linux-kernel - Re: [PATCH 2/3] net: qrtr: Add MHI transport layer

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: <20200401082000.GA15627@Mani-XPS-13-9360>
Date:   Wed, 1 Apr 2020 13:50:00 +0530
From:   Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
To:     Bjorn Andersson <bjorn.andersson@...aro.org>
Cc:     gregkh@...uxfoundation.org, davem@...emloft.net,
        smohanad@...eaurora.org, jhugo@...eaurora.org,
        kvalo@...eaurora.org, hemantk@...eaurora.org,
        linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
        netdev@...r.kernel.org, clew@...eaurora.org
Subject: Re: [PATCH 2/3] net: qrtr: Add MHI transport layer

On Wed, Apr 01, 2020 at 12:10:23AM -0700, Bjorn Andersson wrote:
> On Tue 31 Mar 23:44 PDT 2020, Manivannan Sadhasivam wrote:
> > diff --git a/net/qrtr/mhi.c b/net/qrtr/mhi.c
> [..]
> > +static void qcom_mhi_qrtr_ul_callback(struct mhi_device *mhi_dev,
> > +				      struct mhi_result *mhi_res)
> > +{
> > +	struct sk_buff *skb = (struct sk_buff *)mhi_res->buf_addr;
> > +
> > +	consume_skb(skb);
> > +	if (skb->sk)
> > +		sock_put(skb->sk);
> 
> Don't you need to do this in opposite order, to avoid a use after free?
> 

I thought about it but the socket refcounting postulates in net/sock.h states:

"sk_free is called from any context: process, BH, IRQ. When it is called,
socket has no references from outside -> sk_free may release descendant
resources allocated by the socket, but to the time when it is called, socket
is NOT referenced by any hash tables, lists etc."

Here the sock it still referenced by skb, so I don't exactly know if we can
release the socket using sock_put() before consume_skb(). But on the other hand,
once skb is freed then accessing its member is clearly a use after free issue.

Maybe someone can clarify this?

Thanks,
Mani

> Regards,
> Bjorn