Merge tag 'for-linus-4.4-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip

Pull xen bug fixes from David Vrabel:
 - XSA-155 security fixes to backend drivers.
 - XSA-157 security fixes to pciback.

* tag 'for-linus-4.4-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  xen-pciback: fix up cleanup path when alloc fails
  xen/pciback: Don't allow MSI-X ops if PCI_COMMAND_MEMORY is not set.
  xen/pciback: For XEN_PCI_OP_disable_msi[|x] only disable if device has MSI(X) enabled.
  xen/pciback: Do not install an IRQ handler for MSI interrupts.
  xen/pciback: Return error on XEN_PCI_OP_enable_msix when device has MSI or MSI-X enabled
  xen/pciback: Return error on XEN_PCI_OP_enable_msi when device has MSI or MSI-X enabled
  xen/pciback: Save xen_pci_op commands before processing it
  xen-scsiback: safely copy requests
  xen-blkback: read from indirect descriptors only once
  xen-blkback: only read request operation from shared ring once
  xen-netback: use RING_COPY_REQUEST() throughout
  xen-netback: don't use last request to determine minimum Tx credit
  xen: Add RING_COPY_REQUEST()
  xen/x86/pvh: Use HVM's flush_tlb_others op
  xen: Resume PMU from non-atomic context
  xen/events/fifo: Consume unprocessed events when a CPU dies

arch/x86/xen/mmu.c

@@ -2495,14 +2495,9 @@ void __init xen_init_mmu_ops(void)
 {
 	x86_init.paging.pagetable_init = xen_pagetable_init;
 
-	/* Optimization - we can use the HVM one but it has no idea which
-	 * VCPUs are descheduled - which means that it will needlessly IPI
-	 * them. Xen knows so let it do the job.
-	 */
-	if (xen_feature(XENFEAT_auto_translated_physmap)) {
-		pv_mmu_ops.flush_tlb_others = xen_flush_tlb_others;
+	if (xen_feature(XENFEAT_auto_translated_physmap))
 		return;
-	}
+
 	pv_mmu_ops = xen_mmu_ops;
 
 	memset(dummy_mapping, 0xff, PAGE_SIZE);

arch/x86/xen/suspend.c

@@ -68,26 +68,16 @@ static void xen_pv_post_suspend(int suspend_cancelled)
 
 void xen_arch_pre_suspend(void)
 {
-	int cpu;
-
-	for_each_online_cpu(cpu)
-		xen_pmu_finish(cpu);
-
 	if (xen_pv_domain())
 		xen_pv_pre_suspend();
 }
 
 void xen_arch_post_suspend(int cancelled)
 {
-	int cpu;
-
 	if (xen_pv_domain())
 		xen_pv_post_suspend(cancelled);
 	else
 		xen_hvm_post_suspend(cancelled);
-
-	for_each_online_cpu(cpu)
-		xen_pmu_init(cpu);
 }
 
 static void xen_vcpu_notify_restore(void *data)
@@ -106,10 +96,20 @@ static void xen_vcpu_notify_suspend(void *data)
 
 void xen_arch_resume(void)
 {
+	int cpu;
+
 	on_each_cpu(xen_vcpu_notify_restore, NULL, 1);
+
+	for_each_online_cpu(cpu)
+		xen_pmu_init(cpu);
 }
 
 void xen_arch_suspend(void)
 {
+	int cpu;
+
+	for_each_online_cpu(cpu)
+		xen_pmu_finish(cpu);
+
 	on_each_cpu(xen_vcpu_notify_suspend, NULL, 1);
 }

drivers/block/xen-blkback/blkback.c

@@ -950,6 +950,8 @@ static int xen_blkbk_parse_indirect(struct blkif_request *req,
 		goto unmap;
 
 	for (n = 0, i = 0; n < nseg; n++) {
+		uint8_t first_sect, last_sect;
+
 		if ((n % SEGS_PER_INDIRECT_FRAME) == 0) {
 			/* Map indirect segments */
 			if (segments)
@@ -957,15 +959,18 @@ static int xen_blkbk_parse_indirect(struct blkif_request *req,
 			segments = kmap_atomic(pages[n/SEGS_PER_INDIRECT_FRAME]->page);
 		}
 		i = n % SEGS_PER_INDIRECT_FRAME;
+
 		pending_req->segments[n]->gref = segments[i].gref;
-		seg[n].nsec = segments[i].last_sect -
-			segments[i].first_sect + 1;
-		seg[n].offset = (segments[i].first_sect << 9);
-		if ((segments[i].last_sect >= (XEN_PAGE_SIZE >> 9)) ||
-		    (segments[i].last_sect < segments[i].first_sect)) {
+
+		first_sect = READ_ONCE(segments[i].first_sect);
+		last_sect = READ_ONCE(segments[i].last_sect);
+		if (last_sect >= (XEN_PAGE_SIZE >> 9) || last_sect < first_sect) {
 			rc = -EINVAL;
 			goto unmap;
 		}
+
+		seg[n].nsec = last_sect - first_sect + 1;
+		seg[n].offset = first_sect << 9;
 		preq->nr_sects += seg[n].nsec;
 	}
 

