1 |
On 07/05/2014 16:12, James wrote: |
2 |
> Alan McKinnon <alan.mckinnon <at> gmail.com> writes: |
3 |
> |
4 |
> |
5 |
>>> This is retarded, and I'm too old to do that now, so I went shopping |
6 |
>>> for some script/tool/code to do it for me. In fact, I do not know |
7 |
>>> why the integrity check is not fully integrated into ftp. rsync. |
8 |
>>> or whatever the download tool is? |
9 |
> |
10 |
> |
11 |
>> Perhaps I'm just old and retarded myself, but portage already does what |
12 |
>> you want. |
13 |
> |
14 |
> |
15 |
> Well certainly portage is one common methods to download, and we can explore |
16 |
> that thread. But, I was thinking more general. Late last night (too late) I |
17 |
> decided to download 'lilblue'. I poked around on several gentoo mirrors and |
18 |
> could not find it. So with my google hat on, I found it on a non typical |
19 |
> (mirror) server. The download was slow (300K) so naturally, I became |
20 |
> suspicious. I checked manually, but it was late and I was tired..... |
21 |
> |
22 |
> |
23 |
> The download was kicked off from the web browser (seamonkey). Now that I |
24 |
> think about it, there are a myriad of ways to download sources. What I was |
25 |
> suggesting (inquiring?) is that a command line tool could be readily |
26 |
> developed (if it did not already exist) to simple check any download |
27 |
> against the published data (keys/hashes/etc) depending on what is in the |
28 |
> local dir where the download lands (is stored). It could be used with |
29 |
> protage files too. |
30 |
> But why not just use a simple script: |
31 |
> |
32 |
> <scriptname> package.just.downloaded package.just.downloaded.DIGESTS |
33 |
> |
34 |
> |
35 |
> |
36 |
> But then I got to questioning the integrity of both the downloaded sources |
37 |
> and the digest originating on the same server........ Probably not a good |
38 |
> idea either? So the digest should come from elsewhere? Maybe pull the digest |
39 |
> from a certificated (pontificated?) (gentoo controlled) server and not |
40 |
> somebody's (low priority managed) public server. Or maybe a master list of |
41 |
> digests (hashes) could be included on every (hardened) gentoo box? |
42 |
> |
43 |
> |
44 |
> |
45 |
> It seems *everything is hacked* now. Certainly the NSA has fessed up to that |
46 |
> as have others. Sure it may be just "good business" but the brightest minds |
47 |
> now days are mostly focused on security comprimises, particularly offensive |
48 |
> strategies, imho. So it seems to me, there is probably a "fly in the |
49 |
> ointment" common to what everyone is doing on a semi regular basis. To me |
50 |
> this sort of (justified/unjustified) paranoia should be incorporated into |
51 |
> the entire "hardened" effort at gentoo, imho, if not on a wider basis. |
52 |
> |
53 |
> |
54 |
> So please continue the "protage" thread discussion, but also a wider thread |
55 |
> concerning other source downloads. Afterall, *if" you can inject* into |
56 |
> sources, which are then compiled, who checks under the under_garments? If |
57 |
> you read about "The rat" the most secure implementation had/has tainted it's |
58 |
> very soul. [1] |
59 |
> |
60 |
> |
61 |
> |
62 |
> curiously, |
63 |
> James |
64 |
> |
65 |
> [1] |
66 |
> http://arstechnica.com/information-technology/2014/04/openssl-code-beyond-repair-claims-creator-of-libressl-fork/ |
67 |
|
68 |
Thanks, now I understand better the question you are asking. |
69 |
|
70 |
I don't think it can be solved at all in the general case, for two reasons. |
71 |
|
72 |
One, the internet and it's core protocols are inherently not worthy of |
73 |
trust. There just isn't any way to prove that traffic is what it claims |
74 |
to be and no crypto verification built into the core of it. You either |
75 |
trust the traffic or you don't, but there's nothing inherent in the |
76 |
traffic to help you decide. So, all the download protocols have security |
77 |
checking bolted on afterwards by individual apps. These apps may or may |
78 |
not be compatible with each other and may or may not do their checks |
79 |
similarly from one protocol to the next. Somebody would have to garner |
80 |
enough support so that all the major projects doing file and data |
81 |
transfers agree on some way to implement crypto checks. Good luck with |
82 |
that :-) if they do agree on something, we have the second problem. |
83 |
|
84 |
Internet downloads have an inherent problem - you download an unknown |
85 |
bunch of bits from somewhere and can't fully trust the result. You can |
86 |
check hashes against the downloaded file, but you have to get them from |
87 |
somewhere. And the method to get them is the same as getting the data |
88 |
file itself - a bunch of bits from somewhere and you can't trust it. How |
89 |
can you download trusted hash data from a source where you don't trust |
90 |
the regular downloads? Can't work; two no trusts don't make a one trust. |
91 |
|
92 |
And who's global hash store of all known hashes of all known |
93 |
downloadables would you trust anyway? The NSAs? :-) |
94 |
|
95 |
Best you can do is make something for the specific case. The Gentoo tree |
96 |
and distfiles can be GPG signed and if you agree to trust Gentoo's keys |
97 |
then you are good to go and it can be automated (which is the easy bit |
98 |
btw). |
99 |
|
100 |
For the general case/ I can't see that work at all. I trust Gentoo with |
101 |
Gentoo, but I don't see myself ever trusting $ARB_3RD_PARTY with $EVERYTHING |
102 |
|
103 |
-- |
104 |
Alan McKinnon |
105 |
alan.mckinnon@×××××.com |