When we hot unplug a virtserialport and then try to hot plug again,
it fails:
(qemu) chardev-add socket,id=serial0,path=/tmp/serial0,server,nowait
(qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
chardev=serial0,id=serial0,name=serial0
(qemu) device_del serial0
(qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
chardev=serial0,id=serial0,name=serial0
kernel error:
virtio-ports vport2p2: Error allocating inbufs
qemu error:
virtio-serial-bus: Guest failure in adding port 2 for device \
virtio-serial0.0
This happens because buffers for the in_vq are allocated when the port is
added but are not released when the port is unplugged.
They are only released when virtconsole is removed (see a7a69ec0d8e4)
To avoid the problem and to be symmetric, we could allocate all the buffers
in init_vqs() as they are released in remove_vqs(), but it sounds like
a waste of memory.
Rather than that, this patch changes add_port() logic to ignore ENOSPC
error in fill_queue(), which means queue has already been filled.
Fixes: a7a69ec0d8e4 ("virtio_console: free buffers after reset")
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Laurent Vivier <[email protected]>
---
Notes:
v3: add a comment about ENOSPC error
v2: making fill_queue return int and testing return code for -ENOSPC
drivers/char/virtio_console.c | 28 +++++++++++++---------------
1 file changed, 13 insertions(+), 15 deletions(-)
diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index 7270e7b69262..3259426f01dc 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -1325,24 +1325,24 @@ static void set_console_size(struct port *port, u16 rows, u16 cols)
port->cons.ws.ws_col = cols;
}
-static unsigned int fill_queue(struct virtqueue *vq, spinlock_t *lock)
+static int fill_queue(struct virtqueue *vq, spinlock_t *lock)
{
struct port_buffer *buf;
- unsigned int nr_added_bufs;
+ int nr_added_bufs;
int ret;
nr_added_bufs = 0;
do {
buf = alloc_buf(vq->vdev, PAGE_SIZE, 0);
if (!buf)
- break;
+ return -ENOMEM;
spin_lock_irq(lock);
ret = add_inbuf(vq, buf);
if (ret < 0) {
spin_unlock_irq(lock);
free_buf(buf, true);
- break;
+ return ret;
}
nr_added_bufs++;
spin_unlock_irq(lock);
@@ -1362,7 +1362,6 @@ static int add_port(struct ports_device *portdev, u32 id)
char debugfs_name[16];
struct port *port;
dev_t devt;
- unsigned int nr_added_bufs;
int err;
port = kmalloc(sizeof(*port), GFP_KERNEL);
@@ -1421,11 +1420,13 @@ static int add_port(struct ports_device *portdev, u32 id)
spin_lock_init(&port->outvq_lock);
init_waitqueue_head(&port->waitqueue);
- /* Fill the in_vq with buffers so the host can send us data. */
- nr_added_bufs = fill_queue(port->in_vq, &port->inbuf_lock);
- if (!nr_added_bufs) {
+ /* We can safely ignore ENOSPC because it means
+ * the queue already has buffers. Buffers are removed
+ * only by virtcons_remove(), not by unplug_port()
+ */
+ err = fill_queue(port->in_vq, &port->inbuf_lock);
+ if (err < 0 && err != -ENOSPC) {
dev_err(port->dev, "Error allocating inbufs\n");
- err = -ENOMEM;
goto free_device;
}
@@ -2059,14 +2060,11 @@ static int virtcons_probe(struct virtio_device *vdev)
INIT_WORK(&portdev->control_work, &control_work_handler);
if (multiport) {
- unsigned int nr_added_bufs;
-
spin_lock_init(&portdev->c_ivq_lock);
spin_lock_init(&portdev->c_ovq_lock);
- nr_added_bufs = fill_queue(portdev->c_ivq,
- &portdev->c_ivq_lock);
- if (!nr_added_bufs) {
+ err = fill_queue(portdev->c_ivq, &portdev->c_ivq_lock);
+ if (err < 0) {
dev_err(&vdev->dev,
"Error allocating buffers for control queue\n");
/*
@@ -2077,7 +2075,7 @@ static int virtcons_probe(struct virtio_device *vdev)
VIRTIO_CONSOLE_DEVICE_READY, 0);
/* Device was functional: we need full cleanup. */
virtcons_remove(vdev);
- return -ENOMEM;
+ return err;
}
} else {
/*
--
2.23.0
On Thu, 2019-11-14 at 13:25 +0100, Laurent Vivier wrote:
> When we hot unplug a virtserialport and then try to hot plug again,
> it fails:
>
> (qemu) chardev-add socket,id=serial0,path=/tmp/serial0,server,nowait
> (qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
> chardev=serial0,id=serial0,name=serial0
> (qemu) device_del serial0
> (qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
> chardev=serial0,id=serial0,name=serial0
> kernel error:
> virtio-ports vport2p2: Error allocating inbufs
> qemu error:
> virtio-serial-bus: Guest failure in adding port 2 for device \
> virtio-serial0.0
>
> This happens because buffers for the in_vq are allocated when the
> port is
> added but are not released when the port is unplugged.
