tech-toolchain archive

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

Re: Confused about MKBUILD=yes

On Sun, Mar 07, 2010 at 01:33:43AM -0500, Richard Hansen wrote:
> While investigating a segfault in libpam on the netbsd-5 branch, I  
> discovered that MKDEBUG=yes (see mk.conf(5)):
>   * does not install debugging symbols for libraries
>   * does not install any debug symbols at all (because it does not cause 
> -g to be passed to gcc)
>   * does not strip debug symbols from installed binaries
> This behavior is contrary to my expectations, and arguably contrary to  
> the documentation of MKDEBUG in mk.conf(5).

At least the first point sounds like a bug; the second one is somewhat
arguable (I would say it ought to change the default for $(DBG), but
someone else might say otherwise); and the third one is quite

> I'm new to NetBSD, and I'm not familiar with the standard ways the  
> NetBSD developer community debugs base libraries.  Is the approach I  
> tried to take obsolete?  Is this a bug/incomplete feature that has been  
> or is planned to be addressed in current?

I don't know. I suspect it's something nobody's thought about much in
years. The last time I needed to step through libc I built a copy with
DBG=-g and loaded it using LD_PRELOAD, which worked fine.

David A. Holland

Home | Main Index | Thread Index | Old Index