Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.debian.user > #196072 > unrolled thread

Possible for full-disk encryption to encrypt /boot as well?

Started byRobert Dodier <robert.dodier@gmail.com>
First post2018-05-25 21:00 +0200
Last post2018-05-27 19:40 +0200
Articles 14 — 8 participants

Back to article view | Back to linux.debian.user


Contents

  Possible for full-disk encryption to encrypt /boot as well? Robert Dodier <robert.dodier@gmail.com> - 2018-05-25 21:00 +0200
    Re: Possible for full-disk encryption to encrypt /boot as well? Pascal Hambourg <pascal@plouf.fr.eu.org> - 2018-05-25 21:50 +0200
      Re: Possible for full-disk encryption to encrypt /boot as well? Robert Dodier <robert.dodier@gmail.com> - 2018-05-26 02:40 +0200
        Re: Possible for full-disk encryption to encrypt /boot as well? Abdullah Ramazanoğlu <ar018@yahoo.com> - 2018-05-26 03:20 +0200
        Re: Possible for full-disk encryption to encrypt /boot as well? Pascal Hambourg <pascal@plouf.fr.eu.org> - 2018-05-26 10:20 +0200
          Re: Possible for full-disk encryption to encrypt /boot as well? Robert Dodier <robert.dodier@gmail.com> - 2018-05-26 18:00 +0200
            Re: Possible for full-disk encryption to encrypt /boot as well? Curt <curty@free.fr> - 2018-05-26 19:00 +0200
              Re: Possible for full-disk encryption to encrypt /boot as well? David Wright <deblis@lionunicorn.co.uk> - 2018-05-26 21:10 +0200
    Re: Possible for full-disk encryption to encrypt /boot as well? Celejar <celejar@gmail.com> - 2018-05-25 22:00 +0200
    Re: Possible for full-disk encryption to encrypt /boot as well? David Christensen <dpchrist@holgerdanske.com> - 2018-05-27 02:30 +0200
      Re: Possible for full-disk encryption to encrypt /boot as well? Paul Johnson <baloo@ursamundi.org> - 2018-05-27 06:20 +0200
        Re: Possible for full-disk encryption to encrypt /boot as well? David Christensen <dpchrist@holgerdanske.com> - 2018-05-27 07:50 +0200
          Re: Possible for full-disk encryption to encrypt /boot as well? Paul Johnson <baloo@ursamundi.org> - 2018-05-27 08:50 +0200
            Re: Possible for full-disk encryption to encrypt /boot as well? David Christensen <dpchrist@holgerdanske.com> - 2018-05-27 19:40 +0200

#196072 — Possible for full-disk encryption to encrypt /boot as well?

FromRobert Dodier <robert.dodier@gmail.com>
Date2018-05-25 21:00 +0200
SubjectPossible for full-disk encryption to encrypt /boot as well?
Message-ID<vTq6t-1wc-1@gated-at.bofh.it>
Hello,

I'm working with Debian 9. I gather that there is a full-disk
encryption option for the standard Debian installer, which, as I
understand it, does not include encrypting /boot. (The system I'm
working on wasn't encrypted when it was installed, so the system would
have to be reinstalled, that's OK.)

The only description of encrypting /boot that I was able to find is:
https://gist.github.com/ppmathis/ccfbfce86484dc61834c1f17568d7b80
I wonder if there is any simpler approach.

Is it possible that, perhaps, other Linux distributions have an option
for encrypting /boot? I wasn't able to find any information about
that.

It's OK if the answer to these questions is no, I'm just trying to
sort out the feasibility of encrypting /boot.

Thanks in advance for any info, I appreciate it very much.

Robert Dodier

[toc] | [next] | [standalone]


#196074

FromPascal Hambourg <pascal@plouf.fr.eu.org>
Date2018-05-25 21:50 +0200
Message-ID<vTqSS-21j-7@gated-at.bofh.it>
In reply to#196072
Le 25/05/2018 à 20:55, Robert Dodier a écrit :
> 
> I'm working with Debian 9. I gather that there is a full-disk
> encryption option for the standard Debian installer, which, as I
> understand it, does not include encrypting /boot. (The system I'm
> working on wasn't encrypted when it was installed, so the system would
> have to be reinstalled, that's OK.)
> 
> The only description of encrypting /boot that I was able to find is:
> https://gist.github.com/ppmathis/ccfbfce86484dc61834c1f17568d7b80
> I wonder if there is any simpler approach.
> 
> Is it possible that, perhaps, other Linux distributions have an option
> for encrypting /boot? I wasn't able to find any information about
> that.

