linux-kernel - Re: [PATCH net v2] net: ethernet: ti: am65-cpsw: Fix NAPI registration sequence

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <02d685e2aa8721a119f528bde2f4ec9533101663.camel@siemens.com>
Date: Tue, 11 Mar 2025 13:53:10 +0000
From: "Sverdlin, Alexander" <alexander.sverdlin@...mens.com>
To: "andrew+netdev@...n.ch" <andrew+netdev@...n.ch>, "s-vadapalli@...com"
	<s-vadapalli@...com>, "davem@...emloft.net" <davem@...emloft.net>,
	"dan.carpenter@...aro.org" <dan.carpenter@...aro.org>, "jpanis@...libre.com"
	<jpanis@...libre.com>, "c-vankar@...com" <c-vankar@...com>,
	"pabeni@...hat.com" <pabeni@...hat.com>, "horms@...nel.org"
	<horms@...nel.org>, "edumazet@...gle.com" <edumazet@...gle.com>,
	"rogerq@...nel.org" <rogerq@...nel.org>, "kuba@...nel.org" <kuba@...nel.org>
CC: "linux-arm-kernel@...ts.infradead.org"
	<linux-arm-kernel@...ts.infradead.org>, "vigneshr@...com" <vigneshr@...com>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>, "srk@...com"
	<srk@...com>
Subject: Re: [PATCH net v2] net: ethernet: ti: am65-cpsw: Fix NAPI
 registration sequence

Hi Siddharth!

On Tue, 2025-03-11 at 18:31 +0530, Siddharth Vadapalli wrote:
> From: Vignesh Raghavendra <vigneshr@...com>
> 
> Registering the interrupts for TX or RX DMA Channels prior to registering
> their respective NAPI callbacks can result in a NULL pointer dereference.
> This is seen in practice as a random occurrence since it depends on the
> randomness associated with the generation of traffic by Linux and the
> reception of traffic from the wire.
> 
> Fixes: 681eb2beb3ef ("net: ethernet: ti: am65-cpsw: ensure proper channel cleanup in error path")
> Signed-off-by: Vignesh Raghavendra <vigneshr@...com>
> Co-developed-by: Siddharth Vadapalli <s-vadapalli@...com>
> Signed-off-by: Siddharth Vadapalli <s-vadapalli@...com>

...

> v1 of this patch is at:
> https://lore.kernel.org/all/20250311061214.4111634-1-s-vadapalli@ti.com/
> Changes since v1:
> - Based on the feedback provided by Alexander Sverdlin <alexander.sverdlin@...mens.com>
>   the patch has been updated to account for the cleanup path in terms of an imbalance
>   between the number of successful netif_napi_add_tx/netif_napi_add calls and the
>   number of successful devm_request_irq() calls. In the event of an error, we will
>   always have one extra successful netif_napi_add_tx/netif_napi_add that needs to be
>   cleaned up before we clean an equal number of netif_napi_add_tx/netif_napi_add and
>   devm_request_irq.

...

> --- a/drivers/net/ethernet/ti/am65-cpsw-nuss.c
> +++ b/drivers/net/ethernet/ti/am65-cpsw-nuss.c
> @@ -2569,6 +2570,9 @@ static int am65_cpsw_nuss_init_rx_chns(struct am65_cpsw_common *common)
>  			     HRTIMER_MODE_REL_PINNED);
>  		flow->rx_hrtimer.function = &am65_cpsw_nuss_rx_timer_callback;
>  
> +		netif_napi_add(common->dma_ndev, &flow->napi_rx,
> +			       am65_cpsw_nuss_rx_poll);
> +
>  		ret = devm_request_irq(dev, flow->irq,
>  				       am65_cpsw_nuss_rx_irq,
>  				       IRQF_TRIGGER_HIGH,
> @@ -2579,9 +2583,6 @@ static int am65_cpsw_nuss_init_rx_chns(struct am65_cpsw_common *common)
>  			flow->irq = -EINVAL;
>  			goto err_flow;
>  		}
> -
> -		netif_napi_add(common->dma_ndev, &flow->napi_rx,
> -			       am65_cpsw_nuss_rx_poll);
>  	}
>  
>  	/* setup classifier to route priorities to flows */
> @@ -2590,10 +2591,11 @@ static int am65_cpsw_nuss_init_rx_chns(struct am65_cpsw_common *common)
>  	return 0;
>  
>  err_flow:
> -	for (--i; i >= 0 ; i--) {
> +	netif_napi_del(&flow->napi_rx);

There are totally 3 "goto err_flow;" instances, so if k3_udma_glue_rx_flow_init() or
k3_udma_glue_rx_get_irq() would fail on the first iteration, we would come here without
a single call to netif_napi_add().

> +	for (--i; i >= 0; i--) {
>  		flow = &rx_chn->flows[i];
> -		netif_napi_del(&flow->napi_rx);
>  		devm_free_irq(dev, flow->irq, flow);
> +		netif_napi_del(&flow->napi_rx);
>  	}
>  
>  err:

-- 
Alexander Sverdlin
Siemens AG
www.siemens.com