[FFmpeg-devel] [rfc] ffmpeg security issue mailing list

[Paul B Mahol]

On 2/9/17, Michael Niedermayer <michael at niedermayer.cc> wrote:
> On Thu, Feb 09, 2017 at 08:25:43AM +0100, wm4 wrote:
>> On Wed, 8 Feb 2017 22:07:24 +0100
>> Michael Niedermayer <michael at niedermayer.cc> wrote:
>>
>> > Hi all
>> >
>> > On Sat, Aug 08, 2015 at 03:51:11AM +0200, Michael Niedermayer wrote:
>> > > On Fri, Aug 07, 2015 at 07:46:55PM -0400, compn wrote:
>> > > > hello,
>> > > >
>> > > > some of you know that we have a list for security / CVE issues.
>> > > > some of you did not know this.
>> > > >
>> > > > i think it is a private list due to not wanting people to make
>> > > > exploits
>> > > > before we have a chance to fix them. of course, if no one is
>> > > > subscribed
>> > > > to review/fix issues then they will never get fixed.
>> > > >
>> > > > so if you are a regular developer who wants access to this list,
>> > > > please
>> > > > speak up.
>> > > >
>> > > > i do not run nor admin the security email/list (nor do i know who
>> > > > does)
>> > > > so please dont ask me questions about it.
>> > >
>> > > I guess, i "de facto" admin the security "email/list".
>> > > if someone wants to help with security issues, mail me
>> > >
>> > > but there are no open security issues and if there was one i very
>> > > likely would fix it ASAP
>> >
>> > A small update due to never? before seen interrest in ffmpeg-security
>> > in the recent weeks/months
>> >
>> > How to get on the ffmpeg-security "list"
>> >
>> > People working on security in FFmpeg, thats maybe fixing many coverity
>> > issues, backporingt fixes to releases, maintaining FFmpeg releases, ...
>> > have an obsession with fixing bugs about undefined behavior or bugs
>> > about crashes and race conditions on trac. Or an obsession with testing
>> > every bugfix and who want and need access to ffmpeg-security should
>> > be on ffmpeg-security
>> > In short people on ffmpeg-security should need to be on ffmpeg-security
>> > If you fall in this kind of category, please mail me
>> >
>> > Or someone who reviews commits and obtains CVE#s for everything that
>> > could be exploitable ...
>> >
>> > I dont think we should give access to ffmpeg-security to everyone who
>> > wants to be on the list. This is of course something the community
>> > has to decide and not me, iam just err-ing on the safe side and am very
>> > restrictive on who is added.
>> >
>> > About the content i must warn you the list is really not very
>> > interresting as in trying to find together with debian someone at
>> > chromium who knows what the CVEs they registered about FFmpeg actually
>> > are about ... and then it embarassingly is a patch on ffmpeg-devel
>> > that is stuck in review and not applied and now i can redo the releases
>> > ...
>> > ... Where are the people caring about security ? why did they not
>> > pick these 2 public patches up, change what they felt needs changing
>> > and pushed them ?
>> > and there are the fuzz samples that need more than 20sec, these are
>> > the main type of reported issue recently after ive succeeded to stop
>> > the oom kind.
>> >
>> > Also there are no open security(*) issues i know of, and if there would
>> > be i likely would fix them ASAP. Not saying that help is unwelcome
>> > or that its impossible for me to make a mistake or miss something ...
>> >
>> > (*) I assume here that fuzz samples taking more than 20sec or integer
>> > overflows in DSP code arent security issues. Iam working on fixing
>> > these too but for this category there are open issues.
>> >
>> > PS: If you want access to the oss-fuzz reports, they all seem
>> > automatically public 7 days after being fixed
>> >
>> > [...]
>> >
>>
>> I'd like to get on the ffmpeg-security mailing list to review patches.
>
> Thats appreciated, though theres a problem, there rarely are patches
> on that "list". Besides there is no mailing list this is just a mail
> alias
>
> if i search for "~cffmpeg-security ~b\\+\\+\\+" i see only 54 matches
> in the whole history of the list in my inbox most of which are
> duplicates in quotes of replies
> so maybe there were less than 20 patches ever posted to that list.
> also patches tend to be CC-ed to developers knowing the code or commit
> related to a issue, like ronald and james for the http fix in december
> or paul and martin for the exr patch in august
>
> If the community wants me to add every FFmpeg maintainer who wants
> to be on the alias, i can do that. But in the absence of a clear
> community decission (poll/vote) on the inclusion criteria iam reluctant
> to add anyone without a strong reason. There occasionally is
> information or files posted that could be used in the construction of
> an exploit prior to everyone updating, so the fewer addresses it is
> sent to the better.

So others are sending CVE reports directly to you?