linux-kernel - Re: [PATCH] dma-buf: fix use-after-free in dmabuffs

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: <20200512084534.GA3557007@kroah.com>
Date:   Tue, 12 May 2020 10:45:34 +0200
From:   Greg KH <gregkh@...uxfoundation.org>
To:     Charan Teja Kalla <charante@...eaurora.org>
Cc:     dri-devel@...ts.freedesktop.org, linaro-mm-sig@...ts.linaro.org,
        linux-kernel@...r.kernel.org, vinmenon@...eaurora.org,
        sumit.semwal@...aro.org, ghackmann@...gle.com, fengc@...gle.com,
        linux-media@...r.kernel.org
Subject: Re: [PATCH] dma-buf: fix use-after-free in dmabuffs_dname

On Tue, May 12, 2020 at 10:43:18AM +0530, Charan Teja Kalla wrote:
> > Ok, but watch out, now you have 2 different reference counts for the
> > same structure.  Keeping them coordinated is almost always an impossible
> > task so you need to only rely on one.  If you can't use the file api,
> > just drop all of the reference counting logic in there and only use the
> > kref one.
> 
> I feel that changing the refcount logic now to dma-buf objects involve
> changes in
> 
> the core dma-buf framework. NO? Instead, how about passing the user passed
> name directly
> 
> in the ->d_fsdata inplace of dmabuf object? Because we just need user passed
> name in the
> 
> dmabuffs_dname(). With this we can avoid the need for extra refcount on
> dmabuf.

Odd formatting :(

> Posted patch-V2: https://lkml.org/lkml/2020/5/8/158

Please just post links to lore.kernel.org, we have no control over
lkml.org at all.

I'll go review that patch now...

greg k-h