trim support (discard)

trim support (discard)

[Jens Kasten]

Hi list,

It is the first time the i write to this list.
I asked on the ext4 ml and they told me to ask here.

I use the kernel 2.6.37 with cryptsetup and lvm on a intel ssd. 
In the logfile after booting i see this entry:
EXT4-fs (dm-1): re-mounted. Opts: discard,errors=remount-ro
EXT4-fs (sda1): mounted filesystem with ordered data mode. Opts: discard
EXT4-fs (dm-3): mounted filesystem with ordered data mode. Opts: discard
EXT4-fs (dm-2): mounted filesystem with ordered data mode. Opts: discard
EXT4-fs warning (device dm-1): ext4_issue_discard:2619: discard not
supported, disabling
EXT4-fs warning (device dm-3): ext4_issue_discard:2619: discard not
supported, disabling
EXT4-fs warning (device dm-3): ext4_issue_discard:2619: discard not
supported, disabling
EXT4-fs warning (device dm-3): ext4_issue_discard:2619: discard not
supported, disabling
EXT4-fs (sda1): re-mounted. Opts: discard,commit=0
EXT4-fs (dm-3): re-mounted. Opts: discard,commit=0
EXT4-fs (dm-2): re-mounted. Opts: discard,commit=0

The mount command shows:
/dev/mapper/System2-root_ssd on / type ext2
(rw,noatime,errors=remount-ro,barrier=1,stripe=32,data=ordered,discard)
and the corresponding entry from fstab:
UUID="635a1ad9-9ebb-4d1e-b8c3-30e11148bd92"     /                   ext4
noatime,discard,errors=remount-ro               0       1

So, no discard support so far.
Then i try the wiper.sh script and i get:
wiper.sh: Linux SATA SSD TRIM utility, version 3.1, by Mark Lord.
/dev/dm-1: unable to reliably determine the underlying physical device
name, aborting

Its there in near future the discard support on the LVM?

Viele Grüsse
Jens Kasten

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Re: trim support (discard)

[Milan Broz]

On 01/08/2011 08:45 PM, Jens Kasten wrote:
> 
> I use the kernel 2.6.37 with cryptsetup and lvm on a intel ssd. 

Device-mapper targets supports discards in 2.6.37 with exception
of dm-crypt target (that's why you see that message).

Supporting discards in dm-crypt means that information
about unused blocks is leaked (SSD ususally returns zeroes instead
of expected random data for discarded blocks) and it it can be
security problem in some situations.

We will probably add discard support in dm-crypt later but as optional
parameter (admin must decide that leaking allocated block info
is not problem and must explicitly enable it).
Probably 2.6.39 material.

> Then i try the wiper.sh script and i get:
> wiper.sh: Linux SATA SSD TRIM utility, version 3.1, by Mark Lord.
> /dev/dm-1: unable to reliably determine the underlying physical device
> name, aborting

That seems like bug in that script btw, it should understand DM devices.

> Its there in near future the discard support on the LVM?

LVM/DM itself should fully support discards now.
Exception is stacking with dm-crypt devices.

Milan

Re: trim support (discard)

[Jens Kasten]

Am Sonntag, den 09.01.2011, 14:28 +0100 schrieb Milan Broz:
> On 01/08/2011 08:45 PM, Jens Kasten wrote:
> > 
> > I use the kernel 2.6.37 with cryptsetup and lvm on a intel ssd. 
> 
> Device-mapper targets supports discards in 2.6.37 with exception
> of dm-crypt target (that's why you see that message).
> 
> Supporting discards in dm-crypt means that information
> about unused blocks is leaked (SSD ususally returns zeroes instead
> of expected random data for discarded blocks) and it it can be
> security problem in some situations.

Thank you, for this information.
About this i have never thought.
   
> We will probably add discard support in dm-crypt later but as optional
> parameter (admin must decide that leaking allocated block info
> is not problem and must explicitly enable it).
> Probably 2.6.39 material.
> 
> > Then i try the wiper.sh script and i get:
> > wiper.sh: Linux SATA SSD TRIM utility, version 3.1, by Mark Lord.
> > /dev/dm-1: unable to reliably determine the underlying physical device
> > name, aborting

Ok, I take a look on the script.

> That seems like bug in that script btw, it should understand DM devices.
> 
> > Its there in near future the discard support on the LVM?
> 
> LVM/DM itself should fully support discards now.
> Exception is stacking with dm-crypt devices.
> 
> Milan
> 
> --
> dm-devel mailing list
> dm-devel@redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel