Bug#860268: .desktop files can hide malware in Nautilus

[Micah Lee]

Package: nautilus
Version: 3.22.3-1

There is a bug in Nautilus that makes it possible to disguise a
malicious script as an innocent document, like a PDF or ODT, that gets
executed when the user opens it.

The upstream nautilus issue [1] has already been resolved, and will be
released in nautilus 3.24. But since this is an important security
issue, I think this patch should be backported so that it's fixed in
older versions of Debian.

See this blog post [2] for more about how this bug allows attackers to
compromise the security-focused Debian-based distro Subgraph.

[1] https://bugzilla.gnome.org/show_bug.cgi?id=777991
[2]
https://micahflee.com/2017/04/breaking-the-security-model-of-subgraph-os/

[intrigeri]

Hi!

Post by Micah Lee
The upstream nautilus issue [1] has already been resolved, and will be
released in nautilus 3.24. But since this is an important security
issue, I think this patch should be backported so that it's fixed in
older versions of Debian.

Thanks for raising this issue in Debian!

Is there any plan upstream to backport this fix to their 3.22.x
branch, and/or to request a CVE?

Did you personally check whether it's straightforward to backport the
fix to 3.22?

Cheers,

[Phil Wyett]

Post by intrigeri
Hi!

Post by Micah Lee
The upstream nautilus issue [1] has already been resolved, and will be
released in nautilus 3.24. But since this is an important security
issue, I think this patch should be backported so that it's fixed in
older versions of Debian.

Thanks for raising this issue in Debian!
Is there any plan upstream to backport this fix to their 3.22.x
branch, and/or to request a CVE?
Did you personally check whether it's straightforward to backport the
fix to 3.22?
Cheers,

Hi,

Seeing this bug. I have backported from the upstream patch (hash issue with
upstream diff) for testing purposes and all looks good. If anyone wishes to
test, a debdiff is attached.

The debdiff is prepared with a 'stretch-pu' in mind.

If any edits are required, please do not hesitate to let me know.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

Twitter: kathenasorg

Instagram: kathenasorg

[Donncha O'Cearbhaill]

Hi,

Thank you Phil for providing a backport patch. What is the next step
needed to get this fix released as a backport? The .desktop security
issue is widely know and can be exploited in the wild [1]. IMO this
fixed should be made available as soon as possible.

Regards,
Donncha

[1] https://github.com/freedomofpress/securedrop/issues/2238

[intrigeri]

Control: tag -1 + security

Post by Donncha O'Cearbhaill
Thank you Phil for providing a backport patch. What is the next step
needed to get this fix released as a backport? The .desktop security
issue is widely know and can be exploited in the wild [1]. IMO this
fixed should be made available as soon as possible.

IMO the next step is to find out the answer to "Is there any plan
upstream to backport this fix to their 3.22.x branch, and/or to
request a CVE?": if this problem is as severe as it sounds, then it
should be tracked as a security issue and fixed cross-distro, rather
than patched in only the distros that are lucky enough to have users
who care about such things.

[Donncha O'Cearbhaill]

Post by intrigeri
Control: tag -1 + security

Post by Donncha O'Cearbhaill
Thank you Phil for providing a backport patch. What is the next step
needed to get this fix released as a backport? The .desktop security
issue is widely know and can be exploited in the wild [1]. IMO this
fixed should be made available as soon as possible.

IMO the next step is to find out the answer to "Is there any plan
upstream to backport this fix to their 3.22.x branch, and/or to
request a CVE?": if this problem is as severe as it sounds, then it
should be tracked as a security issue and fixed cross-distro, rather
than patched in only the distros that are lucky enough to have users
who care about such things.

The upstream developer has indicated that he willing to make a 3.22.x
release if a backport patch is provided. I've sent him a link to Phil
Wyett's debdiff which I hope is acceptable.

I will also file a CVE request for this issue which should help to
coordinate the release of this fix for other distros.