>
> They are only released when virtconsole is removed (see a7a69ec0d8e4)
>
> To avoid the problem and to be symmetric, we could allocate all the
> buffers
> in init_vqs() as they are released in remove_vqs(), but it sounds
> like
> a waste of memory.
>
> Rather than that, this patch changes add_port() logic to ignore
> ENOSPC
> error in fill_queue(), which means queue has already been filled.
>
> Fixes: a7a69ec0d8e4 ("virtio_console: free buffers after reset")
> Cc: [email protected]
> Cc: [email protected]
> Signed-off-by: Laurent Vivier <[email protected]>
Reviewed-by: Amit Shah <[email protected]>
Thanks!
> ---
>
> Notes:
> v3: add a comment about ENOSPC error
> v2: making fill_queue return int and testing return code for
> -ENOSPC
>
> drivers/char/virtio_console.c | 28 +++++++++++++---------------
> 1 file changed, 13 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/char/virtio_console.c
> b/drivers/char/virtio_console.c
> index 7270e7b69262..3259426f01dc 100644
> --- a/drivers/char/virtio_console.c
> +++ b/drivers/char/virtio_console.c
> @@ -1325,24 +1325,24 @@ static void set_console_size(struct port
> *port, u16 rows, u16 cols)
> port->cons.ws.ws_col = cols;
> }
>
> -static unsigned int fill_queue(struct virtqueue *vq, spinlock_t
> *lock)
> +static int fill_queue(struct virtqueue *vq, spinlock_t *lock)
> {
> struct port_buffer *buf;
> - unsigned int nr_added_bufs;
> + int nr_added_bufs;
> int ret;
>
> nr_added_bufs = 0;
> do {
> buf = alloc_buf(vq->vdev, PAGE_SIZE, 0);
> if (!buf)
> - break;
> + return -ENOMEM;
>
> spin_lock_irq(lock);
> ret = add_inbuf(vq, buf);
> if (ret < 0) {
> spin_unlock_irq(lock);
> free_buf(buf, true);
> - break;
> + return ret;
> }
> nr_added_bufs++;
> spin_unlock_irq(lock);
> @@ -1362,7 +1362,6 @@ static int add_port(struct ports_device
> *portdev, u32 id)
> char debugfs_name[16];
> struct port *port;
> dev_t devt;
> - unsigned int nr_added_bufs;
> int err;
>
> port = kmalloc(sizeof(*port), GFP_KERNEL);
> @@ -1421,11 +1420,13 @@ static int add_port(struct ports_device
> *portdev, u32 id)
> spin_lock_init(&port->outvq_lock);
> init_waitqueue_head(&port->waitqueue);
>
> - /* Fill the in_vq with buffers so the host can send us data. */
> - nr_added_bufs = fill_queue(port->in_vq, &port->inbuf_lock);
> - if (!nr_added_bufs) {
> + /* We can safely ignore ENOSPC because it means
> + * the queue already has buffers. Buffers are removed
> + * only by virtcons_remove(), not by unplug_port()
> + */
> + err = fill_queue(port->in_vq, &port->inbuf_lock);
> + if (err < 0 && err != -ENOSPC) {
> dev_err(port->dev, "Error allocating inbufs\n");
> - err = -ENOMEM;
> goto free_device;
> }
>
> @@ -2059,14 +2060,11 @@ static int virtcons_probe(struct
> virtio_device *vdev)
> INIT_WORK(&portdev->control_work, &control_work_handler);
>
> if (multiport) {
> - unsigned int nr_added_bufs;
> -
> spin_lock_init(&portdev->c_ivq_lock);
> spin_lock_init(&portdev->c_ovq_lock);
>
> - nr_added_bufs = fill_queue(portdev->c_ivq,
> - &portdev->c_ivq_lock);
> - if (!nr_added_bufs) {
> + err = fill_queue(portdev->c_ivq, &portdev->c_ivq_lock);
> + if (err < 0) {
> dev_err(&vdev->dev,
> "Error allocating buffers for control
> queue\n");
> /*
> @@ -2077,7 +2075,7 @@ static int virtcons_probe(struct virtio_device
> *vdev)
> VIRTIO_CONSOLE_DEVICE_READY,
> 0);
> /* Device was functional: we need full cleanup.
