DSM-G600, DNS-3xx and NSA-220 Hack Forum

Unfortunately no one can be told what fun_plug is - you have to see it for yourself.

You are not logged in.

Announcement

#1 2011-07-14 11:58:35

pavelc
Member
Registered: 2011-05-03
Posts: 10

i try compile tthsum

I added compiling capabilities to my DNS-323

funpkg -i binutils-2.18.50.0.1-4.tgz automake-1.10.1-2.tgz autoconf-2.61-2.tgz flex-2.5.33-5.tgz bison-2.3-3.tgz patch-2.5.9-3.tgz gcc-4.1-2.tgz gettext-0.17-2.tgz make-3.81-3.tgz uclibc-0.9.29-7.tgz kernel-headers-2.6.9.1-2.tgz

I downloaded tthsum-1.2.1 from http://tthsum.devs.nu/

modified Makefile

CFLAGS += -fsigned-char -Os -mtune=arm9e -march=armv5te # was -O3 -march=native
LDFLAGS += -fsigned-char -Os -mtune=arm9e -march=armv5te #was -O3 -march=native

It compiled. Passed programs tests
But counts tth wrong.

How do i know what is problem?
In LITTLE-BIG-ENDIANS, signed-unsigned, in options or something else in sources:)?

Offline

 

#2 2011-07-15 15:59:02

pavelc
Member
Registered: 2011-05-03
Posts: 10

Re: i try compile tthsum

i discussed with author of program
he add some new tests.
And program works fine on my win32/athlon (as i know it has also little endian)
and failed on dns-323

Code:

#gcc -v
Using built-in specs.
Target: arm-unknown-linux-uclibc
Configured with: /home/bob/funplug/work/gcc-4.1/gcc-4_1-branch/configure --enable-languages=c,c++ --enable-__cxa_atexit --with-local-prefix=/ffp --disable-multilib --disable-libssp --disable-libgomp --disable-libmudflap --disable-tls --with-float=soft --disable-bootstrap
Thread model: posix
gcc version 4.1.3 20080414 (prerelease)

# uname -a
Linux bob 2.6.12.6-arm1 #32 Wed Jun 24 15:19:48 CST 2009 armv5tejl unknown

Maybe i forgot some compile keys?

Last edited by pavelc (2011-07-15 15:59:55)

Offline

 

#3 2011-07-17 01:53:56

pavelc
Member
Registered: 2011-05-03
Posts: 10

Re: i try compile tthsum

author solved problem
problem was in 8bit/64bit alignment
ARM needs compiler key -DREQUIRE_ALIGNMENT also
(work good with -fsigned-char and without -O keys)
from sources
/* The tiger_bp function uses unaligned 64 bit memory access on little
* endian platforms. Tests on my x86_64 incur a 3.5% performance penalty
* when we force aligned access. However, some platforms need alignment
* to function properly (like the ARM).
*
* I tested doing 1024 byte reads on memory aligned such that it was
* properly aligned for the tiger_bp function, but that caused the
* system time to double: 6% performance penalty. */

p.s. http://tthsum.devs.nu/  tthsum-1.3.0-rc2 works well without any modifications

Last edited by pavelc (2011-07-20 01:51:49)

Offline

 

Board footer

Powered by PunBB
© Copyright 2002–2010 PunBB