GNU bug report logs - #35591
Segfault on flatpak remote-add

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 35591 in the body.
You can then email your comments to 35591 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: "Jonathan Frederickson" <jonathan <at> terracrypt.net>
To: bug-guix <at> gnu.org
Subject: Segfault on flatpak remote-add
Date: Sun, 05 May 2019 16:05:25 -0400

I'm attempting to use Flatpak on my Guix system, and I'm experiencing segfaults when attempting to add a remote repo. Provided below are the output of the command I'm attempting to run, as well as some info about my system.

=====
jfred <at> lambdacrypt ~$ flatpak remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo

Note that the directories

'/var/lib/flatpak/exports/share'
'/home/jfred/.local/share/flatpak/exports/share'

are not in the search path set by the XDG_DATA_DIRS environment variable, so
applications installed by Flatpak may not appear on your desktop until the
session is restarted.

Segmentation fault
=====
Current guix commit:

jfred <at> lambdacrypt ~$ guix describe -f channels
(list (channel
        (name 'guix)
        (url "https://git.savannah.gnu.org/git/guix.git")
        (commit
          "187f9636f176d62bf869b787ad410e3a528aa1a0")))
=====
Current user package manifest:

jfred <at> lambdacrypt ~$ cat guix.scm
(specifications->manifest
 '("glibc-utf8-locales"
   ;;# Dotfiles management
   "stow"
   "rofi"
   "gnupg"
   "emacs-password-store"
   "pcsc-lite"
   ;;# Hacking
   "emacs"
   "git"
   "guile <at> 2.2"
   "font-adobe-source-code-pro"
   "emacs-guix"
   ;;# Devops
   "docker"
   "docker-cli"
   "docker-compose"
   ;; System Administration
   "htop"
   "curl"
   ;;# Desktop
   "gnome-tweak-tool"
   "icecat"
   "luakit"
   "flatpak"
   "evolution"
   ;; Gaming
   "tintin++"
   "powwow"))

Message #8 received at 35591 <at> debbugs.gnu.org (full text, mbox):

From: Mark H Weaver <mhw <at> netris.org>
To: "Jonathan Frederickson" <jonathan <at> terracrypt.net>
Cc: 35591 <at> debbugs.gnu.org
Subject: Re: bug#35591: Segfault on flatpak remote-add
Date: Sun, 05 May 2019 18:36:06 -0400

Hi Jonathan,

"Jonathan Frederickson" <jonathan <at> terracrypt.net> writes:
> I'm attempting to use Flatpak on my Guix system, and I'm experiencing
> segfaults when attempting to add a remote repo. Provided below are the
> output of the command I'm attempting to run, as well as some info
> about my system.

My knowledge of Flatpak is weak, but in general, foreign pre-compiled
binaries will not work on Guix unless they are statically linked.
Traditional dynamically-linked executables expect to find a dynamic
linker and system shared libraries in /lib or /lib64, which do not exist
on a Guix system.  It is also fairly common for foreign binaries to
depend on standard files and programs in /usr, /bin, and /sbin, none of
which are present on a Guix system except for /bin/sh.

Having said that, if you must run foreign pre-compiled binaries on Guix,
there are various hacks to do it.  One approach is to install symlinks
in /lib*, /usr, /bin, /sbin, etc, pointing to the needed components in
/run/current-system.  Another is to modify the pre-compiled binaries to
look for components in the right place using 'patchelf'.  We use the
latter approach to bootstrap GHC (Glasgow Haskell Compiler), which can
only be compiled by itself.

I'm sorry I don't have a better answer for you, but Guix was not
designed to run foreign pre-compiled binaries.  If that's important to
you, it might be that Guix is not a good fit for you.

However, if your goal is to run Guix-compiled programs in a sandbox,
that's something we are certainly very interested in, and we could use
help.  We have "guix container" as a building block, and I know that
some Guix users have experimented with running e.g. GNU IceCat within a
container.  I don't know the current status of those efforts.  Hopefully
someone more knowledgable in this area will chime in.

     Regards,
       Mark

Message #11 received at 35591 <at> debbugs.gnu.org (full text, mbox):

From: Jonathan Frederickson <jonathan <at> terracrypt.net>
To: Mark H Weaver <mhw <at> netris.org>
Cc: 35591 <at> debbugs.gnu.org
Subject: Re: bug#35591: Segfault on flatpak remote-add
Date: Sun, 05 May 2019 18:47:57 -0400

[Message part 1 (text/plain, inline)]

On Sunday, May 5, 2019 6:36:06 PM EDT Mark H Weaver wrote:
> My knowledge of Flatpak is weak, but in general, foreign pre-compiled
> binaries will not work on Guix unless they are statically linked.
> Traditional dynamically-linked executables expect to find a dynamic
> linker and system shared libraries in /lib or /lib64, which do not exist
> on a Guix system.  It is also fairly common for foreign binaries to
> depend on standard files and programs in /usr, /bin, and /sbin, none of
> which are present on a Guix system except for /bin/sh.

I believe Flatpak applications are statically linked. (Or rather, individual 
applications may be dynamically linked, but they are eventually combined with 
a full OS image similarly to how Docker works.)

> I'm sorry I don't have a better answer for you, but Guix was not
> designed to run foreign pre-compiled binaries.  If that's important to
> you, it might be that Guix is not a good fit for you.

