Re: MONITOR not catching MCH3601

[Barbara Morris <bmorris@xxxxxxxxxx>]

rob@xxxxxxxxx wrote:
> ...
> Regarding API's, especially list APIs.  I seem to recall a Barbara tip 
> from a couple years ago that when using list apis a common error like 
> you're having occurs on a loop to process each entry.  Something people 
> often did, which mostly worked, but occasionally bombed like you're 
> getting.  To fix it you just set your looping up differently.  Sort of 
> like the debate on whether to use a priming read.  Getting kind of busy 
> and a quick scan of some source files for Barbara wasn't turning it up. 

Rob, that list-API coding error usually shows up as MCH0602, pointer 
offset not valid.

It's not really relevant to this thread any more, but here's the way of 
coding the list-API loop that could lead to the offset error:

  pEntry = address of first list item;   // this is bad code
  for i = 1 to number of list items;     // this is bad code
     process the entry based on pEntry;  // this is bad code
     pEntry += size of each list entry;  // here is the badness
  endfor;                                // this is bad code

By incrementing the pointer at the end of the loop, there's a 
possibility of trying to point past the end of the physical storage 
after processing the last entry.  Even if it falls out of the loop right 
after setting pEntry, and never actually tries to access the storage 
based on pEntry, it could get MCH0602 just from trying to set the pointer.

Here's a safe way to code it:

  offset = 0;
  for i = 1 to number of list items;
     pEntry = address of first list item
            + offset;
     process the entry based on pEntry;
     offset += size of each list entry;
  endfor;