Newbie - Mac G5 Dual PPC 970 @ 2Ghz

[Ian Chapman]

Hi,

I have FreeBSD 9.0 installed on a G5 mac using the PPC64 kernel. 
Initially when doing a portsnap fetch and extract it was taking hours to 
perform the verify phase. It didn't hang because I could see it 
decompressing each tarfile but it was taking a long time - I left it 
running for several hours and in the end I actually commented out the 
following in portsnap

fetch_snapshot_verify() {
         #while read F; do
         #       if [ "`gunzip -c snap/${F} | ${SHA256} -q`" != ${F} ]; then
         #               echo "snapshot corrupt."
         #               return 1
         #       fi
         #done
         return 0
}

which at least allowed me to get ports going although it's not an ideal 
solution, but the system was running terribly slow. At least spawning 
processors was horrendously slow. I recompiled the kernel and disabled 
these options as they seemed to be enabled by default for the 64bit 
kernel but not the 32bit.

#options        KDB                     #Enable the kernel debugger
#options        DDB                     #Support DDB
#options        DEADLKRES               #Enable the deadlock resolver
#options        INVARIANTS              #Enable calls of extra sanity 
checking
#options        INVARIANT_SUPPORT       #Extra sanity checks of internal 
structures, required by INVARIANTS
#options        WITNESS                 #Enable checks to detect 
deadlocks and cycles
#options        WITNESS_SKIPSPIN        #Don't run witness on spinlocks 
for speed
#options        MALLOC_DEBUG_MAXZONES=8 # Separate malloc(9) zones

and it appears and feels much faster now. I was wondering if these 
debugging options are on for a reason or whether it was an oversight?

Secondly I noticed the system always booted up with a processor speed of 
1333mhz. I searched the archives and found that auto-booting FreeBSD 
from OF and not using either the OF prompt or the 'Alt' menu brought the 
system up at the full speed of 2Ghz. I guess that what's powerd doesn't 
work either - no dynamic frequency scaling for this mac.

Anyway the real stumbling block is getting X to work. I did see a 
similar issue in the archives but wondered if anyone at resolve it. My 
graphics card is as follows:

vgapci0 at pci0:0:16:0:	class=0x030000 card=0x41521002 chip=0x41521002 
rev=0x00 hdr=0x00
     vendor     = 'ATI Technologies Inc'
     device     = 'RV350 AR [Radeon 9600]'
     class      = display
     subclass   = VGA
     bar   [10] = type Prefetchable Memory, range 32, base 0xb0000000, 
size 268435456, enabled
     bar   [14] = type I/O Port, range 32, base 0x400, size 256, enabled
     bar   [18] = type Memory, range 32, base 0xa0000000, size 65536, 
enabled
     cap 02[58] = AGP v3 8x 4x SBA disabled
     cap 01[50] = powerspec 2  supports D0 D1 D2 D3  current D0

When starting X with the radeon driver it always segfaults with the 
following:

Caught signal 10 (Bus error: 10). Server aborting

Curiously dmesg shows this:

pid 17612 (Xorg), uid 0: exited on signal 6 (core dumped)

I've tried just about every incarnation of xorg.conf with the radeon 
driver but the result is always the same. I also installed radeontool 
from ports and interesting when running it as follows:

# radeontool regs
RADEON_DAC_CNTL=ac160ff0a000000
Bus error (core dumped)

I also downloaded the latest version of radeontool (1.6.3) and it does 
the following:

# ./radeontool regs
RADEON_DAC_CNTL	ff60c10a
RADEON_DAC_EXT_CNTL	00000000
RADEON_DAC_MACRO_CNTL	00000806
RADEON_DAC_CNTL2	00000000
RADEON_TV_DAC_CNTL	00000103
RADEON_DISP_OUTPUT_CNTL	00000000
RADEON_CONFIG_MEMSIZE	08000000
RADEON_AUX_SC_CNTL	00000000
RADEON_CRTC_EXT_CNTL	00000000
RADEON_CRTC_GEN_CNTL	03000200
RADEON_CRTC2_GEN_CNTL	02000200
Bus error (core dumped)

So it gets further but the end result is the same. I guess reading a 
particular register on the graphics card is causing the bus error. I 
don't believe the hardware is faulty as the system runs perfectly stable 
in Leopard, no hangs, crashes or graphical artefacts etc. So where can I 
go from here? Any ideas how I could fix this?




-- 
Ian Chapman.