[gentoo-user] SSDs with TRIM issues

[gentoo-user] SSDs with TRIM issues

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 600 bytes --]

/* devices that don't properly handle queued TRIM commands */
Micron_M500*
Crucial_CT*M500*
Micron_M5[15]0*
Crucial_CT*M550*
Crucial_CT*MX100*
Samsung SSD 8*

from drivers/ata/libata-core.c

More info:
http://linux.slashdot.org/story/15/06/16/201217/trim-and-linux-tread-cautiously-and-keep-backups-handy

Also, my samsung 840 is slow (multisect is off and cant be enabled with
hdparm). Tried update the bios but the boot cd crashed on my machine (AMD
chipset related) so i will have to put it in another machine and try
again...

Samsung EVOs also have a firmware bug that causes them to slow down.

[-- Attachment #2: Type: text/html, Size: 822 bytes --]

Re: [gentoo-user] SSDs with TRIM issues

[Rich Freeman]

On Tue, Jun 16, 2015 at 10:24 PM, Adam Carter <adamcarter3@gmail.com> wrote:
> /* devices that don't properly handle queued TRIM commands */
> Micron_M500*
> Crucial_CT*M500*
> Micron_M5[15]0*
> Crucial_CT*M550*
> Crucial_CT*MX100*
> Samsung SSD 8*
>

As somebody pointed out on Slashdot, this is basically all the models
that otherwise seem to perform really well.

The flip side of this is that nobody can really say that other models
of drives aren't on the list because they actually work, vs not being
on the list because nobody has bothered to test them extensively
enough to detect a problem.

Also, I believe they're trying to be conservative with the
black-listing. There could very well be drives that aren't affected
that match these patterns, and I believe this is supposed to get
sorted out once the drive vendors get their acts together.

Also, I don't know which stable kernels have the blacklist.  It
doesn't look like this list is in 3.18.14 which is the most recent
longterm in Gentoo (3.18.15 is out but not yet in gentoo-sources).  I
went ahead and disabled fstrim and discard in the meantime.


-- 
Rich

Re: [gentoo-user] SSDs with TRIM issues

[mario_reja]

On Wed, Jun 17, 2015 at 12:24:55PM +1000, Adam Carter wrote:
> /* devices that don't properly handle queued TRIM commands */
> Micron_M500*
> Crucial_CT*M500*
> Micron_M5[15]0*
> Crucial_CT*M550*
> Crucial_CT*MX100*
> Samsung SSD 8*
> 
> from drivers/ata/libata-core.c
> 
> More info:
> http://linux.slashdot.org/story/15/06/16/201217/trim-and-linux-tread-cautiously-and-keep-backups-handy
> 
> Also, my samsung 840 is slow (multisect is off and cant be enabled with
> hdparm). Tried update the bios but the boot cd crashed on my machine (AMD
> chipset related) so i will have to put it in another machine and try
> again...
> 
> Samsung EVOs also have a firmware bug that causes them to slow down.

Regarding Crucial SSDs: MU02 firmware isn't blacklisted.

--
Mario

Re: [gentoo-user] SSDs with TRIM issues

[Bruce Hill]

On Wed, Jun 17, 2015 at 12:24:55PM +1000, Adam Carter wrote:
> /* devices that don't properly handle queued TRIM commands */
> Micron_M500*
> Crucial_CT*M500*
> Micron_M5[15]0*
> Crucial_CT*M550*
> Crucial_CT*MX100*
> Samsung SSD 8*
> 
> from drivers/ata/libata-core.c

Then read further:

         * Whitelist drives that are known to reliably return zeroes
         * after TRIM.

         * The intel 510 drive has buggy DRAT/RZAT. Explicitly exclude
         * that model before whitelisting all other intel SSDs.
         */
        { "INTEL*SSDSC2MH*",            NULL,   0, },

        { "Micron*",                    NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, },
        { "Crucial*",                   NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, },
        { "INTEL*SSD*",                 NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, },
        { "SSD*INTEL*",                 NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, },
        { "Samsung*SSD*",               NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, },
        { "SAMSUNG*SSD*",               NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, },
        { "ST[1248][0248]0[FH]*",       NULL,   ATA_HORKAGE_ZERO_AFTER_TRIM, }

> More info:
> http://linux.slashdot.org/story/15/06/16/201217/trim-and-linux-tread-cautiously-and-keep-backups-handy
> 
> Also, my samsung 840 is slow (multisect is off and cant be enabled with
> hdparm). Tried update the bios but the boot cd crashed on my machine (AMD
> chipset related) so i will have to put it in another machine and try
> again...
> 
> Samsung EVOs also have a firmware bug that causes them to slow down.

I chose long ago not to use queued TRIM (when OCZs were the rage).

From a cursory read this sounds like *only* ext4 fs. I have XFS on all my
SSDs. Two are Samsung 840 Pro Series, and 850 Pro Series. The 840 I've had
over a year and it is golden. The 850 I've had only a couple of months, and
due to work requirements it is running Windows atm. Still, there is nothing
like what I read in this article happening with either.

I also have a Micron C400 RealSSD mSATA 32GB -- no problems with it, either.
It's running Funtoo atm with VFAT /boot for UEFI, swap, all other XFS.

Re: [gentoo-user] SSDs with TRIM issues

[Bruce Hill]

On Wed, Jun 17, 2015 at 03:16:16PM -0400, Rich Freeman wrote:
> 
> Also, I don't know which stable kernels have the blacklist.  It
> doesn't look like this list is in 3.18.14 which is the most recent
> longterm in Gentoo (3.18.15 is out but not yet in gentoo-sources).  I
> went ahead and disabled fstrim and discard in the meantime.

mingdao@headache ~ $ uname -a
Linux headache 4.0.5 #1 SMP Mon Jun 15 17:36:18 CDT 2015 x86_64 Intel(R) Core(TM) i7-4712HQ CPU @ 2.30GHz GenuineIntel GNU/Linux

This one has the blacklist, but all my kernel source comes from kernel.org