[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#852201: RFS: mlucas/14.2-1



Adam Borowski <kilobyte@angband.pl> writes:

> On Sun, Jan 22, 2017 at 11:34:55PM +0100, Adam Borowski wrote:
>> On Sun, Jan 22, 2017 at 09:10:33PM +0800, Alex Vong wrote:
>> > * Package name    : mlucas
>> >   Version         : 14.2-1
>> 
>> Alas, it fails to build for me, tested on amd64 and armhf:
>> FAIL: fermat-test.test
>> ./test-driver: line 107: 29967 Segmentation fault (core dumped) "$@"
>> > $log_file 2>&1
>> 
>> On amd64 it then SIGILLs in some subsequent tests, despite being supposed to
>> detect CPU used, but that goes after the segfault so that's probably either
>> a result or unrelated.
>
> On the other hand, it did succeed on real i386 (non-nx Pentium 4, rather
> than hardware emulation on amd64), which suggests it's not about depending
> on a modern instruction.  Of course, considering how long the i386 build
> takes compared to amd64 or armhf, I don't think anyone wants to run
> CPU-intensive tasks like this on such a crummy arch.
>
Indeed, it should at least be buildable on amd64 to be useful. I have
contacted upstream and ask for help.

I guess the segfault is due to some bugs in the #threads detection code
since we have:
  WARN: CY_THREADS = 8 exceeds number of cores = 6
  WARN: At line 342 of file .././src/radix1024_ditN_cy_dif1.c:
  n_div_nwt%CY_THREADS != 0
but I am not absolutely sure.

Yes, I think it is a good idea to call it 14.2~pre since some more bugs
might get fixed on the real release.

Thanks for the testing! I will reply when the issues are (believed to
be) fixed.

>
> Meow!

Attachment: signature.asc
Description: PGP signature


Reply to: