Re: {Patch]: Giving access to pinfo after seteuid and exec

[Christopher Faylor <cgf at redhat dot com>]

On Thu, Sep 25, 2003 at 10:56:41PM -0400, Pierre A. Humblet wrote:
>At 10:23 PM 9/25/2003 -0400, you wrote:
>>On Thu, Sep 25, 2003 at 10:17:22PM -0400, Christopher Faylor wrote:
>>>I'll check in the rest of the spawn.cc stuff with some modifications.  I see
>>>I missed some cases with the addition of _P_SYSTEM.
>>
>>I'm sorry.  Long day.  I'm checking in the non-acl related stuff.  I'll leave
>>the rest for Corinna's review.
>>
>OK.
>
>By the way, in the case      
>if (!child)
>	{
>	  syscall_printf ("pinfo failed");
>	  set_errno (EAGAIN);
>
>the EAGAIN is masking the true reason that should have been set by the
>pinfo code. That may be Posix, but it's not helpful from a debugging
>point of view.

If it's a cygwin problem there should be enough clues in strace or
whatever.  SUSv3 says that there should be two possible errnos on error
-- EAGAIN or ENOMEM.  If we were strictly correct we could check for the
ENOMEM case and not overwrite it.

cgf