[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <318a13f6-da2d-4ba4-a51c-2f8444444bbf@linux.dev>
Date: Tue, 10 Sep 2024 10:24:36 -0400
From: Sean Anderson <sean.anderson@...ux.dev>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Radhey Shyam Pandey <radhey.shyam.pandey@....com>,
"David S . Miller" <davem@...emloft.net>, Eric Dumazet
<edumazet@...gle.com>, Paolo Abeni <pabeni@...hat.com>,
netdev@...r.kernel.org, Michal Simek <michal.simek@....com>,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
Andy Chiu <andy.chiu@...ive.com>, Daniel Borkmann <daniel@...earbox.net>
Subject: Re: [PATCH net v2] net: xilinx: axienet: Fix packet counting
On 9/9/24 21:00, Jakub Kicinski wrote:
> On Fri, 6 Sep 2024 12:42:27 -0400 Sean Anderson wrote:
>> axienet_free_tx_chain returns the number of DMA descriptors it's
>> handled. However, axienet_tx_poll treats the return as the number of
>> packets. When scatter-gather SKBs are enabled, a single packet may use
>> multiple DMA descriptors, which causes incorrect packet counts. Fix this
>> by explicitly keepting track of the number of packets processed as
>> separate from the DMA descriptors.
>>
>> Fixes: 8a3b7a252dca ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
>> Signed-off-by: Sean Anderson <sean.anderson@...ux.dev>
>
>> diff --git a/drivers/net/ethernet/xilinx/xilinx_axienet_main.c b/drivers/net/ethernet/xilinx/xilinx_axienet_main.c
>> index 9aeb7b9f3ae4..556033849d55 100644
>> --- a/drivers/net/ethernet/xilinx/xilinx_axienet_main.c
>> +++ b/drivers/net/ethernet/xilinx/xilinx_axienet_main.c
>> @@ -670,21 +670,21 @@ static int axienet_device_reset(struct net_device *ndev)
>> * @force: Whether to clean descriptors even if not complete
>> * @sizep: Pointer to a u32 filled with the total sum of all bytes
>> * in all cleaned-up descriptors. Ignored if NULL.
>> - * @budget: NAPI budget (use 0 when not called from NAPI poll)
>> + * @budget: NAPI budget (use INT_MAX when not called from NAPI poll)
>
> use INT_MAX and force=true when ... ?
> To make sure the dependency is clear.
> But actually...
>
>> *
>> * Would either be called after a successful transmit operation, or after
>> * there was an error when setting up the chain.
>> - * Returns the number of descriptors handled.
>> + * Returns the number of packets handled.
>> */
>> static int axienet_free_tx_chain(struct axienet_local *lp, u32 first_bd,
>> int nr_bds, bool force, u32 *sizep, int budget)
>> {
>> struct axidma_bd *cur_p;
>> unsigned int status;
>> + int i, packets = 0;
>> dma_addr_t phys;
>> - int i;
>>
>> - for (i = 0; i < nr_bds; i++) {
>> + for (i = 0; i < nr_bds && packets < budget; i++) {
>
> why are you doing this? To make sure drivers doesn't complete more
> than "budget" Tx skbs? The budget is really for Rx, for Tx you can
> use a reasonable fixed value, independent of what budget core
> passes in, e.g. 128. See:
> https://www.kernel.org/doc/html/next/networking/napi.html#datapath-api
I read this but it was unclear to me because it seems oriented towards
"combined" NAPI instances, while we have separate instances for RX and
TX. So even for TX-only instances, we can ignore budget?
>> cur_p = &lp->tx_bd_v[(first_bd + i) % lp->tx_bd_num];
>> status = cur_p->status;
>>
>> @@ -701,8 +701,10 @@ static int axienet_free_tx_chain(struct axienet_local *lp, u32 first_bd,
>> (cur_p->cntrl & XAXIDMA_BD_CTRL_LENGTH_MASK),
>> DMA_TO_DEVICE);
>>
>> - if (cur_p->skb && (status & XAXIDMA_BD_STS_COMPLETE_MASK))
>> - napi_consume_skb(cur_p->skb, budget);
>> + if (cur_p->skb && (status & XAXIDMA_BD_STS_COMPLETE_MASK)) {
>> + napi_consume_skb(cur_p->skb, force ? 0 : budget);
>> + packets++;
>> + }
>>
>> cur_p->app0 = 0;
>> cur_p->app1 = 0;
>> @@ -718,7 +720,13 @@ static int axienet_free_tx_chain(struct axienet_local *lp, u32 first_bd,
>> *sizep += status & XAXIDMA_BD_STS_ACTUAL_LEN_MASK;
>> }
>>
>> - return i;
>> + if (!force) {
>> + lp->tx_bd_ci += i;
>> + if (lp->tx_bd_ci >= lp->tx_bd_num)
>> + lp->tx_bd_ci %= lp->tx_bd_num;
>> + }
>
> Moving this chunk into axienet_free_tx_chain() is a noop, right?
> Please avoid code cleanups in fixes.
The relevant variable (number of descriptors handled) is no longer
returned to axienet_tx_poll, so it can't update the current descriptor
properly.
--Sean
>> + return packets;
>> }
>>
>> /**
Powered by blists - more mailing lists