Discussion related to cado-nfs

[Robert Balfour <rhb11931@gmail.com>]

Dear Paul et al,

Any progress on finding the source of this bug?

Best regards,

Robert

On Mon, Jul 26, 2021 at 3:32 PM Paul Zimmermann <Paul.Zimmermann@inria.fr> wrote:

       Hi,

I am able to reproduce the issue. It seems to happen since the following
commit:

commit c3079ff3ce67dd801915dbd335b210e8570bea68
Author: Emmanuel Thomé <Emmanuel.Thome@inria.fr>
Date:   Sat Apr 3 15:58:22 2021 +0200

    fopen_maybe_compressed: always use a temp file when writing

Emmanuel, do you have any clue?

Paul

> From: Robert Balfour <rhb11931@gmail.com>
> Date: Mon, 26 Jul 2021 03:06:11 +0100
>
>
> [1:multipart/alternative Hide]
>
>
> [1/1:text/plain Show]
>
>
> [1/2:text/html Hide Save:noname (3kB)]
>
> Dear all,
>
> In recent revisions of CADO (revision 4ef6307d and the latest revision) I am
> seeing a bug, on at least some numbers, when dup1 tries to process more than
> 2^26 = 67108864 relations. dup1.stdout.1 ends abruptly with
>
> # Read 67108864 relations in 135.7s -- 62.5 MB/s -- 494715.5 rels/s
>
> while dup1.stderr.1 ends with the following:
>
> code BUG() : condition rc == 0 failed in split_iter_open_next_file at
> /store/charybdis/cado_4ef6307d/cado-nfs/filter/dup1.c:111 -- Abort
>
> I tried altering the contents of dup1.filelist.1 and found that the bug
> appears precisely when the files contain more than 2^26 relations. So that
> you can attempt to reproduce this, I have attached a parameter file that
> produces the bug when run with the 149-digit number
>
> N =
> 30451074702779672217829979074588808155199316436755065947683534426904549869449262327316787277317820182752090699957549708141192736670950500076802512839
>
>
> I have also heard from another user who has experienced the same bug with
> revision 90dd8db3.
>
> Best regards,
> Robert Balfour
>
> [2:application/octet-stream Show Save:c150.params (2kB)]
>