Subject: port-sparc/3048: heavy mfs I/O causes panic
To: None <gnats-bugs@gnats.netbsd.org>
From: None <Robert-Byrnes@deshaw.com>
List: netbsd-bugs
Date: 12/19/1996 22:15:33
>Number:         3048
>Category:       port-sparc
>Synopsis:       heavy mfs I/O causes panic
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    gnats-admin (GNATS administrator)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Dec 19 14:20:00 1996
>Last-Modified:
>Originator:     byrnes@deshaw.com
>Organization:
D. E. Shaw & Co.
>Release:        1.2
>Environment:
System: NetBSD sparx.nyc.deshaw.com 1.2B NetBSD 1.2B (SPARX) #3: Sat Dec 7 22:23:49 EST 1996 root@sparx.nyc.deshaw.com:/tmp_mnt/src/NetBSD/cvsroot/src/sys/arch/sparc/compile/SPARX sparc


>Description:
	When I do enough I/O in an mfs filesystem, I get:
	----------------------------------------
	panic: cluster_rbuild: too much memory
	syncing disks... 37 37 35 31 26 23 23 23 23 23 23 23 23 23 23 23 23 23 23 23 giving up
	Frame pointer is at 0xf9dd4ae8
	Call traceback:
	  pc = f80d5d98  args = (0, 90900fe3, f8104400, f9dd4b68, 2, 21, f9dd4b50) fp = 0xf9dd4b50
	  pc = f80234cc  args = (100, 0, 1, f944f000, f8686348, 2000, f9dd4bb8) fp = 0xf9dd4bb8
	  pc = f8039184  args = (100, 2e, f000, 0, 0, 2000, f9dd4c20) fp = 0xf9dd4c20
	  pc = f8038e00  args = (f8c9adf4, f8f2f000, f8c9a584, 1, 2d, 26b80, f9dd4c90) fp = 0xf9dd4c90
	  pc = f80a2e2c  args = (f86d5a80, f8c9adf4, 100000, 2d, 2000, ffffffff, f9dd4d28) fp = 0xf9dd4d28
	  pc = f8041fd4  args = (0, f9dd4e98, 0, f9dd4d9c, f8739180, 60060, f9dd4da8) fp = 0xf9dd4da8
	  pc = f8024afc  args = (f8777480, f9dd4e98, f8739180, f8041f24, f7ffabc8, f810e000, f9dd4e28) fp = 0xf9dd4e28
	  pc = f80dd660  args = (f8735200, f9dd4f28, f9dd4f20, f8024a38, f80f7b44, f80c6434, f9dd4ec0) fp = 0xf9dd4ec0
	  pc = f80058f4  args = (3, f9dd4fb0, 0, e, 0, 58c2c, f9dd4f50) fp = 0xf9dd4f50
	  pc = c598  args = (3, f7ffabc8, 4120, 0, 60028, ffffffff, f7ffab68) fp = 0xf7ffab68
	----------------------------------------
	I don't think that I was running out of swap space ...
>How-To-Repeat:
	Do things like this:
		cd /tmp
		cvs checkout ssh
		make
>Fix:

>Audit-Trail:
>Unformatted: