Hi,

unfortunately there is still a problem and I presume this kernel
doesn't have any debug support as I couldn't find the sparc_pci_debug
symbol. I should probably set up my own build environment here. 
I'll probably set up a NetBSD/i386 box for this.

trace below.
##########################

hme1 at pci0 dev 1 function 1: address 08:00:20:e6:d6:b0                
nsphy0 at hme1 phy 1: DP83840 10/100 media interface, rev. 1
nsphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
hme1: using vector 33 for interrupt                         
data memory error type 32 sfsr=0 sfva=ff17641c afsr=88000000 afva=1fe01000a00 tf=0x1809120
panic: trap: memory error                                                                 
kdb breakpoint at 125bda0
Stopped in pid 0 (swapper) at   cpu_Debugger+0x4:       nop
db> trace                                                  
data_access_error(32, ff17641c, 0, 1fe01000a00, 100b258, 1809120) at data_access
_error+0xa0                                                                     
data_error(ff3f1092, 1d, 1802000, 1fe010009ec, 4, 100b0b4) at data_error+0x4
confaddr_ok(26e5b00, a00, 1809490, 0, 140a400, 1405c00) at confaddr_ok+0x9c 
pci_conf_read(26e5b00, a00, 0, 200, 50a0021, 0) at pci_conf_read+0x10      
pci_probe_bus(2703a80, 8, ffff, 2, 2703d00, 2703c80) at pci_probe_bus+0xf8
pciattach(12b5180, 2703a80, 18097f0, 121699c, 2725f90, 0) at pciattach+0x200
config_attach(2703a80, 1406f38, 18097f0, 0, 140a400, 1405c00) at config_attach+0
x380                                                                            
config_found_sm(26e5b00, 18097f0, 1223f54, 0, 2703d00, 2703e00) at config_found_
sm+0x24                                                                         
psycho_attach(2703f00, 26e5b00, 1809ab0, 1223e44, 2725fc0, 1000000) at psycho_at
tach+0x100                                                                      
config_attach(26e5b00, 1406f38, 1809ab0, 0, 140a400, 1405c00) at config_attach+0
x380                                                                            
config_found_sm(2703f00, 1809ab0, 124cb80, 0, 1809ad0, 130f400) at config_found_
sm+0x24                                                                         
mainbus_attach(130f7b0, 1416870, 130f750, 130f7a8, 130f798, 130f760) at mainbus_
attach+0x39c                                                                    
config_attach(2703f00, 1406f38, 0, 0, 140a400, 1405c00) at config_attach+0x380
config_rootfound(130f630, 0, 140d2d0, 0, 2704e00, 1298238) at config_rootfound+0
x44                                                                             
cpu_configure(14767c0, 14767e0, 1407068, 0, 0, 12983e0) at cpu_configure+0x3c
configure(1499400, 1499000, 1499000, 1497400, fffffffffffffc00, 3ae4e000) at con
figure+0x50                                                                     
main(1472400, 1475400, 149bc00, 14a1000, 14a1420, 14a13e8) at main+0x410
cpu_initialize(fffffffffffffff8, 10, ffffffffffffffff, f007601c, f007601c, 1009d
88) at cpu_initialize+0x378                                                     



> 
> 
> 
> i've placed a kernel that works on my U5 at:
> 
> 	ftp.netbsd.org:/pub/NetBSD/arch/sparc64/kernels/NetBSD-1.5R
> 
> this hopefully should work for you.
> 
> 
> .mrg.
> 
> 
> ps: there is another issue i've noticed looking at this problem.  the
> onboard `hme' probes as `hme1' on the e250 because of the hard coded
> hme0 for the U5/U10 onboard.  perhaps the right solution would be to
> force the two simba busses on the U5/U10 (US-IIi) port to be numbered
> the other way?  then everyone would have the onboard hme on pci1.