I have managed to use the standard Debian installer to install with 
encrypted /boot (either including or excluding /boot/grub) but it was 
far from straightforward. I had to perform some steps with the embedded 
shell. Also, the installer insists that /boot should not be encrypted. 
Jessie's installer was quite easy to trick (just put /boot on LVM on 
encrypted volume). But the trick did not work with Stretch's installer, 
so I had to create a dummy /boot.

> It's OK if the answer to these questions is no, I'm just trying to
> sort out the feasibility of encrypting /boot.

Why do you want an encrypted /boot ? It does not usually contain any 
sensitive information. Encrypted /boot is not tamper-proof unless extra 
steps are taken to protect the first stage boot such as booting from 
write-protected, authenticated or removable media.

[toc] | [prev] | [next] | [standalone]


#196078

FromRobert Dodier <robert.dodier@gmail.com>
Date2018-05-26 02:40 +0200
Message-ID<vTvpv-4HK-3@gated-at.bofh.it>
In reply to#196074
On Fri, May 25, 2018 at 12:44 PM, Pascal Hambourg
<pascal@plouf.fr.eu.org> wrote:

> Why do you want an encrypted /boot ? It does not usually contain any
> sensitive information. Encrypted /boot is not tamper-proof unless extra
> steps are taken to protect the first stage boot such as booting from
> write-protected, authenticated or removable media.

Thanks for your reply. I am working for an organization which requires
computers to be full disk encrypted. They support Windows, but if I
want to run Linux, I'm on my own. So to be precise I need something
which is strictly comparable to whatever is provided by Symantec full
disk encryption for Windows. If I can achieve that, I'll be in
business.

It may be true that encrypted /boot is not really relevant, but I
don't want to try to argue with tech support staff that a system with
unencrypted /boot is close enough. I feel like it needs to be all or
nothing in order for me to press this issue with them.

Thank you very much for your comments, and all the best.
Robert Dodier

[toc] | [prev] | [next] | [standalone]


#196082

FromAbdullah Ramazanoğlu <ar018@yahoo.com>
Date2018-05-26 03:20 +0200
Message-ID<vTw2e-59f-3@gated-at.bofh.it>
In reply to#196078
On Fri, 25 May 2018 17:33:54 -0700 Robert Dodier said:

> It may be true that encrypted /boot is not really relevant, but I
> don't want to try to argue with tech support staff that a system with
> unencrypted /boot is close enough. I feel like it needs to be all or
> nothing in order for me to press this issue with them.

Or you can offer them a setup more than full-disk encryption: Put the
(unencrypted) /boot on a USB stick, and the rest of the system on fully
encrypted disk.

* With full disk encryption (Linux or Windows) all one needs is the pass-phrase.
* With /boot on USB stick, one needs both the key (the stick) and the
  pass-phrase.

At any rate, it is technically "full disk encryption" - there's nothing
unencrypted on hard drive, is there? :)

Regards
-- 
Abdullah Ramazanoğlu

[toc] | [prev] | [next] | [standalone]


#196086

FromPascal Hambourg <pascal@plouf.fr.eu.org>
Date2018-05-26 10:20 +0200
Message-ID<vTCAF-OH-5@gated-at.bofh.it>
In reply to#196078
Le 26/05/2018 à 02:33, Robert Dodier a écrit :
> On Fri, May 25, 2018 at 12:44 PM, Pascal Hambourg
> <pascal@plouf.fr.eu.org> wrote:
> 
>> Why do you want an encrypted /boot ? It does not usually contain any
>> sensitive information. Encrypted /boot is not tamper-proof unless extra
>> steps are taken to protect the first stage boot such as booting from
>> write-protected, authenticated or removable media.
> 
> Thanks for your reply. I am working for an organization which requires
> computers to be full disk encrypted. They support Windows, but if I
> want to run Linux, I'm on my own. So to be precise I need something
> which is strictly comparable to whatever is provided by Symantec full
> disk encryption for Windows. If I can achieve that, I'll be in
> business.

