Hi Sakari,
On Thu, Feb 16, 2023 at 10:12:22AM +0200, Sakari Ailus wrote:
> On Wed, Feb 15, 2023 at 09:23:05PM +0200, Laurent Pinchart wrote:
> > On Wed, Feb 15, 2023 at 02:00:46PM +0200, Sakari Ailus wrote:
> > > On Fri, Feb 10, 2023 at 12:34:01AM +0200, Laurent Pinchart wrote:
> > > > Thank you for the patch.
> > >
> > > Thanks for the review!
> > >
> > > This was indeed hastily written, to help debugging a particular issue. But
> > > I hope it'll be useful for other purposes, too. V4L2 async is about to get
> > > more complicated soon.
> >
> > Could it get simpler instead ? :-) Maybe one day v4l2-async may cross
> > the threshold of how much pain I can bear, and I'll rewrite it...
>
> I wish it could, but often supporting complex needs is complicated.
> "Simplicated" is not even a proper word after all. Let's see.
Don't try to lure me into rewriting it sooner than later ;-)
> > > > On Fri, Feb 10, 2023 at 12:16:34AM +0200, Sakari Ailus wrote:
> > > > > Just add some debug prints for V4L2 async sub-device matching process. These
> > > > > might come useful in figuring out why things don't work as expected.
> > > > >
> > > > > Signed-off-by: Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx>
> > > > > ---
> > > > > Frieder,
> > > > >
> > > > > Can you try this? It prints what is being matched with what. Perhaps this
> > > > > could be merged in a bit more refined form if it proves useful.
> > > > >
> > > > > Not tested in any way.
> > > > >
> > > > > drivers/media/v4l2-core/v4l2-async.c | 41 ++++++++++++++++++++++++----
> > > > > 1 file changed, 36 insertions(+), 5 deletions(-)
> > > > >
> > > > > diff --git a/drivers/media/v4l2-core/v4l2-async.c b/drivers/media/v4l2-core/v4l2-async.c
> > > > > index 2f1b718a9189..6c13a9488415 100644
> > > > > --- a/drivers/media/v4l2-core/v4l2-async.c
> > > > > +++ b/drivers/media/v4l2-core/v4l2-async.c
> > > > > @@ -86,13 +86,18 @@ match_fwnode_one(struct v4l2_async_notifier *notifier,
> > > > > bool sd_fwnode_is_ep;
> > > > > struct device *dev;
> > > > >
> > > > > + dev_dbg(sd->dev, "async fwnode match %pfw vs. %pfw\n", sd_fwnode,
> > > > > + asd->match.fwnode);
> > > >
> > > > Let's be more explicit:
> > > >
> > > > dev_dbg(sd->dev, "async fwnode match: need %pfw, trying %pfw\n",
> > > > sd_fwnode, asd->match.fwnode);
> > > >
> > > > (feel free to adjust, as long as we differentiate what we're looking for
> > > > from what we're testing)
> > >
> > > Yes.
> > >
> > > > > +
> > > > > /*
> > > > > * Both the subdev and the async subdev can provide either an endpoint
> > > > > * fwnode or a device fwnode. Start with the simple case of direct
> > > > > * fwnode matching.
> > > > > */
> > > > > - if (sd_fwnode == asd->match.fwnode)
> > > > > + if (sd_fwnode == asd->match.fwnode) {
> > > > > + dev_dbg(sd->dev, "true\n");
> > > >
> > > > dev_dbg(sd->dev, "direct match found\n");
> > > >
> > > > > return true;
> > > > > + }
> > > > >
> > > > > /*
> > > > > * Otherwise, check if the sd fwnode and the asd fwnode refer to an
> > > > > @@ -105,8 +110,12 @@ match_fwnode_one(struct v4l2_async_notifier *notifier,
> > > > > sd_fwnode_is_ep = fwnode_graph_is_endpoint(sd_fwnode);
> > > > > asd_fwnode_is_ep = fwnode_graph_is_endpoint(asd->match.fwnode);
> > > > >
> > > > > - if (sd_fwnode_is_ep == asd_fwnode_is_ep)
> > > > > + dev_dbg(sd->dev, "async fwnode match %pfw vs. %pfw\n", sd_fwnode,
> > > > > + asd->match.fwnode);
> > > >
> > > > You've already printed this above, no need to repeat it.
