Discussion related to cado-nfs

[lunaluan9555 <lunaluan9555@gmail.com>]

Dear Authors, 

1- About MPI

At first I download an up-to-date version of cado-nfs from gitlab. Before the MPI part, some other building errors occurred, maybe there's something wrong with gf2x? I put the output in attachment. 

Then I only modified `linalg/bwc/check_mpi_problems.c` according to commit b8a0b293 in my old version (downloaded on 18, Mar) and now it works well. 

2. About JL method and GMP-ECM.

I check the /tmp/cado.xxxx/p40.descent.stderr.1 and /tmp/cado.xxxx/p40.descent.stdout.1, that's not helpful for me. I put them into attachment. I'm a beginner on DLP and maybe the parameters I chose by random are not proper. I think maybe this bug is not that important...?

But the ECM is quite annoying :-<. I think there's definitely something wrong with my old ECM lib. Version 6.4.4 seems too old to be compatible with my linux kernel, then I tried the up-to-date version on  https://gitlab.inria.fr/zimmerma/ecm/-/tree/master . The new version works well itself, but still cannot be caught by cado. The `make cmake` command could find libecm.so, but it failed to locate `ecm.h` (It's just there!), and also failed to print GMP-ECM version. 

I even restart the server, and that doesn't work either. It's quite annoying. Maybe I can try it on other machines later, or I will try to install ECM in other path.

I'm not sure which version of GMP-ECM is stable. I could not access the official website http://ecm.gforge.inria.fr/, maybe the website crashed or because of the GFW :-( . Could you provide a stable version (or which you are using)? 

3. Another (small) problem. 

In `cado-nfs.py` line 261-264, Sometimes python reports bug like:

```

File "./cado-nfs.py", line 263, in  
opportunistic_generator = pow(int(target), ilogt*cof, p)
ValueError: pow() 2nd argument cannot be negative when 3rd argument specified
```

This is because `ilogt` has negative value. I don't know if this result is legal, I just commented it out temporarily. The original `xgcd()` algorithm cannot always guarantee a positive coefficient, is here something special?

That's all the progress so far. Thanks.

On 4/11/2022 08:12，Emmanuel Thomé<Emmanuel.Thome@inria.fr> wrote：

Hi,

1 - it's super weird that your cmake build didn't catch the ECM config.
It usually works, I don't know what you did. In config/gmpecm.cmake, the
HAVE_GMPECM variable is clearly defined, and that should expose the
descent_init_Fp target.

2 -  The error "'GeneralClass' object  has no attribute 'initfacu'" is
caused by the file
/tmp/cado.dx6kanr3/p40.descent.3092759267...1165803190.upper.init
being empty (and the error message should be more polite).

I was able to reproduce a misbehaviour, but not the same as yours. It
seems that some parameters are really badly tuned for these small sizes,
and the initialization of the descent fails, or at least takes ages,
which has no reason to be. (pushing the tasks.descent.init_maxeff
parameter to a larger value such as 20000000 could help it to
at least produce something useful, but generally the time taken for this
initialization doesn't look right). Further down the road (but it is
independent), some of the stuff that I added recently doesn't seem to
work well in this case, and I would need to check that.

However, in your case, the reason why the file
/tmp/cado.dx6kanr3/p40.descent.3092759267...1165803190.upper.init is
empty is entirely different. It appears that the descent_init_Fp binary
ended with a seg fault:

Error:Individual logarithm: Command line was: lab124-RH2288H-V3/scripts/descent.py --target 30927592678837481655171185231165803190 --gfpext 1 --prefix p40 --datadir /tmp/cado.dx6kanr3 --cadobindir lab124-RH2288H-V3 --descent-hint /home/lab124/NFS_CADO/cado-development/cado-nfs-master/parameters/dlp/p40.hint --init-I 11 --init-ncurves 10 --init-lpb 20 --init-lim 7000 --init-mfb 40 --init-tkewness 7000 --I 11 --lpb0 14 --lpb1 14 --mfb0 28 --mfb1 28 --lim0 7000 --lim1 7000 --ell 340282366920938463463374607431768645989 > /tmp/cado.dx6kanr3/p40.descent.stdout.1 2> /tmp/cado.dx6kanr3/p40.descent.stderr.1
Error:Individual logarithm: Stderr output (last 10 lines only) follow (stored in file /tmp/cado.dx6kanr3/p40.descent.stderr.1):
Error:Individual logarithm:     double free or corruption (fasttop)

We're going to investigate that on our side, and it seems likely that we
will find the problem. However, further input from your side could be
valuable, too.  Could you please send us the files:
/tmp/cado.dx6kanr3/p40.descent.stderr.1
/tmp/cado.dx6kanr3/p40.descent.stdout.1

As well as a result of a rerun of the descent_init_Fp binary (the full
command line to it should be in the stdout file above). If you feel like
it, a gdb stacktrace on a debug build could be helpful, too.

Thanks!

E.

[  4%] Building gf2x
make[3]: warning: jobserver unavailable: using -j1.  Add '+' to parent make 
rule.
Making all in lowlevel
Making all in src
Making all in .
In file included from 
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:100:0,
                 from 
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x.c:39:
./gf2x/gf2x_mul1.h:1:1: error: expected identifier or ¡®(¡¯ before ¡®.¡¯ token
 ../../lowlevel/mul1cl.c
 ^
In file included from 
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x.c:39:0:
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:47:1:
 warning: ¡®gf2x_mul1¡¯ used but never defined
 gf2x_mul1(unsigned long *c, unsigned long a, unsigned long b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:50:1:
 warning: ¡®gf2x_mul_1_n¡¯ used but never defined
 gf2x_mul_1_n(unsigned long *cp, const unsigned long *bp, long sb, unsigned 
long a)
 ^~~~~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:53:1:
 warning: ¡®gf2x_addmul_1_n¡¯ used but never defined
 gf2x_addmul_1_n(unsigned long *dp,
 ^~~~~~~~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:58:1:
 warning: ¡®gf2x_mul2¡¯ used but never defined
 gf2x_mul2(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:61:1:
 warning: ¡®gf2x_mul3¡¯ used but never defined
 gf2x_mul3(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:64:1:
 warning: ¡®gf2x_mul4¡¯ used but never defined
 gf2x_mul4(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:67:1:
 warning: ¡®gf2x_mul5¡¯ used but never defined
 gf2x_mul5(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:70:1:
 warning: ¡®gf2x_mul6¡¯ used but never defined
 gf2x_mul6(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:73:1:
 warning: ¡®gf2x_mul7¡¯ used but never defined
 gf2x_mul7(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:76:1:
 warning: ¡®gf2x_mul8¡¯ used but never defined
 gf2x_mul8(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
/home/lab124/NFS_CADO/cado-development/cado-nfs-master/gf2x/gf2x/gf2x-small.h:79:1:
 warning: ¡®gf2x_mul9¡¯ used but never defined
 gf2x_mul9(unsigned long *c, const unsigned long *a, const unsigned long *b)
 ^~~~~~~~~
Makefile:725: recipe for target 'libgf2x_local_a-gf2x.o' failed

Attachment: p40.descent.stderr.1
Description: Binary data

Attachment: p40.descent.stdout.1
Description: Binary data