Eric W. Biederman wrote:
> I got annoyed with memtest86 a while ago and started scratching itches.
> 
> I have a preliminary patch available that I will send to Chris Brady
> as soon as I do as I sleep on it and can do a final review with
> a fresh mind.
> 
> As of 2.9 memtest86 does BIOS calls to get the memory size, this
> broke netbooting.

I could not get the 2.9 release (01/march/2002 from teresaudio.com) to
work with the small memory machines here - 16 meg on 486's.

It would detect 640K of memory using the default MemMap of e801-Std and
would then proceed to Testing: 88K - 1024K  552K     with errors occuring
on test #4. I assume the errors because it was testing the "wrong" area?
Resetting the Memory Sizing to Probe would hang the machine.

Using this newer version (memtest.2.9-eb1) resolves that problem. It
detects 16M and proceeds to Testing: 96K - 16M 16M.
Is this to do with the the MemMap detection change to e88-Std ?

However, restarting the test ( c => 8 ) will hang the machine.

In trying to see if there was a pattern to this hang I can only say
that....
        If I restart the test instantly (straight after it starts
        running) odds are good that it will restart as intended. It will
        hang eventually though.
        
        If the time between pressing (c) then (8) is minimal I'll get a
        couple of successes before hanging. (seemingly)
        
        And for something a little less vague :-) If I restart during
        Test #3 It will spontaneously reboot. So far that's been
        repeatable at 5 times out of 5
        
        If I leave it alone (prefered option!?) it passes all tests
        (which this machine did using 2.8)

When compiled without serial support....
        the console still hangs
        
        Test #3 still reboots although it changed to 4 reboots and 2
        hangs out of 6 attempts.

A quick check of 2.8 and these problems don't occur.
Memtest 2.8 detected 16M and would test 76K - 16M under the default (Std)
test.

Performing this on another machine (P120 - 48M) showed the same
behaviour to the restarting problem  (only the NIC, with BootRom, was
transferred over).
MemMap e820-Std         Memory 48M      Testing: 96K - 48M  48M

HTH

-- 
Cheers
Glenn

http://members.optushome.com.au/graybeard/linux/netboot.html

Reply via email to