[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20241005145717.302575-1-jdamato@fastly.com>
Date: Sat, 5 Oct 2024 14:57:15 +0000
From: Joe Damato <jdamato@...tly.com>
To: netdev@...r.kernel.org
Cc: Joe Damato <jdamato@...tly.com>,
"David S. Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>,
linux-kernel@...r.kernel.org (open list),
Michael Chan <mchan@...adcom.com>,
Paolo Abeni <pabeni@...hat.com>,
Pavan Chebbi <pavan.chebbi@...adcom.com>
Subject: [RFC net-next v2 0/2] tg3: Link IRQs, NAPIs, and queues
Greetings:
This RFC v3 follows from a previous RFC [1] submission which I noticed
had an issue in patch 2.
Patch 1 is changed to wrap a long line at 80 characters. No functional
changes. As such, I retained Pavan Chebbi's Reviewed-by.
Patch 2 in RFC v2 had an issue where it used the index into tp->irq_cnt
as the rxq or txq index; this is incorrect. It does not need seem that
tg3 assigns explicit queue index to struct tg3_napi, so the least
invasive change I could think of included two running counters in
tg3_napi_enable and tg3_napi_disable.
This is required because netif_queue_set_napi expected the queue index
(0 to real_num_[rt]x_queues) to be passed in to associate queues IDs
with NAPI IDs. tg2_napi_disable is modified in the reverse order;
counting down queue indices.
I am open to other suggestions on implementation from broadcom, but
thought that this was the least disruptive change.
I've tested this change on my tg3 hardware and it seems to work, see
commit message for examples of how to test.
Thanks,
Joe
[1]: https://lore.kernel.org/netdev/20240925162048.16208-1-jdamato@fastly.com/
Joe Damato (2):
tg3: Link IRQs to NAPI instances
tg3: Link queues to NAPIs
drivers/net/ethernet/broadcom/tg3.c | 45 ++++++++++++++++++++++++-----
1 file changed, 38 insertions(+), 7 deletions(-)
--
2.25.1
Powered by blists - more mailing lists