drivers/block/xen-blkback/common.h

@@ -408,8 +408,8 @@ static inline void blkif_get_x86_32_req(struct blkif_request *dst,
 					struct blkif_x86_32_request *src)
 {
 	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST, j;
-	dst->operation = src->operation;
-	switch (src->operation) {
+	dst->operation = READ_ONCE(src->operation);
+	switch (dst->operation) {
 	case BLKIF_OP_READ:
 	case BLKIF_OP_WRITE:
 	case BLKIF_OP_WRITE_BARRIER:
@@ -456,8 +456,8 @@ static inline void blkif_get_x86_64_req(struct blkif_request *dst,
 					struct blkif_x86_64_request *src)
 {
 	int i, n = BLKIF_MAX_SEGMENTS_PER_REQUEST, j;
-	dst->operation = src->operation;
-	switch (src->operation) {
+	dst->operation = READ_ONCE(src->operation);
+	switch (dst->operation) {
 	case BLKIF_OP_READ:
 	case BLKIF_OP_WRITE:
 	case BLKIF_OP_WRITE_BARRIER:

drivers/net/xen-netback/netback.c

@@ -258,18 +258,18 @@ static struct xenvif_rx_meta *get_next_rx_buffer(struct xenvif_queue *queue,
 						 struct netrx_pending_operations *npo)
 {
 	struct xenvif_rx_meta *meta;
-	struct xen_netif_rx_request *req;
+	struct xen_netif_rx_request req;
 
-	req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
+	RING_COPY_REQUEST(&queue->rx, queue->rx.req_cons++, &req);
 
 	meta = npo->meta + npo->meta_prod++;
 	meta->gso_type = XEN_NETIF_GSO_TYPE_NONE;
 	meta->gso_size = 0;
 	meta->size = 0;
-	meta->id = req->id;
+	meta->id = req.id;
 
 	npo->copy_off = 0;
-	npo->copy_gref = req->gref;
+	npo->copy_gref = req.gref;
 
 	return meta;
 }
@@ -424,7 +424,7 @@ static int xenvif_gop_skb(struct sk_buff *skb,
 	struct xenvif *vif = netdev_priv(skb->dev);
 	int nr_frags = skb_shinfo(skb)->nr_frags;
 	int i;
-	struct xen_netif_rx_request *req;
+	struct xen_netif_rx_request req;
 	struct xenvif_rx_meta *meta;
 	unsigned char *data;
 	int head = 1;
@@ -443,15 +443,15 @@ static int xenvif_gop_skb(struct sk_buff *skb,
 
 	/* Set up a GSO prefix descriptor, if necessary */
 	if ((1 << gso_type) & vif->gso_prefix_mask) {
-		req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
+		RING_COPY_REQUEST(&queue->rx, queue->rx.req_cons++, &req);
 		meta = npo->meta + npo->meta_prod++;
 		meta->gso_type = gso_type;
 		meta->gso_size = skb_shinfo(skb)->gso_size;
 		meta->size = 0;
-		meta->id = req->id;
+		meta->id = req.id;
 	}
 
-	req = RING_GET_REQUEST(&queue->rx, queue->rx.req_cons++);
+	RING_COPY_REQUEST(&queue->rx, queue->rx.req_cons++, &req);
 	meta = npo->meta + npo->meta_prod++;
 
 	if ((1 << gso_type) & vif->gso_mask) {
@@ -463,9 +463,9 @@ static int xenvif_gop_skb(struct sk_buff *skb,
 	}
 
 	meta->size = 0;
-	meta->id = req->id;
+	meta->id = req.id;
 	npo->copy_off = 0;
-	npo->copy_gref = req->gref;
+	npo->copy_gref = req.gref;
 
 	data = skb->data;
 	while (data < skb_tail_pointer(skb)) {
@@ -679,9 +679,7 @@ static void tx_add_credit(struct xenvif_queue *queue)
 	 * Allow a burst big enough to transmit a jumbo packet of up to 128kB.
 	 * Otherwise the interface can seize up due to insufficient credit.
 	 */
-	max_burst = RING_GET_REQUEST(&queue->tx, queue->tx.req_cons)->size;
-	max_burst = min(max_burst, 131072UL);
-	max_burst = max(max_burst, queue->credit_bytes);
+	max_burst = max(131072UL, queue->credit_bytes);
 
