are explicitly specified as little-endian */
+	reg = virtio_read_device_config_4(vsc, VIRTIO_BALLOON_CONFIG_NUM_PAGES);
+	sc->sc_npages = letoh32(reg);
+	reg = virtio_read_device_config_4(vsc, VIRTIO_BALLOON_CONFIG_ACTUAL);
+	sc->sc_actual = letoh32(reg);
+	VIOMBDEBUG(sc, "sc->sc_npages %u, sc->sc_actual %u\n",
+		   sc->sc_npages, sc->sc_actual);
+}
+
+int
+viomb_vq_dequeue(struct virtqueue *vq)
+{
+	struct virtio_softc *vsc = vq->vq_owner;
+	struct viomb_softc *sc = (struct viomb_softc *)vsc->sc_child;
+	int r, slot;
+
+	r = virtio_dequeue(vsc, vq, &slot, NULL);
+	if (r != 0) {
+		printf("%s: dequeue failed, errno %d\n",
+		       DEVNAME(sc), r);
+		return(r);
+	}
+	virtio_dequeue_commit(vq, slot);
+	return(0);
+}
+
+/*
+ * interrupt handling for vq's
+ */
+int
+viomb_inflate_intr(struct virtqueue *vq)
+{
+	struct virtio_softc *vsc = vq->vq_owner;
+	struct viomb_softc *sc = (struct viomb_softc *)vsc->sc_child;
+	struct balloon_req *b;
+	struct vm_page *p;
+	u_int64_t nvpages;
+
+	if (viomb_vq_dequeue(vq))
+		return(1);
+
+	b = &sc->sc_req;
+	nvpages = b->bl_nentries;
+	bus_dmamap_sync(vsc->sc_dmat, b->bl_dmamap,
+			offsetof(struct balloon_req, bl_pages),
+			sizeof(u_int32_t) * nvpages,
+			BUS_DMASYNC_POSTWRITE);
+	while (!TAILQ_EMPTY(&b->bl_pglist)) {
+		p = TAILQ_FIRST(&b->bl_pglist);
+		TAILQ_REMOVE(&b->bl_pglist, p, pageq);
+		TAILQ_INSERT_TAIL(&sc->sc_balloon_pages, p, pageq);
+	}
+	VIOMBDEBUG(sc, "updating sc->sc_actual from %lu to %lu\n",
+		   sc->sc_actual, sc->sc_actual + nvpages);
+	virtio_write_device_config_4(vsc, VIRTIO_BALLOON_CONFIG_ACTUAL,
+				     sc->sc_actual + nvpages);
+	viomb_read_config(sc);
+	/* if we have more work to do, add it to the task list */
+	if (sc->sc_npages > sc->sc_actual && sc->sc_workq_queued == 0){
+		workq_queue_task(viomb_workq, &sc->sc_task, 0,
+				 viomb_worker, sc, NULL);
+		sc->sc_workq_queued = 1;
+	}
+	return (1);
+}
+
+int
+viomb_deflate_intr(struct virtqueue *vq)
+{
+	struct virtio_softc *vsc = vq->vq_owner;
+	struct viomb_softc *sc = (struct viomb_softc *)vsc->sc_child;
+	struct balloon_req *b;
+	u_int64_t nvpages;
+
+	if (viomb_vq_dequeue(vq))
+		return(1);
+
+	b = &sc->sc_req;
+	nvpages = b->bl_nentries;
+	bus_dmamap_sync(vsc->sc_dmat, b->bl_dmamap,
+			offsetof(struct balloon_req, bl_pages),
+			sizeof(u_int32_t) * nvpages,
+			BUS_DMASYNC_POSTWRITE);
+
+	if (vsc->sc_features & VIRTIO_BALLOON_F_MUST_TELL_HOST)
+		uvm_pglistfree(&b->bl_pglist);
+
+	VIOMBDEBUG(sc, "updating sc->sc_actual from %lu to %lu\n",
+		sc->sc_actual, sc->sc_actual - nvpages);
+	virtio_write_device_config_4(vsc, VIRTIO_BALLOON_CONFIG_ACTUAL,
+				     sc->sc_actual - nvpages);
+	viomb_read_config(sc);
+
+	/* if we have more work to do, add it to tasks list */
+	if (sc->sc_npages < sc->sc_actual && sc->sc_workq_queued == 0){
+		workq_queue_task(viomb_workq, &sc->sc_task, 0,
+				 viomb_worker, sc, NULL);
+		sc->sc_workq_queued = 1;
+	}
+	return(1);
+}