50 years in filesystems: 1984 BSD FFS

* https://web.archive.org/web/20170213221835/http://www.nobius...

For example: It's fairly well recorded that FreeBSD switched to Mewburn rc, which it got from NetBSD, in 2002. But what isn't so well recorded, that one can only really find out from the version control history, is that this was actually a regression in one aspect, as FreeBSD had some years earlier finally got rid of an old backwards-compatibility mechanism, that Mewburn rc promptly restored.

* http://jdebp.info/FGA/rc.local-is-history.html

One of the interesting things about FreeBSD et al. is that one can reach back decades in version control and look at this stuff.

Another example: Linux people often talk about BSD-style options to their "ps" command. But they actually aren't, and BSD has not documented any such things for the entire lifetime of any "ps" command on Linux. Inspecting BSD version control history one can find where the BSD "ps" switched to getopt(3) and changed its manual, and it was in 1990, a year before Linux even existed.

There's also a series of videos from classes McKusick taught. Back in the 90s/early 2000s I attended some viewings of these at the apartment of some High Voltage Software employees back in IL. They'd convinced their employer to foot the bill on buying the full VHS set (I think it was $1k/tape), and hosted a bunch of viewing "parties".

Whenever McKusick had to explain some virtual memory details, he'd have vi as the running process and some user started up emacs, then he'd walk through all the minutia of paging and swap. It was quite amusing at the time, he was a master at throwing fuel on the emacs vs. vi religious wars.

I was never a huge freebsd fan, but these books were very well done and McKusick a tremendous asset to that community.

It was more the opposite IME (90s-era linux): fsck after a crash was boring and time-consuming, and we grew increasingly impatient as capacities grew.

Prior to having a journaling filesystem and log replay @ reboot/mount, the filesystem was often plain conservatively implemented using methods like two-phase-commits and arguably kept more consistent due to the repeated exhaustive checking process at every dirty mount.

We went through a phase of putting a lot of trust into the hardware keeping things consistent/no-bitrot and more or less assuming everything was fine after a journal replay. Only relatively recently did we start getting checksums of metadata to detect such failures, that's still not a universal feature.

The lack of frequent fscks creates opportunity for hidden corruption from either bugs or hardware errors to go unnoticed until it's far too late to recover from... There's a reason filesystems tend to have a max mount count before a fsck is forced, even if not dirty.

Journaling did take off in a big way, with NTFS, HFS+, ext3, etc. supporting it. This mostly removed the long fsck on most modern operating systems.

Of course, traditional filesystems + journaling are now rapidly being replaced by CoW filesystems.