HN Explorer

9 comments

[ 3.3 ms ] story [ 34.4 ms ] thread
beanjuiceII
ok? it looks like there were not that many commits in that many days so what does this mean?
mkeeter
Finding bugs in the Hubris kernel is rare enough that we have a running joke about resetting the “days since last kernel bug” timer.

I decided to make this joke into an actual docs page; because HN has enjoyed posts about Hubris in the past [1], I figured this might be of interest!

Many of the individual bugs are terrifying dives into corner cases of an embedded OS. This one is particularly good reading: https://github.com/oxidecomputer/hubris/issues/1134

[1] https://news.ycombinator.com/item?id=29390751

p1necone
It means they made a cool status page that lists all the bugs that have been found in the Hubris kernel, along with a bunch of detail about fixing them.

Not everything has to satisfy some arbitrary measure of usefulness to be cool/fun to look at/work on.

hiddencost
Neither of the last two bugs has any appreciable consequences besides (1) breaking a debug tool, or (2) making crashes possible in the future.

>460 days without any serious issues...

mkeeter
The second-from-most-recent bug, "Leases cannot span MPU regions", did in fact cause immediate failures in our network switch image!

Cliff wrote this up in a blog post, which also had substantial discussion:

Who killed the network switch? A Hubris Bug Story (2024) - https://news.ycombinator.com/item?id=39813365 - March 2024 (128 comments)

anitil
I wish I was still working in the embedded world so I could give this a whirl!