>Number:         4326
>Category:       port-alpha
>Synopsis:       Upping maxusers past 16 makes kernel fail in boot
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    gnats-admin (GNATS administrator)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Oct 23 12:05:03 1997
>Last-Modified:
>Originator:     Charlie Root
>Organization:
	Clinet Ltd
>Release:        supped 21th october, I think, has been around longer
>Environment:
	
System: NetBSD proxy2.clinet.fi 1.3_ALPHA NetBSD 1.3_ALPHA (CLINETALPHA) #17: Thu Oct 23 21:32:05 EEST 1997 root@proxy2.clinet.fi:/usr/src/sys/arch/alpha/compile/CLINETALPHA alpha

512M, 5 wide-SCSI disks.  Intended to be a proxy server

>Description:

When trying to up maxusers in kernel config (plain GENERIC config),
kernel breaks up.  At 32 it panics with

panic: pmap_enter_ptpage
	: can't get KPT page
halted

(before reporting memory sizes and such)

at 64 (and up) kernel reports memory sizes and locks up.

Experimenting with kernel parameters I noticed that I can up NMBCLUSTERS 
but making NPROC 2048 causes same deadlock.

I hope to get around this for a proxy server as I am able to get more
NMBCLUSTERS and max *size limits, which probably is the primary problem
with a proxy server.  For other users one probably wants larger process
table?

>How-To-Repeat:

Modify GENERIC with maxusers larger than 16.

>Fix:

>Audit-Trail:
>Unformatted: