Keyboard Shortcuts
Likes
- H390-Vm
- Messages
Search
|
Re: How To Capture CP SPOOL Command Output
#VMCE
On Sun, May 8, 2022 at 10:13 PM, Joe Monk wrote:
As I understand EXECIO, if youre supplying a CP command on the same line, you have to use the STRING option...Joe, Thank you verymuch! I read the online help, but didn't understand. I guess I will also go look at other EXECIO documentation as well. I ended up with the following, which gets me the spoolid I wanted. 'EXECIO * CP ( STRING SPOOL CONSOLE STOP CLOSE TERM' ?????????????????????????????????????????????????? ? PULL . . Spoolid .???????????????????????????????? ? SAY Spoolid???????????????????????????????????????? ?... Mark S. |
|
Re: How To Capture CP SPOOL Command Output
#VMCE
As I understand EXECIO, if youre supplying a CP command on the same line, you have to use the STRING option... As an example: EXECIO * CP (SKIP STRING Q userid Which would issue the CP command Q userid Joe On Sun, May 8, 2022 at 8:07 PM Mark A. Stevens via <marXtevens=[email protected]> wrote: I am trying to capture the line generated by CP when I close the spooled console, so I can get the spoolid. The following EXEC is pared down to the minimum, yet EXECIO doesn't seem to stack the output. |
|
How To Capture CP SPOOL Command Output
#VMCE
I am trying to capture the line generated by CP when I close the spooled console, so I can get the spoolid. The following EXEC is pared down to the minimum, yet EXECIO doesn't seem to stack the output.
// /* FileName:? REXPARS EXEC??????????????????????????????????????????? */ ADDRESS 'COMMAND'????????????????????? /* Strict command evaluation.? */ PARSE UPPER ARG fn ft fm . /* 'CP SPOOL CONSOLE START TO * NOTERM' */ 'CP SPOOL CONSOLE START TO *' 'SCAN' fn ft fm /* 'CP SPOOL CONSOLE STOP CLOSE TERM' */ 'EXECIO * CP SPOOL CONSOLE STOP CLOSE TERM' PULL Spoolid SAY Spoolid RETURN I get the following, abbreviated,output. It sits at VM READ until I type something and hit enter (the PULL) and echoes it. rexpars SOURCE?? G27AJCL? H Enter pattern: add name= ? File: SOURCE G27AJCL H1. 16: ./ ADD NAME=ASMGF1???????? ? 2213: ./ ADD NAME=ASMGASM??????? ? ... 41849: ./ ADD NAME=ASMGRTA????????????? ? 42071: ./ ADD NAME=ASMGUP?????????????? ? Invalid operand: SPOOL. x X Ready(00024); T=1.76/2.64 18:42:57 I have tried the commands just entered at the console, as well. In the second run, below, I actually tried 'SPOOL CONSOLE STOP CLOSE TERM ( STACK', hoping against hope that CMS would intervene and stack the output, but it got swallowed up, without working. CP SPOOL CONSOLE START TO * NOTERM Enter pattern: CON FILE 1731? TO? XMAS???? COPY 01 NOHOLD Ready; T=0.01/0.01 18:47:20 CP SPOOL CONSOLE START TO * NOTERM Enter pattern: CON FILE 1732? TO? XMAS???? COPY 01 NOHOLD Ready; T=0.01/0.01 18:48:05 Using SFBROWSE to see what happened ... ... 42071:??????????????????????????????? ./ ADD NAME=ASMGUP??????????????????? ? Ready; T=1.35/1.90 18:47:47?????????? ? SPOOL CONSOLE STOP CLOSE TERM ( STACK ? (??????? NOT IN CP DIRECTORY????????? ? Ready(00053); T=0.01/0.01 18:47:57??? ? SPOOL CONSOLE STOP CLOSE TERM???????? ? * * *? END OF FILE? * * *????????????? Any suggestions, including "Just go away," are appreciated :-) ?... Mark S. |
|
Re: ECMODE on in every user directory entry
¿ªÔÆÌåÓýKris, ? Thanks for that. I had forgotten the trick but it works just fine. I used ¡°set input¡± to map ¡°~¡±(Tilde) ?to 0x15 and created a file with EDIT with one line ? CP DEFINE STOR 10M~CP SET ECMODE ON~CP IPL CMS PARM AUTOCR ? And that works just fine. I used EDIT as EE uses full screen IO and so by-passes the input character mapping. ? Dave ? From: [email protected] <[email protected]> On Behalf Of kris.buelens@...
