mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Stjepan Poljak <stjepan.poljak@gmail.com>
To: "Michael S. Tsirkin" <mst@redhat.com>, Jason Wang <jasowangio@gmail.com>
Cc: "Xuan Zhuo" <xuanzhuo@linux.alibaba.com>,
	"Eugenio Pérez" <eperezma@redhat.com>,
	"Richard Weinberger" <richard@nod.at>,
	"Anton Ivanov" <anton.ivanov@cambridgegreys.com>,
	"Johannes Berg" <johannes@sipsolutions.net>,
	"Cornelia Huck" <cohuck@redhat.com>,
	"Halil Pasic" <pasic@linux.ibm.com>,
	"Eric Farman" <farman@linux.ibm.com>,
	"Heiko Carstens" <hca@linux.ibm.com>,
	"Vasily Gorbik" <gor@linux.ibm.com>,
	"Alexander Gordeev" <agordeev@linux.ibm.com>,
	"Christian Borntraeger" <borntraeger@linux.ibm.com>,
	"Sven Schnelle" <svens@linux.ibm.com>,
	"Alexander Graf" <graf@amazon.com>,
	virtualization@lists.linux.dev, linux-um@lists.infradead.org,
	linux-s390@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH] virtio_ring: return an error pointer from vring_create_virtqueue()
Date: Thu,  5 Nov 2026 20:34:25 +0100	[thread overview]
Message-ID: <20261105193426.2999631-1-stjepan.poljak@gmail.com> (raw)

vring_create_virtqueue() and vring_create_virtqueue_map() return NULL
on any failure, so the error returned from vring_alloc_queue_split()
is lost and turned into -ENOMEM. This is especially misleading if
vring_alloc_queue_split() rejects a queue size with -EINVAL (i.e. if
queue size is not a power of two):

  virtio_blk virtio0: Bad virtqueue length 100
  virtio_blk virtio0: probe with driver virtio_blk failed with error -12

So the error points whoever debugs it at an allocation failure that
never happened.

Return an ERR_PTR() instead: the error from vring_alloc_queue_split()
or vring_alloc_queue_packed(), or -ENOMEM when allocating the
virtqueue itself fails. Make the callers, virtio_pci (modern and
legacy), virtio_mmio, virtio_vdpa, virtio_ccw and virtio_uml, check
with IS_ERR() and pass the error on. The error path in virtio_ccw
deletes a non-NULL vq, so clear vq there before taking it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Stjepan Poljak <stjepan.poljak@gmail.com>
---
Tested on x86_64 with a virtio-mmio virtio-blk device offering a
100-entry queue: the probe now fails with -22 instead of -12, and a
128-entry queue probes as before. The other transports are only
compile-tested (x86_64, ARCH=um, s390 with clang, W=1).

No Fixes: tag, since the error has been -ENOMEM since 2010 and nothing
regressed.

This conflicts trivially with patch 07/12 of Alexander Graf's "virtio:
support devices that own their virtqueue memory" v2:
https://lore.kernel.org/virtualization/20260818211425.91009-1-graf@amazon.com/
---
 arch/um/drivers/virtio_uml.c       |  4 ++--
 drivers/s390/virtio/virtio_ccw.c   |  5 +++--
 drivers/virtio/virtio_mmio.c       |  4 ++--
 drivers/virtio/virtio_pci_legacy.c |  4 ++--
 drivers/virtio/virtio_pci_modern.c |  4 ++--
 drivers/virtio/virtio_ring.c       | 12 +++++++-----
 drivers/virtio/virtio_vdpa.c       |  4 ++--
 include/linux/virtio_ring.h        |  6 +++++-
 8 files changed, 25 insertions(+), 18 deletions(-)

diff --git a/arch/um/drivers/virtio_uml.c b/arch/um/drivers/virtio_uml.c
index 881c34098a74..8b645523277f 100644
--- a/arch/um/drivers/virtio_uml.c
+++ b/arch/um/drivers/virtio_uml.c
@@ -984,8 +984,8 @@ static struct virtqueue *vu_setup_vq(struct virtio_device *vdev,
 