I don't know how Symantec's "full" disk encryption works, but AFAIK a 
boot disk cannot be fully encrypted, unless the platform firmware can 
handle the encryption format. IIUC Coreboot or Libreboot may do it. 
Otherwise the bootstrap code must not be encrypted so that the platform 
firmware can load and run it. Even if /boot is encrypted, the first 
stage of the boot loader (GRUB boot image and core image if using 
BIOS/legacy boot, the EFI partition if using EFI boot) cannot not be 
encrypted.

[toc] | [prev] | [next] | [standalone]


#196094

FromRobert Dodier <robert.dodier@gmail.com>
Date2018-05-26 18:00 +0200
Message-ID<vTJLP-5b0-1@gated-at.bofh.it>
In reply to#196086
On Sat, May 26, 2018 at 1:16 AM, Pascal Hambourg <pascal@plouf.fr.eu.org> wrote:

> I don't know how Symantec's "full" disk encryption works, but AFAIK a boot
> disk cannot be fully encrypted,

Yes, this is an important question -- what, exactly, is provided by
Symantec here, so that I can look for something to do the same for
Linux. But not surprisingly I haven't been able to find a careful
description -- so far all I have found is some marketing material. I
will keep looking.

best,
Robert Dodier

[toc] | [prev] | [next] | [standalone]


#196096

FromCurt <curty@free.fr>
Date2018-05-26 19:00 +0200
Message-ID<vTKHU-5Le-5@gated-at.bofh.it>
In reply to#196094
On 2018-05-26, Robert Dodier <robert.dodier@gmail.com> wrote:
> On Sat, May 26, 2018 at 1:16 AM, Pascal Hambourg <pascal@plouf.fr.eu.org> wrote:
>
>> I don't know how Symantec's "full" disk encryption works, but AFAIK a boot
>> disk cannot be fully encrypted,
>
> Yes, this is an important question -- what, exactly, is provided by
> Symantec here, so that I can look for something to do the same for
> Linux. But not surprisingly I haven't been able to find a careful
> description -- so far all I have found is some marketing material. I
> will keep looking.

They seem to be saying the boot loader is decrypted prior to the point
at which it begins execution (a "pre-boot environment" is installed that
prompts the user for pass phrase, etc.)

 https://www.symantec.com/content/en/us/enterprise/white_papers/b-pgp_how_wholedisk_encryption_works_WP_21158817.en-us.pdf
> best,
> Robert Dodier
>
>


-- 

[toc] | [prev] | [next] | [standalone]


#196105

FromDavid Wright <deblis@lionunicorn.co.uk>
Date2018-05-26 21:10 +0200
Message-ID<vTMJH-7ej-7@gated-at.bofh.it>
In reply to#196096
On Sat 26 May 2018 at 16:51:56 (+0000), Curt wrote:
> On 2018-05-26, Robert Dodier <robert.dodier@gmail.com> wrote:
> > On Sat, May 26, 2018 at 1:16 AM, Pascal Hambourg <pascal@plouf.fr.eu.org> wrote:
> >
> >> I don't know how Symantec's "full" disk encryption works, but AFAIK a boot
> >> disk cannot be fully encrypted,
> >
> > Yes, this is an important question -- what, exactly, is provided by
> > Symantec here, so that I can look for something to do the same for
> > Linux. But not surprisingly I haven't been able to find a careful
> > description -- so far all I have found is some marketing material. I
> > will keep looking.
> 
> They seem to be saying the boot loader is decrypted prior to the point
> at which it begins execution (a "pre-boot environment" is installed that
> prompts the user for pass phrase, etc.)
> 
>  https://www.symantec.com/content/en/us/enterprise/white_papers/b-pgp_how_wholedisk_encryption_works_WP_21158817.en-us.pdf

which says:

  "A boot sequence executes during the startup process of Microsoft®
  Windows, Apple Mac OS X, or Linux® operating systems. The boot
  system is the initial set of operations that the computer performs
  when it is switched on. A boot loader (or a bootstrap loader) is a
  short computer program that loads the main operating system for the
  computer. The boot loader first looks at a boot record or partition
  table, which is the logical area “zero” (or starting point) of the
  disk drive.

  Whole disk encryption modifies the zero point area of the drive.
  A computer protected with Whole Disk Encryption presents a modified
  “pre-boot” environment (Figure 1) to the user."

