Fix two subtle problems in PPC32 RTLD. The first is a concurrency issue

where long PLT calls in multi-threaded environments could end up with
incorrect jmptab values. The second is that, after the addition of extended
PLT support, I forgot to update the PLT icache synchronization code to cover
the extended PLT instead of just the basic PLT.

MFC after:	10 days

1 files changed, 5 insertions, 2 deletions

diff --git a/libexec/rtld-elf/powerpc/reloc.c b/libexec/rtld-elf/powerpc/reloc.c
index c90852f0f1ac..84feff36ab35 100644
--- a/libexec/rtld-elf/powerpc/reloc.c
+++ b/libexec/rtld-elf/powerpc/reloc.c
@@ -38,6 +38,7 @@
 #include <string.h>
 #include <unistd.h>
 #include <machine/cpu.h>
+#include <machine/cpufunc.h>
 #include <machine/md_var.h>
 
 #include "debug.h"
@@ -481,6 +482,7 @@ reloc_jmpslot(Elf_Addr *wherep, Elf_Addr target, const Obj_Entry *defobj,
 
 		jmptab = obj->pltgot + JMPTAB_BASE(N);
 		jmptab[reloff] = target;
+		powerpc_mb(); /* Order jmptab update before next changes */
 
 		if (reloff < PLT_EXTENDED_BEGIN) {
 			/* for extended PLT entries, we keep the old code */
@@ -489,7 +491,8 @@ reloc_jmpslot(Elf_Addr *wherep, Elf_Addr target, const Obj_Entry *defobj,
 
 			/* li   r11,reloff */
 			/* b    pltcall  # use indirect pltcall routine */
-			wherep[0] = 0x39600000 | reloff;
+
+			/* first instruction same as before */
 			wherep[1] = 0x48000000 | (distance & 0x03fffffc);
 			__syncicache(wherep, 8);
 		}
@@ -577,7 +580,7 @@ init_pltgot(Obj_Entry *obj)
 	 * Sync the icache for the byte range represented by the
 	 * trampoline routines and call slots.
 	 */
-	__syncicache(pltcall, 72 + N * 8);
+	__syncicache(obj->pltgot, JMPTAB_BASE(N)*4);
 }
 
 void