qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] configure flag for compilation question..


From: Martin Bochnig
Subject: Re: [Qemu-devel] configure flag for compilation question..
Date: Sat, 14 Oct 2006 07:23:07 +0200
User-agent: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:1.7) Gecko/20041221

Ben Taylor wrote:

>---- Martin Bochnig <address@hidden> wrote: 
>  
>
>>Ben Taylor wrote:
>>
>>    
>>
>>>---- Ishwar Rattan <address@hidden> wrote: 
>>> 
>>>
>>>      
>>>
>>>>Trying to compile qemu on amd64 based Solaris.
>>>>
>>>>I do not have write permission to /usr/local subtree
>>>>
>>>>./configure --libdir=other-path --prefix=not-ustlocal
>>>>is fine
>>>>but make always generates binaries that want to find
>>>>/usr/local/lib/libSDL-1.2.so.0 etc. (checked with ldd).
>>>>
>>>>What is the way out for this sticky point?
>>>>   
>>>>
>>>>        
>>>>
>>>Manually add "-L/usr/local/lib -R/usr/local/lib" 
>>>
>>>      
>>>
>>Rather "-L/his/home/sdl_amd64/lib -R/his/home/sdl_amd64/lib"
>>    
>>
>
>Yeah, that.  brain fart when I was typing that.
>
>  
>
>>Because I doubt, an amd64 version of libSDL is currently present in
>>/usr/local/lib/amd64
>>(and he doesn't have w access).
>>    
>>
>
>Right. I knew what I meant, I just didn't type it right.
>
>  
>
>>This method is btw not really "new" to me, see my posting from a few
>>hours ago:
>>http://www.opensolaris.org/jive/thread.jspa?threadID=15448&tstart=0
>>
>>
>>    
>>
>>>to the Makefile for the link
>>>phase so it will correctly add those paths to the library lookup.  If I had
>>>a code base to look at this instance, I could tell you where.  You could
>>>also add those flags to Makefile.target in the SOLARIS specific areas,
>>>which would probably make more sense.
>>>
>>>As Martin indicated, setting LD_LIBRARY_PATH may get you a running
>>>binary, but LD_LIBRARY_PATH is the wrong answer for Solaris.
>>>
>>>Ben
>>> 
>>>
>>>      
>>>
>>>"LD_LIBRARY_PATH is the wrong answer for Solaris"
>>>      
>>>
>>???
>>--->>  Weak statement.
>>    
>>
>
>LD_LIBRARY_PATH is the wrong answer for Solaris.  It's great for
>trying to fix problems like this when the app won't run with a particular
>set of libraries, or you're testing with new libraries and you don't
>want to recompile  (I've done this before).  Too many people use
>it as a panacea for poorly compiled/configured Open source software.
>In this case, he might just set the LD_LIBRARY_PATH and never
>recompile it, even though it won't work if it uses the default LIB path.
>  
>

No further comment, except this one:

