Netdev List
 help / color / mirror / Atom feed
* [PATCH net v3] selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings
@ 2026-11-13  2:27 Mohsin Bashir
  2026-11-16  2:29 ` netdev-bot+sashiko
  0 siblings, 1 reply; 3+ messages in thread
From: Mohsin Bashir @ 2026-11-13  2:27 UTC (permalink / raw)
  To: netdev
  Cc: andrew+netdev, edumazet, kuba, pabeni, shuah, maxime.chevallier,
	linux-kselftest, hmohsin

From: Mohsin Bashir <hmohsin@meta.com>

pause_setup() saves the initial link settings before tests run. Devices
that report link settings but do not support changing them can reject even
a no-op restore with EOPNOTSUPP. That turns pause tests that use
pause_setup() into cleanup failures.

Fix this by queuing link-setting restores only after a link-setting
request has succeeded. Keep pause_test_support coverage for devices that
cannot change link settings by falling back to forced MAC pause on the
current link. Run the extra link autoneg on/off checks only when those
link setting requests are accepted. Tests that force link autoneg off,
including pause_autoneg_link_autoneg(), now queue their own autoneg-on
restore after the force request succeeds.

Results on fbnic:

  Before: # Totals: pass:0 fail:4 xfail:0 xpass:0 skip:24 error:0
  After:  # Totals: pass:4 fail:0 xfail:0 xpass:0 skip:24 error:0

Fixes: cbd8a536cd6d ("selftests: drv-net: Introduce a selftest for ethtool flow control")
Signed-off-by: Mohsin Bashir <hmohsin@meta.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
---
Changelog:
v3:
- Drop the remaining unconditional autoneg-on restore from pause_setup()
  and only queue link-setting restores after a link-setting request succeeds.
- Since pause_setup() no longer restores autoneg on, make pause_test_support()
  and pause_autoneg_link_autoneg() queue their own autoneg-on restores after
  they successfully force the link.

v2: https://lore.kernel.org/netdev/20261103190622.3981970-1-mohsin.bashr@gmail.com
v1: https://lore.kernel.org/netdev/20261030173516.1002250-1-mohsin.bashr@gmail.com
---
 .../testing/selftests/drivers/net/hw/pause.py | 56 ++++++++++---------
 1 file changed, 30 insertions(+), 26 deletions(-)

diff --git a/tools/testing/selftests/drivers/net/hw/pause.py b/tools/testing/selftests/drivers/net/hw/pause.py
index ca4a55e59c30..c3398c7452e0 100755
--- a/tools/testing/selftests/drivers/net/hw/pause.py
+++ b/tools/testing/selftests/drivers/net/hw/pause.py
@@ -402,16 +402,13 @@ def pause_setup(cfg):
     if "auto-negotiation" not in link:
         raise KsftSkipEx(f"{cfg.ifname} doesn't report its link modes")
 
-    if link["auto-negotiation"]:
-        defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")
-    elif "speed" in link and "duplex" in link:
-        defer(cmd, f"ethtool -s {cfg.ifname} autoneg off speed {link['speed']} "
-                   f"duplex {link['duplex'].lower()}")
-
     # Local interface admin up, link aneg on
     ip(f"link set {cfg.ifname} up")
     if link["supports-auto-negotiation"] and not link["auto-negotiation"]:
         ethtool(f"-s {cfg.ifname} autoneg on")
+        if "speed" in link and "duplex" in link:
+            defer(cmd, f"ethtool -s {cfg.ifname} autoneg off speed {link['speed']} "
+                       f"duplex {link['duplex'].lower()}")
 
     # Get remote pause params
     ret, params = ethtool_ret(f"-a {cfg.remote_ifname}", host=cfg.remote)
@@ -426,18 +423,14 @@ def pause_setup(cfg):
     if "auto-negotiation" not in link:
         raise KsftSkipEx(f"{cfg.remote_ifname} doesn't report its link modes")
 
-    if link["auto-negotiation"]:
-        defer(cmd, f"ethtool -s {cfg.remote_ifname} autoneg on",
-              host=cfg.remote)
-    elif "speed" in link and "duplex" in link:
-        defer(cmd, f"ethtool -s {cfg.remote_ifname} autoneg off "
-                   f"speed {link['speed']} duplex {link['duplex'].lower()}",
-                   host=cfg.remote)
-
     # Remote interface admin up, link aneg on
     ip(f"link set {cfg.remote_ifname} up", host=cfg.remote)
     if link["supports-auto-negotiation"] and not link["auto-negotiation"]:
         ethtool(f"-s {cfg.remote_ifname} autoneg on", host=cfg.remote)
+        if "speed" in link and "duplex" in link:
+            defer(cmd, f"ethtool -s {cfg.remote_ifname} autoneg off "
+                       f"speed {link['speed']} duplex {link['duplex'].lower()}",
+                       host=cfg.remote)
 
     wait_for_link(cfg)
 
@@ -470,9 +463,12 @@ def pause_test_support(cfg, pauseparams):
 
     forced = forced_link_settings(cfg)
     _, supported = get_local_pause_supported(cfg)
+    link_config_tested = False
 
     # We check that what we can configure in the pause params matches what we
-    # support under various contditions : Link aneg on/off, pause aneg on/off
+    # support under various conditions: link aneg on/off, and pause aneg
+    # on/off. If no link configuration change is supported, fall back to
+    # checking forced MAC pause on the current link.
     ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg on", is_get = False)
     if ret != 0:
         ksft_pr("link autoneg on refused, not tested")
@@ -481,18 +477,23 @@ def pause_test_support(cfg, pauseparams):
                                "link autoneg on")
         expect_pauseparams_set(cfg, pauseparams, True, supported,
                                "link autoneg on")
+        link_config_tested = True
 
     if not forced:
         ksft_pr("link speed unknown, the forced link is not tested")
-        return
-
-    ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}", is_get = False)
-    if ret != 0:
-        ksft_pr(f"link autoneg off {forced} refused, not tested")
-        return
-
-    expect_pauseparams_set(cfg, pauseparams, False, supported, "link autoneg off")
-    expect_pauseparams_set(cfg, pauseparams, True, supported, "link autoneg off")
+    else:
+        ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}", is_get = False)
+        if ret != 0:
+            ksft_pr(f"link autoneg off {forced} refused, not tested")
+        else:
+            if link_config_tested:
+                defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")
+            expect_pauseparams_set(cfg, pauseparams, False, supported, "link autoneg off")
+            expect_pauseparams_set(cfg, pauseparams, True, supported, "link autoneg off")
+            link_config_tested = True
+
+    if not link_config_tested:
+        expect_pauseparams_set(cfg, pauseparams, False, supported, "current link")
 
 @ksft_variants(pauseparams_variants)
 @ksft_disruptive
@@ -902,11 +903,14 @@ def pause_autoneg_link_autoneg(cfg):
                          is_get=False, host=cfg.remote)
     if ret != 0:
         raise KsftSkipEx(f"Can't force the peer's link at {forced}")
+    remote_restore = defer(cmd, f"ethtool -s {cfg.remote_ifname} autoneg on",
+                           host=cfg.remote)
 
     ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}",
                          is_get=False)
     if ret != 0:
         raise KsftSkipEx(f"Can't force the link at {forced}")
+    local_restore = defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")
 
     try:
         wait_for_aneg(cfg)
@@ -917,8 +921,8 @@ def pause_autoneg_link_autoneg(cfg):
     check_local_pauseparams(cfg, True, rx, tx)
 
     # Re-enable autoneg
-    ethtool(f"-s {cfg.remote_ifname} autoneg on", host=cfg.remote)
-    ethtool(f"-s {cfg.ifname} autoneg on")
+    remote_restore.exec()
+    local_restore.exec()
     wait_for_aneg(cfg)
 
     # Pause autoneg must still be on
-- 
2.53.0-Meta


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH net v3] selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings
  2026-11-13  2:27 [PATCH net v3] selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings Mohsin Bashir
@ 2026-11-16  2:29 ` netdev-bot+sashiko
  2026-11-16 16:53   ` Mohsin Bashir
  0 siblings, 1 reply; 3+ messages in thread
From: netdev-bot+sashiko @ 2026-11-16  2:29 UTC (permalink / raw)
  To: mohsin.bashr
  Cc: netdev, andrew+netdev, edumazet, kuba, pabeni, shuah,
	maxime.chevallier, linux-kselftest, hmohsin

Thank you for your contribution! Sashiko AI review found 1 potential
issue(s) to consider.

