Gumstix Verdex not booting with NetBSD 7.0 and newer

[smesgr <smesgr%gmail.com@localhost>]

Hi everybody,

I got my hands on some older Gumstix Verdex boards (with 120pol 
connector) and want to give NetBSD a go. Sadly I figured out the more 
current NetBSD versions are not running with my boards.

While NetBSD 6.0 & 6.1.5 seem to work:

GUM> pinit on
GUM> fatload ide 0:1 0xa0200000 netbsd-GUMSTIX615.bin
GUM> go 0xa0200000 console=ttyS0,19200 busheader=netcf
## Starting application at 0xA0200000 ...

NetBSD/evbarm (gumstix) booting ...
system serial: 0xffff0015c900afa2
[ Kernel symbol table missing! ]
Loaded initial symtab at 0xc05f6e30, strtab at 0xc06322ac, # entries 15084
pmap_postinit: Allocated 35 static L1 descriptor tables
Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
    2006, 2007, 2008, 2009, 2010, 2011, 2012
    The NetBSD Foundation, Inc.  All rights reserved.
Copyright (c) 1982, 1986, 1989, 1991, 1993
    The Regents of the University of California.  All rights reserved.

NetBSD 6.1.5 (GUMSTIX)
total memory = 128 MB
avail memory = 118 MB
cprng kernel: WARNING insufficient entropy at creation.
mainbus0 (root)
cpu0 at mainbus0: PXA27x rev 8 (XScale core)
cpu0: DC enabled IC enabled WB enabled LABT branch prediction enabled
cpu0: 32KB/32B 32-way Instruction cache
cpu0: 32KB/32B 32-way write-back-locking Data cache
pxaip0 at mainbus0: Onchip Peripheral Bus
pxaip0: CPU clock = 623.972 MHz
[...]

NetBSD 7.0 & 7.0.2 are failing with an odd message:

GUM> pinit on
GUM> fatload ide 0:1 0xa0200000 netbsd-GUMSTIX702.bin
GUM> go 0xa0200000 console=ttyS0,19200 busheader=netcf
## Starting application at 0xA0200000 ...

NetBSD/evbarm (gumstix) booting ...
system serial: 0xffff0015c900afa2
panic: kernel diagnostic assertion "ram_size <= KERNEL_VM_BASE - 
KERNEL_BASE" failed: file 
"/usr/src/sys/arch/evbarm/gumstix/gumstix_machdep.c", line 487

NetBSD-current from 2016-12-31 doesn't start at all - hanging at

## Starting application at 0xA0200000 ...

after some digging I reverted this commit for the gumstix_start.S and 
got the previous 7.0ish error message again:

http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/evbarm/gumstix/gumstix_start.S?rev=1.14&content-type=text/x-cvsweb-markup

my ARM assembly knowledge is a bit rusty. The main changes are register 
swap from r4 to ip. But another register has changed for non CPU_CORTEX-path

mcr     p15, 0, r1, c2, c0, 0   /* Set TTB */ <<< r1 was r0 before
mcr     p15, 0, r1, c8, c7, 0   /* Flush TLB */ <<< r1 was r0 before

the other change from 6.0 to 7.0 which could do something with ram_size 
is imho.

http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/evbarm/gumstix/gumstix_start.S?rev=1.10&content-type=text/x-cvsweb-markup

If somebody with more ARM assembler knowledge could help me - I would 
appreciate it.

Best Regards

Stephan