[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <31ac7255-1c45-bb90-316d-751cb3186946@gmail.com>
Date: Wed, 21 Apr 2021 11:57:36 +0100
From: Edward Cree <ecree.xilinx@...il.com>
To: Yury Vostrikov <mon@...ormed.ru>
Cc: "David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org,
Martin Habets <habetsm.xilinx@...il.com>
Subject: Re: [PATCH] net: sfc: Fix kernel panic introduced by commit
044588b96372 ("sfc: define inner/outer csum offload TXQ types")
On 20/04/2021 20:42, Yury Vostrikov wrote:
> efx_get_tx_queue(efx, qid / EFX_MAX_TXQ_PER_CHANNEL,
> qid % EFX_MAX_TXQ_PER_CHANNEL);
>
> This uses qid / EFX_MAX_TXQ_PER_CHANNEL as index inside
> efx_nic->channels[] and qid % EFX_MAX_TXQ_PER_CHANNEL as index inside
> channel->tx_queue_be_type[].
>
> Indexing into bitset mapping with modulo operation seems to oversight
> from the previous refactoring.
This should be fixed by 5b1faa92289b ("sfc: farch: fix TX queue lookup
in TX flush done handling") and 83b09a180741 ("sfc: farch: fix TX queue
lookup in TX event handling"), which were applied to 'net' this morning.
Do you see any call sites that I've missed and that remain broken?
(The one in ef100_ev_tx() is technically incorrect, but qlabel is always
0 there so it shouldn't matter.)
-ed
Powered by blists - more mailing lists