It's not the most important thing to me; it may end up being one more reason 
for me to start attempting to package things for Guix! But it would still be 
nice to be able to run the same tooling that e.g. some Gnome developers are 
using to publish their apps across distros. (And the fact that Flatpak is 
packaged for Guix seems to imply that it should be possible to use it here.)

> However, if your goal is to run Guix-compiled programs in a sandbox,
> that's something we are certainly very interested in, and we could use
> help.  We have "guix container" as a building block, and I know that
> some Guix users have experimented with running e.g. GNU IceCat within a
> container.  I don't know the current status of those efforts.  Hopefully
> someone more knowledgable in this area will chime in.

While this isn't my immediate goal, this is also something that I'm interested 
in, if someone else knows more about this!

[signature.asc (application/pgp-signature, inline)]

Message #14 received at 35591 <at> debbugs.gnu.org (full text, mbox):

From: Ben Sturmfels <ben <at> sturm.com.au>
To: "Jonathan Frederickson" <jonathan <at> terracrypt.net>
Cc: 35591 <at> debbugs.gnu.org
Subject: Re: bug#35591: Segfault on flatpak remote-add
Date: Thu, 31 Oct 2019 21:50:18 +1100

[Message part 1 (text/plain, inline)]

On Sun, 05 May 2019, Jonathan Frederickson wrote:

> I'm attempting to use Flatpak on my Guix system, and I'm experiencing
> segfaults when attempting to add a remote repo. Provided below are the
> output of the command I'm attempting to run, as well as some info
> about my system.

I learnt on #guix IRC that flatpak works when run with the --user flag:

$ flatpak --user remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo
$ flatpak install flathub org.gnome.Builder
$ flatpak run org.gnome.Builder

The segfaulting behaviour is still a bug in my opinion, because it's
extremely user un-friendly, and the requirement for --user is not
obvious.

Regards,
Ben

[signature.asc (application/pgp-signature, inline)]

Message #17 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: bug-guix <at> gnu.org
Cc: 35591 <at> debbugs.gnu.org, Jonathan Frederickson <jonathan <at> terracrypt.net>
Subject: Re: bug#35591: Segfault on flatpak remote-add
Date: Mon, 11 May 2020 22:23:39 +0200

Ben Sturmfels <ben <at> sturm.com.au> writes:

> On Sun, 05 May 2019, Jonathan Frederickson wrote:
>
>> I'm attempting to use Flatpak on my Guix system, and I'm experiencing
>> segfaults when attempting to add a remote repo. Provided below are the
>> output of the command I'm attempting to run, as well as some info
>> about my system.
>
> I learnt on #guix IRC that flatpak works when run with the --user flag:
>
> $ flatpak --user remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo
> $ flatpak install flathub org.gnome.Builder
> $ flatpak run org.gnome.Builder
>
> The segfaulting behaviour is still a bug in my opinion, because it's
> extremely user un-friendly, and the requirement for --user is not
> obvious.

Yes, segfaults are always bugs, unless you forced the program to
load garbage.

I traced the execution of flatpak and saw that it segfaults right after
failing to created /var/lib/flatpak.  Sure enough, after creating it
manually the program no longer segfaults.

-- 
Ricardo

Message #23 received at 35591 <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: 35591 <at> debbugs.gnu.org
Cc: Jonathan Frederickson <jonathan <at> terracrypt.net>
Subject: Re: bug#35591: Segfault on flatpak remote-add
Date: Mon, 11 May 2020 23:38:05 +0200

Ricardo Wurmus <rekado <at> elephly.net> writes:

> Ben Sturmfels <ben <at> sturm.com.au> writes:
>
>> On Sun, 05 May 2019, Jonathan Frederickson wrote:
>>
>>> I'm attempting to use Flatpak on my Guix system, and I'm experiencing
>>> segfaults when attempting to add a remote repo. Provided below are the
>>> output of the command I'm attempting to run, as well as some info
>>> about my system.
>>
>> I learnt on #guix IRC that flatpak works when run with the --user flag:
>>
>> $ flatpak --user remote-add --if-not-exists flathub https://flathub.org/repo/flathub.flatpakrepo
>> $ flatpak install flathub org.gnome.Builder
>> $ flatpak run org.gnome.Builder
>>
>> The segfaulting behaviour is still a bug in my opinion, because it's
>> extremely user un-friendly, and the requirement for --user is not
>> obvious.
>
> Yes, segfaults are always bugs, unless you forced the program to
> load garbage.
>
> I traced the execution of flatpak and saw that it segfaults right after
> failing to created /var/lib/flatpak.  Sure enough, after creating it
> manually the program no longer segfaults.

I reported this here:

    https://github.com/flatpak/flatpak/issues/3612

I’d say we close this issue here, because the segfault isn’t our fault.

Thoughts?

--
Ricardo

Message #30 received at 35591-done <at> debbugs.gnu.org (full text, mbox):

From: Ricardo Wurmus <rekado <at> elephly.net>
To: 35591-done <at> debbugs.gnu.org
Cc: Jonathan Frederickson <jonathan <at> terracrypt.net>
Subject: Re: bug#35591: Segfault on flatpak remote-add
Date: Thu, 14 May 2020 07:26:35 +0200

Ricardo Wurmus <rekado <at> elephly.net> writes:

> I reported this here:
>
>     https://github.com/flatpak/flatpak/issues/3612
>
> I’d say we close this issue here, because the segfault isn’t our fault.

Upstream has acknowledged and addressed the bug.

Closing.

-- 
Ricardo

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.