Critical: 0 · High: 0 · Medium: 1 · Low: 0

If you believe none of the issues are legitimate, please reply to each
issue of Medium severity or higher and explain why the AI is wrong.
If any of them are legitimate and you plan to respin, please let patchwork
know by including "pw-bot: cr" as a separate line at the end of your reply
(one such reply per series is enough).

- [Medium] The patch removes the unconditional restore of the pre-test
  link state from pause_setup(), and pause_test_support() does not…

--

selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings

This patch changes pause.py so that link-setting restores are queued only
after a link-setting request has succeeded. Devices that cannot change link
settings therefore no longer fail during cleanup. If no link configuration
change is accepted, pause_test_support() falls back to checking forced MAC
pause on the current link, and pause_autoneg_link_autoneg() now queues its
own autoneg-on restores.

> diff --git a/tools/testing/selftests/drivers/net/hw/pause.py b/tools/testing/selftests/drivers/net/hw/pause.py
> index ca4a55e59c30f..c3398c7452e01 100755
> --- a/tools/testing/selftests/drivers/net/hw/pause.py
> +++ b/tools/testing/selftests/drivers/net/hw/pause.py
> @@ -402,16 +402,13 @@ def pause_setup(cfg):
>      if "auto-negotiation" not in link:
>          raise KsftSkipEx(f"{cfg.ifname} doesn't report its link modes")
>  
> -    if link["auto-negotiation"]:
> -        defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")
> -    elif "speed" in link and "duplex" in link:
> -        defer(cmd, f"ethtool -s {cfg.ifname} autoneg off speed {link['speed']} "
> -                   f"duplex {link['duplex'].lower()}")
> -
>      # Local interface admin up, link aneg on
>      ip(f"link set {cfg.ifname} up")
>      if link["supports-auto-negotiation"] and not link["auto-negotiation"]:
>          ethtool(f"-s {cfg.ifname} autoneg on")
> +        if "speed" in link and "duplex" in link:
> +            defer(cmd, f"ethtool -s {cfg.ifname} autoneg off speed {link['speed']} "
> +                       f"duplex {link['duplex'].lower()}")

[ ... ]

> @@ -470,9 +463,12 @@ def pause_test_support(cfg, pauseparams):
>  
>      forced = forced_link_settings(cfg)
>      _, supported = get_local_pause_supported(cfg)
> +    link_config_tested = False
>  
>      # We check that what we can configure in the pause params matches what we
> -    # support under various contditions : Link aneg on/off, pause aneg on/off
> +    # support under various conditions: link aneg on/off, and pause aneg
> +    # on/off. If no link configuration change is supported, fall back to
> +    # checking forced MAC pause on the current link.
>      ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg on", is_get = False)
>      if ret != 0:
>          ksft_pr("link autoneg on refused, not tested")
> @@ -481,18 +477,23 @@ def pause_test_support(cfg, pauseparams):
>                                 "link autoneg on")
>          expect_pauseparams_set(cfg, pauseparams, True, supported,
>                                 "link autoneg on")
> +        link_config_tested = True
>  
>      if not forced:
>          ksft_pr("link speed unknown, the forced link is not tested")
> -        return
> -
> -    ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}", is_get = False)
> -    if ret != 0:
> -        ksft_pr(f"link autoneg off {forced} refused, not tested")
> -        return
> -
> -    expect_pauseparams_set(cfg, pauseparams, False, supported, "link autoneg off")
> -    expect_pauseparams_set(cfg, pauseparams, True, supported, "link autoneg off")
> +    else:
> +        ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}", is_get = False)
> +        if ret != 0:
> +            ksft_pr(f"link autoneg off {forced} refused, not tested")
> +        else:
> +            if link_config_tested:
> +                defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")

[Severity: Medium]
Could this leave the local interface autonegotiating when it started with
a forced link?

pause_setup() now queues a restore only when it enables autoneg itself,
which happens when supports-auto-negotiation is true and auto-negotiation
is false. If the port starts forced and reports
supports-auto-negotiation: false, pause_setup() queues nothing.

pause_test_support() then sends:

    ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg on", is_get = False)

If that succeeds, no restore is queued for it. The ethtool netlink core
does not check the Autoneg supported bit. In net/ethtool/linkmodes.c,
ethnl_update_linkmodes() only does:

    ethnl_update_u8(&lsettings->autoneg, tb[ETHTOOL_A_LINKMODES_AUTONEG], mod);

The change is then passed to ->set_link_ksettings(), so the driver alone
decides whether to accept it.

