Avoid wrong unwind info in clone child

Avoid wrong unwind info in clone child

[Andreas Schwab]

The unwind info attached to the clone function is usually wrong for the
child.  This changes the ppc(64) implementation to terminate the FDE
before the syscall, similar to the i386 implementation.

Andreas.

2006-04-03  Andreas Schwab  <schwab@suse.de>

	* sysdeps/unix/sysv/linux/powerpc/powerpc32/clone.S: Terminate FDE
	before syscall.
	* sysdeps/unix/sysv/linux/powerpc/powerpc64/clone.S: Likewise.

--- sysdeps/unix/sysv/linux/powerpc/powerpc32/clone.S.~1.7.~	2006-01-11 10:24:55.000000000 +0100
+++ sysdeps/unix/sysv/linux/powerpc/powerpc32/clone.S	2006-03-26 20:35:33.000000000 +0200
@@ -84,6 +84,10 @@ ENTRY (BP_SYM (__clone))
 	mr	r6,r8
 	mr	r7,r9
 
+	/* End FDE now, because in the child the unwind info will be
+	   wrong.  */
+	cfi_endproc
+
 	/* Do the call.  */
 	DO_CALL(SYS_ify(clone))
 
@@ -138,6 +142,8 @@ L(parent):
 L(badargs):
 	li	r3,EINVAL
 	b	__syscall_error@local
+
+	cfi_startproc
 END (BP_SYM (__clone))
 
 weak_alias (BP_SYM (__clone), BP_SYM (clone))
--- sysdeps/unix/sysv/linux/powerpc/powerpc64/clone.S.~1.10.~	2006-03-17 11:37:26.000000000 +0100
+++ sysdeps/unix/sysv/linux/powerpc/powerpc64/clone.S	2006-03-31 15:45:48.000000000 +0200
@@ -81,6 +81,10 @@ ENTRY (BP_SYM (__clone))
 	mr	r6,r8
 	mr	r7,r9
 
+	/* End FDE now, because in the child the unwind info will be
+	   wrong.  */
+	cfi_endproc
+
 	/* Do the call.  */
 	DO_CALL(SYS_ify(clone))
 
@@ -132,6 +136,8 @@ L(parent):
 L(badargs):
 	li	r3,EINVAL
 	b	JUMPTARGET(__syscall_error)
+
+	cfi_startproc
 END (BP_SYM (__clone))
 
 weak_alias (BP_SYM (__clone), BP_SYM (clone))

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Avoid wrong unwind info in clone child

[Roland McGrath]

> The unwind info attached to the clone function is usually wrong for the
> child.  This changes the ppc(64) implementation to terminate the FDE
> before the syscall, similar to the i386 implementation.

That looks reasonable.  I put it in.  It would be nice if we had a test for
this, though it might be hard to put together in machine-independent C.


Thanks,
Roland