 	vq = vring_create_virtqueue(index, num, PAGE_SIZE, vdev, true, true,
 				    ctx, vu_notify, callback, info->name);
-	if (!vq) {
-		rc = -ENOMEM;
+	if (IS_ERR(vq)) {
+		rc = PTR_ERR(vq);
 		goto error_create;
 	}
 	vq->priv = info;
diff --git a/drivers/s390/virtio/virtio_ccw.c b/drivers/s390/virtio/virtio_ccw.c
index 552d77998012..c5a30f0a2bc2 100644
--- a/drivers/s390/virtio/virtio_ccw.c
+++ b/drivers/s390/virtio/virtio_ccw.c
@@ -590,10 +590,11 @@ static struct virtqueue *virtio_ccw_setup_vq(struct virtio_device *vdev,
 				    vdev, true, may_reduce, ctx,
 				    notify, callback, name);
 
-	if (!vq) {
+	if (IS_ERR(vq)) {
 		/* For now, we fail if we can't get the requested size. */
 		dev_warn(&vcdev->cdev->dev, "no vq\n");
-		err = -ENOMEM;
+		err = PTR_ERR(vq);
+		vq = NULL;
 		goto out_err;
 	}
 
diff --git a/drivers/virtio/virtio_mmio.c b/drivers/virtio/virtio_mmio.c
index e2ec0ec01d67..c403a025006e 100644
--- a/drivers/virtio/virtio_mmio.c
+++ b/drivers/virtio/virtio_mmio.c
@@ -396,8 +396,8 @@ static struct virtqueue *vm_setup_vq(struct virtio_device *vdev, unsigned int in
 	/* Create the vring */
 	vq = vring_create_virtqueue(index, num, VIRTIO_MMIO_VRING_ALIGN, vdev,
 				 true, true, ctx, notify, callback, name);
-	if (!vq) {
-		err = -ENOMEM;
+	if (IS_ERR(vq)) {
+		err = PTR_ERR(vq);
 		goto error_new_virtqueue;
 	}
 
diff --git a/drivers/virtio/virtio_pci_legacy.c b/drivers/virtio/virtio_pci_legacy.c
index 8115aa39e01e..9d94fffff344 100644
--- a/drivers/virtio/virtio_pci_legacy.c
+++ b/drivers/virtio/virtio_pci_legacy.c
@@ -130,8 +130,8 @@ static struct virtqueue *setup_vq(struct virtio_pci_device *vp_dev,
 				    VIRTIO_PCI_VRING_ALIGN, &vp_dev->vdev,
 				    true, false, ctx,
 				    vp_notify, callback, name);
-	if (!vq)
-		return ERR_PTR(-ENOMEM);
+	if (IS_ERR(vq))
+		return vq;
 
 	vq->num_max = num;
 
diff --git a/drivers/virtio/virtio_pci_modern.c b/drivers/virtio/virtio_pci_modern.c
index 922e2df027de..28ccb2130e27 100644
--- a/drivers/virtio/virtio_pci_modern.c
+++ b/drivers/virtio/virtio_pci_modern.c
@@ -720,8 +720,8 @@ static struct virtqueue *setup_vq(struct virtio_pci_device *vp_dev,
 				    SMP_CACHE_BYTES, &vp_dev->vdev,
 				    true, true, ctx,
 				    notify, callback, name);
-	if (!vq)
-		return ERR_PTR(-ENOMEM);
+	if (IS_ERR(vq))
+		return vq;
 
 	vq->num_max = num;
 
