Re: ptrace single-stepping change breaks Wine

[Jesse Allen <the3dfxdude@gmail.com>]

On Wed, 29 Dec 2004 16:44:05 -0800 (PST), Linus Torvalds
<torvalds@osdl.org> wrote:
> 
> That still leaves the problem of the clearing of TF_MASK. I _appears_ that
> the problem is that TF was set both by Wine doing a PTRACE_SINGLESTEP
> (since PT_DTRACE is set) _and_ the application having TF enabled in eflags
> from before (since it seems to want to keep it enabled).
> 
> How about something like the attachment for the TF_MASK issue (replacing
> your "don't clear" code)? The comments should make the intention pretty
> obvious, but I have equally obviously not actually _tested_ any of this..
> 

The following patch works for me.  I still have to remove
TIF_SINGLESTEP test.  I also went ahead and tried adding the other fix
on the ptrace_notify call and _TIF_WORK_MASK, but for some reason,
changing _TIF_WORK_MASK seems to break the program now.  This patch
does fix the TF clearing problem.

Jesse


--- linux/arch/i386/kernel/ptrace.c	2004-12-29 14:10:34.000000000 -0700
+++ linux-mod/arch/i386/kernel/ptrace.c	2004-12-29 20:50:00.000000000 -0700
@@ -142,18 +142,31 @@
 {
 	long eflags;
 
+	/*
+	 * Always set TIF_SINGLESTEP - this guarantees that 
+	 * we single-step system calls etc.. 
+	 */
 	set_tsk_thread_flag(child, TIF_SINGLESTEP);
+
+	/*
+	 * If TF was already set, don't do anything else
+	 */
 	eflags = get_stack_long(child, EFL_OFFSET);
+	if (eflags & TRAP_FLAG)
+		return;
 	put_stack_long(child, EFL_OFFSET, eflags | TRAP_FLAG);
 	child->ptrace |= PT_DTRACE;
 }
 
 static void clear_singlestep(struct task_struct *child)
 {
+	/* Always clear TIF_SINGLESTEP... */
+	clear_tsk_thread_flag(child, TIF_SINGLESTEP);
+
+	/* But touch TF only if it was set by us.. */
 	if (child->ptrace & PT_DTRACE) {
 		long eflags;
 
-		clear_tsk_thread_flag(child, TIF_SINGLESTEP);
 		eflags = get_stack_long(child, EFL_OFFSET);
 		put_stack_long(child, EFL_OFFSET, eflags & ~TRAP_FLAG);
 		child->ptrace &= ~PT_DTRACE;
@@ -568,15 +581,13 @@
 			audit_syscall_exit(current, regs->eax);
 	}
 
-	if (!test_thread_flag(TIF_SYSCALL_TRACE) &&
-	    !test_thread_flag(TIF_SINGLESTEP))
+	if (!test_thread_flag(TIF_SYSCALL_TRACE))
 		return;
 	if (!(current->ptrace & PT_PTRACED))
 		return;
 	/* the 0x80 provides a way for the tracing parent to distinguish
 	   between a syscall stop and SIGTRAP delivery */
-	ptrace_notify(SIGTRAP | ((current->ptrace & PT_TRACESYSGOOD) &&
-				 !test_thread_flag(TIF_SINGLESTEP) ? 0x80 : 0));
+	ptrace_notify(SIGTRAP | ((current->ptrace & PT_TRACESYSGOOD) ? 0x80 : 0));
 
 	/*
 	 * this isn't the same as continuing with a signal, but it will do