5 Jul 2022 |
Alyssa Ross | hexa: did I accidentally take on OpenSSL responsibility by renaming the package? :P | 15:26:41 |
6 Jul 2022 |
| deep6 joined the room. | 08:00:25 |
stigo | Added Demi Marie's patch from the last oss-sec post in a PR https://github.com/NixOS/nixpkgs/pull/180336 | 12:17:03 |
stigo | It seems legit to me, but I'm not sure if we should wait from some info from upstream.. | 12:18:46 |
stigo | it's a bit annoying that it hasn't gotten a CVE (yet), not sure what to search for to see other discussions about the issue. | 12:26:28 |
@linus:schreibt.jetzt | Hm, signatures wouldn't actually be compressible anyway, right? | 13:56:06 |
stigo | They can, it seems | 13:56:44 |
hexa | https://cr.yp.to/sigs/compress.html | 13:56:53 |
stigo | Or, they can be put in a compressed packet | 13:57:32 |
stigo | "By concatenating the
same signature to itself a very large number of times, and then wrapping
them in a compressed packet, I am able to cause GnuPG to take over a
minute to process an input that is less than 5KB armored." | 13:59:51 |
@linus:schreibt.jetzt | right, but compressed packets are compressed with generic data compression, which is different from what djb describes, and generic data compression wouldn't work well on cryptographic sigs, so disallowing it Should Be Fine™️... right? | 14:02:41 |
@linus:schreibt.jetzt | (I really don't know enough about PGP to be sure of any of this...) | 14:03:11 |
stigo | yes i think that's right, in the gnupg case it seems to be like a signature can be turned into a "zip bomb" | 14:05:04 |
hexa | this is how I understood it as well | 14:11:17 |
hexa | the author of the patch kindly demonstrated that on oss-sec | 14:11:25 |
stigo | You can test it with curl -O https://seclists.org/oss-sec/2022/q3/att-9/decomp-3.bin && gpg --verify decomp-3.bin /dev/null | 14:12:25 |
stigo | (slower if you import the pubkey https://seclists.org/oss-sec/2022/q3/att-9/test-key_cert.bin it seems) | 14:14:04 |
stigo | * You can test it with #POC# curl -O https://seclists.org/oss-sec/2022/q3/att-9/decomp-3.bin && gpg --verify decomp-3.bin /dev/null | 14:16:15 |
stigo | hm, maybe more fun with some very large rsa keys | 14:24:14 |
7 Jul 2022 |
stigo | https://github.com/NixOS/nixpkgs/pull/180336 broke tests in gpgme , can't look at it today, too much work to do. | 08:49:42 |
hexa | Not a good day for me either, sorry | 09:32:39 |
@linus:schreibt.jetzt | hexa: vcunat (thanks!) seems to have taken care of it :) | 09:36:16 |
vcunat | Well, I unblocked the builds, but I haven't really verified whether the test detected some real unexpected consequence. | 09:38:15 |
vcunat | And perhaps notify some appropriate place around upstream or around the patch? | 09:38:50 |
stigo | It's unclear what upstream thinks of this patch/bug -> https://seclists.org/oss-sec/2022/q3/27 | 12:11:10 |
stigo | In reply to @vcunat:matrix.org And perhaps notify some appropriate place around upstream or around the patch? Pinged the patch author on #180336, just in case | 12:49:42 |
hexa | cool. | 12:54:20 |
hexa | https://element.io/blog/element-launches-chatterbox/ | 14:14:06 |
hexa | looks like they're selling that feature | 14:16:27 |
Dandellion | https://github.com/vector-im/chatterbox it's foss, they're selling homeserver hosting as usual (just lower rates for the guest users through the livechat thing) | 16:15:48 |