[Fwd: Re: [uClibc] compiling problem]

[Philippe PEZET]

I must first state that it is necessary to do a "make install" to get
the ELF saned file (from the script file).
doing the full complete sequence "./configure && make && make install"
does work in both glibc and uclibc environments.
problems arise when I try to do a i386 host/build compil.


David Meggy wrote:
> Can you send this to the mailing list
> 
> On Thu, 2004-02-26 at 12:10, Philippe PEZET wrote:
> 
>>I guess root_fs is the latest image : uClibc version is 0.9.24.
>>I would give you further infos :
>>- I'm trying to compile for an i386 (or i486) host and setting either 
>>build=i386 or host=i386 brings this huge saned file.
>>- however trying to compile with plain ./configure (under AMD K7) gives 
>>only a saned script (not an ELF thing) : looking at the saned script 
>>only indicates that probably not all libtools were installed ??
>>
>>David Meggy wrote:
>>
>>>On Thu, 2004-02-26 at 10:30, Philippe PEZET wrote:
>>>
>>>
>>>>I wanted to compile sane-backends (SANE) latest version tarball under 
>>>>uclibc.
>>>>compiling with my regular config (glibc-2.3.2) yields a sane daemon 
>>>>(saned) of size 4.3 kB.
>>>>I downloaded root_fs_i386, mounted it loop, installed sane-backends 
>>>>tarball, chrooted to the mounted directory and compiled sane-backends. 
>>>>sane daemon (saned) is now 4 MB (yes 4000 kB, and still 1.9 MB stripped) !!!
>>>>what am I doing wrong ?
>>>
>>>
>>>Something doesn't sound right at all.  That is an order of magnitude
>>>different.
>>>
>>>First thing to check is whether or not you are using static or dynamic
>>>libraries.  Dynamically linked programs will be much smaller.  Also when
>>>you compile SANE, are optimizing for space?  You should have CFLAGS
>>>contain "-Os".
>>>
>>>Did you compile additional debugging support into the binary?  Maybe
>>>something that couldn't be eliminated by strip?
>>>
>>>The root_fs_i386 you downloaded, is it the latest?  What version of
>>>uClibc does it contain? and how was it compiled?  (The answer to the
>>>first question, will probably lead to the next 2)
>>>
>>>David
>>>
>>