> > > >
> > > > > + if (sd_fwnode_is_ep == asd_fwnode_is_ep) {
> > > > > + dev_dbg(sd->dev, "unmatching node types (false)\n");
> > > >
> > > > dev_dbg(sd->dev, "direct match not found\n");
> > > >
> > > > > return false;
> > > > > + }
> > > > >
> > > > > /*
> > > > > * The sd and asd fwnodes are of different types. Get the device fwnode
> > > > > @@ -120,10 +129,15 @@ match_fwnode_one(struct v4l2_async_notifier *notifier,
> > > > > other_fwnode = sd_fwnode;
> > > > > }
> > > > >
> > > > > + dev_dbg(sd->dev, "async fwnode (compat) match %pfw vs. %pfw\n",
> > > > > + dev_fwnode, other_fwnode);
> > > >
> > > > Same comment as above regarding "vs." not telling which is which.
> > > >
> > > > > +
> > > > > fwnode_handle_put(dev_fwnode);
> > > > >
> > > > > - if (dev_fwnode != other_fwnode)
> > > > > + if (dev_fwnode != other_fwnode) {
> > > > > + dev_dbg(sd->dev, "false\n");
> > > >
> > > > dev_dbg(sd->dev, "compat match not found\n");
> > > >
> > > > > return false;
> > > > > + }
> > > > >
> > > > > /*
> > > > > * We have a heterogeneous match. Retrieve the struct device of the side
> > > > > @@ -143,6 +157,8 @@ match_fwnode_one(struct v4l2_async_notifier *notifier,
> > > > > dev->driver->name);
> > > > > }
> > > > >
> > > > > + dev_dbg(sd->dev, "true\n");
> > > >
> > > > dev_dbg(sd->dev, "compat match found\n");
> > > >
> > > > > +
> > > > > return true;
> > > > > }
> > > > >
> > > > > @@ -255,7 +271,10 @@ v4l2_async_nf_can_complete(struct v4l2_async_notifier *notifier)
> > > > > v4l2_async_find_subdev_notifier(sd);
> > > > >
> > > > > if (subdev_notifier &&
> > > > > - !v4l2_async_nf_can_complete(subdev_notifier))
> > > > > + !v4l2_async_nf_can_complete(subdev_notifier)) {
> > > > > + if (subdev_notifier->sd)
> > > > > + deb_dbg(subdev_notifier->sd->dev,
> > > > > + "cannot complete\n");
> > > >
> > > > I'd add a reference to v4l2-async, either directly in the string, or
> > > > with a "%s: ", __func__ prefix. Otherwise the message will be easy to
> > > > miss. Same in other messages. Maybe a "v4l2-async: " prefix for all
> > > > debug messages ?
> > >
> > > How about just "async: " for all of these? It's shorther, still unique, and
> > > these lines will be long.
> >
> > "async" is a bit vague, I think you should mention V4L2 too. If this
> > ends up printing
> >
> > 3-0010 async: cannot complete
> >
> > someone may wonder what it relates to. We're talking about debugging
> > messages here, let's make sure they improve debugging as much as
> > possible.
