jbk thats yours? #15395 2bba9fd47 Zpool can start allocating from metaslab before TRIMs have completed ?

tsoome: yes

nice.

it's hopefully not something you'll normally come across

yes, I did read the description:)

but still... something that should be fixed

it's been reviewed and approved in OpenZFS as well

(because they appeared to have the same issue)

yep, thats the line from fbsd commit ;)

since you reviewed it, and it's been reviewed by OpenZFS, i'll probably RTI it in the next couple of days unless someone pipes up with an issue in the interim

jbk: Pretty terrifying tbh haha

Not doing anything to improve my general fear of trim

i think it should be better with the fix...

the implementation actually isn't too complicated... just neglected to consider the behavior on how zfs schedules i/os

so writes could skip ahead of the TRIM