Comment by killerstorm
3 months ago
~12 years ago I installed Linux on Fujitsu UH572.
As an "ultrabook" it came with 16GB of fast SSD which could be used as a cache via some Windows-specific Intel feature, while main storage was a slower spinning disk.
As Linux did not support cache feature, I thought I can just format it as ext4 and use as a storage for things which can benefit from more iops, e.g. running DB tests. And as SSD technology was still rather new at that time, I started with running some IO benchmarks.
Well, it survived formatting into ext4 and few minutes of that IO benchmark, then it became permanently unresponsive.
My guess is that wear-leveling algorithm was designed specifically for the FS originally had (some version of FAT?), and different FS caused it to corrupt some internal data structure, so SSD's controller firmware went into panic on each boot.
Unfortunately, this added few minutes delay to boot as Linux tried to communicate with unresponsive SSD controller, but otherwise laptop worked fine...
Similar-enough story on an Asus similarly branded "ultrabook", I figured I would put my whole fedora root partition on that cache, which worked wonders for a few months, until I started having random freezes and crashes that I later diagnosed thanks to `rpm -V` to be data corruptions. At that point the hinges were dead, pulling the screen away from the chassis during operations and dangerously exposing a mainboard PCB. That's when I replaced it with a ThinkPad T that's still going like new after 8 years :-)