Upstream bug: https://bugzilla.gnome.org/show_bug.cgi?id=777991

[Donncha O'Cearbhaill]

The upstream developer has now indicated that they will not be
backporting the fix to 3.22.x. They have a policy of not backporting
fixes which involve UI changes in stable branches.

Will Debian backport this issue themselves? I have requested a CVE which
I hope will help other distros to coordinate their fixes.

Upstream bug: https://bugzilla.gnome.org/show_bug.cgi?id=777991

Post by intrigeri
Control: tag -1 + security

Post by Donncha O'Cearbhaill
Thank you Phil for providing a backport patch. What is the next step
needed to get this fix released as a backport? The .desktop security
issue is widely know and can be exploited in the wild [1]. IMO this
fixed should be made available as soon as possible.

IMO the next step is to find out the answer to "Is there any plan
upstream to backport this fix to their 3.22.x branch, and/or to
request a CVE?": if this problem is as severe as it sounds, then it
should be tracked as a security issue and fixed cross-distro, rather
than patched in only the distros that are lucky enough to have users
who care about such things.

[Jeremy Bicha]

Post by Donncha O'Cearbhaill
The upstream developer has now indicated that they will not be
backporting the fix to 3.22.x. They have a policy of not backporting
fixes which involve UI changes in stable branches.
Will Debian backport this issue themselves? I have requested a CVE which
I hope will help other distros to coordinate their fixes.

It's not just a UI change but a translatable string change. The new
dialog that users will have to use to mark .desktop's as trusted will
be untranslated.

Therefore, if you want this feature, you will need to use Nautilus >=
3.24 which means you will need to upgrade to buster.

Thanks,
Jeremy Bicha

[Donncha O'Cearbhaill]

Post by Jeremy Bicha
It's not just a UI change but a translatable string change. The new
dialog that users will have to use to mark .desktop's as trusted will
be untranslated.
Therefore, if you want this feature, you will need to use Nautilus >=
3.24 which means you will need to upgrade to buster.

I understand backporting is more difficult when there are user facing UI
and localisation changes. AFAIK the only new translatable string in the
patch is "Trust and _Launch". Would it be possible to include the
translations for that string with this backport patch?

Personally I don't consider this change a *feature*, it is a fix for a
serious security issue affecting Debian stable users (and Tails). The
issue is trivially exploitable against the default configuration.

Video demonstrating the issue:
https://twitter.com/bleidl/status/851969179980845056
More information and an example:
https://github.com/DonnchaC/desktop-file-social-engineering

[Phil Wyett]

Post by Donncha O'Cearbhaill

Post by Jeremy Bicha
It's not just a UI change but a translatable string change. The new
dialog that users will have to use to mark .desktop's as trusted will
be untranslated.
Therefore, if you want this feature, you will need to use Nautilus >=
3.24 which means you will need to upgrade to buster.

I understand backporting is more difficult when there are user facing UI
and localisation changes. AFAIK the only new translatable string in the
patch is "Trust and _Launch". Would it be possible to include the
translations for that string with this backport patch?
Personally I don't consider this change a *feature*, it is a fix for a
serious security issue affecting Debian stable users (and Tails). The
issue is trivially exploitable against the default configuration.
https://twitter.com/bleidl/status/851969179980845056
https://github.com/DonnchaC/desktop-file-social-engineering

Hi,

Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a 'here it is,
all done' patch ready for submission as a stable update.

I am a little busy at the moment, but if I can help here, I will.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

Github: https://github.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

[Donncha O'Cearbhaill]

Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a 'here it is,
all done' patch ready for submission as a stable update.
I am a little busy at the moment, but if I can help here, I will.
Regards
Phil

Hi,

I have cherry-picked the translations for the string "Trust and _Launch"
and created an updated patch and debdiff containing those strings in the
respective .po files.

Unfortunately it looks like the Debian package does not rebuild the
.gmo/.mo files from the .po files during the build. Instead it uses the
pre-built .gmo files which have be include in the upstream release. As a
result the added translation are not included with the built package.

