• Time Nut (unregistered)

    PHP 7.4, 8.0, 8.2 maybe EOL'd but at least one vendor has back ported the fixes and had them out the same day as the fixes for 8.4 and above in their supported stream.

  • no (unregistered) in reply to Time Nut

    official backports for 7.4, 8.0, and 8.1 were released at the same time as the fixes for non-EOL version, albeit only in the form of source code:

  • (nodebb)

    Re: PHP2Shell - exploitation is decidedly non-trivial. Even the "we've seen it in the wild" example was against a very specific configuration, which anyone is unlikely to use. It involves using one type of apache worker, and configured to have a single worker process, not the multiple you'd usually configure (and is the default). Oh, and the artificially limited the OOM limit lower than the default as well.

    So, sure, some people might be vulnerable, and an attacker might get things to align between the races involved, the OOM setting, the exact versions of PHP and other libraries etc, but it does seem vanishingly unlikely.

    The biggest issue might be with some embedded devices using PHP, and which no longer receive security updates. Someone could hammer away at those with a botnet until they fall through luck.

  • Darren (unregistered)

    What makes the eMudhra situation even worse (for them) is they claimed that their processes were audited and given a pass. Probably a bit like the Cyber Essentials 'audit' where you answer a couple of questions and the only thing stopping you passing is if the cheque bounces...

    One thing that stood out for me with the PHP vulnerability was the blind panic it brought about in senior management / board-level. Because they'd heard about it on the news they had to be seen to be doing something and so the IT departments were given even more work to do to show that they were handling the PHP issue - either through patching, mitigations or the fact that they weren't actually using PHP - and that all had to be documented. In triplicate. With a short animated presentation, as we all know senior management are far too important to actually read something and can only respond to traffic light graphics.

  • 516052 (unregistered) in reply to Darren

    Well to be fair, figuring out how to writhe every single last drop of profit from the stone that are your overworked underpaid workforce is exhausting work. And it only gets harder the more you crush. Its diminishing returns all the way down yet you have no choice but to chase every last cent. And that's in the civilized world. I can't imagine how it must be for the poor CEOs in a place with no or weak worker protections. Those poor people must be spending every waking and sleeping hours of their day thinking up new evils to inflict.

    So can you really blame them for not caring about the actual work being done as long as all the rights are green?

  • LongtimeLurker (unregistered)

    Did anyone else have "Sharpened Focus on Core Enterprise Solutions" on their buzzword bingo card?

  • Commander Javik (unregistered)

    eMudhra didn't generate impossibly insecure keys

    Why yes, yes they did.

    They performed a key signing ceremony and generated appropriately strong keys. Then they used an in-house tool to pack/unpack the keys

    Err... no. That's not how it's done.

    Compliant root CAs have to use the HSM and private keys must not only be generated in there but they must never leave it. Instead, whatever needs signing goes into HSM and signed shit comes out of it. Private key NEVER comes out of it. Heck, even my Yubikey 5 NFC won't let you export private key, and if you want to use it for code signing you must generate attestation and show it as a proof that private key has never left it.

    So no, they are fucking liars. They generated keys outside HSM then botched the import or worse, they never even had a HSM to begin with. That they were allowed to operate boggles the mind.

  • erffrfez (unregistered)

    in a similar fashion to "Trust Delivered" means opposite from what the two words suggest, I can speak after first hand knowledge and say that companies / services with names of "Secure Parking" and "Spotless Painting" quite assuredly are not either of these things

  • Old Timer (unregistered) in reply to erffrfez

    My first rule of food nomenclanture: if it's got the name of place in it, that's not what they eat in that place; French Toast: not French. Hamburger: not in the style of Hamburg. "American Dogs”: not Corn Dogs.

Leave a comment on “Security Fourth”

Log In or post as a guest

Replying to comment #705871:

« Return to Article