[VxW] Re: System restart debugging problem

[John]

Hello,

> Thank's for your quick feedback.
> I need some time to obtain necessary environment in order to try your
> suggestions.
> sysToMonitor() monitor approach seems to be usefull, but requires to
> reflash my BSP. I have just requested necessary stuff to do that.
> Using of user reserved RAM however seems to be unfortunately
> unacceptable on my environment, since after system restart I could not
> reestablish connection from Tornado to my target without switching
> power off and losing RAM content :(

Try asking the hardware guys for a reset switch as well... Also, for
debugging these types of problem I find that the target shell is
infinitely more reliable that the host tools (the first thing to fail
in most serious crash scenarios is the WDB connection :-( )

Also, you do not need to reboot after the failure if you are using a
VxWorks bootrom. You can dump memory from the bootrom prompt as well.
If you are using a kernel programmed directly into flash, you might
want to think about whether it would be worthwhile to switch to
network booting while you debug this problem.

> However I will try flash memory to store crash status.
> I have doubts if it's safe using flash memory driver on reboot /
> interrupt / exception contexts. Since flash memory driver have been
> written not by myself and I do not have direct contact to these guys,
> but I have got sources of the driver, what the vxWorks in driver could
> be dangerous? I guess limitations generally shall be the same as for
> ISRs. Isn't it?

I'm not sure I would want to be writing to flash for this - apart from
anything else, it will increase the wear on the flash block you write
to. It is also slow, so it will impact the overall timing of the
system, potentially masking the problem.

HTH,
John...

=====
Rate Corporate America at http://exec-ratings.bluedonkey.org