Discussion related to cado-nfs

[Emmanuel Thomé <emmanuel.thome@gmail.com>]

Hi Jason,

Thanks for the report.

These build failures have indeed been introduced by recent changes,
and will get fixed. Cado-nfs was compiling and running on several
platforms previously, it should remain so.

I'll handle the shift count parts.

fgetc_unlocked needs a simple cmake test.

For pthread_tryjoin_np, it's less obvious, since the functionality can
probably not be emulated trivially.

Best,

E.

On Sun, Nov 25, 2012 at 2:29 AM, Jason Papadopoulos <jasonp@boo.net> wrote:
> Hello,
>
> I'm getting some build problems when attempting to compile git master
> of CADO-NFS:
>
> utils/relation.c now needs fgetc_unlocked, which neither OS X or
> cygwin have
>
> After working around that, in both OS X and cygwin I get
>
> CMakeFiles/purge.dir/purge.c.o: In function 
> `prempt_scan_relations_pass_two':
> /home/jasonp/cado/filter/purge.c:1946: undefined reference to
> `_pthread_tryjoin_np'
>
> On 32-bit x86 I get
>
> In file included from /home/jasonp/cado/linalg/bwc/mpfq/abase_vbase.c:6:0:
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_4.h: In function
> `abase_p_4_reduce':
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_4.h:819:5: warning: right shift
> count
>>= width of type
> In file included from /home/jasonp/cado/linalg/bwc/mpfq/abase_vbase.c:8:0:
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.h: In function
> `abase_p_1_reduce':
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.h:819:5: warning: right shift
> count
>>= width of type
>
> In file included from /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.c:4:0:
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.h: In function
> `abase_p_1_reduce':
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.h:819:5: warning: right shift
> count
>>= width of type
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.c: In function
> `abase_p_1_vec_conv_u
> r_ks':
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.c:648:5: warning: comparison
> between
>  signed and unsigned integer expressions
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.c:656:5: warning: comparison
> between
>  signed and unsigned integer expressions
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.c:671:5: warning: comparison
> between
>  signed and unsigned integer expressions
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.c: In function
> `abase_p_1_wrapper_ve
> c_scal_mul':
> /home/jasonp/cado/linalg/bwc/mpfq/fixmp_x86_32.h:14:3: error: can't find a
> regis
> ter in class `GENERAL_REGS' while reloading `asm'
> /home/jasonp/cado/linalg/bwc/mpfq/fixmp_x86_32.h:14:3: error: `asm'
> operand has
> impossible constraints
> linalg/bwc/CMakeFiles/bwc_mpfq.dir/build.make:169: recipe for target
> `linalg/bwc
> /CMakeFiles/bwc_mpfq.dir/mpfq/abase_p_1.c.o' failed
> make[2]: *** [linalg/bwc/CMakeFiles/bwc_mpfq.dir/mpfq/abase_p_1.c.o] Error 1
>
> and finally
>
> In file included from /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1_t.h:6:0,
>                  from /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1_t.c:4:
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.h: In function
> `abase_p_1_reduce':
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1.h:819:5: warning: right shift
> count
>>= width of type
> /home/jasonp/cado/linalg/bwc/mpfq/abase_p_1_t.c: In function
> `abase_p_1_p_1_dotp
> rod':
> /home/jasonp/cado/linalg/bwc/mpfq/fixmp_x86_32.h:14:3: error: can't find a
> regis
> ter in class `GENERAL_REGS' while reloading `asm'
> /home/jasonp/cado/linalg/bwc/mpfq/fixmp_x86_32.h:14:3: error: `asm'
> operand has
> impossible constraints
> linalg/bwc/CMakeFiles/bwc_mpfq.dir/build.make:192: recipe for target
> `linalg/bwc
> /CMakeFiles/bwc_mpfq.dir/mpfq/abase_p_1_t.c.o' failed
> make[2]: *** [linalg/bwc/CMakeFiles/bwc_mpfq.dir/mpfq/abase_p_1_t.c.o]
> Error 1
>
>
> Note that it isn't a big deal for me if support for 32-bit x86 is too
> cumbersome in the current codebase.
>
> jasonp
>
>
>
>
> _______________________________________________
> Cado-nfs-discuss mailing list
> Cado-nfs-discuss@lists.gforge.inria.fr
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/cado-nfs-discuss