[Debian-science-sagemath] cypari2 hangs on ppc64el when pari stack overflows

Tobias Hansen thansen at debian.org
Mon Oct 29 09:30:50 GMT 2018


Hi,

sorry for the late answer, for some reason I received your mail only today.


On 10/06/2018 11:47 AM, Jeroen Demeyer wrote:
> On 2018-10-06 11:36, Tobias Hansen wrote:
>> Hi,
>>
>> sagemath 8.3 currently fails to build on ppc64el
>
> Is that a typo for ppc64le or is that something else? I have
>
> $ uname -a
> Linux sardonis 4.4.0-124-generic #148-Ubuntu SMP Wed May 2 13:02:22 UTC 2018 ppc64le ppc64le ppc64le GNU/Linux

It's the same. uname shows the POSIX machine name, ppc64el is the Debian architecture name. It's like x86_64 vs amd64.

>
>> import cypari2
>> pari = cypari2.Pari()
>> pari.allocatemem(200000)
>> pari('x^2 + 10^100 + 1').nfinit()
>
> This doesn't actually stack overflow for me (PARI-2.11, cypari2-1.2.1, cysignals-1.7.2):
>
> >>> import cypari2
> >>> pari = cypari2.Pari()
> >>> pari.allocatemem(200000)
> PARI stack size set to 200000 bytes, maximum size set to 8060928
> >>> pari('x^2 + 10^100 + 1').nfinit()
> *** _aprcl_step4_worker: Warning: increasing stack size to 400000.
> *** _aprcl_step4_worker: Warning: increasing stack size to 800000.
> [x^2 + 10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001, [0, 1], -40000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000004, 1, [Mat([1, 0.E-134 + 1.00000000000000 E50*I]), [1, 1.00000000000000 E50; 1, -1.00000000000000 E50], [1, 100000000000000000000000000000000000000000000000000; 1, -100000000000000000000000000000000000000000000000000], [2, 0; 0, -20000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000002], [20000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000002, 0; 0, 2], [10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001, 0; 0, -1], [10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001, [0,
> -10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001; 1, 0]], [2, 73, 137, 401, 1201, 1601, 1676321, 5964848081, 129694419029057750551385771184564274499075700947656757821537291527196801]], [0.E-134 + 1.00000000000000 E50*I], [1, x], [1, 0; 0, 1], [1, 0, 0, -10000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001; 0, 1, 1, 0]]
>
>
> Cheers,
> Jeroen (cypari2 and cysignals maintainer) 

It might be partly random. The test succeeded for 8.3-1, failed for 8.3-2. Then I disabled it for 8.3-3 and 8.3-4 and enabled it again for 8.4-1 and 8.4-2 where it succeeded.

Best,
Tobias



More information about the Debian-science-sagemath mailing list