git.kernel.org and lore.kernel.org now require proof of work. Big thanks to Anubis developers for filling this important need.

@monsieuricon what does this actually mean? I saw a loading bar that quickly disappeared before I could read anything

@cafkafk It means you're not a bot. :)

@cafkafk @monsieuricon essentially if it thinks the browser is a scraper, it runs a small script (similar to how Bitcoin mining works) to prove the client really cares about the content

@natty @cafkafk @monsieuricon this. Your computer is being given nonsense waste work to do before the page is loaded.

This incurs a small burden on individual users, but for scrapers making hundreds of thousands of requests it has a substantial dollar cost and acts as a disincentive for abuse.

@monsieuricon oh no I thought I'd get welcomed by anime girl on kernel pages :'(

@raito It's certainly an interesting topic, and the main reason I swapped the default graphics for generic emojis is really two-fold: 1. the tech community in general has a complicated history when it comes to objectifying women, so while having a feminine mascot greet you may be seen as a positive thing in one light, it can also be interpreted poorly. 2. "sad" and "happy" are emotions that can be difficult to judge for some people. I wanted a very blunt "working," "passed," "failed" graphics.

@monsieuricon is b4 mbox broken by this by any chance? I can't download:

```
> b4 mbox -n bla.mbox 20250403-quartal-kaltstart-eb56df61e784@brauner
Looking up https://lore.kernel.org/r/20250403-quartal-kaltstart-eb56df61e784@brauner
Grabbing thread from lore.kernel.org/all/20250403-quartal-kaltstart-eb56df61e784@brauner/T/20250403-quartal-kaltstart-eb56df61e784@brauner/t.mbox.gz
That message-id is not known.
```

@brauner It shouldn't be, and I can't replicate that failure.

@monsieuricon weird, thank you. I'll check again tomorrow if the problem still persists.

@cafkafk @monsieuricon

Is there any way to stop it before redirecting, for the curious?

@brauner @monsieuricon The link didn't work on my web browser, too (the page was saying "page not found"), a few hours ago. It works now, though.

@monsieuricon Meant to reply to your mail but forgot - I had trouble with someone else's anubis protected page not loading but the kernel ones work fine :)

@monsieuricon I see you replaced the logo, but didn't remove the text crediting the original logo's artist.

@monsieuricon @raito not only is this one of the better answers I have read on mastodon (friendly tone, good explanation of your decision), but I also missed how the anime girls expressions might be difficult to parse.
Also, I think usage in such a core part in the opensource ecosystem might be enough publicity for the Anubis dev, even without it's "logo". Did you ask Xe if removing the Anubis character for kernel.org is fine?

@monsieuricon wait is this supposed to affect link previews? Moshidom shows it with "Making sure you're not a bot!"

@monsieuricon thank you for putting that in place.

I have a strange issue when using a former web.git.kernel.org link with a shorter SHA:

- this one is not working: https://web.git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=4b74726c01b7 → no repository found

- shorter OK: https://web.git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=4b74726c01b

- longer OK: https://web.git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=4b74726c01b7a

- without "web." is OK: https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=4b74726c01b7

Any idea? :)

@monsieuricon I'm honestly not sure what's happening I get:

```
> b4 shazam -H --merge-base=HEAD --sloppy-trailers 20250403-work-pidfd-fixes-v1-0-a123b6ed6716@kernel.org
Looking up https://lore.kernel.org/r/20250403-work-pidfd-fixes-v1-0-a123b6ed6716@kernel.org
Grabbing thread from lore.kernel.org/all/20250403-work-pidfd-fixes-v1-0-a123b6ed6716@kernel.org/T/20250403-work-pidfd-fixes-v1-0-a123b6ed6716@kernel.org/t.mbox.gz
That message-id is not known.
```

It doesn't matter what message I use.

@brauner I have one other report of this. Could you send it to tools@kernel.org and include the output of `git config --get-regexp b4.*`

@brauner Also, should start working for you now as-is.

@monsieuricon Hm, I'm not sure I want to share that all so widely. Can I send it just to you?

@brauner No worries, I was already able to fix that anyway. So, unless you still have this problem, I would just move on at this point. :)

@monsieuricon It works for me again now. If you can explain it here I would like to know what happened. :)

@brauner The anubis bot protection that I put in place yesterday required remapping some of the mountpoints, such as the legacy /r/. Internally, b4 has been using /all/ instead of /r/ for a while, but people who had b4.midmask set to a URL with /r/ in it experienced problems. Fixing the /r/ mount fixed the problem.

@monsieuricon Thanks for the explanation! Based on your explanation: what is the canonical URL we should use for Link: tags? https://lore.kernel.org/r or /all?

@brauner You can continue to use /r/ in URLs, or just omit /r/ entirely. Don't use /all/.

@monsieuricon how did you manage to use Anubis with cgit? It only works with HTTP and does not support CGI forwarding

https://github.com/TecharoHQ/anubis/issues/11

@cybertailor we serve cgit via uwsgi

@monsieuricon thanks for the heads-up

@monsieuricon so it's been almost a week, I'm wondering, did Anubis actually help with reducing load on the infrastructure?

@fstachura Yes, dramatically. Currently, it is the most effective way of dealing with crawler bots.

@monsieuricon BPF CI sets the user agent to "BPFCIBot/1.0 (bpf@vger.kernel.org)" in wget, and that turns out to trigger the challenge flow of the proof-of-work proxy. Without it the download goes through just fine.

Is there a better alternative than simply dropping the custom user agent?

https://github.com/libbpf/ci/pull/178/files#diff-295db3ad3a1f3c6b2584b6078bc42f8184b2ae7fa5e634c6d352100b1b3ac927L38

@shunghsiyu just call it BPFCI/1.0. It's obviously a bot anyway and removing it won't trigger the challenge.

@monsieuricon got it, thanks!