bug#43518: Guix substitute crash in procedure raise-exception: wrong type agument in position 1: #f

[Ludovic Courtès]

Hi,

Maxim Cournoyer <maxim.cournoyer@gmail.com> skribis:

> downloading from 
> https://ci.guix.gnu.org/nar/6m9zimmw8p6gbc1yfbg454c1r587b7h4-gcc-10.2.0.tar.xz
>  ...
[…]
>  gcc-10.2.0.tar.xz  74.3MiB                                     1.1MiB/s 
> 00:25 [#######           ]  39.1%Backtrace:
[…]
> In unknown file:
>            5 (display "@ substituter-succeeded 
> /gnu/store/r06j3ms57z4mzfpdzfclsi3i9hr4184g-module-imp…" …)
> In guix/status.scm:
>    699:16  4 (write! _ _ _)
>     613:6  3 (_ (download-progress 
> "/gnu/store/6m9zimmw8p6gbc1yfbg454c1r587b7h4-gcc-10.2.0.tar.xz" # …) …)
> In guix/progress.scm:
>    213:14  2 (display-download-progress "6m9zimmw8p6gbc1yfbg454c1r58@" _ 
> #:start-time _ #:transferred _ …)
> In ice-9/boot-9.scm:
>   1669:16  1 (raise-exception _ #:continuable? _)
>   1669:16  0 (raise-exception _ #:continuable? _)
>
> ice-9/boot-9.scm:1669:16: In procedure raise-exception:
> In procedure =: Wrong type argument in position 1: #f

If you have a case that’s reproducible, please take advantage of it and
add ‘pk’ calls in (guix progress) to see what happens.  Here it seems
that ‘transferred’ (in ‘display-download-progress’) is #f.

The #f would come from this clause in (guix status):

     ('download-progress item uri
                         (= string->number size)
                         (= string->number transferred))  ;<- here

That in turn comes from ‘progress-reporter/trace’, called from (guix
scripts substitute).

It looks like this issue happens when max-jobs is (noticeably) greater
than 1, as if the client was receiving intermingled “@
download-progress” traces, and thus not getting the right ‘transferred’
value.

Could you try to reproduce it while ‘guix’ is running under strace?
Like:

  strace -o log -s 200 guix build whatever --max-jobs=20

?

Then, if you reproduce the error, we’ll look for intermingled traces in
the log.

Thanks,
Ludo’.