bash-3.00# pwd
/opt/SUNWqemu/bin
bash-3.00# ls -al
total 676
dr-xr-xr-x   5 root     bin         1024 Oct 14 06:26 .
drwxr-xr-x   5 root     bin          512 Oct 14 06:26 ..
lrwxrwxrwx   1 root     root           9 Oct 14 06:26 JQEMU.jar -> jqemu.jar
lrwxrwxrwx   1 root     root          10 Oct 14 06:26 JQEMU.jnlp ->
jqemu.jnlp
lrwxrwxrwx   1 root     root          20 Oct 14 06:26
SWING-LAYOUT-1.0.jar -> swing-layout-1.0.jar
-r-xr-xr-x   1 root     bin           48 Oct 14 03:23 jqemu-launcher-gui
-r-xr-xr-x   1 root     bin       150854 Oct 14 03:30 jqemu.jar
-r-xr-xr-x   1 root     bin          524 Oct 14 03:32 jqemu.jnlp
lrwxrwxrwx   1 root     root          18 Oct 14 06:26 qemu ->
./sparcv8plus/qemu
lrwxrwxrwx   1 root     root          22 Oct 14 06:26 qemu-img ->
./sparcv8plus/qemu-img
lrwxrwxrwx   1 root     root          29 Oct 14 06:26 qemu-system-586 ->
./sparcv8plus/qemu-system-586
lrwxrwxrwx   1 root     root          29 Oct 14 06:26 qemu-system-686 ->
./sparcv8plus/qemu-system-686
lrwxrwxrwx   1 root     root          29 Oct 14 06:26 qemu-system-arm ->
./sparcv8plus/qemu-system-arm
lrwxrwxrwx   1 root     root          30 Oct 14 06:26 qemu-system-mips
-> ./sparcv8plus/qemu-system-mips
lrwxrwxrwx   1 root     root          32 Oct 14 06:26 qemu-system-mipsel
-> ./sparcv8plus/qemu-system-mipsel
lrwxrwxrwx   1 root     root          29 Oct 14 06:26 qemu-system-ppc ->
./sparcv8plus/qemu-system-ppc
lrwxrwxrwx   1 root     root          31 Oct 14 06:26 qemu-system-sparc
-> ./sparcv8plus/qemu-system-sparc
lrwxrwxrwx   1 root     root          32 Oct 14 06:26 qemu-system-x86_64
-> ./sparcv8plus/qemu-system-x86_64
-r-xr-xr-x   1 root     bin         1591 Oct 14 00:00 sdl-config
dr-xr-xr-x   2 root     bin          512 Oct 14 06:26 sparcv7
dr-xr-xr-x   2 root     bin          512 Oct 14 06:26 sparcv8
dr-xr-xr-x   2 root     bin          512 Oct 14 06:26 sparcv8plus
-r-xr-xr-x   1 root     bin       140545 Oct 14 03:30 swing-layout-1.0.jar
-r-xr-xr-x   1 root     bin          332 Oct 14 05:39 test-qemu
-r-xr-xr-x   1 root     bin          252 Oct 14 05:40 test-qemu-system-arm
-r-xr-xr-x   1 root     bin          887 Oct 14 04:49 test-qemu-system-mips
-r-xr-xr-x   1 root     bin          319 Oct 14 04:54 test-qemu-system-sparc
bash-3.00# ldd */qemu-system-mips|grep SDL
        libSDL-1.2.so.0 =>       /opt/SUNWqemu/lib/sparcv7/libSDL-1.2.so.0
        libSDL-1.2.so.0 =>       /opt/SUNWqemu/lib/sparcv8/libSDL-1.2.so.0
        libSDL-1.2.so.0 =>      
/opt/SUNWqemu/lib/sparcv8plus/libSDL-1.2.so.0
bash-3.00# ldd */qemu-system-mips|grep libz
        libz.so =>       /opt/SUNWqemu/lib/sparcv7/libz.so
        libz.so =>       /opt/SUNWqemu/lib/sparcv8/libz.so
        libz.so =>       /opt/SUNWqemu/lib/sparcv8plus/libz.so
bash-3.00# uname -a
SunOS mb1x-ws1 5.11 snv_41 sun4u sparc SUNW,Sun-Fire-280R
bash-3.00# isainfo -v
64-bit sparcv9 applications
        vis2 vis
32-bit sparc applications
        vis2 vis v8plus div32 mul32
bash-3.00#

> 
>  
>
>>It has its [dis]advantages.
>>Namely that the paths to a library are _not_ hardwired.
>>    
>>
>
>But then that requires a dependency on knowing where your
>libraries are, and could possibly create an imcombaility by
>overriding a library path for some other application that is
>also leaning on LD_LIBRARY_PATH.  Obviously you can
>get around this by encapsulating each application requiring
>LD_LIBRARY_PATH, thereby negating the need for a user
>or system wide LD_LIBRARY_PATH variable.
>
>  
>
>>That's the exactly reason, why I did recommend it in this very scenario.
>>    
>>
>
>In this case, I can agree for the purposes of a test.  But it also indicates
>that some work is required for the Solaris port if other library paths
>need to be linked in during the compliation/link phase.
>
>  
>
>>And I would do it again for Ishwar's current needs.
>>    
>>
>
>Works for me.
>
>Ben
>
>
>  
>





reply via email to

[Prev in Thread] Current Thread [Next in Thread]