GNU bug report logs - #32478
pcscd service activation causes boot failure

Previous Next

Package: guix;

Reported by: Chris Marusich <cmmarusich <at> gmail.com>

Date: Mon, 20 Aug 2018 07:42:02 UTC

Severity: normal

Done: Chris Marusich <cmmarusich <at> gmail.com>

Bug is archived. No further changes may be made.

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 32478 in the body.
You can then email your comments to 32478 AT debbugs.gnu.org in the normal way.

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

View this report as an mbox folder, status mbox, maintainer mbox


Report forwarded to bug-guix <at> gnu.org:
bug#32478; Package guix. (Mon, 20 Aug 2018 07:42:02 GMT) Full text and rfc822 format available.

Acknowledgement sent to Chris Marusich <cmmarusich <at> gmail.com>:
New bug report received and forwarded. Copy sent to bug-guix <at> gnu.org. (Mon, 20 Aug 2018 07:42:02 GMT) Full text and rfc822 format available.

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

From: Chris Marusich <cmmarusich <at> gmail.com>
To: bug-guix <at> gnu.org
Cc: Arun Isaac <arunisaac <at> systemreboot.net>
Subject: pcscd service activation causes boot failure
Date: Mon, 20 Aug 2018 00:26:38 -0700
[Message part 1 (text/plain, inline)]
Hi,

Commit de30205ba0f63eb987097a9f47b6e4fd38cd9044 added a pcscd service
(hooray!).  However, its activation procedure doesn't always work.  The
system test passes, and the activation procedure works the first time
you boot, but if you reboot, it will fail because the symlink it creates
already exists.  The attached patch fixes the problem.

To reproduce the bug, use the attached operating system configuration
file (bare-bones.scm), and run:

  guix system vm-image bare-bones.scm

This takes a long time to finish (up to a few hours, depending on your
system).  Once it's done, copy the image out of the store:

  cp $the_store_path /tmp/qemu-image

Then run it:

  sudo qemu-system-x86_64 -smp cpus=1 -net user -net nic,model=virtio -enable-kvm  -m 2048 /tmp/qemu-image 

Observe how the VM boots successfully.  Now log in as root (no password)
and reboot.  Observe that the boot process fails now because the symlink
created by the pcscd activation service already exists.

The attached patch fixes the issue by using an idempotent procedure to
create the symlink.  If there are no objections, I'll push the patch to
master about 24 hours from now.  There is a fairly long comment in my
patch because although I wanted to re-use the switch-symlinks procedure
from (guix utils), I couldn't figure out how to do it.  Ideas regarding
this are welcome!  To be specific, I tried to use (guix utils), but then
I got the following error message at boot time (see my comment for more
information):

--8<---------------cut here---------------start------------->8---
loading '/gnu/store/f4ng1dlpm7q74vssbb049vpf2gvw3n3r-system/boot'...
[    2.175072] random: fast init done
making '/gnu/store/f4ng1dlpm7q74vssbb049vpf2gvw3n3r-system' the current system..
.
setting up setuid programs in '/run/setuid-programs'...
populating /etc from /gnu/store/n6cwz1hlmjylva2xrv61njl68g6c8k5l-etc...
usermod: no changes
usermod: no changes
usermod: no changes
ERROR: In procedure dynamic-func:
In procedure dynamic-pointer: Symbol not found: strverscmp

Entering a new prompt.  Type `,bt' for a backtrace or `,q' to continue.
GNU Guile 2.2.3
Copyright (C) 1995-2017 Free Software Foundation, Inc.

Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.

Enter `,help' for help.
scheme@(guix utils)> ,bt
In gnu/build/linux-boot.scm:
   530:13 22 (_)
In unknown file:
          21 (primitive-load "/gnu/store/f4ng1dlpm7q74vssbb049vpf2gv·")
In ice-9/eval.scm:
    619:8 20 (_ #f)
In unknown file:
          19 (primitive-load "/gnu/store/b6yg0pkp835a3zky9sj53yyjxac·")
In ice-9/boot-9.scm:
   260:13 18 (for-each #<procedure primitive-load (_)> _)
In unknown file:
          17 (primitive-load "/gnu/store/v42zaw2hjwxy5wnd0dwa6j245nr·")
In ice-9/eval.scm:
   721:20 16 (primitive-eval (begin (use-modules (guix build #) #) ·))
In ice-9/psyntax.scm:
  1235:36 15 (expand-top-sequence ((begin (use-modules (# # ·) ·) ·)) ·)
  1182:24 14 (parse _ (("placeholder" placeholder)) ((top) #(# # ·)) ·)
  1182:24 13 (parse _ (("placeholder" placeholder)) ((top) #(# # ·)) ·)
   285:10 12 (parse _ (("placeholder" placeholder)) (()) _ c&e (eval) ·)
In ice-9/boot-9.scm:
  3365:20 11 (process-use-modules _)
   222:29 10 (map1 (((guix build utils)) ((guix utils))))
   222:17  9 (map1 (((guix utils))))
  3366:31  8 (_ ((guix utils)))
  2788:17  7 (resolve-interface (guix utils) #:select _ #:hide _ # _ ·)
  2714:10  6 (_ (guix utils) _ _ #:ensure _)
  2982:16  5 (try-module-autoload _ _)
   2312:4  4 (save-module-excursion #<procedure 2de04e0 at ice-9/boo·>)
  3002:22  3 (_)
In unknown file:
           2 (primitive-load-path "guix/utils" #<procedure 2abb480 a·>)
In guix/utils.scm:
   485:24  1 (_)
In unknown file:
           0 (dynamic-func "strverscmp" #<dynamic-object #f>)
scheme@(guix utils)> [   53.048203] random: crng init done
--8<---------------cut here---------------end--------------->8---

