make check fails on openSUSE 10.3 (x86_64) about gperftools HOT 4 CLOSED

What machine architecture are you running on: intel, powerpc, other?  See the 
top of
the INSTALL file for details about what tests do not yet pass on what systems.

If you could include the full output of the make check output, that may also be
helpful in figuring out what's going wrong.

The x86_64 architecture which is mentioned in the subject is an Intel 64-bit
architecture.

For the full output of make check, see also the attachment make-check-output.txt

Update: when running the command below, all 22 tests pass:
$ ./configure --enable-frame-pointer && make && make check
...
===================
All 22 tests passed
===================

Can the configure script please be modified such that on x86_64
--enable-frame-pointer is enabled automatically ?

Doh! -- I missed the subject line when looking for the machine info.  Sorry 
about that.

} Update: when running the command below, all 22 tests pass:
} $ ./configure --enable-frame-pointer && make && make check

Ah, I'm glad that fixed it.

} Can the configure script please be modified such that on x86_64
} --enable-frame-pointer is enabled automatically ?

Alas, it's not that simple; if you use --enable-frame-pointer, you also need to
ensure the other libraries you link in, including libc and libc++, have frame
pointers enabled.  This is not something the configure script can determine (as 
far
as I know).  It also requires that you compile all applications that link in
-ltcmalloc with -fno-omit-frame-pointers.  Overall, this is a brittle setting 
that
requires user awareness to work properly; it's not something best done 
automatically.

That said, --enable-frame-pointer should not be necessary if you're using 
libunwind,
which it sounds like you were doing.  Perhaps you were running into some of the
problems mentioned in the '64 bit notes' in the README.  It's hard to tell from 
the
logs you included unfortunately: I can see that the relevant tests are 
segfaulting,
but can't tell much about why.

If I have a chance I'll see I can reproduce this locally on my own x86_64 
development
machine.  Even if so, I'm not sure how much we can do to improve the situation. 
 The
fact x86_64 omits the frame pointer by default makes life difficult for us!  
But I
know libunwind is being actively developed in its own right, and improvements 
there
should benefit these tests as well.