|
On Wed, Apr 12, 2006 at 10:08:17AM +1000, James Steward wrote:
Ive now hacked in a linear feedback shift register as you suggested and
in doing that has made the ramtest more explosive on the system. By
that I mean it doesnt even get to detecting a memory error before many
other user space apps segv including init at which point the box is
beyond recovery and requires a power off. This is all with
cachepolicy=writethrough.
*But* with the default cache policy (writeback), I get several apps run
from init scripts faulting - like modprobe for example, however the
ramtest runs without error!
The ramtest does run successfully on my desktop every time.
Your original test program also runs fine on my ARM machines.
Therefore, I think we can probably say that your hardware is faulty in
some way. Did you say that a 2.4 kernel ran fine on this hardware?
If so, does it still run fine with this test program running as well?
Im curious, what was this for?
if ((i & (31 / sizeof(*mem)) == 0)
*(volatile unsigned char *)(&mem[i]);
Guennadi answered that.
-------------------------------------------------------------------
List admin: rel="nofollow" lists.arm.linux.org.uk/mailman/listinfo/linux-arm lists.arm.linux.org.uk/mailman/listinfo/linux-arm
FAQ: rel="nofollow" www.arm.linux.org.uk/mailinglists/faq.php www.arm.linux.org.uk/mailinglists/faq.php
Etiquette: rel="nofollow" www.arm.linux.org.uk/mailinglists/etiquette.php www.arm.linux.org.uk/mailinglists/etiquette.php
|