> */
> virtcons_remove(vdev);
> - return -ENOMEM;
> + return err;
> }
> } else {
> /*
On Tue, Dec 03, 2019 at 03:46:50PM +0100, Amit Shah wrote:
> On Thu, 2019-11-14 at 13:25 +0100, Laurent Vivier wrote:
> > When we hot unplug a virtserialport and then try to hot plug again,
> > it fails:
> >
> > (qemu) chardev-add socket,id=serial0,path=/tmp/serial0,server,nowait
> > (qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
> > chardev=serial0,id=serial0,name=serial0
> > (qemu) device_del serial0
> > (qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
> > chardev=serial0,id=serial0,name=serial0
> > kernel error:
> > virtio-ports vport2p2: Error allocating inbufs
> > qemu error:
> > virtio-serial-bus: Guest failure in adding port 2 for device \
> > virtio-serial0.0
> >
> > This happens because buffers for the in_vq are allocated when the
> > port is
> > added but are not released when the port is unplugged.
> >
> > They are only released when virtconsole is removed (see a7a69ec0d8e4)
> >
> > To avoid the problem and to be symmetric, we could allocate all the
> > buffers
> > in init_vqs() as they are released in remove_vqs(), but it sounds
> > like
> > a waste of memory.
> >
> > Rather than that, this patch changes add_port() logic to ignore
> > ENOSPC
> > error in fill_queue(), which means queue has already been filled.
> >
> > Fixes: a7a69ec0d8e4 ("virtio_console: free buffers after reset")
> > Cc: [email protected]
> > Cc: [email protected]
> > Signed-off-by: Laurent Vivier <[email protected]>
>
> Reviewed-by: Amit Shah <[email protected]>
>
> Thanks!
Thanks, however this has already been merged by Linus.
I can't add the tag retroactively, sorry about that.
For bugfix patches like that, I think we can reasonably
target a turn around of a couple of days, these
shouldn't really need to wait weeks for review.
> > ---
> >
> > Notes:
> > v3: add a comment about ENOSPC error
> > v2: making fill_queue return int and testing return code for
> > -ENOSPC
> >
> > drivers/char/virtio_console.c | 28 +++++++++++++---------------
> > 1 file changed, 13 insertions(+), 15 deletions(-)
> >
> > diff --git a/drivers/char/virtio_console.c
> > b/drivers/char/virtio_console.c
> > index 7270e7b69262..3259426f01dc 100644
> > --- a/drivers/char/virtio_console.c
> > +++ b/drivers/char/virtio_console.c
> > @@ -1325,24 +1325,24 @@ static void set_console_size(struct port
> > *port, u16 rows, u16 cols)
> > port->cons.ws.ws_col = cols;
> > }
> >
> > -static unsigned int fill_queue(struct virtqueue *vq, spinlock_t
> > *lock)
> > +static int fill_queue(struct virtqueue *vq, spinlock_t *lock)
> > {
> > struct port_buffer *buf;
> > - unsigned int nr_added_bufs;
> > + int nr_added_bufs;
> > int ret;
> >
> > nr_added_bufs = 0;
> > do {
> > buf = alloc_buf(vq->vdev, PAGE_SIZE, 0);
> > if (!buf)
> > - break;
> > + return -ENOMEM;
> >
> > spin_lock_irq(lock);
> > ret = add_inbuf(vq, buf);
> > if (ret < 0) {
> > spin_unlock_irq(lock);
> > free_buf(buf, true);
> > - break;
> > + return ret;
> > }
> > nr_added_bufs++;
> > spin_unlock_irq(lock);
> > @@ -1362,7 +1362,6 @@ static int add_port(struct ports_device
> > *portdev, u32 id)
> > char debugfs_name[16];
> > struct port *port;
> > dev_t devt;
> > - unsigned int nr_added_bufs;
> > int err;
> >
> > port = kmalloc(sizeof(*port), GFP_KERNEL);
> > @@ -1421,11 +1420,13 @@ static int add_port(struct ports_device
> > *portdev, u32 id)
> > spin_lock_init(&port->outvq_lock);
> > init_waitqueue_head(&port->waitqueue);
> >
> > - /* Fill the in_vq with buffers so the host can send us data. */
