* [PATCH] hw: virtio-mem: detach the element fromt the virtqueue when error occurs
@ 2020-08-13 16:46 Li Qiang
2020-08-13 17:15 ` David Hildenbrand
0 siblings, 1 reply; 4+ messages in thread
From: Li Qiang @ 2020-08-13 16:46 UTC (permalink / raw)
To: mst, david; +Cc: Li Qiang, liq3ea, qemu-devel
If error occurs while processing the virtio request we should call
'virtqueue_detach_element' to detach the element from the virtqueue
before free the elem.
Signed-off-by: Li Qiang <liq3ea@163.com>
---
hw/virtio/virtio-mem.c | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/hw/virtio/virtio-mem.c b/hw/virtio/virtio-mem.c
index 7740fc613f..5ac6c3ec67 100644
--- a/hw/virtio/virtio-mem.c
+++ b/hw/virtio/virtio-mem.c
@@ -318,8 +318,7 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
if (iov_to_buf(elem->out_sg, elem->out_num, 0, &req, len) < len) {
virtio_error(vdev, "virtio-mem protocol violation: invalid request"
" size: %d", len);
- g_free(elem);
- return;
+ goto out_free;
}
if (iov_size(elem->in_sg, elem->in_num) <
@@ -327,8 +326,7 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
virtio_error(vdev, "virtio-mem protocol violation: not enough space"
" for response: %zu",
iov_size(elem->in_sg, elem->in_num));
- g_free(elem);
- return;
+ goto out_free;
}
type = le16_to_cpu(req.type);
@@ -348,12 +346,15 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
default:
virtio_error(vdev, "virtio-mem protocol violation: unknown request"
" type: %d", type);
- g_free(elem);
- return;
+ goto out_free;
}
g_free(elem);
}
+
+out_free:
+ virtqueue_detach_element(vq, elem, 0);
+ g_free(elem);
}
static void virtio_mem_get_config(VirtIODevice *vdev, uint8_t *config_data)
--
2.17.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] hw: virtio-mem: detach the element fromt the virtqueue when error occurs
2020-08-13 16:46 [PATCH] hw: virtio-mem: detach the element fromt the virtqueue when error occurs Li Qiang
@ 2020-08-13 17:15 ` David Hildenbrand
2020-08-14 1:01 ` Li Qiang
0 siblings, 1 reply; 4+ messages in thread
From: David Hildenbrand @ 2020-08-13 17:15 UTC (permalink / raw)
To: Li Qiang, mst; +Cc: liq3ea, qemu-devel
On 13.08.20 18:46, Li Qiang wrote:
For now we use "virtio-mem:" for the subject, without the "hw: "part.
> If error occurs while processing the virtio request we should call
> 'virtqueue_detach_element' to detach the element from the virtqueue
> before free the elem.
What's the effect of this? In all cases we trigger a virtio_error(), so
do we really have to bother?
>
> Signed-off-by: Li Qiang <liq3ea@163.com>
> ---
> hw/virtio/virtio-mem.c | 13 +++++++------
> 1 file changed, 7 insertions(+), 6 deletions(-)
>
> diff --git a/hw/virtio/virtio-mem.c b/hw/virtio/virtio-mem.c
> index 7740fc613f..5ac6c3ec67 100644
> --- a/hw/virtio/virtio-mem.c
> +++ b/hw/virtio/virtio-mem.c
> @@ -318,8 +318,7 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
> if (iov_to_buf(elem->out_sg, elem->out_num, 0, &req, len) < len) {
> virtio_error(vdev, "virtio-mem protocol violation: invalid request"
> " size: %d", len);
> - g_free(elem);
> - return;
> + goto out_free;
> }
>
> if (iov_size(elem->in_sg, elem->in_num) <
> @@ -327,8 +326,7 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
> virtio_error(vdev, "virtio-mem protocol violation: not enough space"
> " for response: %zu",
> iov_size(elem->in_sg, elem->in_num));
> - g_free(elem);
> - return;
> + goto out_free;
> }
>
> type = le16_to_cpu(req.type);
> @@ -348,12 +346,15 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
> default:
> virtio_error(vdev, "virtio-mem protocol violation: unknown request"
> " type: %d", type);
> - g_free(elem);
> - return;
> + goto out_free;
> }
>
> g_free(elem);
> }
> +
> +out_free:
> + virtqueue_detach_element(vq, elem, 0);
> + g_free(elem);
> }
>
> static void virtio_mem_get_config(VirtIODevice *vdev, uint8_t *config_data)
>
--
Thanks,
David / dhildenb
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] hw: virtio-mem: detach the element fromt the virtqueue when error occurs
2020-08-13 17:15 ` David Hildenbrand
@ 2020-08-14 1:01 ` Li Qiang
2020-08-16 12:10 ` David Hildenbrand
0 siblings, 1 reply; 4+ messages in thread
From: Li Qiang @ 2020-08-14 1:01 UTC (permalink / raw)
To: David Hildenbrand
Cc: Paolo Bonzini, Li Qiang, Qemu Developers, Michael S. Tsirkin
David Hildenbrand <david@redhat.com> 于2020年8月14日周五 上午1:15写道:
>
> On 13.08.20 18:46, Li Qiang wrote:
>
> For now we use "virtio-mem:" for the subject, without the "hw: "part.