To be clear, the above stack trace is NOT related to the bug I am
reporting.  It's just a different problem that made it difficult to
re-use switch-symlinks from (guix utils), which is why in my fix I
decided to just copy the switch-symlinks definition verbatim.

-- 
Chris
[bare-bones.scm (application/octet-stream, attachment)]
[0001-gnu-services-Fix-pcscd-activation-bug.patch (text/x-patch, attachment)]
[signature.asc (application/pgp-signature, inline)]

Information forwarded to bug-guix <at> gnu.org:
bug#32478; Package guix. (Mon, 20 Aug 2018 16:14:01 GMT) Full text and rfc822 format available.

Message #8 received at 32478 <at> debbugs.gnu.org (full text, mbox):

From: ludo <at> gnu.org (Ludovic Courtès)
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: 32478 <at> debbugs.gnu.org
Subject: Re: bug#32478: pcscd service activation causes boot failure
Date: Mon, 20 Aug 2018 18:13:25 +0200
Hello,

Chris Marusich <cmmarusich <at> gmail.com> skribis:

> The attached patch fixes the issue by using an idempotent procedure to
> create the symlink.  If there are no objections, I'll push the patch to
> master about 24 hours from now.  There is a fairly long comment in my
> patch because although I wanted to re-use the switch-symlinks procedure
> from (guix utils), I couldn't figure out how to do it.  Ideas regarding
> this are welcome!  To be specific, I tried to use (guix utils), but then
> I got the following error message at boot time (see my comment for more
> information):
>
> loading '/gnu/store/f4ng1dlpm7q74vssbb049vpf2gvw3n3r-system/boot'...
> [    2.175072] random: fast init done
> making '/gnu/store/f4ng1dlpm7q74vssbb049vpf2gvw3n3r-system' the current system..
> .
> setting up setuid programs in '/run/setuid-programs'...
> populating /etc from /gnu/store/n6cwz1hlmjylva2xrv61njl68g6c8k5l-etc...
> usermod: no changes
> usermod: no changes
> usermod: no changes
> ERROR: In procedure dynamic-func:
> In procedure dynamic-pointer: Symbol not found: strverscmp

This is because (guix utils) pulls in lots of things that are not really
meant to be used on the “build side” (a misnomer in this case, but you
see what I mean ;-)).  Here, it assumes that dlopen support is available
but it’s not because the activation snippet runs on the
statically-linked Guile.

> From 3fa1e930b827aebca2dbbfe84c36cf203f15afda Mon Sep 17 00:00:00 2001
> From: Chris Marusich <cmmarusich <at> gmail.com>
> Date: Mon, 20 Aug 2018 00:16:06 -0700
> Subject: [PATCH] gnu: services: Fix pcscd activation bug.
>
> * gnu/services/security-token.scm (pcscd-activation): Idempotently create the
> /var/lib/pcsc symlink so that it does not fail when it already exists.

LGTM!

> +     (with-imported-modules (source-module-closure
> +                             '((guix build utils)))
> +       #~(begin
> +           (use-modules (guix build utils))
> +           ;; This switch-symlinks procedure was copied from (guix utils).  It
> +           ;; would be nice to re-use the procedure from that module, but if
> +           ;; we add that module to this gexp's imported modules and try to
> +           ;; use it, then this activation gexp can fail when it runs.  To be
> +           ;; specific, if you try to use (guix utils) and then build a VM
> +           ;; with a pcscd-service-type using "guix system vm-image", then
> +           ;; when you boot the VM, it will fail.  It fails because (guix
> +           ;; utils) dynamically links glibc's strverscmp function when
> +           ;; defining the version-compare procedure, and for some reason
> +           ;; strverscmp can't be found.  Perhaps there's a way to fix or
> +           ;; avoid this, but since we don't need the version-compare
> +           ;; procedure here, anyway, it's simpler to just define our own
> +           ;; switch-symlinks procedure instead.
> +           (define (switch-symlinks link target)
> +             (let ((pivot (string-append link ".new")))
> +               (symlink target pivot)
> +               (rename-file pivot link)))

I think “TODO: Deduplicate” would be enough as a comment, or at least
something more concise like “XXX: We cannot use (guix utils) because it
requires a dynamically-linked Guile, hence the duplicate
‘switch-symlinks’.”  I leave it up to you!

