Re: ptrace single-stepping change breaks Wine

From: Linus Torvalds
Date: Wed Dec 29 2004 - 20:57:27 EST




On Wed, 29 Dec 2004, Davide Libenzi wrote:
>
> That test went in to be able to have ptrace single step, to see even the
> instruction following the #int instruction (this was the target of the
> patch itself). I just verified that, in 2.6.8 that does not have such test
> anymore, the single-step-after-int capability is lost.

Ahh. That's because of a separate bug: we have this silly separation of
"_TIF_WORK_MASK" (everything but tracing) and "_TIF_ALLWORK_MASK"
(everything), and the system call stuff takes over _TIF_SINGLESTEP for
some very non-obvious reasons.

I don't see why the system-call code thinks _TIF_SINGLESTEP is special,
but it certainly explains why it doesn't get handled normally.

So the updated patch would look something like the appended.

Will test whether it cleanly handles your test-case. Davide - you also
added the TIF_SINGLESTEP flag to that _TIF_WORK_MASK, can you explain
that?

(And yes, I know you'd sent me the test-program before, but I'm about as
organized as a Performing Seal with Alzheimers..)

Linus

--- 1.23/include/asm-i386/thread_info.h 2004-11-18 23:03:11 -08:00
+++ edited/include/asm-i386/thread_info.h 2004-12-29 17:52:16 -08:00
@@ -153,7 +153,7 @@

/* work to do on interrupt/exception return */
#define _TIF_WORK_MASK \
- (0x0000FFFF & ~(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT|_TIF_SINGLESTEP))
+ (0x0000FFFF & ~(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT))
#define _TIF_ALLWORK_MASK 0x0000FFFF /* work to do on any return to u-space */

/*
--- 1.28/arch/i386/kernel/ptrace.c 2004-11-22 09:44:52 -08:00
+++ edited/arch/i386/kernel/ptrace.c 2004-12-29 17:36:41 -08:00
@@ -568,15 +568,13 @@
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
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/