 	/* Take care that adding a new chunk of credit doesn't wrap to zero. */
 	max_credit = queue->remaining_credit + queue->credit_bytes;
@@ -711,7 +709,7 @@ static void xenvif_tx_err(struct xenvif_queue *queue,
 		spin_unlock_irqrestore(&queue->response_lock, flags);
 		if (cons == end)
 			break;
-		txp = RING_GET_REQUEST(&queue->tx, cons++);
+		RING_COPY_REQUEST(&queue->tx, cons++, txp);
 	} while (1);
 	queue->tx.req_cons = cons;
 }
@@ -778,8 +776,7 @@ static int xenvif_count_requests(struct xenvif_queue *queue,
 		if (drop_err)
 			txp = &dropped_tx;
 
-		memcpy(txp, RING_GET_REQUEST(&queue->tx, cons + slots),
-		       sizeof(*txp));
+		RING_COPY_REQUEST(&queue->tx, cons + slots, txp);
 
 		/* If the guest submitted a frame >= 64 KiB then
 		 * first->size overflowed and following slots will
@@ -1112,8 +1109,7 @@ static int xenvif_get_extras(struct xenvif_queue *queue,
 			return -EBADR;
 		}
 
-		memcpy(&extra, RING_GET_REQUEST(&queue->tx, cons),
-		       sizeof(extra));
+		RING_COPY_REQUEST(&queue->tx, cons, &extra);
 		if (unlikely(!extra.type ||
 			     extra.type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
 			queue->tx.req_cons = ++cons;
@@ -1322,7 +1318,7 @@ static void xenvif_tx_build_gops(struct xenvif_queue *queue,
 
 		idx = queue->tx.req_cons;
 		rmb(); /* Ensure that we see the request before we copy it. */
-		memcpy(&txreq, RING_GET_REQUEST(&queue->tx, idx), sizeof(txreq));
+		RING_COPY_REQUEST(&queue->tx, idx, &txreq);
 
 		/* Credit-based scheduling. */
 		if (txreq.size > queue->remaining_credit &&

drivers/xen/events/events_fifo.c

@@ -281,7 +281,8 @@ static void handle_irq_for_port(unsigned port)
 
 static void consume_one_event(unsigned cpu,
 			      struct evtchn_fifo_control_block *control_block,
-			      unsigned priority, unsigned long *ready)
+			      unsigned priority, unsigned long *ready,
+			      bool drop)
 {
 	struct evtchn_fifo_queue *q = &per_cpu(cpu_queue, cpu);
 	uint32_t head;
@@ -313,13 +314,17 @@ static void consume_one_event(unsigned cpu,
 	if (head == 0)
 		clear_bit(priority, ready);
 
-	if (evtchn_fifo_is_pending(port) && !evtchn_fifo_is_masked(port))
+	if (evtchn_fifo_is_pending(port) && !evtchn_fifo_is_masked(port)) {
+		if (unlikely(drop))
+			pr_warn("Dropping pending event for port %u\n", port);
+		else
 			handle_irq_for_port(port);
+	}
 
 	q->head[priority] = head;
 }
 
-static void evtchn_fifo_handle_events(unsigned cpu)
+static void __evtchn_fifo_handle_events(unsigned cpu, bool drop)
 {
 	struct evtchn_fifo_control_block *control_block;
 	unsigned long ready;
@@ -331,11 +336,16 @@ static void evtchn_fifo_handle_events(unsigned cpu)
 
 	while (ready) {
 		q = find_first_bit(&ready, EVTCHN_FIFO_MAX_QUEUES);
-		consume_one_event(cpu, control_block, q, &ready);
+		consume_one_event(cpu, control_block, q, &ready, drop);
 		ready |= xchg(&control_block->ready, 0);
 	}
 }
 
