Re: [patch] [new testcase] Regression on qsort_cmp [Re: status of Darwin support]

[Paul Pluzhnikov <ppluzhnikov@google.com>]

On Tue, Aug 25, 2009 at 1:47 AM, Jan
Kratochvil<jan.kratochvil@redhat.com> wrote:

> Requesting approval for the testcase check-in.
> Tested on {x86_64,i686}-fedorarawhide-linux-gnu:
> FAIL: gdb.base/solib-overlap.exp: 0x40000000: attach (GDB internal error)
> FAIL: gdb.base/solib-overlap.exp: 0x50000000: attach (GDB internal error)

Thanks for the test. It nicely reproduces the problem, and is fixed by the
"remove assertions and deal with possible unexpected overlaps" patch I am
working on :)

I do have a general question on testing shlibs though:

In order to run this test on x86_64, I had to do this:

-    set lib_flags {debug}
+    set lib_flags {debug additional_flags=-fpic}

Otherwise, I get
  gdb.base/solib-overlap-lib.c.o: relocation R_X86_64_32 against `a
local symbol' can not be used when making a shared object; recompile
with -fPIC
and the test goes UNTESTED.

Shouldn't gdb_compile_shlib be adding '-fpic' automatically?
Or is the above correct fix for that test?

Thanks,
-- 
Paul Pluzhnikov