ptrace(PTRACE_SYSCALL) and data field

ptrace(PTRACE_SYSCALL) and data field

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 332 bytes --]

the ptrace(2) page doesnt say anything about the data field (ignored or 
otherwise).  it seems to be ignored, but i didnt trace the kernel 
exhaustively.  the examples found via google shows that it's ignored.  strace 
usage seems odd (it passes in an irrelevant pointer), which would indicate 
it's ignored as well ...
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: ptrace(PTRACE_SYSCALL) and data field

[Michael Kerrisk]

Hi Mike,

(Roland may be able to help out here.)

On Wed, Mar 18, 2009 at 4:46 PM, Mike Frysinger <vapier-aBrp7R+bbdUdnm+yROfE0A@public.gmane.org> wrote:
> the ptrace(2) page doesnt say anything about the data field (ignored or
> otherwise).  it seems to be ignored, but i didnt trace the kernel
> exhaustively.  the examples found via google shows that it's ignored.  strace
> usage seems odd (it passes in an irrelevant pointer), which would indicate
> it's ignored as well ...

It's true that this looks like a defect in the man page.  But I'm not
sure what the fix is.

In the kernel source file kernel/ptrace.c, function ptrace_resume() we
have the line

            child->exit_code = data;

which seems to be executed by the PRACE_SYSCALL code path.  That
implies that data is treated as a signal number (like PTRACE_CONT).

That seems consistent with the following in the strace.c source file
of strace(1):


                         if (ptrace(PTRACE_SYSCALL, pid, (char *) 1,
                                   WSTOPSIG(status)) < 0) {
                                perror("trace: ptrace(PTRACE_SYSCALL, ...)");
                                cleanup();
                                return -1;
                        }

Did I misunderstand something?  Your thoughts?

Cheers,

Michael
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Roland McGrath]

All of the "resuming" ptracer requests treat the data argument the same.
Those are PTRACE_CONT, PTRACE_SYSCALL, PTRACE_*STEP, PTRACE_SYSEMU*.
The data argument is either 0 or a signal number.
The addr argument is ignored by all these.


Roland
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Michael Kerrisk]

On Mon, Mar 30, 2009 at 11:35 AM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> All of the "resuming" ptracer requests treat the data argument the same.
> Those are PTRACE_CONT, PTRACE_SYSCALL, PTRACE_*STEP, PTRACE_SYSEMU*.
> The data argument is either 0 or a signal number.
> The addr argument is ignored by all these.

Thanks for jumping in Roland.

So, for PTRACE_SYSCALL, the sameantics are like PTRACE_CONT, that is,
it is interpreted as a signal to be sent to the child.  Right?

Cheers,

Michael

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
git://git.kernel.org/pub/scm/docs/man-pages/man-pages.git
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Roland McGrath]

> On Mon, Mar 30, 2009 at 11:35 AM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> > All of the "resuming" ptracer requests treat the data argument the same.
> > Those are PTRACE_CONT, PTRACE_SYSCALL, PTRACE_*STEP, PTRACE_SYSEMU*.
> > The data argument is either 0 or a signal number.
> > The addr argument is ignored by all these.
> 
> Thanks for jumping in Roland.
> 
> So, for PTRACE_SYSCALL, the sameantics are like PTRACE_CONT, that is,
> it is interpreted as a signal to be sent to the child.  Right?

For all those requests, yes.
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Michael Kerrisk]

On Mon, Mar 30, 2009 at 12:34 PM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> On Mon, Mar 30, 2009 at 11:35 AM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> > All of the "resuming" ptracer requests treat the data argument the same.
>> > Those are PTRACE_CONT, PTRACE_SYSCALL, PTRACE_*STEP, PTRACE_SYSEMU*.
>> > The data argument is either 0 or a signal number.
>> > The addr argument is ignored by all these.
>>
>> Thanks for jumping in Roland.
>>
>> So, for PTRACE_SYSCALL, the sameantics are like PTRACE_CONT, that is,
>> it is interpreted as a signal to be sent to the child.  Right?
>
> For all those requests, yes.

Thanks for the confirmation, Roland.

Cheers,

Michael


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
git://git.kernel.org/pub/scm/docs/man-pages/man-pages.git
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Michael Kerrisk]

On Mon, Mar 30, 2009 at 12:34 PM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> On Mon, Mar 30, 2009 at 11:35 AM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> > All of the "resuming" ptracer requests treat the data argument the same.
>> > Those are PTRACE_CONT, PTRACE_SYSCALL, PTRACE_*STEP, PTRACE_SYSEMU*.
>> > The data argument is either 0 or a signal number.
>> > The addr argument is ignored by all these.
>>
>> Thanks for jumping in Roland.
>>
>> So, for PTRACE_SYSCALL, the sameantics are like PTRACE_CONT, that is,
>> it is interpreted as a signal to be sent to the child.  Right?
>
> For all those requests, yes.

Hmmm.  The existing man page includes this text:

       PTRACE_SYSEMU, PTRACE_SYSEMU_SINGLESTEP (since Linux 2.6.14)
              ...
              (addr and data are ignored; ...

which according to what you've said (and also according to my reading
of the source), is wrong.  Correct?

Cheers,

Michael


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
git://git.kernel.org/pub/scm/docs/man-pages/man-pages.git
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Roland McGrath]

> Hmmm.  The existing man page includes this text:
> 
>        PTRACE_SYSEMU, PTRACE_SYSEMU_SINGLESTEP (since Linux 2.6.14)
>               ...
>               (addr and data are ignored; ...
> 
> which according to what you've said (and also according to my reading
> of the source), is wrong.  Correct?

Yes, those also treat the signal argument like PTRACE_CONT et al.


Thanks,
Roland
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Michael Kerrisk]

On Mon, Mar 30, 2009 at 1:07 PM, Roland McGrath <roland-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
>> Hmmm.  The existing man page includes this text:
>>
>>        PTRACE_SYSEMU, PTRACE_SYSEMU_SINGLESTEP (since Linux 2.6.14)
>>               ...
>>               (addr and data are ignored; ...
>>
>> which according to what you've said (and also according to my reading
>> of the source), is wrong.  Correct?
>
> Yes, those also treat the signal argument like PTRACE_CONT et al.

Thanks!


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
git://git.kernel.org/pub/scm/docs/man-pages/man-pages.git
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: ptrace(PTRACE_SYSCALL) and data field

[Michael Kerrisk]

On Wed, Mar 18, 2009 at 4:46 PM, Mike Frysinger <vapier-aBrp7R+bbdUdnm+yROfE0A@public.gmane.org> wrote:
> the ptrace(2) page doesnt say anything about the data field (ignored or
> otherwise).  it seems to be ignored, but i didnt trace the kernel
> exhaustively.  the examples found via google shows that it's ignored.  strace
> usage seems odd (it passes in an irrelevant pointer), which would indicate
> it's ignored as well ...
> -mike

Mike,

For man-pages-3.20, I've upated the PTRACE_SYSCALL documentation to
note that 'data' is used as for PTRACE_CONT.  I've also added similar
text to PTRAC_SYSEMU*.

Cheers,

Michael


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
git://git.kernel.org/pub/scm/docs/man-pages/man-pages.git
man-pages online: http://www.kernel.org/doc/man-pages/online_pages.html
Found a bug? http://www.kernel.org/doc/man-pages/reporting_bugs.html
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html