I'm not sure what is the best way to resolve this:

1. Add gettext build dependency and rebuild the .mo files
3. Ask upstream maintainer to make a 3.22 release contain the patch and
translation
3. Create release without translation for that one string

Phil, I have tested your patch on Tail 3.1 (based on Debian Jessie) and
it is functioning as expected.

[Donncha O'Cearbhaill]

Is there anything that I can do to help get this backport patch
deployed? This issue can be exploited in the wild and I think it should
be fixed as soon as possible.

I am still waiting for a response for my CVE request.

[Phil Wyett]

Post by Donncha O'Cearbhaill

Post by Jeremy Bicha
It's not just a UI change but a translatable string change. The new
dialog that users will have to use to mark .desktop's as trusted will
be untranslated.
Therefore, if you want this feature, you will need to use Nautilus >=
3.24 which means you will need to upgrade to buster.

I understand backporting is more difficult when there are user facing UI
and localisation changes. AFAIK the only new translatable string in the
patch is "Trust and _Launch". Would it be possible to include the
translations for that string with this backport patch?
Personally I don't consider this change a *feature*, it is a fix for a
serious security issue affecting Debian stable users (and Tails). The
issue is trivially exploitable against the default configuration.
https://twitter.com/bleidl/status/851969179980845056
https://github.com/DonnchaC/desktop-file-social-engineering

Hi,
Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a 'here it is,
all done' patch ready for submission as a stable update.
I am a little busy at the moment, but if I can help here, I will.
Regards
Phil

Hi,

Has anyone looked at how Red Hat are approaching this issue? RHEL 7.4 is gnome
3.22 and using nautilus 3.22.3 I believe.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

Github: https://github.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

[Donncha O'Cearbhaill]

Post by Phil Wyett

Hi,
Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a 'here it is,
all done' patch ready for submission as a stable update.
I am a little busy at the moment, but if I can help here, I will.
Regards
Phil

Hi,
Has anyone looked at how Red Hat are approaching this issue? RHEL 7.4 is gnome
3.22 and using nautilus 3.22.3 I believe.
Regards
Phil

The corresponding Red Hat bug is at
https://bugzilla.redhat.com/show_bug.cgi?id=1442231. Unfortunately there
has not been any progress with fixing this issue in RHEL or Fedora 25
either.

Thanks for creating the original patch. I'm not experienced with Debian
packing, but I will try to test your patch later today.

[Phil Wyett]

Post by Donncha O'Cearbhaill

Post by Phil Wyett

Hi,
Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a 'here it is,
all done' patch ready for submission as a stable update.
I am a little busy at the moment, but if I can help here, I will.
Regards
Phil

Hi,
Has anyone looked at how Red Hat are approaching this issue? RHEL 7.4 is gnome
3.22 and using nautilus 3.22.3 I believe.
Regards
Phil

The corresponding Red Hat bug is at
https://bugzilla.redhat.com/show_bug.cgi?id=1442231. Unfortunately there
has not been any progress with fixing this issue in RHEL or Fedora 25
either.
Thanks for creating the original patch. I'm not experienced with Debian
packing, but I will try to test your patch later today.

Hi,

Being that this is tagged against Fedora 27 in Red Hats bugzilla. I have cloned
the bug and assigned it to RHEL 7.4.

https://bugzilla.redhat.com/show_bug.cgi?id=1491425

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

Github: https://github.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

[Donncha O'Cearbhaill]

CVE-2017-14604 has been issued for this vulnerability.

[Donncha O'Cearbhaill]

Hi,
Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a 'here it is,
all done' patch ready for submission as a stable update.
I am a little busy at the moment, but if I can help here, I will.

I have created a backport patch targeting Nautilus 3.22.3 which contains
the cherry-picked translations for the new UI string.

It adds a line to the debian/control file to remove the pre-built .mo
translation files which were included in the upstream source release. I
also needed to add gettext as a build dependency. With this patch the
.mo/.gmo files should be rebuilt with the new strings during the Debian
package build.

I have tested the backported Nautlius package with Tails 3.1 which is
based on Debian stable. The English and localised interface is displayed
correctly.

Ideally this backport would be ready for Tails 3.2 which is schedule to
be released early next week.

