netdev - Re: net_dma: mark broken

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: <CAPcyv4gQWWv-bq57=6BNVNnRCEdiy5dEE09Q2snO7nC76Xmvbw@mail.gmail.com>
Date:	Mon, 30 Dec 2013 12:15:27 -0800
From:	Dan Williams <dan.j.williams@...el.com>
To:	Dave Jones <davej@...hat.com>
Cc:	netdev@...r.kernel.org
Subject: Re: net_dma: mark broken

On Mon, Dec 23, 2013 at 8:21 AM, Dave Jones <davej@...hat.com> wrote:
> On Fri, Dec 20, 2013 at 09:12:28PM +0000, Linux Kernel wrote:
>  > Gitweb:     http://git.kernel.org/linus/;a=commit;h=77873803363c9e831fc1d1e6895c084279090c22
>  > Commit:     77873803363c9e831fc1d1e6895c084279090c22
>  > Parent:     0baf8f6a2ac86c2c40ed0cacab8ea3d17371a1bb
>  > Author:     Dan Williams <dan.j.williams@...el.com>
>  > AuthorDate: Tue Dec 17 10:09:32 2013 -0800
>  > Committer:  Dan Williams <dan.j.williams@...el.com>
>  > CommitDate: Wed Dec 18 12:53:43 2013 -0800
>  >
>  >     net_dma: mark broken
>  >
>  >     net_dma can cause data to be copied to a stale mapping if a
>  >     copy-on-write fault occurs during dma.  The application sees missing
>  >     data.
>
>
> Since this commit, coverity picked up a possible logic contradiction in tcp_rcv_established
> Now that the only thing setting copied_early = 1 is inside an ifdef that won't be set,
> it notes that this code is unreachable..
>
> 5271                        if (!copied_early || tp->rcv_nxt != tp->rcv_wup)
> 5272                                __tcp_ack_snd_check(sk, 0);
>
> I don't understand all the subtleties of that huge function, so another
> set of eyes would be appreciated.  If it's a non-issue, I'll flag it as such
> for coverity so it doesn't get picked up again.

Unreachable?  Or is that the same report for something like "if()
statement has no effect"?  In any event that fix came from
"53240c208776 tcp: Fix possible double-ack w/ user dma" and can be
reverted to just call __tcp_ack_snd_check unconditionally.  I'll fold
it into the net_dma removal patch.

--
Dan
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html