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

Optware when used on nslu2 unslung firmware and few more platforms, it supports both native and cross build.
Right now for dns-323, optware only supports cross compilation with linux PC as host.
But on the other hand, setting up a cross compile environment with optware is very easy to do, very automated, you'll save tons of time compiling.

Which php and web-server are you using?
If you're using ffp web server, it probably won't automagically load optware php.
I suggest you just use optware lighttpd and optware php-fcgi. Test a simple phpinfo page. See http://www.nslu2-linux.org/wiki/HowTo/D … ingFastCGI

You probably should start a separate thread.

Last edited by bzhou (2008-04-08 09:18:56)

The files available from the optware feed are all designed to work together, and to keep all configuration, libraries and binaries under the /opt directory.

It will be easiest for you if you don't install the same package in to fonz fun_plug AND /opt at the same time. Additionally, it will be harder work mixing packages that depend on each other like lighttpd and php - you will need to change more configuration settings and will probably be the first person to ever test such a set up!

Fonz' fun_plug script obviously sets up the paths needed for the packages that he distributes. The simplest way to use services from optware will be for you to make a change to the fun_plug script that sets up the optware /opt mount, adds the optware directories to PATH and then starts the required daemons. (Or if you want to be neater, don't make the changes directly in fun_plug, but put them in a script in the fun_plug.d/start directory and make that script executable.

If you think that you will mainly be using optware packages, then stick to Fonz' 0.4 fun_plug - 0.5 uses it's own ulibc which will take up a little extra memory compared with the 0.4 version which uses the ulibc that comes built in to the DNS323 (I have no idea how much - maybe it's trivial?!), and of course 0.5 is still experimental.

Last edited by sjmac (2008-04-09 10:54:57)

The optware binary has been compiled to expect the libraries to be in /opt/lib, so libraries in ffp or fon_plug.d  shouldn't be a problem, but again, see my suggestion to avoid havin the same thing installed in optware and fun_plug, and also my suggestion to favour fonz. 0.4 version.

You should use /opt/bin if you need to specify the full path - that means that you have to have to have used the 'mount --bind' command to create /opt

Last edited by sjmac (2008-04-09 14:21:44)

rap wrote:

mount -o bind /mnt/HD_a2/ipkg/opt /opt

Yes, that needs to be run once after each reboot if you want any optware installs to work correctly.

rap wrote:

Any other ideas as to what may be wrong?

What do your configurations and startup scripts look like?

How does clutch get started?

I just installed it, but have no idea how to test it ...

A bug has been filed for this here, I suggest you revert to Transmission 1.06 for now.

Problem building mediatomb to test

I've tried to make mediatomb-ipk and hit a few snags.

First was the file.mk change noted in the comment above.

Then I had to upgrade to a more recent automake than I had on my CentOS 4.4 box - I was surprised it was using a native tool - I expected everything to be in the tool-chain. Is it OK/expected to use the local/native configure tools? I grabbed the most recent tar.gzs for m4, autoconf and automake and installed them on my dev machine.

Finally (and now I'm stuck) I get this error at the end of the configure step for mediatomb:

checking tag.h presence... yes
checking for tag.h... yes
checking audioproperties.h usability... yes
checking audioproperties.h presence... yes
checking for audioproperties.h... yes
checking tstring.h usability... yes
checking tstring.h presence... yes
checking for tstring.h... yes
checking textidentificationframe.h usability... yes
checking textidentificationframe.h presence... yes
checking for textidentificationframe.h... yes
checking attachedpictureframe.h usability... yes
checking attachedpictureframe.h presence... yes
checking for attachedpictureframe.h... yes
checking for main in -ltag... no
configure: error: unable to configure taglib support
make: *** [/home/steven/dns323/optware/dns323/builds/mediatomb/.configured] Error 1

The make taglib-stage step seems to have completed (I did make taglib-ipk just to be sure), so what's going on? Does this build for anyone else?

I don't know how automake/autoconf works, but I looked in the configure script and in configure.ac and I couldn't find the definition of the test that was failing ( ac_cv_lib_tag_main ? )

Any clues?

