Re: fsck seg fault failure on vmware -i386?

[yancm%sdf.lonestar.org@localhost]

> On Tue, Feb 09, 2010 at 06:49:07PM +0700, Robert Elz wrote:
>>   | command line.
>>
>> Thanks, that's useful to know.   But I had the impression the desire was
>> to get this to happen with a build using build.sh rather than directly
>> using make - build.sh creates an environment where the entire build
>> happens
>> in a consistent manner, almost regardless of the host environment, just
>> using make allows what is in the environment (like the host's
>> /usr/include)
>> to heavily influence what gets built and how.
>
> You can use .CURDIR for this purpose to set it conditionally...

I'm sorry to be thick, but all the advice I'm getting on doing
this is confusing me.

I want to build a debug version of libc (really just the libc/time
routines).

I only know how to build with build.sh and for the most part
am only superficially aware of what it does.

I have tools in /usr/tools
Object is /usr/obj
I have a complete distribution I build via (in /usr/src):
./build.sh -O /usr/obj -T /usr/tools distribution

What would I need to do to build just a debug version of libc
or libc with debug symbols for libc/time/*.c pieces?

I do not care so much if I use make or build.sh, but if
make, I need explicit directions that make allowances for
making sure I use my consistent tools and distribution files
above.

What do I add to src/lib/libc/Makefile and/or
src/lib/libc/time/Makefile?

Where would the resultant libc.a end up?
I assume it would end up in /usr/obj/lib/libc?

And how do I get my debug version of fsck_ffs to use the debug
version of libc?

What do I do with .CURDIR ? I looked in the build.sh and BUILDING files
and do not really find any clues.

I've got a reproducible core dump, but am struggling to deliver
the information to properly diagnose it.

thanks,
gene