PERFORCE change 152429 for review

[Peter Wemm]

http://perforce.freebsd.org/chv.cgi?CH=152429

Change 152429 by peter at peter_overcee on 2008/11/03 21:53:10

	something in the toolchain on amd64 seems to generate call frame records
	that point to slightly before where valgrind things the text segment started.
	I assume this is PLT or something else.  Ignore it for now.  I don't think
	it can lead to the overlaps that the invariant checks are worried about.

Affected files ...

.. //depot/projects/valgrind/coregrind/m_debuginfo/debuginfo.c#7 edit

Differences ...

==== //depot/projects/valgrind/coregrind/m_debuginfo/debuginfo.c#7 (text+ko) ====

@@ -523,8 +523,11 @@
    di2 = NULL;
    /* invariant (2) */
    if (di->cfsi) {
+//VG_(printf)("cfsi_minavma %#lx cfsi_maxavma %#lx rx_map_avma %#lx rx_map_size %#lx end %#lx\n", di->cfsi_minavma, di->cfsi_maxavma, di->rx_map_avma, di->rx_map_size, di->rx_map_avma + di->rx_map_size);
       vg_assert(di->cfsi_minavma <= di->cfsi_maxavma); /* duh! */
+#if !defined(VGP_amd64_freebsd)	/* gcc does something wierd here */
       vg_assert(di->cfsi_minavma >= di->rx_map_avma);
+#endif
       vg_assert(di->cfsi_maxavma < di->rx_map_avma + di->rx_map_size);
    }
    /* invariants (3) and (4) */