Re: Sparc DFLDSIZ

[Peter Carter <peterc@pbc.ottawa.on.ca>]

At 04:28 PM 6/5/99 -0700, you wrote:
>A make build on my sparc fails building lynx since the conversion to
>egcs.  It's an out-of-virtual-memory failure that is resolved by
>doing a "ulimit -S -d 65536" (I use ksh).  Now 64 Mbytes may be
>more than is needed, but the default 16 clearly isn't enough.
>
>Looking at vmparam.h for the sparc I see:
>
>#ifndef MAXTSIZ
>#define	MAXTSIZ		(16*1024*1024)		/* max text size */
>#endif
>#ifndef DFLDSIZ
>#define	DFLDSIZ		(16*1024*1024)		/* initial data size limit */
>#endif
>#ifndef MAXDSIZ
>#define	MAXDSIZ		(128*1024*1024)		/* max data size */
>#endif
>#ifndef	DFLSSIZ
>#define	DFLSSIZ		(512*1024)		/* initial stack size limit */
>#endif
>#ifndef	MAXSSIZ
>#define	MAXSSIZ		MAXDSIZ			/* max stack size */
>#endif
>
>Is there any reason why DFLDSIZ couldn't be bumped up to 32 Meg?  I
>do not pretend to understand what might be affected by doing this.
>
>Also, for my information, why are these numbers so low when compared to
>something like the i386.  An RTFM pointer would be appreciated.
>
>// marc
>

This is also why I stopped using my sparc for a file server. The fact is,
in file system recovery, there was not enough process memory to handle the
restore for my 18 gig ccd. I got tired of modifying the header file after a
src tree upgrade. As is my machine usually only uses about 20 meg out of
64, due to this oversight.
---
Peter B. Carter (peterc@pbc.ottawa.on.ca)
Pager: 613-751-4660
http://www.pbc.ottawa.on.ca