netdev - Re: [PATCH net,v2] chelsio/chtls: fix memory leaks

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] [day] [month] [year] [list]

Message-ID: <20201027170507.2266e96b@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Tue, 27 Oct 2020 17:05:07 -0700
From:   Jakub Kicinski <kuba@...nel.org>
To:     Vinay Kumar Yadav <vinay.yadav@...lsio.com>
Cc:     netdev@...r.kernel.org, davem@...emloft.net, secdev@...lsio.com
Subject: Re: [PATCH net,v2] chelsio/chtls: fix memory leaks

On Mon, 26 Oct 2020 01:19:36 +0530 Vinay Kumar Yadav wrote:
> Correct skb refcount in alloc_ctrl_skb(), causing skb memleak
> when chtls_send_abort() called with NULL skb.
> Also race between user context and softirq causing memleak,
> consider the call sequence scenario

Sounds like two separate fixes?

> chtls_setkey()         //user context
> chtls_peer_close()
> chtls_abort_req_rss()
> chtls_setkey()         //user context
> 
> work request skb queued in chtls_setkey() won't be freed
> because resources are already cleaned for this connection,
> fix it by not queuing work request while socket is closing.
> 
> v1->v2:
> - fix W=1 warning.
> 
> Fixes: cc35c88ae4db ("crypto : chtls - CPL handler definition")
> Signed-off-by: Vinay Kumar Yadav <vinay.yadav@...lsio.com>
> ---
>  drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_cm.c | 2 +-
>  drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_hw.c | 3 +++
>  2 files changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_cm.c b/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_cm.c
> index 24154816d1d1..63aacc184f68 100644
> --- a/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_cm.c
> +++ b/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_cm.c
> @@ -212,7 +212,7 @@ static struct sk_buff *alloc_ctrl_skb(struct sk_buff *skb, int len)
>  {
>  	if (likely(skb && !skb_shared(skb) && !skb_cloned(skb))) {
>  		__skb_trim(skb, 0);
> -		refcount_add(2, &skb->users);
> +		refcount_inc(&skb->users);

You just switched from adding two refs to adding one.
Was this always leaking the skb?

Also skb_get().

>  	} else {
>  		skb = alloc_skb(len, GFP_KERNEL | __GFP_NOFAIL);
>  	}
> diff --git a/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_hw.c b/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_hw.c
> index f1820aca0d33..62c829023da5 100644
> --- a/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_hw.c
> +++ b/drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls_hw.c
> @@ -383,6 +383,9 @@ int chtls_setkey(struct chtls_sock *csk, u32 keylen,
>  	if (ret)
>  		goto out_notcb;
>  
> +	if (unlikely(csk_flag(sk, CSK_ABORT_SHUTDOWN)))
> +		goto out_notcb;
> +
>  	set_wr_txq(skb, CPL_PRIORITY_DATA, csk->tlshws.txqid);
>  	csk->wr_credits -= DIV_ROUND_UP(len, 16);
>  	csk->wr_unacked += DIV_ROUND_UP(len, 16);