Some drivers, like tg3, do not set combined-count:

$ ethtool -l enp4s0f1
Channel parameters for enp4s0f1:
Pre-set maximums:
RX:		4
TX:		4
Other:		n/a
Combined:	n/a
Current hardware settings:
RX:		4
TX:		1
Other:		n/a
Combined:	n/a

In the case where combined-count is not set, the ethtool netlink code
in the kernel elides the value and the code in the test:

  netnl.channels_get(...)

With a tg3 device, the returned dictionary looks like:

{'header': {'dev-index': 3, 'dev-name': 'enp4s0f1'},
 'rx-max': 4,
 'rx-count': 4,
 'tx-max': 4,
 'tx-count': 1}

Note that the key 'combined-count' is missing. As a result of this
missing key the test raises an exception:

 # Exception|     if channels['combined-count'] == 0:
 # Exception|        ~~~~~~~~^^^^^^^^^^^^^^^^^^
 # Exception| KeyError: 'combined-count'

Change the test to check if 'combined-count' is a key in the dictionary
first and if not assume that this means the driver has separate RX and
TX queues.

With this change, the test now passes successfully on tg3 and mlx5
(which does have a 'combined-count').

Fixes: 1cf270424218 ("net: selftest: add test for netdev netlink queue-get API")
Signed-off-by: Joe Damato <jdamato@fastly.com>
---
 v2:
   - Simplify logic and reduce indentation as suggested by David Wei.
     Retested on both tg3 and mlx5 and test passes as expected.

 v1: https://lore.kernel.org/lkml/20250225181455.224309-1-jdamato@fastly.com/

 tools/testing/selftests/drivers/net/queues.py | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/drivers/net/queues.py b/tools/testing/selftests/drivers/net/queues.py
index 38303da957ee..8a518905a9f9 100755
--- a/tools/testing/selftests/drivers/net/queues.py
+++ b/tools/testing/selftests/drivers/net/queues.py
@@ -45,10 +45,9 @@ def addremove_queues(cfg, nl) -> None:
 
     netnl = EthtoolFamily()
     channels = netnl.channels_get({'header': {'dev-index': cfg.ifindex}})
-    if channels['combined-count'] == 0:
-        rx_type = 'rx'
-    else:
-        rx_type = 'combined'
+    rx_type = 'rx'
+    if channels.get('combined-count', 0) > 0:
+            rx_type = 'combined'
 
     expected = curr_queues - 1
     cmd(f"ethtool -L {cfg.dev['ifname']} {rx_type} {expected}", timeout=10)

base-commit: 8d52da23b6c68a0f6bad83959ebb61a2cf623c4e
-- 
2.43.0

On 2025-02-26 10:19, Joe Damato wrote:
> Some drivers, like tg3, do not set combined-count:
> 
> $ ethtool -l enp4s0f1
> Channel parameters for enp4s0f1:
> Pre-set maximums:
> RX:		4
> TX:		4
> Other:		n/a
> Combined:	n/a
> Current hardware settings:
> RX:		4
> TX:		1
> Other:		n/a
> Combined:	n/a
> 
> In the case where combined-count is not set, the ethtool netlink code
> in the kernel elides the value and the code in the test:
> 
>   netnl.channels_get(...)
> 
> With a tg3 device, the returned dictionary looks like:
> 
> {'header': {'dev-index': 3, 'dev-name': 'enp4s0f1'},
>  'rx-max': 4,
>  'rx-count': 4,
>  'tx-max': 4,
>  'tx-count': 1}
> 
> Note that the key 'combined-count' is missing. As a result of this
> missing key the test raises an exception:
> 
>  # Exception|     if channels['combined-count'] == 0:
>  # Exception|        ~~~~~~~~^^^^^^^^^^^^^^^^^^
>  # Exception| KeyError: 'combined-count'
> 
> Change the test to check if 'combined-count' is a key in the dictionary
> first and if not assume that this means the driver has separate RX and
> TX queues.
> 
> With this change, the test now passes successfully on tg3 and mlx5
> (which does have a 'combined-count').
> 
> Fixes: 1cf270424218 ("net: selftest: add test for netdev netlink queue-get API")
> Signed-off-by: Joe Damato <jdamato@fastly.com>
> ---
>  v2:
>    - Simplify logic and reduce indentation as suggested by David Wei.
>      Retested on both tg3 and mlx5 and test passes as expected.
> 
>  v1: https://lore.kernel.org/lkml/20250225181455.224309-1-jdamato@fastly.com/

Thanks Joe.

Reviewed-by: David Wei <dw@davidwei.uk>