To me, that implies that what they call the "boot[strap] loader" is
the unencrypted firmware, and that the "logical area “zero” (or
starting point)" is the MBR. It says that the MBR has been "modified",
but I don't see where it says that the boot loader has been, nor how
any of the early code can have been *encrypted* until enough has been
executed to read and act upon a passphrase.

Cheers,
David.

[toc] | [prev] | [next] | [standalone]


#196075

FromCelejar <celejar@gmail.com>
Date2018-05-25 22:00 +0200
Message-ID<vTr2y-24A-13@gated-at.bofh.it>
In reply to#196072
On Fri, 25 May 2018 11:55:34 -0700
Robert Dodier <robert.dodier@gmail.com> wrote:

> Hello,
> 
> I'm working with Debian 9. I gather that there is a full-disk
> encryption option for the standard Debian installer, which, as I
> understand it, does not include encrypting /boot. (The system I'm
> working on wasn't encrypted when it was installed, so the system would
> have to be reinstalled, that's OK.)
> 
> The only description of encrypting /boot that I was able to find is:
> https://gist.github.com/ppmathis/ccfbfce86484dc61834c1f17568d7b80
> I wonder if there is any simpler approach.
> 
> Is it possible that, perhaps, other Linux distributions have an option
> for encrypting /boot? I wasn't able to find any information about
> that.

https://wiki.archlinux.org/index.php/Dm-crypt/Encrypting_an_entire_system#Encrypted_boot_partition_.28GRUB.29

Celejar

[toc] | [prev] | [next] | [standalone]


#196116

FromDavid Christensen <dpchrist@holgerdanske.com>
Date2018-05-27 02:30 +0200
Message-ID<vTRJn-1KU-1@gated-at.bofh.it>
In reply to#196072
On 05/25/18 11:55, Robert Dodier wrote:
> I'm working with Debian 9. I gather that there is a full-disk
> encryption option for the standard Debian installer, which, as I
> understand it, does not include encrypting /boot. ...

On 05/25/18 17:33, Robert Dodier wrote:
 > I am working for an organization which requires
 > computers to be full disk encrypted.

Have you considered a self-encrypting drive (self-encrypting drive)?

https://en.wikipedia.org/wiki/Hardware-based_full_disk_encryption


This approach protects everything on the drive, including /boot.


Understand that your computer must have compatible firmware.


David

[toc] | [prev] | [next] | [standalone]


#196120

FromPaul Johnson <baloo@ursamundi.org>
Date2018-05-27 06:20 +0200
Message-ID<vTVjX-43X-1@gated-at.bofh.it>
In reply to#196116

[Multipart message — attachments visible in raw view] — view raw

On Sat, May 26, 2018 at 7:21 PM, David Christensen <
dpchrist@holgerdanske.com> wrote:

> On 05/25/18 11:55, Robert Dodier wrote:
>
>> I'm working with Debian 9. I gather that there is a full-disk
>> encryption option for the standard Debian installer, which, as I
>> understand it, does not include encrypting /boot. ...
>>
>
> On 05/25/18 17:33, Robert Dodier wrote:
> > I am working for an organization which requires
> > computers to be full disk encrypted.
>
> Have you considered a self-encrypting drive (self-encrypting drive)?
>
> https://en.wikipedia.org/wiki/Hardware-based_full_disk_encryption
>
>
> This approach protects everything on the drive, including /boot.
>
>
> Understand that your computer must have compatible firmware.


I'm 99.99% sure (like Dove soap sure) that Symantec full disk encryption
doesn't work this way because I'm just as sure that none of the Dell models
I've ever worked with have this hardware capability.

[toc] | [prev] | [next] | [standalone]


#196121

FromDavid Christensen <dpchrist@holgerdanske.com>
Date2018-05-27 07:50 +0200
Message-ID<vTWJ3-4YR-1@gated-at.bofh.it>
In reply to#196120
On 05/26/18 21:16, Paul Johnson wrote:
> On Sat, May 26, 2018 at 7:21 PM, David Christensen
>> Have you considered a self-encrypting drive ...
> 
> I'm 99.99% sure (like Dove soap sure) that Symantec full disk
> encryption doesn't work this way because I'm just as sure that none
> of the Dell models I've ever worked with have this hardware
> capability.

I have tested Intel 520 Series SSD's with self-encryption in two Dell
laptop models:

- They do not work in Dell Inspiron E1505's (~2007, Core Duo).

- They do work in Dell Latitude E6520's (2011~2012, Sandy Bridge).


