On Sun, 21 Jan 1996, David Gilbert wrote:

> I have found lately that reboot hangs while syncing disks.

Possibly related: Since I upgraded from a November 11th snapshot to a
January 1st one, my i386 has been dropping into the kernel debugger
whenever I attempt to go from multiuser mode to single user mode.  The
crash is always the same, and occurs at the point where I'm expecting
to be asked whether I want /bin/sh or another shell:

	kernel: page fault trap, code=0
	Stopped at _gdt_compact+0x30: movl 0x68(%ecx),%edx

The backtrace is:

	_gdt_compact()
	_gdt_put_slot()
	_tss_free()
	_switch_exit()
	_bpendtsleep()
	_lock_clear_recursive()
	_lock_clear_recursive()
	_vm_pager_get()
	_vm_fault()
	_vm_fault_wire()
	_vm_map_pageable()
	_scheduler()
	_main()

Apart from this, the only other change I see after this upgrade is
that I get much more frequent occurrences of kernel messages like

	com0: 1 silo overflow, 0 ibuf overflows

while receiving data.  Is there more overhead at interrupt time now?
This is with a 16550 FIFO buffered UART at 38400bps on a 25MHz i386.

-tih
-- 
Tom Ivar Helbekkmo
tih@Hamartun.Priv.NO