On Friday 27 February 2009 17:01:46 Bill Hart wrote: > Nothing much is automatic about the fat build system. Note that > removing a function may screw it up because in multiple places in the > code functions are expected to be given in a certain order (which is > defined in multiple places - hopefully the same in all places). > > The whole thing is one giant hack - and that was before I touched it. > > The divexactbyff function was added to the list in configure.in when > the merge was done. You can't just add a function name there, you have > to update all the code in fat.h, fat.c, x86_64-defs.m4, x86-defs.m4, > gmp-impl.h (multiple places) and who knows where else. > > Bill. >
I taken out divexact_byff and put in redc_basecase and sumdiff_n , tested on K8,K10, it broke my K7 (it's half dead anyway , I had to disable 3/4 of the memory just to get kernel to boot!), have to get a new machine , perhaps a nice nehalem! > 2009/2/27 <ja...@njkfrudils.plus.com>: > > On Friday 27 February 2009 16:40:26 ja...@njkfrudils.plus.com wrote: > >> I was going to add redc_basecase to the fat structure when > >> > >> > >> the fat build is broken in trunk rev 1638 > >> > >> ../mpn/m4-ccas --m4=m4 > >> gcc -std=gnu99 -c -DHAVE_CONFIG_H -O2 -m64 -D__GMP_WITHIN_GMP -I.. > >> -DOPERATION_amd64_submul_1 -I. -I. -I.. amd64_submul_1.asm -o > >> amd64_submul_1.o >/dev/null 2>&1 > >> In file included from ../gmp-impl.h:109, > >> from fat.c:31: > >> ../fat.h:44: warning: data definition has no type or storage class > >> ../fat.h:44: warning: type defaults to 'int' in declaration > >> of 'DECL_divexact_byff' > >> ../fat.h:44: warning: parameter names (without types) in function > >> declaration ../fat.h:268: warning: data definition has no type or > >> storage class ../fat.h:268: warning: type defaults to 'int' in > >> declaration of 'DECL_divexact_byff' > >> ../fat.h:268: warning: parameter names (without types) in function > >> declaration ../fat.h:301: warning: data definition has no type or > >> storage class ../fat.h:301: warning: type defaults to 'int' in > >> declaration of 'DECL_divexact_byff' > >> ../fat.h:301: warning: parameter names (without types) in function > >> declaration In file included from fat.c:31: > >> ../gmp-impl.h:844:1: warning: "mpn_divexact_byff" redefined > >> In file included from ../gmp-impl.h:109, > >> from fat.c:31: > >> ../fat.h:42:1: warning: this is the location of the previous definition > >> fat.c: In function '__gmpn_cpuvec_init': > >> fat.c:234: error: 'struct cpuvec_t' has no member named 'divexact_byff' > >> fat.c:234: error: '__gmpn_divexact_byff_core2' undeclared (first use in > >> this function) > >> fat.c:234: error: (Each undeclared identifier is reported only once > >> fat.c:234: error: for each function it appears in.) > >> fat.c:267: error: 'struct cpuvec_t' has no member named 'divexact_byff' > >> fat.c:267: error: '__gmpn_divexact_byff_amd64' undeclared (first use in > >> this function) > >> fat.c:272: error: 'struct cpuvec_t' has no member named 'divexact_byff' > >> fat.c:305: error: 'volatile struct cpuvec_t' has no member > >> named 'divexact_byff' > >> fat.c:305: error: 'struct cpuvec_t' has no member named 'divexact_byff' > >> make[2]: *** [fat.lo] Error 1 > >> make[2]: *** Waiting for unfinished jobs.... > >> make[2]: Leaving directory `/home/jasonmoxham/mpir/mpir/trunk/mpn' > >> make[1]: *** [all-recursive] Error 1 > >> make[1]: Leaving directory `/home/jasonmoxham/mpir/mpir/trunk' > >> make: *** [all] Error 2 > >> cuda1:~/mpir/mpir/trunk$ ls > >> > >> I don't know if this is picked up automatically , or manual , it > >> certainly doesn't pick up redc basecase > > > > We dont use divexact_byff anywhere yet , users are not exposed to it , > > so there no need for it to be in the fat structure. I'll remove the entry > > from configure.in , that should fix it . > > I will put an entry in for redc_basecase , but the other new functions > > are not suitible (some can be macros, others are only defined on some > > arches) > > > > Jason > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en -~----------~----~----~----~------~----~------~--~---