Please let me know if I need to make any further changes.

Regards,
Donncha

[Phil Wyett]

Post by Donncha O'Cearbhaill

Hi,
Please note that the debdiff I provided was essentially a raw backport for
testing and I thought it may have issues. It was never meant as a
'here it
is,
all done' patch ready for submission as a stable update.
I am a little busy at the moment, but if I can help here, I will.

I have created a backport patch targeting Nautilus 3.22.3 which contains
the cherry-picked translations for the new UI string.
It adds a line to the debian/control file to remove the pre-built .mo
translation files which were included in the upstream source release. I
also needed to add gettext as a build dependency. With this patch the
.mo/.gmo files should be rebuilt with the new strings during the Debian
package build.
I have tested the backported Nautlius package with Tails 3.1 which is
based on Debian stable. The English and localised interface is displayed
correctly.
Ideally this backport would be ready for Tails 3.2 which is schedule to
be released early next week.
Please let me know if I need to make any further changes.
Regards,
Donncha

Hi,

Sorry, been busy, so not had chance to get back to this.

Tested on English, German and French and all Ok.

Attached is updated debdiff, adding credit.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Phil Wyett]

Hi,

Now that the CVE (CVE-2017-14604) has been issued and this would (well, if it
ever does) pass into debian as a security update. I have updated the debdiff
accordingly. See attached.

Link to CVE: http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2017-14604

If any tweaks need to be made. Please let me know via this bug report.

If anyone has issues running with this patch applied. Please be sure to add
information to this bug report.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Jeremy Bicha]

I asked on IRC about this so feel free to send the email, Phil or Donncha:

jbicha | carnil: are you going to sponsor #860268 as a security update?
jmm_ | jbicha: yeah, we can fix that via security.debian.org, please
send a mail to ***@security.debian.org, only a few of us are on IRC


Thanks,
Jeremy Bicha

[Phil Wyett]

Post by Jeremy Bicha
jbicha | carnil: are you going to sponsor #860268 as a security update?
jmm_ | jbicha: yeah, we can fix that via security.debian.org, please
Thanks,
Jeremy Bicha

Hi Security Team,

Please accept the attached 'nautilus' debdiff for stretch-security.

Info:

The debdiff is a backport of the fix from upstream[1] and includes translations
for the UI changes.

[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e9979236d31a
8d3bb0

Related debian bug:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=860268

Related upstream bug:

https://bugzilla.gnome.org/show_bug.cgi?id=777991

Related CVE:

https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2017-14604

Debian security tracker:

https://security-tracker.debian.org/tracker/CVE-2017-14604

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Phil Wyett]

Post by Jeremy Bicha
jbicha | carnil: are you going to sponsor #860268 as a security update?
jmm_ | jbicha: yeah, we can fix that via security.debian.org, please
Thanks,
Jeremy Bicha

Hi Security Team,
Please accept the attached 'nautilus' debdiff for stretch-security.
The debdiff is a backport of the fix from upstream[1] and includes
translations
for the UI changes.
[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e9979236d3
1a
8d3bb0
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=860268
https://bugzilla.gnome.org/show_bug.cgi?id=777991
https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2017-14604
https://security-tracker.debian.org/tracker/CVE-2017-14604
Regards
Phil

--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Phil Wyett]

Post by Jeremy Bicha
jbicha | carnil: are you going to sponsor #860268 as a security update?
jmm_ | jbicha: yeah, we can fix that via security.debian.org, please
Thanks,
Jeremy Bicha

Hi Security Team,
Please accept the attached 'nautilus' debdiff for stretch-security.
The debdiff is a backport of the fix from upstream[1] and includes translations
for the UI changes.
[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e9979236
d3
1a
8d3bb0
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=860268
https://bugzilla.gnome.org/show_bug.cgi?id=777991
https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2017-14604
https://security-tracker.debian.org/tracker/CVE-2017-14604
Regards
Phil

Oops... Massive sleep derived error. debdiff has been forwarded to security team
on another email that did not have a massive recipient list and had them on it.

