In a message dated 11/21/2007 9:23:22 A.M. Central Standard Time,  
[EMAIL PROTECTED] writes:
>Nucleus is not going to page fault. And if STARTIO fails before it gets  the
UCB lock, no harm - no foul - since nothing is "in the  middle".
 
So there are zero defects now in the nucleus?  I doubt it.  A  disabled page 
fault is easy to handle - just ABEND the current piece of work and  move right 
along.  What happens next depends on what the current piece of  work was.  
Even if the nucleus refuses to page fault, nevertheless disabled  nucleus code 
can be interrupted and, if the interrupt status is not preserved,  the nucleus 
resumes execution with interrupts now enabled, and I posit that the  results 
of this situation would likely be far worse than if the nucleus  suffered a 
disabled page-fault.
 
Of course the nucleus is designed not to page-fault while disabled.   One way 
that is guaranteed is by being sure that every byte ever touched by  disabled 
code is fixed before doing the disabling operation.  Which is what  I have 
done in all the code I have written that explicitly disables via  STNSM.  And I 
have learned that if I want to do a GETMAIN for a system  subpool while 
disabled and without holding a system lock, that I need to turn on  a super bit 
to 
tell GETMAIN that the CPU is "officially" disabled.  And  when I get a disabled 
page fault ABEND dump, I know what to do to fix my  bug.  STARTIO, I'm sure, 
has an FRR that knows what to do if it fails after  it gets the UCB lock and 
something is "in the middle".
 
I still do not understand the necessity for the warning about  MVS-guaranteed 
disablement's requiring a system lock's being held.  One  result of acquiring 
a system lock is the turning on of a super bit.  But  you can also turn it on 
with an OI instruction.
 
Bill  Fairchild
Franklin, TN





**************************************Check out AOL's list of 2007's hottest 
products.
(http://money.aol.com/special/hot-products-2007?NCID=aoltop00030000000001)

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to