Re: Odd linker errors

Kris Kennaway wrote:
> On Sun, Jul 04, 2004 at 01:57:44AM -0600, Scott Long wrote:
> 
>>Kris Kennaway wrote:
>>
>>>A number of ports are starting to fail with errors like:
>>>
>>>net/clusterit:
>>>
>>>
>>>
>>>>cc -O -pipe -DCLUSTERS -o dsh dsh.o ../common/common.o
>>>>dsh.o(.text+0x86): In function `main':
>>>>: undefined reference to `malloc_options'
>>>
>>The addition of files/patch-dsh-dsh.c seems to be triggering this.
>>It appears to declare the 'malloc_options' symbol.  Looking at stdlib.h,
>>the real name should be '_malloc_options' (and including stdlib.h is the
>>better way to get this declaration).  This is also documented correctly
>>in the malloc(2) manpage.
> 
> 
> Ah, okay..I was checking 4.x's malloc(3), which says:
> 
> char * malloc_options;
> 

Would it make sense to revert HEAD back to 'malloc_options'?  It looks
like it was changed in rev 1.43 of stdlib.h with a rather vague commit
message.

> 
>>>net/mopd:
>>>
>>>
>>>
>>>>file.o(.text+0x623): In function `GetAOutFileInfo':
>>>>: undefined reference to `le32toh'
>>>>file.o(.text+0x63b): In function `GetAOutFileInfo':
>>>>: undefined reference to `le32toh'
>>>
>>I can't reproduce this here.  However, netbooting a VAX probably isn't
>>a terribly common occurrance these days.
> 
> 
> I couldn't even find where it was calling le32toh.
> 
> Kris
> 

Are le32toh and similar functions supposed to be visible to userland?  I 
thought that they were only visible to the kernel.  Either way, 
something isn't right with your environment.

Scott