svn commit: r304928 - in head/lib/libc: amd64/sys i386/sys sys

[Andrey Chernov]

On 29.08.2016 22:10, Konstantin Belousov wrote:
> On Mon, Aug 29, 2016 at 09:16:29PM +0300, Andrey Chernov wrote:
>> Previous text is documented everywhere and describing usual good
>> practice, so it should remains in place, in that means r304928 should be
>> reverted, because replace proper way of doing things with obsoleted
>> feature description.
>>
>> What I suggest is not _replace_ old text with new, but _add_ new text as
>> describing current and not recommended way, with the direct mention that
>> this feature is obsoleted and not recommended for relay on it.
>>
> 
> diff --git a/lib/libc/sys/ptrace.2 b/lib/libc/sys/ptrace.2
> index 48802f4..d406efc 100644
> --- a/lib/libc/sys/ptrace.2
> +++ b/lib/libc/sys/ptrace.2
> @@ -2,7 +2,7 @@
>  .\"	$NetBSD: ptrace.2,v 1.2 1995/02/27 12:35:37 cgd Exp $
>  .\"
>  .\" This file is in the public domain.
> -.Dd August 28, 2016
> +.Dd August 29, 2016
>  .Dt PTRACE 2
>  .Os
>  .Sh NAME
> @@ -900,19 +900,29 @@ argument is ignored.
>  .Pp
>  Additionally, other machine-specific requests can exist.
>  .Sh RETURN VALUES
> +Most requests return 0 on success and \-1 on error.
>  Some requests can cause
>  .Fn ptrace
>  to return
>  \-1
> -as a non-error value; to disambiguate,
> +as a non-error value, among them are
> +.Dv PT_READ_I
> +and
> +.Dv PT_READ_D ,
> +which return the value read from the process memory on success.
> +To disambiguate,
>  .Va errno
> -is set to 0 in the libc wrapper for the
> -.Fn ptrace
> -system call and
> +can be set to 0 before the call and checked afterwards.
> +.Pp
> +Current
>  .Fn ptrace
> -callers can reliably check
> +implementation always sets
> +.Va errno
> +to 0 before calling into kernel, both for historic reasons and for
> +consistency with other operating systems.
> +It is recommended to assign zero to
>  .Va errno
> -for non-zero value afterwards.
> +explicitely for future compatibility.
>  .Sh ERRORS
>  The
>  .Fn ptrace
> 

Ok.