> > - nr_added_bufs = fill_queue(port->in_vq, &port->inbuf_lock);
> > - if (!nr_added_bufs) {
> > + /* We can safely ignore ENOSPC because it means
> > + * the queue already has buffers. Buffers are removed
> > + * only by virtcons_remove(), not by unplug_port()
> > + */
> > + err = fill_queue(port->in_vq, &port->inbuf_lock);
> > + if (err < 0 && err != -ENOSPC) {
> > dev_err(port->dev, "Error allocating inbufs\n");
> > - err = -ENOMEM;
> > goto free_device;
> > }
> >
> > @@ -2059,14 +2060,11 @@ static int virtcons_probe(struct
> > virtio_device *vdev)
> > INIT_WORK(&portdev->control_work, &control_work_handler);
> >
> > if (multiport) {
> > - unsigned int nr_added_bufs;
> > -
> > spin_lock_init(&portdev->c_ivq_lock);
> > spin_lock_init(&portdev->c_ovq_lock);
> >
> > - nr_added_bufs = fill_queue(portdev->c_ivq,
> > - &portdev->c_ivq_lock);
> > - if (!nr_added_bufs) {
> > + err = fill_queue(portdev->c_ivq, &portdev->c_ivq_lock);
> > + if (err < 0) {
> > dev_err(&vdev->dev,
> > "Error allocating buffers for control
> > queue\n");
> > /*
> > @@ -2077,7 +2075,7 @@ static int virtcons_probe(struct virtio_device
> > *vdev)
> > VIRTIO_CONSOLE_DEVICE_READY,
> > 0);
> > /* Device was functional: we need full cleanup.
> > */
> > virtcons_remove(vdev);
> > - return -ENOMEM;
> > + return err;
> > }
> > } else {
> > /*
On Tue, 2019-12-03 at 10:42 -0500, Michael S. Tsirkin wrote:
> On Tue, Dec 03, 2019 at 03:46:50PM +0100, Amit Shah wrote:
> > On Thu, 2019-11-14 at 13:25 +0100, Laurent Vivier wrote:
> > > When we hot unplug a virtserialport and then try to hot plug
> > > again,
> > > it fails:
> > >
> > > (qemu) chardev-add
> > > socket,id=serial0,path=/tmp/serial0,server,nowait
> > > (qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
> > > chardev=serial0,id=serial0,name=serial0
> > > (qemu) device_del serial0
> > > (qemu) device_add virtserialport,bus=virtio-serial0.0,nr=2,\
> > > chardev=serial0,id=serial0,name=serial0
> > > kernel error:
> > > virtio-ports vport2p2: Error allocating inbufs
> > > qemu error:
> > > virtio-serial-bus: Guest failure in adding port 2 for device \
> > > virtio-serial0.0
> > >
> > > This happens because buffers for the in_vq are allocated when the
> > > port is
> > > added but are not released when the port is unplugged.
> > >
> > > They are only released when virtconsole is removed (see
> > > a7a69ec0d8e4)
> > >
> > > To avoid the problem and to be symmetric, we could allocate all
> > > the
> > > buffers
> > > in init_vqs() as they are released in remove_vqs(), but it sounds
> > > like
> > > a waste of memory.
> > >
> > > Rather than that, this patch changes add_port() logic to ignore
> > > ENOSPC
> > > error in fill_queue(), which means queue has already been filled.
> > >
> > > Fixes: a7a69ec0d8e4 ("virtio_console: free buffers after reset")
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Signed-off-by: Laurent Vivier <[email protected]>
> >
> > Reviewed-by: Amit Shah <[email protected]>
> >
> > Thanks!
>
>
> Thanks, however this has already been merged by Linus.
> I can't add the tag retroactively, sorry about that.
Right, no problem - but I wanted to ensure it's on-list :)
>
> For bugfix patches like that, I think we can reasonably
> target a turn around of a couple of days, these
> shouldn't really need to wait weeks for review.
Sure, thanks for picking it up fast enough. Life happens, etc., and
it's not always possible to reply in a couple of days. Since we had
already covered the main comments in v1 and v2, v3 wasn't going to need
much attention anyway.