Sent: 01 May 2022 16:05 To: [email protected] Subject: Re: [h390-vm] ECMODE on in every user directory entry ? > &CONTROL OFF > &PUSH CP IPL 123 > CP SET ECMODE ON -- |
|
Re: ECMODE on in every user directory entry
I believe its to do with timers. I see that the WAITC user command on the various n-pack uses the interval timer which needs ECMODE...Out of curiosity, Dave, why did such packages require ECMODE? ... which is why I suggested setting it on for "normal users".... Looking at Pages 166-167 detail several differences in the behaviour of the timers depending on the setting of ECMODE. We also required TIMER REAL. This does impose quite a load on CP as it has to do a lot more checking of timers for VMs in virtual wait state. Harold GrovesteenDave G4UGM |
|
Re: ECMODE on in every user directory entry
> &CONTROL OFF > &PUSH CP IPL 123 > CP SET ECMODE ON Kris Buelens |
|
Re: ECMODE on in every user directory entry
"Does anyone know why ECmode is on for all users in the user directory for VMCE?"
ECMODE is set for only these users in VMCE 1.1.1: CPWATCH DOSSTART DOSVS GCCCMS MAINT MAINTC RSCS MVS These are system install and service machines, or machines dedicated to guest operating systems.? Bob |
|
Re: ECMODE on in every user directory entry
On Sun, 2022-05-01 at 14:28 +0100, Dave Wade wrote:
Out of curiosity, Dave, why did such packages require ECMODE?-----Original Message-----to Harold Grovesteen |
|
Re: ECMODE on in every user directory entry
toggle quoted message
Show quoted text
-----Original Message-----to keep things simple(r).Olaf, One thing that changed the "default" was the widespread introduction of communications products. So I wrote X25 software for VM/CMS which was used in UK Universities. At the same time in other countries TCP/IP was starting to be used. Both these packages require ECMODE to be set on for any user that runs a communications product in their VM. Dave --else to \X/ have kids to make his activity cost neutral." -The BOFHfalu.nl@rhialto |
|
Re: ECMODE on in every user directory entry
On Sat 30 Apr 2022 at 20:32:16 +0100, Dave Wade wrote:
Historically the answer is ?it depends?. The tendency was to leaveIndeed. In my recollection (which may be imperfect of course), we computer science students (mid-late '80s, VM/SP) got our VMs running in BC mode and I have no reason to believe other users were different. I think we were told about EC mode but since some of our exercises were about using SIO for low-level disk I/O, it was recommended not to use it, to keep things simple(r). -Olaf. -- ___ "Buying carbon credits is a bit like a serial killer paying someone else to \X/ have kids to make his activity cost neutral." -The BOFH falu.nl@rhialto |
|
Re: ECMODE on in every user directory entry
On Sat, 30 Apr 2022 at 17:28, Joe Monk <joemonk64@...> wrote:
It is extremely common to first IPL CMS in the MVS VM, and have it set things up before issuing the IPL of MVS. Notably, coupling virtual CTCAs pretty much has to be done in a script (EXEC, REXX) once the VM exists. Naturally the first of a pair of VMs to be instantiated will be unable to COUPLE to another that doesn't yet exist, so the second one will have to do the work. And this can continue to more than two, of course. This is still the way things are done in the modern zVM world with guest z/OS systems, though of course ECMODE doesn't come into it. Tony H. |
|
Re: ECMODE on in every user directory entry
On Sat, 30 Apr 2022 at 17:50, Dave Wade <dave.g4ugm@...> wrote:
Its a CP command. CP commands work with nothing IPL d in the VM. In fact if CMS is IPLdI was going to point that out. One may IPL CMS in a VM that¡¯s going to run MVS or other OS so that you can tweak the configuration in ways that can¡¯t be done in the directory. For example, an EXEC with the following commands:- &CONTROL OFF Will set ECMODE ON and then IPL from device 123.Isn't this where the (undocumented but everybody knows about it?) ability to put a X'15' byte in the command gets used? So you have a single command line with SET ECMODE ON;IPL 123 where the semi-colon is actually a X'15' (newline) byte. Or am I remembering a different context? Tony H. |
