[dm-crypt] Does dm-crypt support journaling filesystem transactional guarantees?

[dm-crypt] Does dm-crypt support journaling filesystem transactional guarantees?

[Yang Zhang]

I'm considering using ext4 on encrypted LVM (which uses LUKS and
dm-crypt). Will the transactional guarantees in ext4's journaling be
preserved?

Bonus: where may I find authoritative information on this (besides the source)?

Thanks in advance for any answers.

Re: [dm-crypt] Does dm-crypt support journaling filesystem transactional guarantees?

[Milan Broz]

On 04/24/2012 08:45 AM, Yang Zhang wrote:
> I'm considering using ext4 on encrypted LVM (which uses LUKS and
> dm-crypt). Will the transactional guarantees in ext4's journaling be
> preserved?

yes. dm-crypt operates in block layer, so it is filesystem responsibility
to properly set needed bits for IO (flush cache, FUA - force unit access)
and dmcrypt (device-mapper in general and block layer) must process them.

(read http://lwn.net/Articles/400541/ for more info).

In short, ext4, xfs, btrfs and similar filesystems supports
all features over dmcrypt.

(Except very old kernels, but most of stable distros backports patches.)
 
> Bonus: where may I find authoritative information on this (besides the source)?
The source is your best friend :-)

basically this is the first commit introducing real barrier
(later replaced with FUA) support
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=647c7db14ef9cacc4ccb3683e206b61f0de6dc2b

You can also use blktrace to see how are fs requests propagated
to physical media.

Milan

Re: [dm-crypt] Does dm-crypt support journaling filesystem transactional guarantees?

[Arno Wagner]

On Tue, Apr 24, 2012 at 09:47:42AM +0200, Milan Broz wrote:
> On 04/24/2012 08:45 AM, Yang Zhang wrote:
> > I'm considering using ext4 on encrypted LVM (which uses LUKS and
> > dm-crypt). Will the transactional guarantees in ext4's journaling be
> > preserved?
> 
> yes. dm-crypt operates in block layer, so it is filesystem responsibility
> to properly set needed bits for IO (flush cache, FUA - force unit access)
> and dmcrypt (device-mapper in general and block layer) must process them.
> 
> (read http://lwn.net/Articles/400541/ for more info).
> 
> In short, ext4, xfs, btrfs and similar filesystems supports
> all features over dmcrypt.
> 
> (Except very old kernels, but most of stable distros backports patches.)

See also FAQ item 2.10, which I just have updated to reflect this 
information ;-)

Arno
  


> > Bonus: where may I find authoritative information on this (besides the source)?
> The source is your best friend :-)
> 
> basically this is the first commit introducing real barrier
> (later replaced with FUA) support
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=647c7db14ef9cacc4ccb3683e206b61f0de6dc2b
> 
> You can also use blktrace to see how are fs requests propagated
> to physical media.
> 
> Milan
> _______________________________________________
> dm-crypt mailing list
> dm-crypt@saout.de
> http://www.saout.de/mailman/listinfo/dm-crypt
> 

-- 
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: arno@wagner.name 
GnuPG:  ID: 1E25338F  FP: 0C30 5782 9D93 F785 E79C  0296 797F 6B50 1E25 338F
----
One of the painful things about our time is that those who feel certainty 
are stupid, and those with any imagination and understanding are filled 
with doubt and indecision. -- Bertrand Russell 

Re: [dm-crypt] Does dm-crypt support journaling filesystem transactional guarantees?

[Yang Zhang]

On Tue, Apr 24, 2012 at 1:14 AM, Arno Wagner <arno@wagner.name> wrote:
> On Tue, Apr 24, 2012 at 09:47:42AM +0200, Milan Broz wrote:
>> On 04/24/2012 08:45 AM, Yang Zhang wrote:
>> > I'm considering using ext4 on encrypted LVM (which uses LUKS and
>> > dm-crypt). Will the transactional guarantees in ext4's journaling be
>> > preserved?
>>
>> yes. dm-crypt operates in block layer, so it is filesystem responsibility
>> to properly set needed bits for IO (flush cache, FUA - force unit access)
>> and dmcrypt (device-mapper in general and block layer) must process them.
>>
>> (read http://lwn.net/Articles/400541/ for more info).
>>
>> In short, ext4, xfs, btrfs and similar filesystems supports
>> all features over dmcrypt.
>>
>> (Except very old kernels, but most of stable distros backports patches.)
>
> See also FAQ item 2.10, which I just have updated to reflect this
> information ;-)

At first I was confused what this was referring to - for posterity,
turns out this is the (very useful) cryptsetup FAQ at
http://code.google.com/p/cryptsetup/wiki/FrequentlyAskedQuestions

Thanks!

>
> Arno
>
>
>
>> > Bonus: where may I find authoritative information on this (besides the source)?
>> The source is your best friend :-)
>>
>> basically this is the first commit introducing real barrier
>> (later replaced with FUA) support
>> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=647c7db14ef9cacc4ccb3683e206b61f0de6dc2b
>>
>> You can also use blktrace to see how are fs requests propagated
>> to physical media.
>>
>> Milan
>> _______________________________________________
>> dm-crypt mailing list
>> dm-crypt@saout.de
>> http://www.saout.de/mailman/listinfo/dm-crypt
>>
>
> --
> Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email: arno@wagner.name
> GnuPG:  ID: 1E25338F  FP: 0C30 5782 9D93 F785 E79C  0296 797F 6B50 1E25 338F
> ----
> One of the painful things about our time is that those who feel certainty
> are stupid, and those with any imagination and understanding are filled
> with doubt and indecision. -- Bertrand Russell
> _______________________________________________
> dm-crypt mailing list
> dm-crypt@saout.de
> http://www.saout.de/mailman/listinfo/dm-crypt



-- 
Yang Zhang
http://yz.mit.edu/