+static void evtchn_fifo_handle_events(unsigned cpu)
+{
+	__evtchn_fifo_handle_events(cpu, false);
+}
+
 static void evtchn_fifo_resume(void)
 {
 	unsigned cpu;
@@ -420,6 +430,9 @@ static int evtchn_fifo_cpu_notification(struct notifier_block *self,
 		if (!per_cpu(cpu_control_block, cpu))
 			ret = evtchn_fifo_alloc_control_block(cpu);
 		break;
+	case CPU_DEAD:
+		__evtchn_fifo_handle_events(cpu, true);
+		break;
 	default:
 		break;
 	}

drivers/xen/xen-pciback/pciback.h

@@ -37,6 +37,7 @@ struct xen_pcibk_device {
 	struct xen_pci_sharedinfo *sh_info;
 	unsigned long flags;
 	struct work_struct op_work;
+	struct xen_pci_op op;
 };
 
 struct xen_pcibk_dev_data {

drivers/xen/xen-pciback/pciback_ops.c

@@ -70,6 +70,13 @@ static void xen_pcibk_control_isr(struct pci_dev *dev, int reset)
 		enable ? "enable" : "disable");
 
 	if (enable) {
+		/*
+		 * The MSI or MSI-X should not have an IRQ handler. Otherwise
+		 * if the guest terminates we BUG_ON in free_msi_irqs.
+		 */
+		if (dev->msi_enabled || dev->msix_enabled)
+			goto out;
+
 		rc = request_irq(dev_data->irq,
 				xen_pcibk_guest_interrupt, IRQF_SHARED,
 				dev_data->irq_name, dev);
@@ -144,6 +151,11 @@ int xen_pcibk_enable_msi(struct xen_pcibk_device *pdev,
 	if (unlikely(verbose_request))
 		printk(KERN_DEBUG DRV_NAME ": %s: enable MSI\n", pci_name(dev));
 
+	if (dev->msi_enabled)
+		status = -EALREADY;
+	else if (dev->msix_enabled)
+		status = -ENXIO;
+	else
 		status = pci_enable_msi(dev);
 
 	if (status) {
@@ -173,20 +185,23 @@ static
 int xen_pcibk_disable_msi(struct xen_pcibk_device *pdev,
 			  struct pci_dev *dev, struct xen_pci_op *op)
 {
-	struct xen_pcibk_dev_data *dev_data;
-
 	if (unlikely(verbose_request))
 		printk(KERN_DEBUG DRV_NAME ": %s: disable MSI\n",
 		       pci_name(dev));
+
+	if (dev->msi_enabled) {
+		struct xen_pcibk_dev_data *dev_data;
+
 		pci_disable_msi(dev);
 
+		dev_data = pci_get_drvdata(dev);
+		if (dev_data)
+			dev_data->ack_intr = 1;
+	}
 	op->value = dev->irq ? xen_pirq_from_irq(dev->irq) : 0;
 	if (unlikely(verbose_request))
 		printk(KERN_DEBUG DRV_NAME ": %s: MSI: %d\n", pci_name(dev),
 			op->value);
-	dev_data = pci_get_drvdata(dev);
-	if (dev_data)
-		dev_data->ack_intr = 1;
 	return 0;
 }
 
@@ -197,13 +212,26 @@ int xen_pcibk_enable_msix(struct xen_pcibk_device *pdev,
 	struct xen_pcibk_dev_data *dev_data;
 	int i, result;
 	struct msix_entry *entries;
+	u16 cmd;
 
 	if (unlikely(verbose_request))
 		printk(KERN_DEBUG DRV_NAME ": %s: enable MSI-X\n",
 		       pci_name(dev));
+
 	if (op->value > SH_INFO_MAX_VEC)
 		return -EINVAL;
 
+	if (dev->msix_enabled)
+		return -EALREADY;
+
+	/*
+	 * PCI_COMMAND_MEMORY must be enabled, otherwise we may not be able
+	 * to access the BARs where the MSI-X entries reside.
+	 */
+	pci_read_config_word(dev, PCI_COMMAND, &cmd);
+	if (dev->msi_enabled || !(cmd & PCI_COMMAND_MEMORY))
+		return -ENXIO;
+
 	entries = kmalloc(op->value * sizeof(*entries), GFP_KERNEL);
 	if (entries == NULL)
 		return -ENOMEM;
@@ -245,23 +273,27 @@ static
 int xen_pcibk_disable_msix(struct xen_pcibk_device *pdev,
 			   struct pci_dev *dev, struct xen_pci_op *op)
 {
-	struct xen_pcibk_dev_data *dev_data;
 	if (unlikely(verbose_request))
 		printk(KERN_DEBUG DRV_NAME ": %s: disable MSI-X\n",
 			pci_name(dev));
+
+	if (dev->msix_enabled) {
+		struct xen_pcibk_dev_data *dev_data;
+
 		pci_disable_msix(dev);
 
