On Mon, May 28, 2012 at 7:22 AM, Hans Verkuil hverk...@xs4all.nl wrote:
In practice it seems that the easiest approach is not to clean up anything in
the
disconnect, just take the lock, do the bare minimum necessary for the
disconnect,
unregister the video nodes, unlock and end with
On Tue 29 May 2012 14:05:08 Ezequiel Garcia wrote:
On Mon, May 28, 2012 at 7:22 AM, Hans Verkuil hverk...@xs4all.nl wrote:
In practice it seems that the easiest approach is not to clean up anything
in the
disconnect, just take the lock, do the bare minimum necessary for the
On Sun May 27 2012 23:20:21 Ezequiel Garcia wrote:
Hi Hans,
On Sat, May 26, 2012 at 2:05 PM, Hans Verkuil hverk...@xs4all.nl wrote:
Always test your driver using the v4l2-compliance test tool that it part of
v4l-utils! If it passes that, then your code will be in really good shape!
Hi Hans,
On Sat, May 26, 2012 at 2:05 PM, Hans Verkuil hverk...@xs4all.nl wrote:
Always test your driver using the v4l2-compliance test tool that it part of
v4l-utils! If it passes that, then your code will be in really good shape!
You're right. I'll add v4l2-compliance output in the next
Hi Sylwester,
On Sat, May 26, 2012 at 4:31 PM, Sylwester Nawrocki snj...@gmail.com wrote:
You can drop this line, it's overwritten with KERNEL_VERSION in v4l2-ioctl.c.
Also I could imagine there might be better names, than dev, for
capabilities.
Yes, indeed. Starting with cap.
+
Hi Ezequiel,
On 05/27/2012 11:32 PM, Ezequiel Garcia wrote:
+static int buffer_prepare(struct vb2_buffer *vb)
+{
+struct stk1160 *dev = vb2_get_drv_priv(vb-vb2_queue);
+struct stk1160_buffer *buf =
+container_of(vb, struct stk1160_buffer, vb);
+
+/* If the
(Ezequiel, your original CC list was mangled, so I'm reposting this)
Thanks!
Here is a quick run through of the code.
Always test your driver using the v4l2-compliance test tool that it part of
v4l-utils! If it passes that, then your code will be in really good shape!
On Sat May 26 2012
Hi Hans,
Thanks for your review (I'm a bit amazed at how fast you went through
the code :).
I'll address your excellent comments soon.
I'm still unsure about a numbre of things. Two of them:
1. It seems to mee tracing is not too nice and
I wasn't really sure how to handle it: dev_xxx, pr_xxx,
Hi Ezequiel,
On 05/26/2012 07:50 PM, Hans Verkuil wrote:
+/*
+ * IOCTL vidioc handling
+ */
+static int vidioc_reqbufs(struct file *file, void *priv,
+ struct v4l2_requestbuffers *p)
+{
+struct stk1160 *dev = video_drvdata(file);
+return