>
> These messages include the device name which already implies what it is
> about, so I'm frankly not concerned about this. Also these messages tend to
> occur in series. Having them shorter, instead, is a small plus. An example:
>
> [ 5.716093] dw9807 i2c-PRP0001:01: async: matching for notifier \_SB.PCI0.CIO2, sd \_SB.PCI0.I2C2.VCM0
> [ 5.716109] dw9807 i2c-PRP0001:01: async: fwnode match: need \_SB.PCI0.I2C2.VCM0, trying \_SB_.PCI0.I2C2.CAM0/port@0/endpoint@0
> [ 5.716123] dw9807 i2c-PRP0001:01: async: fwnode compat match, need \_SB_.PCI0.I2C2.CAM0, trying \_SB.PCI0.I2C2.VCM0
> [ 5.716131] dw9807 i2c-PRP0001:01: async: compat match not found
> [ 30.333138] imx258 i2c-SONY258A:00: async: matching for notifier \_SB.PCI0.CIO2, sd \_SB.PCI0.I2C2.CAM0
> [ 30.333154] imx258 i2c-SONY258A:00: async: fwnode match: need \_SB.PCI0.I2C2.CAM0, trying \_SB_.PCI0.I2C2.CAM0/port@0/endpoint@0
> [ 30.333168] imx258 i2c-SONY258A:00: async: fwnode compat match, need \_SB_.PCI0.I2C2.CAM0, trying \_SB.PCI0.I2C2.CAM0
> [ 30.333175] imx258 i2c-SONY258A:00: async: compat match not found
> [ 30.333178] imx258 i2c-SONY258A:00: async: trying secondary fwnode match
> [ 30.333181] imx258 i2c-SONY258A:00: async: fwnode match: need \_SB_.PCI0.I2C2.CAM0, trying \_SB_.PCI0.I2C2.CAM0/port@0/endpoint@0
> [ 30.333189] imx258 i2c-SONY258A:00: async: fwnode compat match, need \_SB_.PCI0.I2C2.CAM0, trying \_SB_.PCI0.I2C2.CAM0
> [ 30.333196] imx258 i2c-SONY258A:00: async: compat match found
> [ 30.333214] dw9807 i2c-PRP0001:01: async: matching for notifier \_SB.PCI0.I2C2.CAM0, sd \_SB.PCI0.I2C2.VCM0
> [ 30.333225] dw9807 i2c-PRP0001:01: async: fwnode match: need \_SB.PCI0.I2C2.VCM0, trying \_SB.PCI0.I2C2.VCM0
> [ 30.333235] dw9807 i2c-PRP0001:01: async: direct match found
> [ 30.333248] dw9807 i2c-PRP0001:01: async: bound to i2c-SONY258A:00's notifier (ret 0)
> [ 30.333252] ipu3-cio2 0000:00:14.3: async: complete
I would still prefer v4l2-async, but I won't nack the patch just for
that.
> > > > > return false;
> > > > > }
> > > > >
> > > > > @@ -273,18 +292,27 @@ v4l2_async_nf_try_complete(struct v4l2_async_notifier *notifier)
> > > > > if (!list_empty(¬ifier->waiting))
> > > > > return 0;
> > > > >
> > > > > + if (notifier->sd)
> > > > > + deb_dbg(notifier->sd->dev, "trying to complete\n");
> > > > > +
> > > > > /* Check the entire notifier tree; find the root notifier first. */
> > > > > while (notifier->parent)
> > > > > notifier = notifier->parent;
> > > > >
> > > > > /* This is root if it has v4l2_dev. */
> > > > > - if (!notifier->v4l2_dev)
> > > > > + if (!notifier->v4l2_dev) {
> > > > > + if (notifier->sd)
> > > > > + deb_dbg(notifier->sd->dev,
> > > > > + "V4L2 device not available\n");
> > > > > return 0;
> > > > > + }
> > > > >
> > > > > /* Is everything ready? */
> > > > > if (!v4l2_async_nf_can_complete(notifier))
> > > > > return 0;
> > > > >
> > > > > + deb_dbg(notifier->sd->dev, "complete\n");
> > > >
> > > > You guard against notifier->sd being NULL above, but not here. At least
> > > > one of the two is wrong.
> > >
> > > I'll add a helper function to get the device safely.
> > >
> > > > > +
> > > > > return v4l2_async_nf_call_complete(notifier);
> > > > > }
> > > > >
> > > > > @@ -386,6 +414,9 @@ v4l2_async_nf_try_all_subdevs(struct v4l2_async_notifier *notifier)
> > > > > continue;
> > > > >
> > > > > ret = v4l2_async_match_notify(notifier, v4l2_dev, sd, asd);
> > > > > + deb_dbg(sd->dev, "bound to %s's notifier (ret %d)\n",
> > > > > + notifier->sd ? dev_name(notifier->sd->dev) : "no-dev",
> > > > > + ret);
> > > > > if (ret < 0)
> > > > > return ret;
> > > > >
--
Regards,
Laurent Pinchart
