Conference azur::mcc

1692.0. "DECmcc V1.1 Kernel patch - request for testing assistance " by RALLYE::MERRIFIELD () Tue Oct 22 1991 17:25

	On multi-processor machines, Event putters in DECmcc V1.1 sometimes 
cannot access memory held by slower CPUs.  This may cause alarm notification
to cease to function and cause "secondary interlock failure" messages to show
up in the MCC_DNA4_EVL.LOG file.

	The problem will be corrected in the next release (V1.2), but a patch 
to significantly reduce the likelihood of getting these failures  for V1.1 
is also desirable.  Because the development group does not have access to 
multi-processor machines to either reproduce the problem, or test the fix, 
I'm requesting assistance from members of our noting communitee.  If you 
believe that you have seen this problem on your machine, could you please 
apply the attached patch and report back to this notes file as to whether or 
not it fixed your problem?  I'd like to get a patch out to our customer base, 
but we need to test the patch first.


!------------------cut here------------------
! MCCBMS011_PATCH_KERNEL_5.COM
! On multiprocessor machines, Event Putters sometimes cannot access memory
! held by slower CPUs.  This increases the number of retries by a factor of 10.
!
! Directions:
! $ SET DEFAULT SYS$COMMON:[SYSLIB]
! $ PATCH @MCCBMS011_PATCH_KERNEL_5.COM !(This file)
! $ INSTALL REPLACE SYS$COMMON:[SYSLIB]MCC_KERNEL_SHR.EXE
MCC_KERNEL_SHR.EXE
SET ECO 5
REPLACE/INST
0008DC12
'MOVZWL  #0400,B^0F8(FP)'
'PUSHAL  B^0F8(FP)'
'PUSHAB  (R2)'
'PUSHL   R1'
'CALLS   #03,@L^000F18B4'
EXIT
'MOVL    #100000,B^0F8(FP)'
'PUSHAL  B^0F8(FP)'
'PUSHAB  (R2)'
'PUSHL   R1'
'CALLS   #03,@L^000F18B4'
EXIT
UPDATE
EXIT