News for hackers. Stuff that matters.

A comment on Transforming a QLC SSD into an SLC SSD. Parent comment

Re:Transforming a QLC SSD into an SLC SSD by howbadisthat on Sunday May 19, 2024 @10:32AM (#40407222)
What happens if the remaining space is TRIMMED but routinely accessed? (for example by dd, read only)
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by wtallis on Sunday May 19, 2024 @12:00PM (#40407774)
If a logical block address is not mapped to any physical flash memory addresses, then the SSD can return zeros for a read request immediately, without touching the flash.
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by stavros on Sunday May 19, 2024 @02:54PM (#40408959)
Does the mapping happen on first write? Is TRIM then a command that signals the SSD to unmap that block?
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by BeeOnRope on Sunday May 19, 2024 @04:55PM (#40409680)
Yes.
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by Dwedit on Sunday May 19, 2024 @12:38PM (#40408066)
You read zeroes.
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by BeeOnRope on Sunday May 19, 2024 @05:04PM (#40409731)
Not guaranteed by default for NVMe drives. There's an NVMe feature bit for "Read Zero After TRIM" which if set for a drive guarantees this behavior but many drives of interest (2024) do not set this.
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by Dwedit on Sunday May 19, 2024 @06:26PM (#40410214)
Hmmm, when I quick-formatted a drive (which TRIMs the whole thing), then tried reading it back in a disk hex editor, I just saw zeroes.
[ Parent | on HN ]
Re:Transforming a QLC SSD into an SLC SSD by BeeOnRope on Sunday May 19, 2024 @07:41PM (#40410595)
It seems like that would be a likely common behavior for the FTL, but it wasn't guaranteed by the spec, which is why they added this NVMe flag so that you can actually rely on it.
[ Parent | on HN ]