On Aug 7, 2005, at 4:42 PM, Nick Holland wrote:
Your description sounds just like a memory problem, the more your
install, the more memory the system uses. MAKEDEV uses a LOT of RAM.
Well, I started by pulling out all the core, and swapping them in 2
at a time (2 @16Mb each). At first I was having issues with all
combinations, and started to get worried. Then I stumbled on it:
there was one sure thing I could do that would always crash the
system: `ifconfig -a`
I pulled out the qfe SBUS card, and it's stable as can be, with all
core tested in pairs of 2, and with all 128Mb installed.
I've added [EMAIL PROTECTED], as this may be platform-specific.
Strange thing is, the card works. I used it under Solaris to
transfer the 3.7 boot floppy image in. I used it with the boot
floppy to grab all the files via FTP. The installation process would
sometimes crash during FTP using the installer, but never if I
escaped to shell and ran FTP manually. Also, the install process
would fail at MAKEDEV, which I presume only "uses" the network if
making network devices.
Boot into the installed system, and I could ping the local IP.
Sometimes I would get "no route to host" pinging another node on the
subnet, and sometimes it would crash. `ifconfig =a` would always crash.
I searched, and there are open qfe bugs for sparc64 (3928/sparc64,
3960/sparc64), but this is sun4m/SS20 sparc. There's (4218/kernel),
but that just says "not working." None of these reflect the severity
(immediate drop into the debugger) I experienced.
I also found this:
http://archive.openbsd.nu/?ml=openbsd-sparc&a=2005-02&t=659794
I've been trying for the last few days to build a router w/ a
Sparc20 to get
into our lab from our desks. The hardware on hand happens to be
a Sparc20
w/ 1 le interface and a Quad card that are hme interfaces. From
inside the
box, I can ping the interface, but can't seem to ping anything
outside of that.
There are also a few entries in the dmesg that seem to indicate
something more
than my inability to pull my head out (thought I'm pretty sure
that's the problem).
...
> If you comment out ukphy from your kernel and recompile it
should start working.
I have no problem with doing this, but I'd like a second opinion if
possible. Is this the "correct fix"? Is this "known," under the
other bug reports?
Thanks,
JMF
