[bugs #10113] Patch for kvm_code on FreeBSD 5 to allow running without /proc

[Adam Fedor]

This mail is an automated notification from the bugs tracker
 of the project: GNUstep.

/**************************************************************************/
[bugs #10113] Latest Modifications:

Changes by: 
                Adam Fedor <fedor@gnu.org>
'Date: 
                Thu 08/26/2004 at 14:48 (US/Mountain)

            What     | Removed                   | Added
---------------------------------------------------------------------------
          Resolution | None                      | Fixed
              Status | Open                      | Closed


------------------ Additional Follow-up Comments ----------------------------
* Source/NSProcessInfo.m ([NSProcessInfo +load]): Get environment
        variables via environ even #ifdef HAVE_KVM (fix/based on #10113).






/**************************************************************************/
[bugs #10113] Full Item Snapshot:

URL: <http://savannah.gnu.org/bugs/?func=detailitem&item_id=10113>
Project: GNUstep
Submitted by: Pete French
On: Mon 08/23/2004 at 15:34

Category:  Base/Foundation
Severity:  7 - Major
Item Group:  Bug
Resolution:  Fixed
Privacy:  Public
Assigned to:  fedor
Status:  Closed


Summary:  Patch for kvm_code on FreeBSD 5 to allow running without /proc

Original Submission:  The KVM code used by FreeBSD to read argv and envp 
requires
/proc to be mounted for the envrionemtne to read properly.
This is not true by default on FreeBSD system of 5 and above.

The attached patch alters the code to always use the static
environment pointer on FreeBSD, even when using KVM. This
allows the code to work properly on FreeBSD 5, whilst also
working on previous versions.

Follow-up Comments
------------------


-------------------------------------------------------
Date: Thu 08/26/2004 at 14:48       By: Adam Fedor <fedor>
* Source/NSProcessInfo.m ([NSProcessInfo +load]): Get environment
        variables via environ even #ifdef HAVE_KVM (fix/based on #10113).

-------------------------------------------------------
Date: Tue 08/24/2004 at 09:20       By: Pete French <minusbat>
If we check kvm_getenvv at runtime then it will always produce an error to 
stderr when it fails. There seems to be no way of suppressing this - and I dont 
think its goot to have code producing errors every time it runs. Thats just 
going to lead to umpteen questions and bug reports.

If we check to see if it works at compile time and set an approrpiate #ifdef 
then we have a situation where code compiled on an earlir FreeBSD system will 
not run on a later system - which breaks backward compatibility in a very bad 
way. It also means that code compiled on a FreeBSD system with /proc mounted 
will not then run on other systems where it is dismounted... again a very bad 
thing as a lot of developers (myself included) run our development systems with 
/proc mounted as standard.

I didnt like the idea of the #ifdef FreeBSD ether - but as it stands the kvm 
code only works properly on FreeBSD anyway, so it didnt seem too much of a 
loss. We could go for the option of always using environ though. I think that 
is always supposed to work everywhere... whatever you think is most portbale 
really 

-------------------------------------------------------
Date: Tue 08/24/2004 at 04:10       By: Adam Fedor <fedor>
it seems a shame to use  #ifdef __FREEBSD__ if we could have the problem on 
other systems. Why not just check if kvm_getenv works, and if not, use the 
environ variable?  In fact, maybe we should just always use the environ 
variable since that should work everywhere?






File Attachments
-------------------

-------------------------------------------------------
Date: Mon 08/23/2004 at 15:34  Name: kvm_fbsd.cdiff.gz.uue  Size: 1.14KB   By: 
minusbat

http://savannah.gnu.org/bugs/download.php?item_id=10113&amp;item_file_id=1596






For detailed info, follow this link:
<http://savannah.gnu.org/bugs/?func=detailitem&item_id=10113>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/