David

[toc] | [prev] | [next] | [standalone]


#196122

FromPaul Johnson <baloo@ursamundi.org>
Date2018-05-27 08:50 +0200
Message-ID<vTXF8-5z6-1@gated-at.bofh.it>
In reply to#196121

[Multipart message — attachments visible in raw view] — view raw

On Sun, May 27, 2018 at 12:43 AM, David Christensen <
dpchrist@holgerdanske.com> wrote:

> On 05/26/18 21:16, Paul Johnson wrote:
>
>> On Sat, May 26, 2018 at 7:21 PM, David Christensen
>>
>>> Have you considered a self-encrypting drive ...
>>>
>>
>> I'm 99.99% sure (like Dove soap sure) that Symantec full disk
>> encryption doesn't work this way because I'm just as sure that none
>> of the Dell models I've ever worked with have this hardware
>> capability.
>>
>
> I have tested Intel 520 Series SSD's with self-encryption in two Dell
> laptop models:
>
> - They do not work in Dell Inspiron E1505's (~2007, Core Duo).
>
> - They do work in Dell Latitude E6520's (2011~2012, Sandy Bridge).


 Right, but is the basic system unencrypted or not?  I very, very, very
highly suspect not, since it seems fairly obviously to me (given that all
the Dell hardware stuff has finished and it's handed off to whatever is on
the boot device) it's handing off to encrypted software off the hard drive
to bootstrap the rest by the time anything Symantec gets involved with.

[toc] | [prev] | [next] | [standalone]


#196144

FromDavid Christensen <dpchrist@holgerdanske.com>
Date2018-05-27 19:40 +0200
Message-ID<vU7Oa-3vV-11@gated-at.bofh.it>
In reply to#196122
On 05/26/18 23:40, Paul Johnson wrote:
> On Sun, May 27, 2018 at 12:43 AM, David Christensen <
> dpchrist@holgerdanske.com> wrote:
> 
>> On 05/26/18 21:16, Paul Johnson wrote:
>>
>>> On Sat, May 26, 2018 at 7:21 PM, David Christensen
>>>
>>>> Have you considered a self-encrypting drive ...
>>>>
>>>
>>> I'm 99.99% sure (like Dove soap sure) that Symantec full disk
>>> encryption doesn't work this way because I'm just as sure that none
>>> of the Dell models I've ever worked with have this hardware
>>> capability.
>>>
>>
>> I have tested Intel 520 Series SSD's with self-encryption in two Dell
>> laptop models:
>>
>> - They do not work in Dell Inspiron E1505's (~2007, Core Duo).
>>
>> - They do work in Dell Latitude E6520's (2011~2012, Sandy Bridge).
> 
> 
>   Right, but is the basic system unencrypted or not?  I very, very, very
> highly suspect not, since it seems fairly obviously to me (given that all
> the Dell hardware stuff has finished and it's handed off to whatever is on
> the boot device) it's handing off to encrypted software off the hard drive
> to bootstrap the rest by the time anything Symantec gets involved with.
> 

AFAIK self-encrypting drives (SED) have hardware encryption/ decryption 
engines (ASIC's) built into the on-board microcontroller.  When the 
system boots, there is a protocol for the motherboard firmware to 
display a prompt on the console for the SED passphrase, which is then 
passed to the SED.  If the passphrase is correct, the SED then encrypts/ 
decrypts data writes/ reads transparently (e.g. the SED acts like an 
unencrypted drive).  If the passphrase is wrong, the SED will refuse to 
write/ read data, because it cannot encrypt/ decrypt without the 
passphrase.  If someone pulls the platters or memory chips out of the 
SED, puts them into a test jig, and reads the magnetic bubbles/ 
transistors, they will obtain ciphertext.  They must know the encryption 
algorithm, passphrase, and other pieces of information (nonces, etc.) to 
decrypt the ciphertext and obtain the plaintext.


Once the SED has been unlocked, the administrator can build additional 
layers of hardware and/or software encryption on top.  FreeOTFE and 
dm-crypt/LUKS are examples of software encryption.  I believe BitLocker 
is also software encryption, but allows keys to be stored in a Trusted 
Platform Module (TPM).  I am not familiar with Symantec encryption products.


David

[toc] | [prev] | [standalone]


Back to top | Article view | linux.debian.user


csiph-web