News for hackers. Stuff that matters.

A comment on History of 127/8 as localhost/loopback addresses (2021). Parent comment

Re:History of 127/8 as localhost/loopback addresses (2021) by blibble on Wednesday May 25, 2022 @08:34AM (#31503617)
that seems completely insane

that logic has been hard-coded inside every OS and vast amounts of software for 30+ years

and to gain less than 0.8%?

they'd do better targeting the reserved multicast range

[ Parent | on HN ]
Re:History of 127/8 as localhost/loopback addresses (2021) by greyface- on Wednesday May 25, 2022 @10:20AM (#31504902)
> they'd do better targeting the reserved multicast range

https://www.ietf.org/archive/id/draft-schoen-intarea-unicast...

[ Parent | on HN ]
Re:History of 127/8 as localhost/loopback addresses (2021) by justsomehnguy on Wednesday May 25, 2022 @10:12AM (#31504782)
> inside every OS and vast amounts of software for 30+ years

Ha. Ha. Ha.

> that logic has been hard-coded inside

The thing is what some things are literally hardcoded. I have a whole bunch of network gear circa 2010 in production.

It does support IPv6, but for the most part it is extremely rudimentary. What do you want from a box what implements SSH by running a telnet session over a no-auth SSH session?

And while I can run the latest Windows/Linux/Whatever on my server gear, the network gear is not that simple to upgrade.

[ Parent | on HN ]
Re:History of 127/8 as localhost/loopback addresses (2021) by vlan0 on Wednesday May 25, 2022 @05:11PM (#31510613)
>I have a whole bunch of network gear circa 2010 in production

I ran into this 10 years ago. I decided to roll out IPv6 to clients. Router docs said it was supported. Then I watched CPU usage peg because every packet hit the CPU. Turns out it was implemented in software; no ASIC support. Was quite the buzz kill.

[ Parent | on HN ]
Re:History of 127/8 as localhost/loopback addresses (2021) by p_l on Wednesday May 25, 2022 @11:03PM (#31513912)
Now consider that there are network components in the wild that have literal latch on first octet equaling 127 which triggers packet drop, and this proposal could drop hw acceleration for v4 too
[ Parent | on HN ]