If the driver accepts it and the later autoneg off {forced} request also
succeeds, the only restore queued is the hardcoded autoneg on above.
Cleanup then leaves the link in autoneg mode.

Before this patch, pause_setup() queued "autoneg off speed X duplex Y"
first. That restore ran last and put back the original forced settings.

The commit message says:

  Fix this by queuing link-setting restores only after a link-setting
  request has succeeded.

The first autoneg on request in pause_test_support() can succeed and
change the link, but no restore is queued for it. This only happens with a
driver that accepts enabling autoneg without advertising Autoneg support.

Should pause_test_support() restore the link state it saw before the test
instead of a fixed autoneg on?

> +            expect_pauseparams_set(cfg, pauseparams, False, supported, "link autoneg off")
> +            expect_pauseparams_set(cfg, pauseparams, True, supported, "link autoneg off")
> +            link_config_tested = True
> +
> +    if not link_config_tested:
> +        expect_pauseparams_set(cfg, pauseparams, False, supported, "current link")

[ ... ]

-- 
Sashiko AI review · https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20261113022717.1914724-1-mohsin.bashr%40gmail.com

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH net v3] selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings
  2026-11-16  2:29 ` netdev-bot+sashiko
@ 2026-11-16 16:53   ` Mohsin Bashir
  0 siblings, 0 replies; 3+ messages in thread
From: Mohsin Bashir @ 2026-11-16 16:53 UTC (permalink / raw)
  To: netdev-bot+sashiko
  Cc: netdev, andrew+netdev, edumazet, kuba, pabeni, shuah,
	maxime.chevallier, linux-kselftest, hmohsin



