RE: [Intel-wired-lan] [PATCH net v5 3/4] iavf: send MAC change request synchronously

-----Original Message-----
From: Intel-wired-lan <redacted> On Behalf Of
Przemek Kitszel
Sent: Wednesday, April 29, 2026 3:04 PM
To: Jose Ignacio Tornos Martinez <redacted>; Loktionov, Aleksandr
[off-list ref]
Cc: Nguyen, Anthony L <anthony.l.nguyen@intel.com>; davem@davemloft.net;
edumazet@google.com; horms@kernel.org; intel-wired-lan@lists.osuosl.org;
Keller, Jacob E [off-list ref]; jesse.brandeburg@intel.com;
kuba@kernel.org; netdev@vger.kernel.org; pabeni@redhat.com;
stable@vger.kernel.org
Subject: Re: [Intel-wired-lan] [PATCH net v5 3/4] iavf: send MAC change request
synchronously

On 4/29/26 14:00, Jose Ignacio Tornos Martinez wrote:

quoted

Hello Aleksandr,

quoted

I think continue at the end of the cycle is redundant.

That continue is intentional; without it, if timeout expires but there
are still messages in the queue, we give up without processing them.
The

Alex is right,
"continue" causes to check the condition clause of while loop, also for do-while

quoted

message we're waiting for might be in the queue and not a lot of
messages stored are expected.
That continue reduces possible false timeouts (because the expected
message could be stored in the queue) while keeping the delay minimal.
The timeout is really just an estimate, and I don't think it needs to
be very precise.

with that said, current code is correct

removing the redundant "if" could be done while applying (if that will be the only
nitpick left)

after more thinking:
in theory, not checking the time but processing next message if there were any
pending on the previous message could cause infinite loop (to fix that we should
stop refreshing "pending" value after the timeout, but only decrementing it - but I
think that this would be needless complication)

My Reviewed-by still holds

quoted

Thanks

Best regards
Jose Ignacio

Tested-by: Rafal Romanowski <redacted>