[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

[tor-dev] Re: [ROOT CAUSE & BULLETPROOF FIX] Why Tor 0.4.9.16 Fails Under HTTP-Wait BEGIN_DIR Floods & The 5-Layer Patch



Hi Adam,

Great to see the progress!

Our November 9 report [1] covered 110 active windows on 93 instrumented Tor 0.4.9.15 relays. Time-weighted averages were about 50,000 admitted BEGIN_DIR streams/s, 49,000 streams closed while waiting for HTTP/s, and 98% main-thread CPU.

Separately, the HSDir limit refused about half of HSDir GET requests in GET-largest windows.

Both our HTTP-wait closure and BEGIN_DIR cell-position counters came from our custom patch; stock .15/.16 does not export them. We have requested and offered this plus more improved upstream metrics [2].

Three corrections:

HTTP-wait describes the classification at closure. It does not establish whether bytes previously arrived, waiting duration, or closure cause. Closure rates do not measure concurrent connections.

Position 101+ means the 101st or later BEGIN_DIR cell on a circuit; it does not identify an attack’s starting position.

Separately from our published analysis, the unconditional 16 KB input-buffer claim does not match stock .16. buf_new() allocates an empty buffer structure; payload chunks are allocated when needed [3].

Did our analysis inform section 1? If so, please cite it and preserve these distinctions as they'll impact measurements and mitigations.

[1] Our measurements:
https://1aeo.com/blog/tor-begindir-hsdir-mitigation-pressure.html

[2] Our upstream metrics request:
https://gitlab.torproject.org/tpo/core/tor/-/work_items/41460

[3] Tor 0.4.9.16 buffer allocation:
https://gitlab.torproject.org/tpo/core/tor/-/blob/tor-0.4.9.16/src/lib/buf/buffers.c#L363-370


On Friday, November 13th, 2026 at 11:02 AM, forest via tor-dev <tor-dev@xxxxxxxxxxxxxxxxxxxx> wrote:

> On 2026-11-11 15:13, keevesadam98--- via tor-dev wrote:
> > The testbench source code and verification harness are fully
> > documented. We are happy to submit this logic as a formal merge
> > request to upstream Tor core (tpo/core/tor).
>
> Please do. If you've verified that it works, please create a merge
> request. Many of our relays are falling due to this continued attack.
>
> Regards,
> forest

Attachment: publickey - tor@1aeo.com - 0x9288289B.asc
Description: application/pgp-keys

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
tor-dev mailing list -- tor-dev@xxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to tor-dev-leave@xxxxxxxxxxxxxxxxxxxx