SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Wind River going up, up, up!

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: ZARAH who wrote (1590)8/5/1997 10:00:00 AM
From: Allen Benn   of 10309
 
More on the Pathfinder computer glitch.

I think everyone now agrees that the problem had to do with a task functioning with the wrong priority, which backed up processing, causing the need for periodic resets. NASA indicated they changed the priority of the aberrant task to solve the problem - hence the correct conclusion that the problem was on the application side. QED.

Actually, what may have happened has deeper implications.

NASA used the default scheduling configuration of VxWorks. When the system
was more heavily loaded with new application tasks, a situation known as
priority inversion hampered the operating system's ability to complete
all the task operations within the apportioned time. Priority inversion
occurs when a high priority task contends for a resource currently "owned"
by a lower priority task while an intermediate priority task executes, thereby
allowing the priority inversion condition to persist indefinitely. Over
time, the high priority task fails to gain sufficient access to the CPU,
which in the pathfinder case led to the application resetting itself. The
fix was not to up the priority of the low-priority task (which generally
needed to be low priority), but simply to utilize a unique feature of
the WIND microkernel to resolve the priority inversion by changing the
configuration options of the resource in question. This amounted to
toggling a single bit from earth, since the resource is controlled in
VxWorks with a semaphore.

Yes, the problem was on the application side. Yes, it required changing
task priorities. But what stands out is that the problem was only solvable
because the commercial RTOS NASA decided to use for Pathfinder just
happened to have an important feature NASA thought wouldn't be needed. Had
NASA developed their own RTOS, the needed feature most certainly would have
been missing. Not only did the commercial software do the job cheaper and
faster, but now we know, better, too.

Allen

PS - As an exercise for the reader, try to figure out how NASA diagnosed this heretofore unnoticed problem with the Pathfinder far away on another planet.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext