* Jan Klemkow <[email protected]> [2014-06-01 23:38]:
> On Fri, May 30, 2014 at 09:51:24PM +0100, Stuart Henderson wrote:
> > On 2014/05/30 22:43, Jan Klemkow wrote:
> > > On Fri, May 30, 2014 at 10:26:43PM +0200, Jan Klemkow wrote:
> > > > On Fri, May 30, 2014 at 01:49:55PM +0200, Henning Brauer wrote:
> > > > > * Stuart Henderson <[email protected]> [2014-05-28 12:31]:
> > > > > > the old port also had this...is this or something like it still 
> > > > > > needed?
> > > > > > 
> > > > > > # datasize limit in 'run' files is too low for ld.so
> > > > > > # to be able to pull in libc
> > > > > > LDFLAGS+=       -static
> > > > > 
> > > > > that is everything but smart, it makes MUCH more sense to increase the
> > > > > datasize limits (the softlimit calls) in the run scripts.
> > > > 
> > > > Henning is right.  It is stupid to compile this port static cause of the
> > > > datasize limit.  So I removed it from the port.
> > 
> > If something is still needed for this, then it probably needs some
> > kind of instructions somewhere...
> 
> I talked with Henning about the reason of that static compiling hack.
> As I understand him, there are some scripts from djb which sets the
> datasize limit to a low value and that causes some crashes.  So I add
> an install notice to the port:
> 
>    Please notice, there may be some scripts that manipulate the datasize
>    limit of tcpclient/tcpserver which may cause a process termination.
>    For more information look at login.conf(5).

not really.

a typical daemontools-style run script looks like this:

#!/bin/sh
exec 2>&1
exec envuidgid tinydns envdir ./env softlimit -d30000 /usr/local/bin/tinydns

since it's djb and just HAS to be different to be different... the
softlimit invocation is the culprit, that number there needs
adjustment, the defaults from more than a decade ago (when no
randomization, eating a little virtual mem, took place) just don't cut
it any more. 

Reply via email to