yasm

[Jason Moxham]

Hi

At some stage we will want to upgrade our included yasm to the latest , we
should be able to just put the new one in our yasm subdirectory.We did make
some changes to it , but most are not applicable now ( they were to do with
disabling make install,check which we now do another way) , so eliminating
these changes and an autotools change , we are left with

Files ./config/config.sub and /root/mpir/mpir/trunk/yasm/config/config.sub
differ
diff mpir/mpir/trunk/yasm/config/config.sub config/config.sub
1076c1076
< *-unknown | *-pc)
---
> *-unknown)

This is compairing the current mpir against mpirsvn-rev 1379
Assuming there are no changes before rev 1379 , we are set to go.

I propose changing this trunk , not on a branch ?

Jason

[Bill Hart]

Yep, let's do it. We can deal with any fallout as it arises.

Bill.

2009/7/25 Jason Moxham <ja...@njkfrudils.plus.com>:

[Jason Moxham]

Whoops , svn yasm doesn't build with my autotools , and the site has slowed to
a crawl , so I cant see what is happening , so trunk doesn't build at the mo.

Jason

[Jason Moxham]

Got it working , had to run autogen first
although building out of the src tree fails , it is a yasm problem , I'll post
to yasm trac

Jason

[Bill Hart]

I still get this on sage.math:

=== configuring in yasm (/home/wbhart/mpir-trunk/yasm)
configure: running /bin/bash ./configure '--prefix=/usr/local'
--cache-file=/dev/null --srcdir=.
configure: error: cannot find install-sh or install.sh in config "."/config
configure: error: ./configure failed for yasm

[Jason Moxham]

I have corrected another glitch , I didn't include all files in yasm , I
thought they were autogenerated.

But I get this error now

ASS: t-fdiv
PASS: t-fdiv_ui
PASS: t-cdiv_ui
gcdext.c:1357: GNU MP assertion failed: tn > 0
/bin/sh: line 4: 7944 Aborted ${dir}$tst
FAIL: t-gcd
PASS: t-gcd_ui
PASS: t-lcm

PASS: t-export
PASS: t-pprime_p
PASS: t-perfpow
=============================================================
1 of 57 tests failed
Please report to http://groups.google.co.uk/group/mpir-devel/
=============================================================

box1

PASSED CC=gcc CXX=g++ configure=
PASSED CC=gcc CXX=g++ configure=--enable-cxx --enable-gmpcompat
FAILED CC=gcc CXX=g++
configure=--enable-cxx --enable-gmpcompat --enable-assert --enable-alloca=debug

[Bill Hart]

I followed rule number 63 in the book of rules on programming:

"When the assert fails, remove the assert."

That fixed it.

[Bill Hart]

I'm afraid I still get this on sage.math:

=== configuring in yasm (/home/wbhart/mpir-trunk/yasm)
configure: running /bin/bash ./configure '--prefix=/usr/local'
--cache-file=/dev/null --srcdir=.
configure: error: cannot find install-sh or install.sh in config "."/config
configure: error: ./configure failed for yasm


Bill.

2009/7/25 Bill Hart <goodwi...@googlemail.com>:

[Jason Moxham]

Yeah , corrected , they were symlinks , not actual files , what a useful
feature , assumes everyone has autotools

[Jason Moxham]

but I get this now , from a generic C build

test_stuff/../../trunk -DOPERATION_sub_err2_n -DNO_ASM -c
sub_err2_n.c -fPIC -DPIC -o .libs/sub_err2_n.o
add_err2_n.c: In function '__gmpn_add_err2_n':
add_err2_n.c:51: error: 'yp' undeclared (first use in this function)
add_err2_n.c:51: error: (Each undeclared identifier is reported only once
add_err2_n.c:51: error: for each function it appears in.)
make[2]: *** [add_err2_n.lo] Error 1
make[2]: *** Waiting for unfinished jobs....

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_sub_err1_n -DNO_ASM -c
sub_err1_n.c -o sub_err1_n.o >/dev/null 2>&1

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_mod_34lsub1 -DNO_ASM -c
mod_34lsub1.c -fPIC -DPIC -o .libs/mod_34lsub1.o
sub_err2_n.c: In function '__gmpn_sub_err2_n':
sub_err2_n.c:52: error: 'yp' undeclared (first use in this function)
sub_err2_n.c:52: error: (Each undeclared identifier is reported only once
sub_err2_n.c:52: error: for each function it appears in.)
make[2]: *** [sub_err2_n.lo] Error 1

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_mod_1 -DNO_ASM -c
mod_1.c -fPIC -DPIC -o .libs/mod_1.o

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_fib2_ui -DNO_ASM -c
fib2_ui.c -o fib2_ui.o >/dev/null 2>&1

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_mod_34lsub1 -DNO_ASM -c
mod_34lsub1.c -o mod_34lsub1.o >/dev/null 2>&1

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_mod_1 -DNO_ASM -c
mod_1.c -o mod_1.o >/dev/null 2>&1

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_mode1o -DNO_ASM -c
mode1o.c -fPIC -DPIC -o .libs/mode1o.o

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I/root/mpir/mpir/branches/test_stuff/../../trunk/mpn -I.. -D__GMP_WITHIN_GMP -I/root/mpir/mpir/branches/test_stuff/../../trunk -DOPERATION_mode1o -DNO_ASM -c
mode1o.c -o mode1o.o >/dev/null 2>&1
make[2]: Leaving directory `/root/mpir/mpir/branches/test_stuff/box1/mpn'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/mpir/mpir/branches/test_stuff/box1'
make: *** [all] Error 2

box1

PASSED CC=gcc CXX=g++ configure=
PASSED CC=gcc CXX=g++ configure=--enable-cxx --enable-gmpcompat
PASSED CC=gcc CXX=g++

configure=--enable-cxx --enable-gmpcompat --enable-assert --enable-alloca=debug

FAILED CC=gcc CXX=g++
configure=--enable-cxx --enable-gmpcompat --enable-assert --enable-alloca=debug --build=none-unknown-linux-gnu

[Bill Hart]

Seem to be some wrong asserts there. I think I fixed them now.

Rule 63 did not apply, as rule 64 takes precedence:

"The same bug fix will never work twice in succession."

[Jason Moxham]

Rule 65 says , there is still more work to do!

make check on build none fails with

/root/mpir/mpir/branches/test_stuff/box1/.libs/libmpir.so: undefined reference
to `__gmpn_subadd_n'
/root/mpir/mpir/branches/test_stuff/box1/.libs/libmpir.so: undefined reference
to `__gmpn_addlsh1_n'

I expect we have an unconditional use of subadd ie no HAVE_NATIVE check around
it

[Bill Hart]

Oh that's in the toom code, I took it out because I thought you said
these functions were defined unconditionally on all arches.

[Jason Moxham]

On Saturday 25 July 2009 11:05:19 Bill Hart wrote:
> Oh that's in the toom code, I took it out because I thought you said
> these functions were defined unconditionally on all arches.

not these functions , but some others , anything in mpn/generic plus any
macros in gmp-impl gmp-h.in
lshift1 rshift1 lshift2 rshift2 zero store