Comment by BoppreH
16 days ago
The more I look at this the angrier I get. This is a CVE that Anthropic claims can get a Remote Code Execution on vulnerable systems, which includes default Ubuntu installs running any sort of Go server like Caddy. Bad enough that I shut down my personal website and self-hosted tools (sorry for any typos, my grammar checker is also down).
And this is how Ubuntu describes the vulnerability to sysadmins:
In the Linux kernel, the following vulnerability has been resolved:
ipv4: free inet_opt and ireq_opt after an RCU grace period
tcp_v4_syn_recv_sock() transfers ownership of ireq->ireq_opt to the
child socket (newinet->inet_opt) without copying it.
Another cpu can concurrently process a retransmitted SYN for the same
request socket, and send a SYNACK from tcp_check_req().
tcp_v4_send_synack() and inet_csk_route_req() read ireq->ireq_opt
under rcu_read_lock() only, and ip_build_and_send_pkt() and
ip_options_build() then read opt->optlen twice.
Note that the SYNACK timer itself is not an issue: it holds a
reference on its request socket, and inet_csk_reqsk_queue_drop()
calls timer_delete_sync() before the child can be freed.
Since commit 079096f103fa (“tcp/dccp: install syn_recv requests
into ehash table”), request sockets are processed without holding
the listener lock, so nothing prevents the child socket from being
freed while the SYNACK is still being built. TCP child sockets do
not have SOCK_RCU_FREE, and inet_sock_destruct() frees inet_opt
with a plain kfree(), leading to a use-after-free in
ip_options_build().
A similar issue exists with request socket migration
(net.ipv4.tcp_migrate_req=1, or a BPF_SK_REUSEPORT_SELECT_OR_MIGRATE
program). reqsk_timer_handler() clones the request socket with
inet_reqsk_clone(), so that the old request socket and its clone
share the same ireq_opt, then reqsk_migrate_reset() clears the
pointer in the old one. Another cpu holding a reference on the old
request socket can still be using these options (sending a SYNACK,
or creating a child in tcp_v4_syn_recv_sock()) when the clone is
freed, and tcp_v4_reqsk_destructor() also uses a plain kfree().
Readers already use RCU, and other paths replacing inet_opt
(do_ip_setsockopt(), cipso_v4_sock_setattr()...) already use
kfree_rcu(). Use kfree_rcu() in inet_sock_destruct() and
tcp_v4_reqsk_destructor() as well.
IPv6 is not affected by the first issue, because tcp_v6_syn_recv_sock()
duplicates the options. tcp_v6_reqsk_destructor() has the same
migration issue with ipv6_opt, which is only set by CALIPSO. This
will be addressed in a separate patch.
How is this in any way, shape, or form acceptable two weeks after the CVE was published?
sorry for any typos, my grammar checker is also down
I assume this is unrelated to the Ubuntu situation, but how does a grammar checker go down? We've been doing offline grammar checking since computers had 64K of RAM.
Self-hosted LanguageTool server[0] with browser addon. 18 gigabytes of n-gram data plus machine learning models. You can catch tpyos on 64K of RAM, but not any deep grammar mistakes, un-idiomatic phrases, or mistakes in common proper names.
[0] https://dev.languagetool.org/http-server.html
Sounds like a lot of heavy lifting for something that is maybe slightly better than spell checking 25 years ago.
Grammatik ran in 64K of RAM on a Z80 running CP/M. (The binary included a Forth interpreter so there was a lot of straight up code compaction.) So yeah, you could catch a lot of common English-language infelicities in a startlingly resource constrained environment... in 1983. Certainly a lot more than typos.
(The reason I remember this bright and shiny data point is I spent a week or three reverse engineering it and getting access to the inner interpreter; the office staff actually used it, though.)
consider harper instead. I'm using it right now and it highlighted first two words of this comment. https://writewithharper.com/
/o\ all this just for what exactly?
7 replies →
> running any sort of Go server like Caddy
It's so much worse than this, and there is a staggering amount of misinformation running rampant about this bug.
MPTCP is not a requirement, it can be exploited merely by sending specially crafted TCP packets with (any) IP options set. No special MPTCP support, web servers or any specific software listeners are even necessary at all to trigger it, just send the packets and boom.
I think the reason people assume MPTCP is a requirement is because that's what the PoC used... but MPTCP itself also doesn't even rely solely on IP options.
Ubuntu also does not even list a single mitigation strategy... at least RHEL did.
And like I said, any normal TCP packet that contains IP (not TCP) options can trigger the bug. I cannot emphasize enough how important and absolutely ridiculous this is.
Like the post says, this is big enough to get a massive amount of people to completely stop trusting Ubuntu entirely and jump ship.
> Ubuntu also does not even list a single mitigation strategy...
I installed newest Ubuntu mainline kernel, it did work for me:
https://kernel.ubuntu.com/mainline/v7.3.1/
I also did patch my FLOSS Go software:
https://github.com/wekan/wekan/commit/e9751b65d4
I'm maintainer of WeKan FLOSS kanban https://wekan.fi , https://github.com/wekan/wekan , and maintainer of FerretDB v1 fork https://github.com/wekan/FerretDB . There is also WeKan Snap package https://snapcraft.io/wekan/ .
But I don't work for Ubuntu, I don't have any answers why Ubuntu does not do something.
I have some publicly facing web servers running Debian and Ubuntu. Specifically, what can I do to test and see if my servers are vulnerable?
Debian (trixie kernel) 6.12.113 was allegedly patched for this bug, but there's also a .115 that's been released in the last few days, upgrade to that and reboot.
For Ubuntu (which I don't use), you'd need a 3rd party updated kernel repo, to use a patched kernel from that.
It's not an instantaneous exploit; it apparently takes a lot of iteration over the network. A lot of cloud network providers are mitigating it by dropping options at the border, but unless you test and confirm that, you should assume it's possible for a motivated-enough attacker to compromise your server.
1 reply →
Someone from Canonical answered about kernel:
https://news.ycombinator.com/item?id=50348688