netdev - Re: [PATCH] net/qla3xxx: switch from 'pci_' to 'dma

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20210118203336.6e7170a1@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Mon, 18 Jan 2021 20:33:36 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Christophe JAILLET <christophe.jaillet@...adoo.fr>
Cc:     GR-Linux-NIC-Dev@...vell.com, davem@...emloft.net,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        kernel-janitors@...r.kernel.org
Subject: Re: [PATCH] net/qla3xxx: switch from 'pci_' to 'dma_' API

On Sun, 17 Jan 2021 09:15:42 +0100 Christophe JAILLET wrote:
> @@ -2525,9 +2519,8 @@ static int ql_alloc_net_req_rsp_queues(struct ql3_adapter *qdev)
>  	wmb();
>  
>  	qdev->req_q_virt_addr =
> -	    pci_alloc_consistent(qdev->pdev,
> -				 (size_t) qdev->req_q_size,
> -				 &qdev->req_q_phy_addr);
> +	    dma_alloc_coherent(&qdev->pdev->dev, (size_t)qdev->req_q_size,
> +			       &qdev->req_q_phy_addr, GFP_KERNEL);
>  
>  	if ((qdev->req_q_virt_addr == NULL) ||
>  	    LS_64BITS(qdev->req_q_phy_addr) & (qdev->req_q_size - 1)) {
> @@ -2536,16 +2529,14 @@ static int ql_alloc_net_req_rsp_queues(struct ql3_adapter *qdev)
>  	}
>  
>  	qdev->rsp_q_virt_addr =
> -	    pci_alloc_consistent(qdev->pdev,
> -				 (size_t) qdev->rsp_q_size,
> -				 &qdev->rsp_q_phy_addr);
> +	    dma_alloc_coherent(&qdev->pdev->dev, (size_t)qdev->rsp_q_size,
> +			       &qdev->rsp_q_phy_addr, GFP_KERNEL);
>  
>  	if ((qdev->rsp_q_virt_addr == NULL) ||
>  	    LS_64BITS(qdev->rsp_q_phy_addr) & (qdev->rsp_q_size - 1)) {
>  		netdev_err(qdev->ndev, "rspQ allocation failed\n");
> -		pci_free_consistent(qdev->pdev, (size_t) qdev->req_q_size,
> -				    qdev->req_q_virt_addr,
> -				    qdev->req_q_phy_addr);
> +		dma_free_coherent(&qdev->pdev->dev, (size_t)qdev->req_q_size,
> +				  qdev->req_q_virt_addr, qdev->req_q_phy_addr);
>  		return -ENOMEM;
>  	}

Something to follow up on - the error handling in this function looks
pretty sketchy. Looks like if the buffer was allocated but these
alignment conditions trigger the buffer is not freed. Happens both for
req and rsp buffer.

Applied, thanks!