Hello Craig!

Thus quoth Craig Swanson to Mike Bilow:
 CS>     LOCK
 CS>      (or can it?)
 CS>     DEC   [EAX]

There is no chance for an interrupt to happen at the location in your
example, since LOCK 
isn't an instruction but an instruction prefix. An instruction is scheduled
at the same 
time with all of its prefixes. So there is absolutely no need for a CLI/STI guard.

There is another problem with your example code. Excerpt from the TASM 3.0
quick reference 
guide:
' Locked access is *not* assured if another CPU is executing an instruction
concurrently 
that has one of the following characteristics:
 - Is not preceded by a LOCK prefix
 - Is not one of the instructions which may have an LOCK prefix (or have it
implicitly)
 - Specifies a memory operand that does not exactly overlap the destination operand.
   Locking is not guaranteed for partial overlap, even if one memory
operand is wholly
   contained within another.'

bye, Detlef

 ! mailto engert{at}ibm.net, 2:2490/2576.1{at}fidonet
--- Sqed/32 1.00/r00108