No, probably it's compiled for the NSLUG2.

@sjmac: Why don't you try compiling with id3lib support instead of taglib? Taglib is relatively slow on our boxes anyway.

sjmac wrote:

I noticed that it doesn't get compiled by default, so I was just trying to find a fix to make it work as part of our "official"/experimental feed.

I will give the ID3 lib a try - do you know what the difference is?

I'm still confused about whether mediatomb.mk should be using my native 'automake' or if it should be using one built in to the opwtare/dns323 toolchain. I'll have another dig around this evening, unless anyone can tell me before then :-)

I just worked on make/id3lib.mk a little and got it build for optware/dns323, it should be in the feed now.

In make/mediatomb.mk, if you force
MEDIATOMB_CONFIG_ARGS=--enable-id3lib --disable-taglib

add DEPENDS and id3lib-stage, mediatomb should build fine.

I got the same error when I --enable-taglib, and I haven't got time to investigate. Looks somewhat related to the id3lib.mk fix. I got the impression that taglib is a bit more actively developed than id3lib, taglib also claims to be faster than id3lib.

Just like make, it's perfectly ok to use automake/autotools on your build machine during cross compilation. Although strictly speaking, to minimize build dependencies, one could/should build these tools under host/ as well. But the version built using optware/dns323/toolchain is only meant to run on dns323.

Hi, i've installes the optware ipkg tool and installed php-thttpd. This is working, i get a nice "hello world" from my index.php.

Now I also want to install php-mysql.

when using ipkg i get the following error for the mysql package.

Installing all prepared tables
Illegal instruction
Installation of system tables failed!

Examine the logs in /opt/var/lib/mysql for more information.
You can also try to start the mysqld daemon with:
/opt/libexec/mysqld --skip-grant &
You can use the command line tool
/opt/bin/mysql to connect to the mysql
database and look at the grant tables:

shell> /opt/bin/mysql -u root mysql
mysql> show tables

Try 'mysqld --help' if you have problems with paths. Using --log
gives you a log in /opt/var/lib/mysql that may be helpful.

The latest information about MySQL is available on the web at
http://www.mysql.com
Please consult the MySQL manual section: 'Problems running mysql_install_db',
and the manual section that describes problems on your OS.
Another information source is the MySQL email archive.
Please check all of the above before mailing us!
And if you do mail us, you MUST use the /opt/bin/mysqlbug script!

I reproduce this by running mysql_install_db

I tried what is suggested:

# /opt/libexec/mysqld --skip-grant &
# /opt/bin/mysql -u root mysql
ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/tmp/mysql.sock' (2)
[1] + Illegal instruction        /opt/libexec/mysqld --skip-grant

... and now I'm stuck, can anybody tell me what i'm doing wrong here?

TIA

rejong wrote:

Hi, i've installes the optware ipkg tool and installed php-thttpd. This is working, i get a nice "hello world" from my index.php.
...

Code:

# /opt/bin/mysql -u root mysql
ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/tmp/mysql.sock' (2)
[1] + Illegal instruction        /opt/libexec/mysqld --skip-grant

... and now I'm stuck, can anybody tell me what i'm doing wrong here?

TIA

It's probably the mysql package and nothing wrong on your side.

A similar problem with openssl was solved by sjmac by lowering optimization level during compilation. I suspect this problem is again related to toolchain, since the same mysql.mk is used to cross build working mysql ipk's for about a dozen optware targets.

What you can help is to provide more information to locate exactly the place "illegal instruction" happens, maybe using strace.

Some one with optware cross build environment setup, who has a dns-323, are in a better position to troubleshoot and fix mysql ipk. I'd think we are pretty close, just need some determined extra push.

bzhou wrote:

What you can help is to provide more information to locate exactly the place "illegal instruction" happens, maybe using strace.

Some one with optware cross build environment setup, who has a dns-323, are in a better position to troubleshoot and fix mysql ipk. I'd think we are pretty close, just need some determined extra push.

I ran the following:

strace /opt/libexec/mysqld 2>&1

result in attachment

---

Attachments:
strace-mysqld.txt, Size: 13,131 bytes, Downloads: 585