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index f683a2d6a200..6dd4a7651bde 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -1421,13 +1421,13 @@ static struct virtqueue *vring_create_virtqueue_split(
 	err = vring_alloc_queue_split(&vring_split, vdev, num, vring_align,
 				      may_reduce_num, map);
 	if (err)
-		return NULL;
+		return ERR_PTR(err);
 
 	vq = __vring_new_virtqueue_split(index, &vring_split, vdev, weak_barriers,
 				   context, notify, callback, name, map);
 	if (!vq) {
 		vring_free_split(&vring_split, vdev, map);
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 	}
 
 	to_vvq(vq)->we_own_ring = true;
@@ -2671,15 +2671,17 @@ static struct virtqueue *vring_create_virtqueue_packed(
 {
 	struct vring_virtqueue_packed vring_packed = {};
 	struct virtqueue *vq;
+	int err;
 
-	if (vring_alloc_queue_packed(&vring_packed, vdev, num, map))
-		return NULL;
+	err = vring_alloc_queue_packed(&vring_packed, vdev, num, map);
+	if (err)
+		return ERR_PTR(err);
 
 	vq = __vring_new_virtqueue_packed(index, &vring_packed, vdev, weak_barriers,
 					context, notify, callback, name, map);
 	if (!vq) {
 		vring_free_packed(&vring_packed, vdev, map);
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 	}
 
 	to_vvq(vq)->we_own_ring = true;
diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
index f9559b6b2e0b..c1b6134b2267 100644
--- a/drivers/virtio/virtio_vdpa.c
+++ b/drivers/virtio/virtio_vdpa.c
@@ -208,8 +208,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 	vq = vring_create_virtqueue_map(index, max_num, align, vdev,
 					true, may_reduce_num, ctx,
 					notify, callback, name, map);
-	if (!vq) {
-		err = -ENOMEM;
+	if (IS_ERR(vq)) {
+		err = PTR_ERR(vq);
 		goto error_new_virtqueue;
 	}
 
diff --git a/include/linux/virtio_ring.h b/include/linux/virtio_ring.h
index c97a12c1cda3..c808ca7e8d58 100644
--- a/include/linux/virtio_ring.h
+++ b/include/linux/virtio_ring.h
@@ -66,6 +66,9 @@ struct device;
  * may_reduce_num is set, then this may allocate a smaller ring than
  * expected.  The caller should query virtqueue_get_vring_size to learn
  * the actual size of the ring.
+ *
+ * Returns the virtqueue or an ERR_PTR() on failure: -EINVAL if num is
+ * not a valid ring size, -ENOMEM if the ring could not be allocated.
  */
 struct virtqueue *vring_create_virtqueue(unsigned int index,
 					 unsigned int num,
@@ -80,7 +83,8 @@ struct virtqueue *vring_create_virtqueue(unsigned int index,
 
 /*
  * Creates a virtqueue and allocates the descriptor ring with per
- * virtqueue mapping operations.
+ * virtqueue mapping operations.  Returns the virtqueue or an ERR_PTR()
+ * on failure, like vring_create_virtqueue().
  */
 struct virtqueue *vring_create_virtqueue_map(unsigned int index,
 					     unsigned int num,

base-commit: 1a577f9402c1780cc1e663f48d5dd54c02051b09
-- 
2.53.0


             reply	other threads:[~2026-11-05 19:34 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-05 19:34 Stjepan Poljak [this message]
2026-11-05 19:43 ` Michael S. Tsirkin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20261105193426.2999631-1-stjepan.poljak@gmail.com \
    --to=stjepan.poljak@gmail.com \
    --cc=agordeev@linux.ibm.com \
    --cc=anton.ivanov@cambridgegreys.com \
    --cc=borntraeger@linux.ibm.com \
    --cc=cohuck@redhat.com \
    --cc=eperezma@redhat.com \
    --cc=farman@linux.ibm.com \
    --cc=gor@linux.ibm.com \
    --cc=graf@amazon.com \
    --cc=hca@linux.ibm.com \
    --cc=jasowangio@gmail.com \
    --cc=johannes@sipsolutions.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=linux-um@lists.infradead.org \
    --cc=mst@redhat.com \
    --cc=pasic@linux.ibm.com \
    --cc=richard@nod.at \
    --cc=svens@linux.ibm.com \
    --cc=virtualization@lists.linux.dev \
    --cc=xuanzhuo@linux.alibaba.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®