SYSIPC/ICC primitives for process with SDA

[Ruslan R. Laishev]

Hi All!

As I undertand there is not any facilities to see what ICC/handles has been allocated for/by process (like SDA's show proc /chan,/fand/buff)?!

So, I'd like to know about any plans of HP VMS-guys on this issue.

Thanks!

[Robert Brooks_1]

I don't support any of the interprocess communications components, so my statement cannot be taken as official, but it's likely there is no plan to enhance SDA in the manner you suggest.

In my experience (and I have added a fair amount to SDA over the years), displays get added and/or modified for debugging purposes during the development of a component. If, after something has been released, it is decided that some additional information would be useful, SDA may get some more attention, but in general, if a component has been available for some time, and SDA has not been enhanced to have knowledge of that component, then it is not likely that support will be added.

Note that VMS Engineering views SDA as pretty much private to VMS Engineering, so while we do document certain commands, the output of those commands could change pretty much without notice (and frequently does).


-- Rob Brooks (I/O Exec group)

[Ruslan R. Laishev]

Thanks, Robert!


In 7.3-2/Alpha IPC$SDA missing HLB file, so it need to spend some time in browsing VMS sources to understand what useful can be obtained with the IPC$SDA stuff.

ICC is the great stuff (!) but debuging of program is enough hard process...

Anyway, let's me hope that the *.HLB file will be added in the VMS, and some additional information about $ICC service will be added in the docs.


Can I ask to explaint some more (than in the docs) about of condition SS$_WRONGSTATE after $ICC_TRANSMITEW ?

[Volker Halle]

Rob,

maybe you could forward a pointer to this thread to the current IPC/ICC maintainer (please include my regards) and ask, whether it would be possible to publish some more detailled info about ICC troubleshooting or maybe write an article about ICC for the next OpenVMS Technical Journal ?


Ruslan,

SS$_WRONGSTATE is declared in the SYS$ICC_TRANSMIT system service, if the ICCPCB$L_STATE of the ICCPCB (internal structure representing the Connection Handle) is not ICCPCB$C_OPEN. This could happen, if the other end of the connection has somehow failed.

You can find the IPCPCBs with SDA like this:

SDA> exa @ICC$GL_ICC_PDB_VECTOR;4*(@SGN$GW_MAXPRCCT&ffff)

will show the ICCPDB vector. Non-zero entries point to ICCPDBs. Each ICCPDB ist 0x40 bytes long and - after the usual header of 0x10 bytes, has 2 links to ICCPABs and then 2 links to ICCPCBs. SDA detects these data structures, but can't format them due to missing symbols.

ICCPCB$L_STATE is at offset 72. of the ICCPCB.

There is a nice diagram of the ICC DATA STRUCTURE RELATIONSHIPS in [IPC]ICC_IDEF.SDL

Volker.

[Ruslan R. Laishev]

Nice suggestion, Volker! VMS-guys, please, add my vote also!


Thanks for the pointers, Volker!