[Discuss] SSD Drives

[dsr at randomstring.org (Dan Ritter)]

On Sat, Jun 20, 2015 at 09:36:41AM -0400, Richard Pieri wrote:
> On 6/20/2015 9:27 AM, Jerry Feldman wrote:
> >At last Wednesday's BLU meeting a discussion was started regarding some
> >issues with the Samsung SSD drives and some bugs in their firmware.
> >In my case I didn't have my laptop, but I have an Intel SSD not that it
> >matters.
> 
> There is a design flaw in the Samsung 840 EVO drives. Only the 840
> EVO drives are affected; the flaw does not affect 840 PRO or any 850
> series or any other vendor's drives. The 19nm NAND chips in 840 EVO
> combined with buggy firmware caused the drives to go into error
> correcting mode when reading data more than 8 weeks old. Samsung has
> since released a firmware update that appears to correct the
> problem. This is based on early reports 8 weeks after the firmware
> release indicating that the performance degradation is no longer an
> issue.

There is a separate issue affecting several Samsung lines and
Crucial's MX500 and MX550 lines: the queued-trim system may
cause data coruption.

Note that queued-trim is a variant on the normal trim support,
and is not used at all on Linux kernels before 3.12 or so.


http://forums.crucial.com/t5/Crucial-SSDs/M500-M5x0-QUEUED-TRIM-data-corruption-alert-mostly-for-Linux/td-p/151028

https://blog.algolia.com/when-solid-state-drives-are-not-that-solid/

-dsr-