On 11/15/26 6:29 PM, netdev-bot+sashiko@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential
> issue(s) to consider.
> 
> Critical: 0 · High: 0 · Medium: 1 · Low: 0
> 
> If you believe none of the issues are legitimate, please reply to each
> issue of Medium severity or higher and explain why the AI is wrong.
> If any of them are legitimate and you plan to respin, please let patchwork
> know by including "pw-bot: cr" as a separate line at the end of your reply
> (one such reply per series is enough).
> 
> - [Medium] The patch removes the unconditional restore of the pre-test
>    link state from pause_setup(), and pause_test_support() does not…
> 
> --
> 
> selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings
> 
> This patch changes pause.py so that link-setting restores are queued only
> after a link-setting request has succeeded. Devices that cannot change link
> settings therefore no longer fail during cleanup. If no link configuration
> change is accepted, pause_test_support() falls back to checking forced MAC
> pause on the current link, and pause_autoneg_link_autoneg() now queues its
> own autoneg-on restores.
> 
>> diff --git a/tools/testing/selftests/drivers/net/hw/pause.py b/tools/testing/selftests/drivers/net/hw/pause.py
>> index ca4a55e59c30f..c3398c7452e01 100755
>> --- a/tools/testing/selftests/drivers/net/hw/pause.py
>> +++ b/tools/testing/selftests/drivers/net/hw/pause.py
>> @@ -402,16 +402,13 @@ def pause_setup(cfg):
>>       if "auto-negotiation" not in link:
>>           raise KsftSkipEx(f"{cfg.ifname} doesn't report its link modes")
>>   
>> -    if link["auto-negotiation"]:
>> -        defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")
>> -    elif "speed" in link and "duplex" in link:
>> -        defer(cmd, f"ethtool -s {cfg.ifname} autoneg off speed {link['speed']} "
>> -                   f"duplex {link['duplex'].lower()}")
>> -
>>       # Local interface admin up, link aneg on
>>       ip(f"link set {cfg.ifname} up")
>>       if link["supports-auto-negotiation"] and not link["auto-negotiation"]:
>>           ethtool(f"-s {cfg.ifname} autoneg on")
>> +        if "speed" in link and "duplex" in link:
>> +            defer(cmd, f"ethtool -s {cfg.ifname} autoneg off speed {link['speed']} "
>> +                       f"duplex {link['duplex'].lower()}")
> 
> [ ... ]
> 
>> @@ -470,9 +463,12 @@ def pause_test_support(cfg, pauseparams):
>>   
>>       forced = forced_link_settings(cfg)
>>       _, supported = get_local_pause_supported(cfg)
>> +    link_config_tested = False
>>   
>>       # We check that what we can configure in the pause params matches what we
>> -    # support under various contditions : Link aneg on/off, pause aneg on/off
>> +    # support under various conditions: link aneg on/off, and pause aneg
>> +    # on/off. If no link configuration change is supported, fall back to
>> +    # checking forced MAC pause on the current link.
>>       ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg on", is_get = False)
>>       if ret != 0:
>>           ksft_pr("link autoneg on refused, not tested")
>> @@ -481,18 +477,23 @@ def pause_test_support(cfg, pauseparams):
>>                                  "link autoneg on")
>>           expect_pauseparams_set(cfg, pauseparams, True, supported,
>>                                  "link autoneg on")
>> +        link_config_tested = True
>>   
>>       if not forced:
>>           ksft_pr("link speed unknown, the forced link is not tested")
>> -        return
>> -
>> -    ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}", is_get = False)
>> -    if ret != 0:
>> -        ksft_pr(f"link autoneg off {forced} refused, not tested")
>> -        return
>> -
>> -    expect_pauseparams_set(cfg, pauseparams, False, supported, "link autoneg off")
>> -    expect_pauseparams_set(cfg, pauseparams, True, supported, "link autoneg off")
>> +    else:
>> +        ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg off {forced}", is_get = False)
>> +        if ret != 0:
>> +            ksft_pr(f"link autoneg off {forced} refused, not tested")
>> +        else:
>> +            if link_config_tested:
>> +                defer(cmd, f"ethtool -s {cfg.ifname} autoneg on")
> 
> [Severity: Medium]
> Could this leave the local interface autonegotiating when it started with
> a forced link?
> 
> pause_setup() now queues a restore only when it enables autoneg itself,
> which happens when supports-auto-negotiation is true and auto-negotiation
> is false. If the port starts forced and reports
> supports-auto-negotiation: false, pause_setup() queues nothing.
> 
> pause_test_support() then sends:
> 
>      ret, _ = ethtool_ret(f"-s {cfg.ifname} autoneg on", is_get = False)
> 
> If that succeeds, no restore is queued for it. The ethtool netlink core
> does not check the Autoneg supported bit. In net/ethtool/linkmodes.c,
> ethnl_update_linkmodes() only does:
> 
>      ethnl_update_u8(&lsettings->autoneg, tb[ETHTOOL_A_LINKMODES_AUTONEG], mod);
> 
> The change is then passed to ->set_link_ksettings(), so the driver alone
> decides whether to accept it.
> 
> If the driver accepts it and the later autoneg off {forced} request also
> succeeds, the only restore queued is the hardcoded autoneg on above.
> Cleanup then leaves the link in autoneg mode.
> 
> Before this patch, pause_setup() queued "autoneg off speed X duplex Y"
> first. That restore ran last and put back the original forced settings.
> 
> The commit message says:
> 
>    Fix this by queuing link-setting restores only after a link-setting
>    request has succeeded.
> 
> The first autoneg on request in pause_test_support() can succeed and
> change the link, but no restore is queued for it. This only happens with a
> driver that accepts enabling autoneg without advertising Autoneg support.
> 
> Should pause_test_support() restore the link state it saw before the test
> instead of a fixed autoneg on?
> 

Thanks for the review. AFAIU this case is only possible if the driver's 
behavior is inconsistent. The case would require driver to report that 
it does not support autoneg and still accept AUTONEG_ENABLE. IMHO 
pause_test_support() is not expected to validate ksetting 
inconsistencies in a driver. I'd leave it up to the maintainers to 
decide if it needs a respin.

>> +            expect_pauseparams_set(cfg, pauseparams, False, supported, "link autoneg off")
>> +            expect_pauseparams_set(cfg, pauseparams, True, supported, "link autoneg off")
>> +            link_config_tested = True
>> +
>> +    if not link_config_tested:
>> +        expect_pauseparams_set(cfg, pauseparams, False, supported, "current link")
> 
> [ ... ]
> 


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2026-11-16 16:53 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-13  2:27 [PATCH net v3] selftests: drv-net: pause: Avoid cleanup failures for unsupported link settings Mohsin Bashir
2026-11-16  2:29 ` netdev-bot+sashiko
2026-11-16 16:53   ` Mohsin Bashir

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox