Guile 3.0.2 crashes with SIGBUS on x86_64 during 'guix pull'

OpenSubmitted by o.rojon.
Details
3 participants
  • Jakub Kądziołka
  • Ludovic Courtès
  • o.rojon
Owner
unassigned
Severity
important
O
O
o.rojon wrote on 4 Jun 21:12 +0200
The program '/gnu/store/foobar/compute-guix-derivation' failed to compute the derivation for guix
(name . Bug guix)(address . bug-guix@gnu.org)
54320f33124f0122ed56fb7e15b17ae4@posteo.net
Hello everyone,
on a freshly installed and reconfigured machine, I receive the following error:# guix pull: error: You found a bug: the program '/gnu/store/kpxami25fi3mrxb37sfbbx2s366chpk5-compute-guix-derivation'# failed to compute the derivation for Guix (version: "790ada9168e0689c1c4607c61cdc1d2dbc327abf"; system: "x86_64-linux";# host version: "398ec3c1e265a3f89ed07987f33b264db82e4080"; pull-version: 1).
I'm not entirely sure if, in this case, it is about RAM, as has been mentioned in this thread: https://issues.guix.gnu.org/41710. I do have 24gb ram, but currently a small swap file, but that might be totally unrelated.
If I can supply you with more information, please do not hesitate to specify how I can aid fixing the issue.
Have a good day,Olivier
L
L
Ludovic Courtès wrote on 4 Jun 22:00 +0200
(address . o.rojon@posteo.net)(address . 41715@debbugs.gnu.org)
87sgfa7g9r.fsf@gnu.org
Hello,
o.rojon@posteo.net skribis:
Toggle quote (9 lines)> on a freshly installed and reconfigured machine, I receive the> following error:> # guix pull: error: You found a bug: the program> '/gnu/store/kpxami25fi3mrxb37sfbbx2s366chpk5-compute-guix-derivation'> # failed to compute the derivation for Guix (version:> "790ada9168e0689c1c4607c61cdc1d2dbc327abf"; system: "x86_64-linux";> # host version: "398ec3c1e265a3f89ed07987f33b264db82e4080";> pull-version: 1).
Is there more info above these lines?
I tried to reproduce it with:
guix time-machine --commit=398ec3c1e265a3f89ed07987f33b264db82e4080 \ -- time-machine --commit=790ada9168e0689c1c4607c61cdc1d2dbc327abf \ -- describe
but it works for me.
Toggle quote (5 lines)> I'm not entirely sure if, in this case, it is about RAM, as has been> mentioned in this thread: https://issues.guix.gnu.org/41710 . I do> have 24gb ram, but currently a small swap file, but that might be> totally unrelated.
24 GiB is more than enough.
Is the problem reproducible if you try again now?
Thanks for reporting the issue,Ludo’.
J
J
Jakub Kądziołka wrote on 5 Jun 00:15 +0200
Re: The program '/gnu/store/foobar/compute-guix-derivation' failed to compute the derivation for guix
(address . 41715@debbugs.gnu.org)
20200604221534.ctileqxszheqs22b@gravity
Hi,
I encountered a similar error message today, but it worked fine afterretrying. After digging around in /var/log/guix/drvs, I found theselines in an appropriately-timestamped log file:
substitute: guix substitute: warning: ci.guix.gnu.org: connection failed: Connection timed out@ build-started /gnu/store/klyq05c1q8jzwnwrlmvy4ma6m2h6hbk0-compute-guix-derivation.drv - x86_64-linux /var/log/guix/drvs/kl//yq05c1q8jzwnwrlmvy4ma6m2h6hbk0-compute-guix-derivation.drv.bz2 3907994@ build-succeeded /gnu/store/klyq05c1q8jzwnwrlmvy4ma6m2h6hbk0-compute-guix-derivation.drv -
I do recall seeing the Connection timed out message, I'd even say it'sprobably the reason why I decided to retry the pull without muchinvestigation.
Regards,Jakub Kądziołka
-----BEGIN PGP SIGNATURE-----
iQIzBAABCAAdFiEE5Xa/ss9usT31cTO54xWnWEYTFWQFAl7ZcoIACgkQ4xWnWEYTFWQdQg/9GEQZomAp308YDv+xsp++RlUpZaiNO7dgaIHb/DGVdSAXAgHLyjpx6wA76QBfckB3hm+HERpKZFKLaE3qy2/l85Pe2gTicp8u/R3SPI5ScVLYelEuNJNxxoQMJN9wauaDuHlfd52LPwu5geAbhMJJ2Ra7/+HE3GtT0+sQdCaZSbp0+uo75sA9566q5D7lOjwHLTbzYRSN0kp9vVlYikleaBQD9x5702vwmcmwaV1R0EQVnwV7YnOHwWh8BIokB/jdWZIW//DkGmySdLqJIvpxulLY88RutWO1IMMNI2MnURxW6u60SL/yNQWttx82D6mOcn7hKWjhnBfA9KEZoBkrzumIkJLfXHNv/De2qaEB3VOYJKo6qK9+yyuWVLcZcJ837d+CUgeD3OpkIFLsF5PCgMjmCnhaysTdbNnb4DZg7xv+uVqSNY8qH4mv17swJFRaKCEmcypOmHZ2Tn2Dbsz6JbpoHVj/BDbeameMRiBIaEI4hh/SDO0sxeGfMrV9i0ebQNb4EGWvzI7zMNyo2JmMJ5NdaZu0Rf54woIxhC7CMXEKZio29h1YNYKNfce6GQ+LQKgCayRzKuVi8mZC7O9onf6bgU1CQe5YyuKcCJZSojYhXKmcHhpx0YC5wVxWBqZSmFNsPBuNuQFNOMl3hLuzqpln3j7x2PvZzMxQ+zhn/Uk==QACt-----END PGP SIGNATURE-----

O
O
o.rojon wrote on 5 Jun 07:54 +0200
Re: bug#41715: The program '/gnu/store/foobar/compute-guix-derivation' failed to compute the derivation for guix
(name . Ludovic Courtès)(address . ludo@gnu.org)
3b6810281efe6055d52232975fd44c91@posteo.net
Hej Ludo,
thanks for the quick reply!
I just tried to pull again and yielded the same result.
Actually, no, there is nothing above these lines. Yet, let me copy/paste the whole command in/output. It is in german but I guess the steps are so common by now that it shouldnt be a problem to understand :)
# user@computer ~$ guix pull# Kanal „guix“ wird vom Git-Repository auf https://git.savannah.gnu.org/git/guix.git“aktualisiert …# Von diesen Kanälen wird erstellt:# guix https://git.savannah.gnu.org/git/guix.git 0713929# guix pull: error: You found a bug: the program '/gnu/store/kpxami25fi3mrxb37sfbbx2s366chpk5-compute-guix-derivation'# failed to compute the derivation for Guix (version: "0713929685d2fd1970df1b1ce238ee7bd6e892f8"; system: "x86_64-linux";# host version: "398ec3c1e265a3f89ed07987f33b264db82e4080"; pull-version: 1).# Please report it by email to <bug-guix@gnu.org>.
Also, now I am sure it has nothing to do with ram. Apart from the already available ram, I resized my swapdevice.
Is there anything else I should test? One thing I could do is to roll-back (I did a upgrade/reconfigure cycle after installation). Should I do that and then pull again?
Greetings,Olivier
On 04.06.2020 22:00, Ludovic Courtès wrote:
Toggle quote (34 lines)> Hello,> > o.rojon@posteo.net skribis:> >> on a freshly installed and reconfigured machine, I receive the>> following error:>> # guix pull: error: You found a bug: the program>> '/gnu/store/kpxami25fi3mrxb37sfbbx2s366chpk5-compute-guix-derivation'>> # failed to compute the derivation for Guix (version:>> "790ada9168e0689c1c4607c61cdc1d2dbc327abf"; system: "x86_64-linux";>> # host version: "398ec3c1e265a3f89ed07987f33b264db82e4080";>> pull-version: 1).> > Is there more info above these lines?> > I tried to reproduce it with:> > guix time-machine --commit=398ec3c1e265a3f89ed07987f33b264db82e4080 \> -- time-machine --commit=790ada9168e0689c1c4607c61cdc1d2dbc327abf \> -- describe> > but it works for me.> >> I'm not entirely sure if, in this case, it is about RAM, as has been>> mentioned in this thread: https://issues.guix.gnu.org/41710 . I do>> have 24gb ram, but currently a small swap file, but that might be>> totally unrelated.> > 24 GiB is more than enough.> > Is the problem reproducible if you try again now?> > Thanks for reporting the issue,> Ludo’.
O
O
o.rojon wrote on 5 Jun 10:26 +0200
(name . Ludovic Courtès)(address . ludo@gnu.org)
6a6c6f638f0a33ec39b9e70843090299@posteo.net
Just to follow up: a roll-back does NOT solve the issue.
What I have tried:1) roll-back via guix system roll-back (to the generation that was created upon system installation)2) roll-back via guix package --roll-back (same)3) (1) + (2) combined4) boot into the generation created upon system installation.
In none of these cases was I able to run 'guix pull'.
Does it help for you to see my system configuration file? Because there isnt much in there apart from very basic packages (ncdu, recutils, some fonts, some icon packs) and a few services (GDM->SDDM login manager, MATE desktop, TOR). Also note that before the re-installation, I have not experienced problems with this configuration (MINUS mate desktop and TOR).
On 04.06.2020 22:00, Ludovic Courtès wrote:
Toggle quote (34 lines)> Hello,> > o.rojon@posteo.net skribis:> >> on a freshly installed and reconfigured machine, I receive the>> following error:>> # guix pull: error: You found a bug: the program>> '/gnu/store/kpxami25fi3mrxb37sfbbx2s366chpk5-compute-guix-derivation'>> # failed to compute the derivation for Guix (version:>> "790ada9168e0689c1c4607c61cdc1d2dbc327abf"; system: "x86_64-linux";>> # host version: "398ec3c1e265a3f89ed07987f33b264db82e4080";>> pull-version: 1).> > Is there more info above these lines?> > I tried to reproduce it with:> > guix time-machine --commit=398ec3c1e265a3f89ed07987f33b264db82e4080 \> -- time-machine --commit=790ada9168e0689c1c4607c61cdc1d2dbc327abf \> -- describe> > but it works for me.> >> I'm not entirely sure if, in this case, it is about RAM, as has been>> mentioned in this thread: https://issues.guix.gnu.org/41710 . I do>> have 24gb ram, but currently a small swap file, but that might be>> totally unrelated.> > 24 GiB is more than enough.> > Is the problem reproducible if you try again now?> > Thanks for reporting the issue,> Ludo’.
L
L
Ludovic Courtès wrote on 5 Jun 18:29 +0200
(address . o.rojon@posteo.net)(address . 41715@debbugs.gnu.org)
87mu5h1noe.fsf@gnu.org
Hi,
o.rojon@posteo.net skribis:
Toggle quote (11 lines)> Just to follow up: a roll-back does NOT solve the issue.>> What I have tried:> 1) roll-back via guix system roll-back (to the generation that was> created upon system installation)> 2) roll-back via guix package --roll-back (same)> 3) (1) + (2) combined> 4) boot into the generation created upon system installation.>> In none of these cases was I able to run 'guix pull'.
Thanks for testing this.
Note that, if “which guix” returns ~/.config/guix/current/bin/guix, thennone of the rollbacks above can have any effect. What could make adifference (but again, that would seem weird to me) is:
guix pull --roll-back
Are you passing extra options to guix-daemon, such as‘--cache-failures’? Or did you enable offloading?
It could also be that a transient failure is causing a silent buildfailure somewhere, as Jakub suggests.
Thanks,Ludo’.
PS: Please keep the bug address Cc’d.
O
O
o.rojon wrote on 6 Jun 02:57 +0200
(name . Ludovic Courtès)(address . ludo@gnu.org)(address . 41715@debbugs.gnu.org)
97da9eb75628d48de3b8ef4abb212bca@posteo.net
Hi Ludo,
guix pull --roll-back did not solve the issue. The only thing I believe to have changed is the "host version" portion of the error message.
'which guix' actually points to ~/.configu/guix/current/bin/guix. Should this command yield a different value?
I have actually not done anything regarding either guix-daemon or offloading. The only things I did after the fresh install were a reconfigure and passing a manifest file to 'guix package -m'.
I may have misunderstood what Jakub wrote because I am unsure what a transient failure is. The portion I believe to have understood essentially says "try again later and it should work".
I found the logs of 'compute-guix-derivation', but I was unable to see anything meaningful opening the .drv.bz2 file in emacs. Am I doing something wrong?
I wont be on my guix machine for some days now but appreciate pointers. :)
Olivier
On 05.06.2020 18:29, Ludovic Courtès wrote:
Toggle quote (34 lines)> Hi,> > o.rojon@posteo.net skribis:> >> Just to follow up: a roll-back does NOT solve the issue.>> >> What I have tried:>> 1) roll-back via guix system roll-back (to the generation that was>> created upon system installation)>> 2) roll-back via guix package --roll-back (same)>> 3) (1) + (2) combined>> 4) boot into the generation created upon system installation.>> >> In none of these cases was I able to run 'guix pull'.> > Thanks for testing this.> > Note that, if “which guix” returns ~/.config/guix/current/bin/guix, > then> none of the rollbacks above can have any effect. What could make a> difference (but again, that would seem weird to me) is:> > guix pull --roll-back> > Are you passing extra options to guix-daemon, such as> ‘--cache-failures’? Or did you enable offloading?> > It could also be that a transient failure is causing a silent build> failure somewhere, as Jakub suggests.> > Thanks,> Ludo’.> > PS: Please keep the bug address Cc’d.
