Of course it would, try to remove the following from sparc/machdep.c @ 353:

	/*
	 * XXX stopgap measure to prevent wasting too much KVM on
	 * the sparsely filled buffer cache.
	 */
	if (nbuf > 128)
		nbuf = 128;

Looks like we are not ready for nbuf != bufpages.

> 	This random lossage with >32Mb wouldn't be anything like
> 
> 	panic cluster_rbuild: too much memory?
> 
> 	I've just upgraded to -current today & have had two panics...
> 	(finger based cut&paste below)
> 
> _cluster_rbuild(f8c1e6c8, f9342000, f8c20450,2 ,3, d280)
> _cluster_read(
> _ffs_read(
> _vn_read(
> _sys_read(
> _syscall(3,
> syscall(3,
> 
> _cluster_rbuild(f8c20600,f8ca2000,f8c1d798,1,be,34b80) at _cluster_rbuild+0x1a4
> _cluster_read(f8679e80,f8c20600,100000,be,2000,ffffffff) at _cluster_read+0x3c8
> _ffs_read(0,f9e8dca8,1000,f875e600,17d000,5c0010) at _ffs_read+0x23c
> _vnode_pager_io(
> _vnode_pager_getpage(
> _vm_pager_get_pages(
> _vm_pager_get(
> _vm_fault(
> _mem_access_fault(
> _normal_mem_fault(
> 
> 
> 
> 		David/abs	david@{mono.org,southern.com,mhm-internet.com}
> 
> Thought for : Loathing the way "Don't cry for me Argentina" has
>   the day   : been bastardized by Madonna into a 'disco' song.
> 
> On Wed, 12 Feb 1997, matthew green wrote:
> 
> > 
> > hi.
> > 
> > i upgraded to -current and now i'm getting a lot of fifo overruns
> > i never used to.  details:
> > 	- ELC
> > 	- was 40M, now 32M [1]
> > 	- X11R6.1
> > 	- pppd with 28k8 modem
> > 
> > i'm talking like at least one per minute.  sometimes as many as 20
> > in a single minute.
> > 
> > [1] i had other random lossage that is gone now.
> > 
> > 
> > .mrg.
> > 
> 
> 


-- 
Juergen Hannken-Illjes - hannken@eis.cs.tu-bs.de - TU Braunschweig (W Germany)