if dsa device supports custom selftests than we should use custom
selftest strings for ethtool.

Signed-off-by: Ante Knezic <ante.knezic@helmholz.de>
---
 net/dsa/user.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/net/dsa/user.c b/net/dsa/user.c
index d438884a4eb0..d0e0d1a2bff7 100644
--- a/net/dsa/user.c
+++ b/net/dsa/user.c
@@ -1072,7 +1072,11 @@ static void dsa_user_get_strings(struct net_device *dev,
 			ds->ops->get_strings(ds, dp->index, stringset,
 					     data + 4 * len);
 	} else if (stringset ==  ETH_SS_TEST) {
-		net_selftest_get_strings(data);
+		if (ds->ops->self_test && ds->ops->get_strings)
+			ds->ops->get_strings(ds, dp->index, stringset,
+					     data);
+		else
+			net_selftest_get_strings(data);
 	}
 
 }
@@ -1123,7 +1127,10 @@ static int dsa_user_get_sset_count(struct net_device *dev, int sset)
 
 		return count + 4;
 	} else if (sset ==  ETH_SS_TEST) {
-		return net_selftest_get_count();
+		if (ds->ops->self_test && ds->ops->get_sset_count)
+			return ds->ops->get_sset_count(ds, dp->index, sset);
+		else
+			return net_selftest_get_count();
 	}
 
 	return -EOPNOTSUPP;
-- 
2.11.0

On Thu, Dec 14, 2023 at 03:21:36PM +0100, Ante Knezic wrote:
> if dsa device supports custom selftests than we should use custom
> selftest strings for ethtool.
> 
> Signed-off-by: Ante Knezic <ante.knezic@helmholz.de>
> ---

I didn't notice when the selftest support was added that there is no
implementation in DSA drivers of custom ds->ops->self_test(). Adding
interfaces with no users is frowned upon, precisely because it doesn't
show the big picture.

You must have noticed this because you do have a driver implementation,
so would you mind posting it together with this fix?

Indeed I do have a custom implementation for the mv88e6xxx chip, but its not
come to state to be posted because of test/chip specifics.

> I didn't notice when the selftest support was added that there is no
> implementation in DSA drivers of custom ds->ops->self_test(). Adding
> interfaces with no users is frowned upon, precisely because it doesn't
> show the big picture.

I was not aware of this, I apologize. If this is the case, perhaps this patch
should wait for the first custom self test implementation and be reposted as
a part of bigger series.

Thanks,
Ante

On Thu, Dec 14, 2023 at 03:47:51PM +0100, Ante Knezic wrote:
> Indeed I do have a custom implementation for the mv88e6xxx chip, but its not
> come to state to be posted because of test/chip specifics.
> 
> > I didn't notice when the selftest support was added that there is no
> > implementation in DSA drivers of custom ds->ops->self_test(). Adding
> > interfaces with no users is frowned upon, precisely because it doesn't
> > show the big picture.
> 
> I was not aware of this, I apologize. If this is the case, perhaps this patch
> should wait for the first custom self test implementation and be reposted as
> a part of bigger series.
> 
> Thanks,
> Ante
> 

I agree this should be resubmitted with a user of the API. Looking
forward to seeing it. Thanks for the understanding.

The only thing I would want to comment on the patch as it is is to avoid
the pattern of:

	if (a)
		return x;
	else
		return y;

and formulate it as:

	if (a)
		return x;

	return y;

instead. The "else" is redundant for an "if" that ends with a return statement.

---
pw-bot: changes-requested