+		dev_data = pci_get_drvdata(dev);
+		if (dev_data)
+			dev_data->ack_intr = 1;
+	}
 	/*
 	 * SR-IOV devices (which don't have any legacy IRQ) have
 	 * an undefined IRQ value of zero.
 	 */
 	op->value = dev->irq ? xen_pirq_from_irq(dev->irq) : 0;
 	if (unlikely(verbose_request))
-		printk(KERN_DEBUG DRV_NAME ": %s: MSI-X: %d\n", pci_name(dev),
-			op->value);
-	dev_data = pci_get_drvdata(dev);
-	if (dev_data)
-		dev_data->ack_intr = 1;
+		printk(KERN_DEBUG DRV_NAME ": %s: MSI-X: %d\n",
+		       pci_name(dev), op->value);
 	return 0;
 }
 #endif
@@ -298,9 +330,11 @@ void xen_pcibk_do_op(struct work_struct *data)
 		container_of(data, struct xen_pcibk_device, op_work);
 	struct pci_dev *dev;
 	struct xen_pcibk_dev_data *dev_data = NULL;
-	struct xen_pci_op *op = &pdev->sh_info->op;
+	struct xen_pci_op *op = &pdev->op;
 	int test_intx = 0;
 
+	*op = pdev->sh_info->op;
+	barrier();
 	dev = xen_pcibk_get_pci_dev(pdev, op->domain, op->bus, op->devfn);
 
 	if (dev == NULL)
@@ -342,6 +376,17 @@ void xen_pcibk_do_op(struct work_struct *data)
 		if ((dev_data->enable_intx != test_intx))
 			xen_pcibk_control_isr(dev, 0 /* no reset */);
 	}
+	pdev->sh_info->op.err = op->err;
+	pdev->sh_info->op.value = op->value;
+#ifdef CONFIG_PCI_MSI
+	if (op->cmd == XEN_PCI_OP_enable_msix && op->err == 0) {
+		unsigned int i;
+
+		for (i = 0; i < op->value; i++)
+			pdev->sh_info->op.msix_entries[i].vector =
+				op->msix_entries[i].vector;
+	}
+#endif
 	/* Tell the driver domain that we're done. */
 	wmb();
 	clear_bit(_XEN_PCIF_active, (unsigned long *)&pdev->sh_info->flags);

drivers/xen/xen-pciback/xenbus.c

@@ -44,7 +44,6 @@ static struct xen_pcibk_device *alloc_pdev(struct xenbus_device *xdev)
 	dev_dbg(&xdev->dev, "allocated pdev @ 0x%p\n", pdev);
 
 	pdev->xdev = xdev;
-	dev_set_drvdata(&xdev->dev, pdev);
 
 	mutex_init(&pdev->dev_lock);
 
@@ -58,6 +57,9 @@ static struct xen_pcibk_device *alloc_pdev(struct xenbus_device *xdev)
 		kfree(pdev);
 		pdev = NULL;
 	}
+
+	dev_set_drvdata(&xdev->dev, pdev);
+
 out:
 	return pdev;
 }

drivers/xen/xen-scsiback.c

@@ -726,7 +726,7 @@ static int scsiback_do_cmd_fn(struct vscsibk_info *info)
 		if (!pending_req)
 			return 1;
 
-		ring_req = *RING_GET_REQUEST(ring, rc);
+		RING_COPY_REQUEST(ring, rc, &ring_req);
 		ring->req_cons = ++rc;
 
 		err = prepare_pending_reqs(info, &ring_req, pending_req);

include/xen/interface/io/ring.h

@@ -181,6 +181,20 @@ struct __name##_back_ring {						\
 #define RING_GET_REQUEST(_r, _idx)					\
     (&((_r)->sring->ring[((_idx) & (RING_SIZE(_r) - 1))].req))
 
+/*
+ * Get a local copy of a request.
+ *
+ * Use this in preference to RING_GET_REQUEST() so all processing is
+ * done on a local copy that cannot be modified by the other end.
+ *
+ * Note that https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58145 may cause this
+ * to be ineffective where _req is a struct which consists of only bitfields.
+ */
+#define RING_COPY_REQUEST(_r, _idx, _req) do {				\
+	/* Use volatile to force the copy into _req. */			\
+	*(_req) = *(volatile typeof(_req))RING_GET_REQUEST(_r, _idx);	\
+} while (0)
+
 #define RING_GET_RESPONSE(_r, _idx)					\
     (&((_r)->sring->ring[((_idx) & (RING_SIZE(_r) - 1))].rsp))