[Bug ld/13991] powerpc-rtems ld failure [regression]

[joel.sherrill at oarcorp dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=13991

--- Comment #6 from Joel Sherrill <joel.sherrill at oarcorp dot com> 2012-04-23 
15:25:57 UTC ---
I managed to narrow down the time frame when this broke to between 2011-07-09
and 2011-07-11:

works - binutils 2.20.1 with rtems 4.10 patches
works - binutils 2.21.1 with no patches
broken - binutils 2.22 with no patches
works - 2.21.52.20110627
works - 2.21.52.20110701
works - 2.21.52.20110708
works - 2.21.52.20110709
broken - 2.21.52.20110710
broken - 2.21.52.20110714

The timestamps for the file changes are so close except for the gold changes
that I can't separate them. Ignoring the gold changes, I see two entries in the
bfd and ld ChangeLogs. This is the single ld ChangeLog entry:

+2011-07-09  H.J. Lu  <address@hidden>
+
+    PR ld/12942
+    * ldlang.c (section_already_linked): Pass "struct already_linked *"
+    to bfd_section_already_linked.
+    (lang_process): Set link_info.loading_lto_outputs before
+    loading LTO outputs.
+    * plugin.c: Include "libbfd.h".
+    (add_symbols): Call bfd_section_already_linked with comdat_key.
+

Since that mentions the phrase "already_linked" and our failure is on already
linked code, this looks like the culprit. I don't see any indication that a
test case like this one was tested for 12942.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.