Thanks,
Ludo’.




Information forwarded to bug-guix <at> gnu.org:
bug#32478; Package guix. (Mon, 20 Aug 2018 20:05:03 GMT) Full text and rfc822 format available.

Message #11 received at 32478 <at> debbugs.gnu.org (full text, mbox):

From: Arun Isaac <arunisaac <at> systemreboot.net>
To: Chris Marusich <cmmarusich <at> gmail.com>
Cc: Ludovic Courtès <ludo <at> gnu.org>, 32478 <at> debbugs.gnu.org
Subject: Re: pcscd service activation causes boot failure
Date: Tue, 21 Aug 2018 01:03:23 +0530
> Commit de30205ba0f63eb987097a9f47b6e4fd38cd9044 added a pcscd service
> (hooray!).  However, its activation procedure doesn't always work.  The
> system test passes, and the activation procedure works the first time
> you boot, but if you reboot, it will fail because the symlink it creates
> already exists.  The attached patch fixes the problem.

Thanks for catching and fixing this! I am still in the process of
building and testing your patch. I ran out of memory and had to `guix
gc' and rebuild again. As you said, it takes a while. But, your patch
LGTM. Please go ahead and push to master at your discretion, after
addressing Ludo's feedback.




Reply sent to Chris Marusich <cmmarusich <at> gmail.com>:
You have taken responsibility. (Tue, 21 Aug 2018 04:53:02 GMT) Full text and rfc822 format available.

Notification sent to Chris Marusich <cmmarusich <at> gmail.com>:
bug acknowledged by developer. (Tue, 21 Aug 2018 04:53:02 GMT) Full text and rfc822 format available.

Message #16 received at 32478-done <at> debbugs.gnu.org (full text, mbox):

From: Chris Marusich <cmmarusich <at> gmail.com>
To: 32478-done <at> debbugs.gnu.org
Cc: Arun Isaac <arunisaac <at> systemreboot.net>, Ludovic
 Courtès <ludo <at> gnu.org>
Subject: Re: bug#32478: pcscd service activation causes boot failure
Date: Mon, 20 Aug 2018 21:52:19 -0700
[Message part 1 (text/plain, inline)]
Hi Ludo and Arun,

ludo <at> gnu.org (Ludovic Courtès) writes:

>> ERROR: In procedure dynamic-func:
>> In procedure dynamic-pointer: Symbol not found: strverscmp
>
> This is because (guix utils) pulls in lots of things that are not really
> meant to be used on the “build side” (a misnomer in this case, but you
> see what I mean ;-)).  Here, it assumes that dlopen support is available
> but it’s not because the activation snippet runs on the
> statically-linked Guile.

That makes sense.  I keep forgetting about this.  :-) Thank you for
reminding me!

> I think “TODO: Deduplicate” would be enough as a comment, or at least
> something more concise like “XXX: We cannot use (guix utils) because it
> requires a dynamically-linked Guile, hence the duplicate
> ‘switch-symlinks’.”  I leave it up to you!

Yes, I agree.  I'll simplify the comment.  I included the long comment
mainly in the hopes that somebody would be able to help me understand
why it wasn't working.

Arun Isaac <arunisaac <at> systemreboot.net> writes:

>> Commit de30205ba0f63eb987097a9f47b6e4fd38cd9044 added a pcscd service
>> (hooray!).  However, its activation procedure doesn't always work.  The
>> system test passes, and the activation procedure works the first time
>> you boot, but if you reboot, it will fail because the symlink it creates
>> already exists.  The attached patch fixes the problem.
>
> Thanks for catching and fixing this! I am still in the process of
> building and testing your patch. I ran out of memory and had to `guix
> gc' and rebuild again. As you said, it takes a while. But, your patch
> LGTM. Please go ahead and push to master at your discretion, after
> addressing Ludo's feedback.

OK!  I've rewritten the comment as Ludo suggested and committed this to
master as 6fb6ac6857df31c55a640eead2a5f79edd7dab14.

Now, I'll close this bug report.

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

bug archived. Request was from Debbugs Internal Request <help-debbugs <at> gnu.org> to internal_control <at> debbugs.gnu.org. (Tue, 18 Sep 2018 11:24:05 GMT) Full text and rfc822 format available.

This bug report was last modified 5 years and 193 days ago.

Previous Next


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