MySQL Bugs: #30846: problem building mysql 5.0.45 from source on IA64

Bug #30846	problem building mysql 5.0.45 from source on IA64
Submitted:	5 Sep 2007 23:01	Modified:	23 Sep 2007 12:14
Reporter:	Don Cohen	Email Updates:
Status:	Not a Bug	Impact on me:	None
Category:	MySQL Server: Compiling	Severity:	S2 (Serious)
Version:	5.0.45	OS:	Linux
Assigned to:		CPU Architecture:	Any

Description:
ending with ...
g++ -O3 -DDBUG_OFF -fno-implicit-templates -fno-exceptions -fno-rtti -rdynamic -\
o .libs/mysql mysql.o readline.o sql_string.o completion_hash.o  ../cmd-line-uti\
ls/libedit/libedit.a -lncurses ../libmysql/.libs/libmysqlclient.so -lcrypt -lnsl\
 -lm -lz -Wl,--rpath -Wl,/home/dcohen/mysql/lib/mysql
collect2: ld terminated with signal 11 [Segmentation fault]
make[2]: *** [mysql] Error 1
make[2]: Leaving directory `/home/dcohen/mysql-5.0.45/client'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/dcohen/mysql-5.0.45'
make: *** [all] Error 2

===
I've stored the entire transcript if that will help, inc. all output from configure as well as make.

How to repeat:
I imagine you need a machine like mine - not all that common.

Thank you for a problem report. Please, send the results of:

gcc --version
uname -a
free

Attach your confgure and build logs also.

dcohen@saturn:~> gcc --version
gcc (GCC) 4.1.0 (SUSE Linux)
Copyright (C) 2006 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

dcohen@saturn:~> uname -a
Linux saturn 2.6.16.27-0.9-default #1 SMP Tue Feb 13 09:35:18 UTC 2007 ia64 ia6\
4 ia64 GNU/Linux
dcohen@saturn:~> free
             total       used       free     shared    buffers     cached
Mem:       3788320    3409664     378656          0          0    1770336
-/+ buffers/cache:    1639328    2148992
Swap:     16779856          0   16779856
dcohen@saturn:~>

Please, send the exact ./configure command you had used.

> Please, send the exact ./configure command you had used.
That's included in the transcript I uploaded:
./configure --prefix=/home/dcohen/mysql

I do not see any files uploaded to this bug report. Please, check again.

transcript of configure (inc preparation)

Attachment: configure-transcript (application/octet-stream, text), 47.22 KiB.

transcript of make attempt

Attachment: make-transcript (application/octet-stream, text), 487.37 KiB.

sorry, the original transcript was just > 500K
now split into two parts

I'm just wondering, is anything happening with this bug?
Is there an estimate of when I should expect to hear something?
Do you have an IA64 machine on which to try a build?
I'd be more than happy to try a pre built binary.
Do you have any suggestions for what I can do to debug this?

We have a SLES9 IA64 box. But it has gcc 3.x.y, not 4.1.0. Can you try to build with older gcc version, 3.3.x or 3.4.x?

Rather than trying to use an older gcc I sent to gcc bug list asking how to fix this version.  The reply was that the problem was in binutils.  I asked the binutils bug address and the first suggestion was to try the most recent version of binutils.  This seems to solve the problem.
I cannot install binutils as root, so I have to use -B in all the calls to gcc.
In order to continue the build I find that I have to edit all the Makefile's that have already been created.  If I want to try a clean build what should I pass in to .configure to add this argument?

Now that I have it all compiled I do make install, and, trying to follow the directions in http://dev.mysql.com/doc/refman/5.0/en/quick-install.html, I do
  bin/mysql_install_db 
(I'm not running as root so no --user=mysql)
The result is
 [ERROR] /home/dcohen/mysql/libexec/mysqld: unknown variable 'ssl-ca=/usr/local/mysql/mysql-test/cacert.pem'
My only guess is that this came from /etc/my.cnf, which I do *NOT* want to use.
In fact I have created my own version of the initialization file.  
The documentation page for mysql_install_db seems not to mention anything that helps me with this.  After looking at the install db script for a while I try
 bin/mysql_install_db --defaults-file=/home/dcohen/mysql/my.cnf

and still get the same error.
I'm hoping someone there knows what this is all about and can tell me what to do about it.

It was fooling me by removing the spaces around the =.
This was coming from my edited my.cnf.
http://dev.mysql.com/doc/refman/5.1/en/mysqld-option-tables.html seems to indicate that ssl-cert (for instance) is allowed in the initialization file but evidently this is not true.  Is that a bug?

Now I find that I cannot reach the server.
There's already a server running on port 3306 so I have to run this one on another port.  I use 3307.  The log file shows me that it's listening there.  But when I try to do
/home/dcohen/mysql/bin/mysqladmin --user=root --port=3307 password ...
or for that matter, 
/home/dcohen/mysql/bin/mysql --user=root --port=3307
I get 
ERROR 1045 (28000): Access denied for user 'root'@'localhost' (using password: NO)
In fact, I can see from the log file that these are being directed at the server on port 3306 !  So these commands are ignoring my port argument.
How can I reach this server.  What is its initial state anyway?  only user is root?  Allowed only from localhost and <machine name>?

Ok, solved that one --protocol=tcp
I guess I should add all these details to your doc pages!

One more problem - I found out why the ssl settings weren't working: the default configure doesn't build ssl !  Why not?  Your pre built binaries contain it (as I think they should).  I think it should even be enabled in the default configuration.  (I'm not proposing to disable non-encrypted communication.)

Anyhow, I consider this bug to now be solved.
I've added a few comments to the documentation.
Let me know if you'd like the results of my build for further distribution.

P.S. how about my other open bugs?

So, looks like this particular problem was not a result of a bug in MySQL's code.