L
L
Ludovic Courtès wrote on 7 Jun 21:47 +0200
(address . o.rojon@posteo.net)(address . 41715@debbugs.gnu.org)
87pnaasloq.fsf@gnu.org
Hi,
o.rojon@posteo.net skribis:
Toggle quote (4 lines)> guix pull --roll-back did not solve the issue. The only thing I> believe to have changed is the "host version" portion of the error> message.
OK, that’s expected.
Toggle quote (4 lines)> 'which guix' actually points to> ~/.configu/guix/current/bin/guix. Should this command yield a> different value?
No.
Toggle quote (8 lines)> I have actually not done anything regarding either guix-daemon or> offloading. The only things I did after the fresh install were a> reconfigure and passing a manifest file to 'guix package -m'.>> I may have misunderstood what Jakub wrote because I am unsure what a> transient failure is. The portion I believe to have understood> essentially says "try again later and it should work".
Yes, that’s the essence of it. :-)
Toggle quote (4 lines)> I found the logs of 'compute-guix-derivation', but I was unable to see> anything meaningful opening the .drv.bz2 file in emacs. Am I doing> something wrong?
Could you try this:
strace -s 500 -o log guix pull
and post the ‘log’ file, compressed, somewhere?
Thanks,Ludo’.
O
O
o.rojon wrote on 8 Jun 16:55 +0200
(name . Ludovic Courtès)(address . ludo@gnu.org)(address . 41715@debbugs.gnu.org)
e1fcc66a4a7d5f89e7135e0459ead070@posteo.net
Hi Ludo,
here is the log uploaded on mediafire: http://www.mediafire.com/file/ldqoi68y88rzrn9/log.bz2/file(note that if you can recommend another uploading service, feel free to!)
Since the log is in german, let me specify at least two of the recurrent phrases in english:"Datei oder Verzeichnis nicht gefunden" means "file or directory not found""Die Ressource ist zur Zeit nicht verfügbar" means "Resource is currently unavailable"
If there is other things that need be translated, dont hesitate to tell me :)
Greetings,Olivier
On 07.06.2020 21:47, Ludovic Courtès wrote:
Toggle quote (38 lines)> Hi,> > o.rojon@posteo.net skribis:> >> guix pull --roll-back did not solve the issue. The only thing I>> believe to have changed is the "host version" portion of the error>> message.> > OK, that’s expected.> >> 'which guix' actually points to>> ~/.configu/guix/current/bin/guix. Should this command yield a>> different value?> > No.> >> I have actually not done anything regarding either guix-daemon or>> offloading. The only things I did after the fresh install were a>> reconfigure and passing a manifest file to 'guix package -m'.>> >> I may have misunderstood what Jakub wrote because I am unsure what a>> transient failure is. The portion I believe to have understood>> essentially says "try again later and it should work".> > Yes, that’s the essence of it. :-)> >> I found the logs of 'compute-guix-derivation', but I was unable to see>> anything meaningful opening the .drv.bz2 file in emacs. Am I doing>> something wrong?> > Could you try this:> > strace -s 500 -o log guix pull> > and post the ‘log’ file, compressed, somewhere?> > Thanks,> Ludo’.
L
L
Ludovic Courtès wrote on 9 Jun 16:30 +0200
(address . o.rojon@posteo.net)(address . 41715@debbugs.gnu.org)
87mu5cia64.fsf@gnu.org
Hi,
o.rojon@posteo.net skribis:
Toggle quote (4 lines)> here is the log uploaded on mediafire:> http://www.mediafire.com/file/ldqoi68y88rzrn9/log.bz2/file (note that> if you can recommend another uploading service, feel free to!)
I don’t know, maybe you could run IPFS.
The log reads:
Toggle snippet (7 lines)clone(child_stack=NULL, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x7f3d44cfee50) = 1498close(17) = 0read(16, "", 1) = 0close(16) = 0wait4(1498, [{WIFSIGNALED(s) && WTERMSIG(s) == SIGBUS}], 0, NULL) = 1498
… which means the ‘compute-guix-derivation’ process crashed with SIGBUS.
Could you run:
ulimit -c unlimited guix pull
That should fail again, but this time there should be a ‘core’ file inthe current directory (or ‘core.’ followed by digits).
Then you can run:
gdb --core=./core
and at the GDB prompt, type:
thread apply all bt
Could you let me know what that returns?
Thanks,Ludo’.
