On Mon, 7 Jul 1997, Andrew Brown wrote:

> what am i gonna break if i raise the size of the (my) msgbuf from
> 
> #define	MSG_BSIZE	(4096 - 3 * sizeof(long))
> 
> to something much larger.  say...
> 
> #define	MSG_BSIZE	(16384 - 3 * sizeof(long))
> 

I have always beleived that MSG_BSIZE should be:

#define       MSG_BSIZE       (NBPG - 3 * sizeof(long))

I did this to my own tree many moons ago when the Amiga port switched from
4K to 8K pages (or was it the other way around?  I forget.)   I felt is
was a bit silly to allocate an entire 8K page for the message buffer but
then only use 1/2 of it. 

It might be a good idea to make this a tuneable, but then there's the
problem with syslog and debuggers.  Of course, you could waste a whole
4 more bytes and embed the size in the message buf structure....

=========================================================================
Eduardo Horvath				eeh@btr.com
"Cliffs are for climbing.  That's why God invented grappling hooks."
					- Benton Frasier