svn commit: r244154 - head/bin/ps

[Konstantin Belousov]

On Mon, Dec 17, 2012 at 02:18:35PM +0100, Pawel Jakub Dawidek wrote:
> On Sat, Dec 15, 2012 at 03:13:00AM +0200, Konstantin Belousov wrote:
> > On Fri, Dec 14, 2012 at 10:42:47PM +0100, Pawel Jakub Dawidek wrote:
> > > In that case I'd prefer to make existing kern.pid_max sysctl read-only
> > > and make it loader tunable. I don't expect there are many users of this
> > > sysctl...
> > > 
> > No, I described you the purpose of the sysctl. Requiring reboot just for
> > running the old binaries is not useful. Please do not break it.
> 
> Kostik. How many users of this sysctl are there? 2? 3? Is it really
> worth introducing yet another sysctl to export PID_MAX from the kernel?
> I think if someone decides to run 1.x binary there will be some
> inconvenience apart from the need to reboot.

I always surprised when users of the old ABIs appear, but trust me,
there is more than you would like.

I do not see why should we break existing feature. The only thing you
need to have on both i386 and amd64 right now for a.out execution is a
suitable COMPAT_ options included into the kernel config. Everything
else is dynamically adjustable, including the binary format support and
knobs needed mostly to allow the compatibility environment. The sysctl
kern.pid_max is one of them, other is security.bsd.map_at_zero.

So I object against breaking this feature. Please add a sysctl you need,
if 99999 should be changed for some reasons.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 834 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-src-head/attachments/20121218/a965d239/attachment.sig>