Re: [PATCH v4 1/2] ptrace: save the type of syscall-stop in ptrace_message

[Dmitry V. Levin]

On Wed, Nov 28, 2018 at 03:20:06PM +0100, Oleg Nesterov wrote:
> On 11/28, Dmitry V. Levin wrote:
> > On Wed, Nov 28, 2018 at 02:49:14PM +0100, Oleg Nesterov wrote:
> > > On 11/28, Dmitry V. Levin wrote:
> > > >
> > > > +/*
> > > > + * These values are stored in task->ptrace_message by tracehook_report_syscall_*
> > > > + * to describe current syscall-stop.
> > > > + *
> > > > + * Values for these constants are chosen so that they do not appear
> > > > + * in task->ptrace_message by other means.
> > > > + */
> > > > +#define PTRACE_EVENTMSG_SYSCALL_ENTRY	0x80000000U
> > > > +#define PTRACE_EVENTMSG_SYSCALL_EXIT	0x90000000U
> > > 
> > > Again, I do not really understand the comment... Why should we care about
> > > "do not appear in task->ptrace_message by other means" ?
> > > 
> > > 2/2 should detect ptrace_report_syscall() case correctly, so we can use any
> > > numbers, say, 1 and 2?
> > > 
> > > If debugger does PTRACE_GETEVENTMSG it should know how to interpet the value
> > > anyway after wait(status).
> > 
> > Given that without this patch the value returned by PTRACE_GETEVENTMSG
> > during syscall stop is undefined, we need two different ptrace_message
> > values that cannot be set by other ptrace events to enable reliable
> > identification of syscall-enter-stop and syscall-exit-stop in userspace:
> > if we make PTRACE_GETEVENTMSG return 0 or any other value routinely set by
> > other ptrace events, it would be hard for userspace to find out whether
> > the kernel implements new semantics or not.
> 
> Hmm, why? Debugger can just do ptrace(PTRACE_GET_SYSCALL_INFO, NULL), if it
> returns EIO then it is not implemented?

The debugger that uses PTRACE_GET_SYSCALL_INFO does not need to call
PTRACE_GETEVENTMSG for syscall stops.
My concern here is the PTRACE_GETEVENTMSG interface itself.  If we use
ptrace_message to implement PTRACE_GET_SYSCALL_INFO and expose
PTRACE_EVENTMSG_SYSCALL_{ENTRY,EXIT} for regular PTRACE_GETEVENTMSG users,
it should have clear semantics.


-- 
ldv

Attachment: signature.asc
Description: PGP signature