[Info-vax] COD inconsistency

[John H. Reinhardt]

On 8/29/2020 4:05 AM, Volker Halle wrote:
>> Worth doing?
> 
> It's not much work, please try this:
> 
> $ ANALYZE/SYS
> SDA> SET PROC OPCOM
> SDA> exa 00040910;50  ! offset for V8.3, may also work for V8.4
> 00000000 00000000 00000000 00000000  ................     00000000.00040910
> 00000001 00000010 00000080 00000000  ................     00000000.00040920
> 00000000 00000000 00000000 00000000  ................     00000000.00040930
> 00040948 00040948 00040940 00040940  @... at ...H...H...     00000000.00040940
> 00000000 00000000 00040950 00040950  P...P...........     00000000.00040950
> 
> 
> The first 2 longwords are COD_ALLOCATED and COD_FREE_COUNT. In an idle OPCOM process, those values should be the same.
> The 3 longwords starting at VA 40924 are constants and could be used with SDA> SEARCH to find this data, should the P0 addresses be different in V8.4.
> 
> The quadwords at 40940 and 40948 are the queue headers of the COD_FREE_QUEUE and COD_BUSY_QUEUE. If the queues are empty, those longwords contain the address of the queue header (as shown above).
> 
> I'm starting to believe, that the values shown in your OPCOM message, are not real counts, but the results of local memory corruption in OPCOM P0 space.
> 
> Volker.
> 

For what it's worth, here is VSI V8.4-2L1 Alpha

$ show system/noproc
OpenVMS V8.4-2L1  on node GORDON   29-AUG-2020 15:24:43.38   Uptime  0 22:40:06
$
$ analy/sys

OpenVMS system analyzer

SDA> SET PROC OPCOM
SDA> exa 00040910;50
00000000 00000000 00000000 00000000  ................     00000000.00040910
00000001 00000010 00000080 00000000  ................     00000000.00040920
00000000 00000000 00000000 00000000  ................     00000000.00040930
00040948 00040948 00040940 00040940  @... at ...H...H...     00000000.00040940
00000000 00000000 00040950 00040950  P...P...........     00000000.00040950


-- 
John H. Reinhardt