[sldev] Static code analysis
Meadhbh Hamrick (Infinity)
infinity at lindenlab.com
Sun Jan 11 23:35:32 PST 2009
i'm a bit more of a fan of fortify's tools, but what the heck, if
scan.coverity.com will scan the released code, then great!
but there are a couple things to note...
a. the good people at coverity want the "official contact" of the open
source project to setup the account for this process. I'm guessing
this probably means Rob or Soft. Speak nicely to Rob about it, he's
already got a bunch on his plate.
b. the SL Viewer makes use of a number of other libraries. for a
"proper" security experience, we should try to get these libraries put
in the scan queue as well.
c. believe it or not, there are bugs that static analysis won't find.
Don't get me wrong, doing a static analysis is a good first step. But
once you do a static analysis and fix the bugs you find, there are
still a large class of vulnerabilities that could be present in the
code. just keep in mind that the idea behind automated tools is not
that they find every vulnerability, but that the free people to track
down the really bad ones.
d. there are other static code analysis tools, some of which are free
as in beer. fortify and coverity will tell you they're not as good,
and it's true to a point.. flawfinder definitely looks for fewer
things than fortify. but it's free. hmm... free tool meets open source
repository... hmm... interested parties can find more than enough to
keep them busy for a while at https://samate.nist.gov/index.php/Source_Code_Security_Analyzers
.
e. it is characteristic of these tools that there is some give and
take between the tool and the tool using mammal behind the keyboard.
many static analysis tools are quite configurable and allow the user
to perform invasive scans that produce an ox-stunning amount of output
or to perform a light scan, which gives little of value. there's a lot
of finesse required to use these tools effectively. this has led some
to consider these tools thinly veiled professional services sales
vehicles.
-cheers
-infinity
On Jan 11, 2009, at 9:09 PM, Gareth Nelson wrote:
> I see no reason why in the meantime the community can't use this tool
> using the free trial
>
> On Mon, Jan 12, 2009 at 4:13 AM, Soft <soft at lindenlab.com> wrote:
>> On Sun, Jan 11, 2009 at 7:57 PM, Jason Giglio
>> <gigstaggart at gmail.com> wrote:
>>> Sheet Spotter wrote:
>>>> I stumbled into a code analysis tool from Coverity that claims to
>>>> identify source code flaws through an elaborate static code
>>>> analysis
>>>> with a lower "false positive" rate than similar tools. Coverity
>>>> seems to
>>>> offer their tool (or their services?) free of charge to open source
>>>> projects.
>>>
>>> I went through this a couple years ago.
>>>
>>> The conclusion of the thread was that Linden Lab already licensed
>>> Coverity internally, and they weren't going to release the results
>>> of
>>> the report to us. There were some vague excuses about security or
>>> something, and that the open source community can't really help fix
>>> those kinds of bugs anyway.
>>
>> The problem is that the Coverity report is generated against the full
>> build, including server components and things where we don't have a
>> license to redistribute code. If we renew our Coverity license
>> (that's
>> up in the air - I'd heard that it's hugely expensive), the plan is to
>> get a separate analysis running against the very same code that's
>> exported, and to export that routinely.
>> _______________________________________________
>> Policies and (un)subscribe information available here:
>> http://wiki.secondlife.com/wiki/SLDev
>> Please read the policies before posting to keep unmoderated posting
>> privileges
>>
>
>
>
> --
> Please avoid sending me Word or PowerPoint attachments.
> See http://www.gnu.org/philosophy/no-word-attachments.html
> _______________________________________________
> Policies and (un)subscribe information available here:
> http://wiki.secondlife.com/wiki/SLDev
> Please read the policies before posting to keep unmoderated posting
> privileges
More information about the SLDev
mailing list