Netbooting Sparc64+HD question (Boot problems resolved without netboot)

Miles Nordin carton at Ivy.NET
Sat Apr 28 15:33:17 UTC 2007

>>>>> "dm" == Didrik Madheden <didrik at> writes:

    dm> Maybe I'll try on netbooting sometime again when I feel more
    dm> confident.

yeah, I still don't know what was wrong.  I guess I would try tcpdump

    >> atapci0: using PIO transfers above 137GB as workaround for
    >> 48bit DMA access bug expect reduced performance

this is a hardware bug on the Ultra 5/10 that's CMD/Silicon Image's
fault.  Sun never added a workaround to their driver---they just
refuse to ship the machines with >80GB.  I didn't know the SunBlade
100 had the bug, too.

In theory, it means exactly what it says.  In practice, I've found IDE
on Ultra 5/10 is so crappy and such a time-sink that I avoid it like
the plague.  On FreeBSD/NetBSD I use SCSI or netbooting, and on
Solaris I use SCSI, FireWire with Oxford 911 firewire cases only, and
the iSCSI initiator to IDE targets in i386 boxes.  Each of those has
its own problems.  (respectively: expense, expense and smartctl
doesn't work, and kernel panics when you unplug the network cable.)

In general I've had so many small problems with IDE on not-i386: I
have some PCI card that will work at Ultra100 speed in NetBSD/i386,
but in NetBSD/macppc it only works at Ultra33.  Some cards don't work
right unless the chip is initialized by their secret proprietary
PeeCee BIOS.  The problems never get fixed or explained.  I don't know
if there is something about the chips that assumes i386 or what.
Reliable storage on noti386 has been a major frustration for me, so I
guess you'll have to try it and see.  good luck. :/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 304 bytes
Desc: not available
Url :

More information about the freebsd-sparc64 mailing list