|
Re: ECMODE on in every user directory entry
¿ªÔÆÌåÓýHi, JUst for grins, I checked out both Bob Adele's 3 pack system that's on my system and the VM/370 R6 Sysgen manual.? They both show only 3 users with the ECMODE on the OPTION card in the User Directory.? MAINT, RSCS and a curious user called ECMODE. Regards, Gary On 4/30/22 18:29, Bob Polmanter wrote:
In most cases, there is no reason to specify ECMODE in each and every user directory entry.? IBM's own VM/370 documentation states this in so many words and also recommends against it for performance reasons.? A typical CMS user doesn't need ECMODE.? The user can do everything CMS supports: editing files, compiling, linking and executing user programs, use CMS and CP commands and more all without ECMODE. |
|
Re: ECMODE on in every user directory entry
In most cases, there is no reason to specify ECMODE in each and every user directory entry.? IBM's own VM/370 documentation states this in so many words and also recommends against it for performance reasons.? A typical CMS user doesn't need ECMODE.? The user can do everything CMS supports: editing files, compiling, linking and executing user programs, use CMS and CP commands and more all without ECMODE.
ECMODE as implemented in VM/370 allows a given virtual machine these capabilities: - use of?virtual channels 6 through F - use of virtual control registers 1 through 15? (only CR 0 is available to a non-ECMODE virtual machine). - can use the DAT bit in the PSW - can turn on the EC bit in the PSW - can use PER bit in the PSW - can issue certain S/370 privileged instructions that would otherwise give an operation exception if issued in a non-ECMODE virtual machine. Unless you have user programming that does one or more of the items above, you do not need to run ECMODE.?? Normally, only guest operating systems would require ECMODE because they do do many of these items. Note that even if ECMODE is turned on but the PSW is in BC mode (bit 12 is zero), the virtual machine is still ECMODE capable; it is simply running in BC mode.? But not the reverse:? If ECMODE is not turned on and the PSW is in BC mode, you cannot enter EC mode by trying to set bit 12 of the PSW to 1. Regarding the performance cost of ECMODE, there are two factors.? The first is the additional real storage consumed so that CP can provide EC mode services to each virtual machine enabled for ECMODE.? This includes the ECBLOK, a system control block used to store the virtual control registers, as well as an increase in the size of the virtual channel block, virtual control unit blocks, and virtual device blocks as needed when the device addresses exceed 5FF.? For Hercules users, this is usually not a problem because we're all running with 16 MB of real memory, but if you can imagine a system from 1972 or 1974 with only 320K or maybe 448K of real memory and 30 or more VM users, a policy to give ECMODE capability to every user could be costly and exacerbate the paging rate. The other factor is the increase in the code path of CP in order to handle ECMODE virtual machines.? There is a lot of code in the system dispatcher that must validate certain bits in the PSW as well as bits and addresses in virtual control registers each time any user is dispatched and given control.? When ECMODE is enabled, CP must do a litany of additional checking.? Its not insignificant.? This does impact Hercules users too, because a longer code path in CP means more instructions must be executed (emulated).? Even ECPS is not immune.? In the Hercules code that provides ECPS support to CP, there are additional checks to determine if EC mode functions are enabled or used and if they are, Hercules ECPS support has to do more work to perform the given function.?? Granted that most of us would never notice the difference at the terminal whether ECMODE was enabled or not, but the above is what ECMODE provides, as well as an indication of what it costs. Regards, Bob |
