Re: [PATCH v3 net 0/8] Fixes for NXP ENETC driver

From: Vladimir Oltean <olteanv@gmail.com>
To: patchwork-bot+netdevbpf@kernel.org
Cc: davem@davemloft.net, kuba@kernel.org, netdev@vger.kernel.org,
	michael@walle.cc, claudiu.manoil@nxp.com,
	alexandru.marginean@nxp.com, andrew@lunn.ch,
	vladimir.oltean@nxp.com
Subject: Re: [PATCH v3 net 0/8] Fixes for NXP ENETC driver
Date: Tue, 2 Mar 2021 13:47:01 +0200	[thread overview]
Message-ID: <20210302114701.agmjacnf6vudbuko@skbuf> (raw)
In-Reply-To: <161463480768.18741.11451492687720022399.git-patchwork-notify@kernel.org>

On Mon, Mar 01, 2021 at 09:40:07PM +0000, patchwork-bot+netdevbpf@kernel.org wrote:
> Hello:
> 
> This series was applied to netdev/net.git (refs/heads/master):
> 
> On Mon,  1 Mar 2021 13:18:10 +0200 you wrote:
> > From: Vladimir Oltean <vladimir.oltean@nxp.com>
> > 
> > This contains an assorted set of fixes collected over the past 2 weeks
> > on the enetc driver. Some are related to VLAN processing, some to
> > physical link settings, some are fixups of previous hardware workarounds,
> > and some are simply zero-day data path bugs that for some reason were
> > never caught or at least identified.
> > 
> > [...]
> 
> Here is the summary with links:
>   - [v3,net,1/8] net: enetc: don't overwrite the RSS indirection table when initializing
>     https://git.kernel.org/netdev/net/c/c646d10dda2d
>   - [v3,net,2/8] net: enetc: initialize RFS/RSS memories for unused ports too
>     https://git.kernel.org/netdev/net/c/3222b5b613db
>   - [v3,net,3/8] net: enetc: take the MDIO lock only once per NAPI poll cycle
>     https://git.kernel.org/netdev/net/c/6d36ecdbc441
>   - [v3,net,4/8] net: enetc: fix incorrect TPID when receiving 802.1ad tagged packets
>     https://git.kernel.org/netdev/net/c/827b6fd04651
>   - [v3,net,5/8] net: enetc: don't disable VLAN filtering in IFF_PROMISC mode
>     https://git.kernel.org/netdev/net/c/a74dbce9d454
>   - [v3,net,6/8] net: enetc: force the RGMII speed and duplex instead of operating in inband mode
>     https://git.kernel.org/netdev/net/c/c76a97218dcb
>   - [v3,net,7/8] net: enetc: remove bogus write to SIRXIDR from enetc_setup_rxbdr
>     https://git.kernel.org/netdev/net/c/96a5223b918c
>   - [v3,net,8/8] net: enetc: keep RX ring consumer index in sync with hardware
>     https://git.kernel.org/netdev/net/c/3a5d12c9be6f
> 
> You are awesome, thank you!
> --
> Deet-doot-dot, I am a bot.
> https://korg.docs.kernel.org/patchwork/pwbot.html

Thanks.
This is somewhat non-ideal, since, as discussed on patch 8/8, I was
planning to resend a new version with the proper Fixes: tag, but
obviously I did not get the chance to do that:
https://patchwork.kernel.org/project/netdevbpf/patch/20210301111818.2081582-9-olteanv@gmail.com/
However, at this point, doing anything at all would be messier than not
doing anything, and according to my calculations, nothing breaks even if
patch 8/8 is backported to kernels containing just the initial commit of
the driver, but not the MDIO workaround. The next_to_use variable will
just be written twice (with the same value) to the RX ring's consumer
index, once in enetc_refill_rx_ring and the second time immediately
afterwards, in enetc_setup_rxbdr.
If I get the chance to NACK backporting of patch 8 to stable kernels
that don't contain commit "enetc: Workaround for MDIO register access
issue" (aka to stable kernels lower than linux-5.9.y) then I'll do that.