> In message <200112091432.fB9EWvF57444@mrynet.com>, you write: 
> 
> -> Could someone please build and test sysinst with the latest cvs code?
> -> Mine segfaults in the following location:
> -> 
> -> Program received signal SIGSEGV, Segmentation fault.
> -> 0x427128 in t_getstr ()
> -> (gdb) back
> -> #0  0x427128 in t_getstr ()
> -> #1  0x402ce4 in init_keyseq () at menu_defs.c:1288
> -> #2  0x40444c in menu_init () at menu_defs.c:1757
> -> #3  0x40651c in main (argc=1, argv=0x7fffebfc)
> ->     at /usr/src/distrib/utils/sysinst/arch/sgimips/../../main.c:114
> -> 
> -> I don't know if this is a platform or curses problem.
> 
> I'd send this by current-users@ as I believe there have been changes in
> curses which required changes to some curses-using apps, with the menu
> compiler used by sysinst being one...
> 
> I'm up to my ears in schoolwork and finishing a project, otherwise I'd 
> give it a try...

Well, I've tracked down the problem (Thank you, google search).  It appears
there are indeed changes going into curses, and a somewhat colourful thread
on tech-userlevel lead me to a patch.  I've applied the patch, and it does
fix things for sysinst.  The changes that actually need to be made are in
/usr/src/usr.bin/menuc/menu_sys.def.  For anyone in need, the basic patch
is at
	http://mail-index.netbsd.org/tech-userlevel/2001/12/07/0000.html
It is complete except for addressing the resulting unused str_size which
causes a build faulure (thanks to -Wall).

I did have a snapshot ready, but that'll have to wait now if I'm to be
proper about it--the patch hasn't been committed yet, and more than just
sysinst is affected.  When they get their ducks in a row on this, I'll
re-roll a snapshot and make it available.  

For everyone's info, I've come to the follow resolutions about my sgimips
installation work thus far:

o	floppy installation doesn't look feasible--I can't for the life of
	me figure out how we'll go about putting a bootable kernel image
	on a floppy without having to be burdoned with the SGI Volume Header.
o	The automated bootable-CD image generation is still being designed.
	Until I finalise this, I'll be rolling bootable CD images by hand.
o	I'm still plagued with serial-console problems.  Quantities of
	output to the serial console while the system is under load (e.g.
	a /usr/src/ build) will result in 

	trap: address error (load or I-fetch) in kernel mode
	status=0x8202, cause=0x10, epc=0x88189a04, vaddr=0x2d2
	pid=9062 cmd=cc usp=0x20202020 ksp=0xca693e88
	Stopped in pid 9062 (cc) at     0x88189a04:     lw      a2,0(v1)
	db> c

	The cheap-fix for this is to avoid using the serial console
	screen for work (I ssh into my machine and work from those
	sessions).
o	Other than the above console issue, the R4400 and R5k INDYs
	have run flawlessly for me for a week now, under heavy load
	and a lot of disk activity.

I'll keep an eye on the patch status and roll out a new snappy and CD
image when the opportunity arises.

-scott

> 
> --rafal
> 
> ----
> Rafal Boni                                              rafal.boni@eDial.com
>  PGP key C7D3024C, print EA49 160D F5E4 C46A 9E91  524E 11E0 7133 C7D3 024C
>    Need to get a hold of me?  http://http.eiab.edial.com/800.edial/rafal
>