>
> > If error occurs while processing the virtio request we should call
> > 'virtqueue_detach_element' to detach the element from the virtqueue
> > before free the elem.
>
> What's the effect of this? In all cases we trigger a virtio_error(), so
> do we really have to bother?
>
Though the 'in_use' will be reset to 0 while reseting the virtio device.
The mapped sglist will not be unammped.
There maybe some undesired behavior. CC Paolo to make a confirmation.
Thanks,
Li Qiang
> >
> > Signed-off-by: Li Qiang <liq3ea@163.com>
> > ---
> > hw/virtio/virtio-mem.c | 13 +++++++------
> > 1 file changed, 7 insertions(+), 6 deletions(-)
> >
> > diff --git a/hw/virtio/virtio-mem.c b/hw/virtio/virtio-mem.c
> > index 7740fc613f..5ac6c3ec67 100644
> > --- a/hw/virtio/virtio-mem.c
> > +++ b/hw/virtio/virtio-mem.c
> > @@ -318,8 +318,7 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
> > if (iov_to_buf(elem->out_sg, elem->out_num, 0, &req, len) < len) {
> > virtio_error(vdev, "virtio-mem protocol violation: invalid request"
> > " size: %d", len);
> > - g_free(elem);
> > - return;
> > + goto out_free;
> > }
> >
> > if (iov_size(elem->in_sg, elem->in_num) <
> > @@ -327,8 +326,7 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
> > virtio_error(vdev, "virtio-mem protocol violation: not enough space"
> > " for response: %zu",
> > iov_size(elem->in_sg, elem->in_num));
> > - g_free(elem);
> > - return;
> > + goto out_free;
> > }
> >
> > type = le16_to_cpu(req.type);
> > @@ -348,12 +346,15 @@ static void virtio_mem_handle_request(VirtIODevice *vdev, VirtQueue *vq)
> > default:
> > virtio_error(vdev, "virtio-mem protocol violation: unknown request"
> > " type: %d", type);
> > - g_free(elem);
> > - return;
> > + goto out_free;
> > }
> >
> > g_free(elem);
> > }
> > +
> > +out_free:
> > + virtqueue_detach_element(vq, elem, 0);
> > + g_free(elem);
> > }
> >
> > static void virtio_mem_get_config(VirtIODevice *vdev, uint8_t *config_data)
> >
>
>
> --
> Thanks,
>
> David / dhildenb
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] hw: virtio-mem: detach the element fromt the virtqueue when error occurs
2020-08-14 1:01 ` Li Qiang
@ 2020-08-16 12:10 ` David Hildenbrand
0 siblings, 0 replies; 4+ messages in thread
From: David Hildenbrand @ 2020-08-16 12:10 UTC (permalink / raw)
To: Li Qiang; +Cc: Paolo Bonzini, Li Qiang, Qemu Developers, Michael S. Tsirkin
On 14.08.20 03:01, Li Qiang wrote:
> David Hildenbrand <david@redhat.com> 于2020年8月14日周五 上午1:15写道:
>>
>> On 13.08.20 18:46, Li Qiang wrote:
>>
>> For now we use "virtio-mem:" for the subject, without the "hw: "part.
>>
>>> If error occurs while processing the virtio request we should call
>>> 'virtqueue_detach_element' to detach the element from the virtqueue
>>> before free the elem.
>>
>> What's the effect of this? In all cases we trigger a virtio_error(), so
>> do we really have to bother?
>>
>
> Though the 'in_use' will be reset to 0 while reseting the virtio device.
> The mapped sglist will not be unammped.
> There maybe some undesired behavior. CC Paolo to make a confirmation.
Looking at hw/virtio/virtio-crypto.c, this seems to be the right thing
to do.
Can you please respin, avoiding adding the label, only inserting the 3
separate virtqueue_detach_element() calls?
Thanks!
--
Thanks,
David / dhildenb
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-08-16 12:11 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-13 16:46 [PATCH] hw: virtio-mem: detach the element fromt the virtqueue when error occurs Li Qiang
2020-08-13 17:15 ` David Hildenbrand
2020-08-14 1:01 ` Li Qiang
2020-08-16 12:10 ` David Hildenbrand
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.