O
O
o.rojon wrote on 12 Jun 07:54 +0200
(name . Ludovic Courtès)(address . ludo@gnu.org)(address . 41715@debbugs.gnu.org)
47c4e596596303893a847212ffa4da9b@posteo.net
Hej Ludo,
crazy, I wouldn't have thought it would go as deep as gdb as quickly ;-)
I followed the steps you mentioned, the results you find here: https://www.mediafire.com/file/g5yz8f3pput8f3w/gdb-output/file . The only part I omitted from the gdb output is the first lines mentioned the license. (Will look at IPFS, maybe next time :))
If you need anything else, dont hesitate to specify. Also, should a reinstall become necessary, that would be annoying but not the end of the world. So if we hit a dead end thats fine by me.
Thanks a lot for your help. So far, I received only good from the guix community :)
Greetings,Olivier
On 09.06.2020 16:30, Ludovic Courtès wrote:
Toggle quote (46 lines)> Hi,> > o.rojon@posteo.net skribis:> >> here is the log uploaded on mediafire:>> http://www.mediafire.com/file/ldqoi68y88rzrn9/log.bz2/file (note that>> if you can recommend another uploading service, feel free to!)> > I don’t know, maybe you could run IPFS.> > The log reads:> > --8<---------------cut here---------------start------------->8---> clone(child_stack=NULL,> flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD,> child_tidptr=0x7f3d44cfee50) = 1498> close(17) = 0> read(16, "", 1) = 0> close(16) = 0> wait4(1498, [{WIFSIGNALED(s) && WTERMSIG(s) == SIGBUS}], 0, NULL) = > 1498> --8<---------------cut here---------------end--------------->8---> > … which means the ‘compute-guix-derivation’ process crashed with > SIGBUS.> > Could you run:> > ulimit -c unlimited> guix pull> > That should fail again, but this time there should be a ‘core’ file in> the current directory (or ‘core.’ followed by digits).> > Then you can run:> > gdb --core=./core> > and at the GDB prompt, type:> > thread apply all bt> > Could you let me know what that returns?> > Thanks,> Ludo’.
L
L
Ludovic Courtès wrote on 12 Jun 16:12 +0200
(address . o.rojon@posteo.net)(address . 41715@debbugs.gnu.org)
87zh9874q9.fsf@gnu.org
Hi Olivier,
o.rojon@posteo.net skribis:
Toggle quote (5 lines)> I followed the steps you mentioned, the results you find here:> https://www.mediafire.com/file/g5yz8f3pput8f3w/gdb-output/file . The> only part I omitted from the gdb output is the first lines mentioned> the license. (Will look at IPFS, maybe next time :))
Since it’s a small file, you can send it as an attachment.
GDB shows:
Toggle snippet (20 lines)Core was generated by `/gnu/store/0m0vd873jp61lcm4xa3ljdgx381qa782-guile-3.0.2/bin/guile --no-auto-com'.Program terminated with signal SIGBUS, Bus error.#0 0x00007f90885ccc78 in ?? ()[Current thread is 1 (LWP 845)](gdb) thread apply all bt
Thread 9 (LWP 853):#0 0x00007f90884d90a4 in ?? ()#1 0x0000000000000001 in ?? ()#2 0x0000000000000001 in ?? ()#3 0x00007f90840c9a30 in ?? ()#4 0x00007f90840c95a0 in ?? ()#5 0x00007f90840ca700 in ?? ()#6 0x00007f90885ac067 in ?? ()#7 0x00007f90845efa80 in ?? ()#8 0x00007f90884fed94 in ?? ()#9 0x0000000000000001 in ?? ()#10 0x00007f90840c95a0 in ?? ()
That means debugging info is lacking. To address that, could you run:
guix build $(guix gc --derivers /gnu/store/0m0vd873jp61lcm4xa3ljdgx381qa782-guile-3.0.2)
and then:
echo 'set debug-file-directory /gnu/store/9lg4gssswn2cwn54p6jjy6nld16ah795-guile-3.0.2-debug/lib/debug' >> ~/.gdbinit
At that point, you can try again to run:
gdb --core=./core
That will hopefully show more useful info.
Thanks in advance!
Ludo’.
L
L
Ludovic Courtès wrote on 17 Jun 14:06 +0200
control message for bug #41715
(address . control@debbugs.gnu.org)
87o8phrj57.fsf@gnu.org
retitle 41715 Guile 3.0.2 crashes with SIGBUS on x86_64 during 'guix pull'quit
L
L
Ludovic Courtès wrote on 17 Jun 14:06 +0200
(address . control@debbugs.gnu.org)
87mu51rj50.fsf@gnu.org
severity 41715 importantquit
?