Apologies for the error.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Yves-Alexis Perez]

Hi Security Team,

Please accept the attached 'nautilus' debdiff for stretch-security.
The debdiff is a backport of the fix from upstream[1] and includes translations
for the UI changes.
[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e9979236d3
1a
8d3bb0

Hi Phil,

the debdiff looks good, but please use +deb9u1 as suffix for the version
number. You may then proceed with the upload to security-master.

Note that since it's the first nautilus security upload to stretch it needs to
be build with -sa.

You can safely upload a source-only upload, but you need to remove the
.buildinfo from the changes file before uploading.

Regards,

[Yves-Alexis Perez]

Post by Yves-Alexis Perez

Hi Security Team,

Please accept the attached 'nautilus' debdiff for stretch-security.
The debdiff is a backport of the fix from upstream[1] and includes translations
for the UI changes.
[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e997
9236d3
1a
8d3bb0

Hi Phil,
the debdiff looks good, but please use +deb9u1 as suffix for the version
number. You may then proceed with the upload to security-master.
Note that since it's the first nautilus security upload to stretch it needs to
be build with -sa.
You can safely upload a source-only upload, but you need to remove the
.buildinfo from the changes file before uploading.

I'll take care of the upload. Do you intend to backport the patches to Jessie?

Regards,

[Phil Wyett]

Post by Yves-Alexis Perez

Post by Yves-Alexis Perez

Hi Security Team,

Please accept the attached 'nautilus' debdiff for stretch-security.
The debdiff is a backport of the fix from upstream[1] and includes translations
for the UI changes.
[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e997
9236d3
1a
8d3bb0

Hi Phil,
the debdiff looks good, but please use +deb9u1 as suffix for the version
number. You may then proceed with the upload to security-master.
Note that since it's the first nautilus security upload to stretch it needs to
be build with -sa.
You can safely upload a source-only upload, but you need to remove the
.buildinfo from the changes file before uploading.

I'll take care of the upload. Do you intend to backport the patches to Jessie?
Regards,

Hi,

I will look at it. But, I just know it will be a nightmare if possible at all. I
shall add info to the bug report probably mod next week.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Phil Wyett]

Post by Yves-Alexis Perez

Post by Yves-Alexis Perez

Hi Security Team,

Please accept the attached 'nautilus' debdiff for stretch-security.
The debdiff is a backport of the fix from upstream[1] and includes translations
for the UI changes.
[1]: https://github.com/GNOME/nautilus/commit/1630f53481f445ada0a455e997
9236d3
1a
8d3bb0

Hi Phil,
the debdiff looks good, but please use +deb9u1 as suffix for the version
number. You may then proceed with the upload to security-master.
Note that since it's the first nautilus security upload to stretch it needs to
be build with -sa.
You can safely upload a source-only upload, but you need to remove the
.buildinfo from the changes file before uploading.

I'll take care of the upload. Do you intend to backport the patches to Jessie?
Regards,

Hi all,

I have looked at both 'jessie' and 'wheezy'. Both are not affected by this
specific issue and have mechanism(s) like stretch (with update) and newer
versions of nautilus that display and require input when confronted with certain
file types.

Screenshot attached showing how 'jessie' and 'wheezy' react to the example
attack desktop file.

If someone else wished to validate this, please feel free.

Regards

Phil
--
*** If this is a mailing list, I am subscribed, no need to CC me.***

Playing the game for the games sake.

Web: https://kathenas.org

GitLab: https://gitlab.com/kathenas

Twitter: kathenasorg

Instagram: kathenasorg

GPG: 1B97 6556 913F 73F3 9C9B 25C4 2961 D9B6 2017 A57A

[Jeremy Bicha]

Post by Phil Wyett
I have looked at both 'jessie' and 'wheezy'. Both are not affected by this
specific issue and have mechanism(s) like stretch (with update) and newer
versions of nautilus that display and require input when confronted with certain
file types.

nautilus 3.22 introduced integrated (almost silent) tarball
decompression support which makes the test case for this vulnerability
a lot simpler.

Thanks,
Jeremy Bicha