|
Re: ECMODE on in every user directory entry
¿ªÔÆÌåÓýJoe, Because that is what the manual says it does. Dave ? From: [email protected] <[email protected]> On Behalf Of Joe Monk
Sent: 30 April 2022 22:35 To: [email protected] Subject: Re: [h390-vm] ECMODE on in every user directory entry ? "Also note that what ECMODE ON does is to change the hardware being emulated by CP." ? how so? ? joe ? On Sat, Apr 30, 2022 at 2:32 PM Dave Wade <dave.g4ugm@...> wrote:
|
|
Re: ECMODE on in every user directory entry
¿ªÔÆÌåÓýJoe, ? Its a CP command. CP commands work with nothing IPL d in the VM. In fact if CMS is IPLd and you do ¡°SET MODE VM¡± you then need to re-ipl CMS because it resets the machine, as do some other CP commands ? However, its normal to have ¡°OPTION ECMODE¡± on the directory for any VM that¡¯s going to run MVS (or another OS that needs it). ? One may IPL CMS in a VM that¡¯s going to run MVS or other OS so that you can tweak the configuration in ways that can¡¯t be done in the directory. You do need to get creative, so if you issue a command that resets the machine, you need to have an IPL command on the console stack. ? For example, an EXEC with the following commands:- ? &CONTROL OFF &PUSH CP IPL 123 CP SET ECMODE ON ? Will set ECMODE ON and then IPL from device 123. ? Dave ? ? From: [email protected] <[email protected]> On Behalf Of Joe Monk
Sent: 30 April 2022 22:28 To: [email protected] Subject: Re: [h390-vm] ECMODE on in every user directory entry ? How would you issue CP SET ECMODE ON if MVS is the IPL'ed OS? CMS is never IPL'ed in the VM. ? Joe ? On Sat, Apr 30, 2022 at 2:02 PM Mark A. Stevens via <marXtevens=[email protected]> wrote:
|
|
Re: ECMODE on in every user directory entry
"Also note that what ECMODE ON does is to change the hardware being emulated by CP." how so? joe On Sat, Apr 30, 2022 at 2:32 PM Dave Wade <dave.g4ugm@...> wrote:
|
|
Re: ECMODE on in every user directory entry
On Sat, Apr 30, 2022 at 04:28 PM, Joe Monk wrote:
How would you issue CP SET ECMODE ON if MVS is the IPL'ed OS? CMS is never IPL'ed in the VM.In that particular case, you would have OPTION ECMODE in the directory entry for the userid that IPLs MVS. Dave's answer was what I needed. Sorry I wasn't clear. ?... Mark S. |
|
Re: ECMODE on in every user directory entry
How would you issue CP SET ECMODE ON if MVS is the IPL'ed OS? CMS is never IPL'ed in the VM. Joe On Sat, Apr 30, 2022 at 2:02 PM Mark A. Stevens via <marXtevens=[email protected]> wrote: On Sat, Apr 30, 2022 at 01:36 PM, Joe Monk wrote: |
|
Re: ECMODE on in every user directory entry
¿ªÔÆÌåÓýMark, ? Historically the answer is ¡°it depends¡±. The tendency was to leave ECMODE off because it can have a performance impact, so on a big system where you had lots of users, you left it off where possible. However, several products need to have ECMODE ON so you sometimes didn¡¯t have any choice¡. ? I suppose, simplistically with ECMODE OFF CP simulates a ¡°360¡± except of course its not a 360 as there is no ASCII bit¡ When you SET ECMODE ON or have the OPTION ECMODE in the directory, which sets it on at logon, then roughly speaking CP simulates an S/370. Which is why you get a reset when changing the setting¡ It¡¯s the classical function <=> performance trade-off. ? I also understand that the overhead depends on the processor. It may be that on machines with VMA or ECPS it may be less¡. ? Also note that what ECMODE ON does is to change the hardware being emulated by CP. ?CMS itself will still only load BCMODE PSWs¡.. ¡ Finally I would set ECMODE ON via an OPTIONS entry in the directory for IDs used by general users¡.. At some point in the future we will need it. ? Dave ? ? ? ? From: [email protected] <[email protected]> On Behalf Of Mark A. Stevens via groups.io
Sent: 30 April 2022 17:44 To: [email protected] Subject: Re: [h390-vm] ECMODE on in every user directory entry ? On Sat, Apr 30, 2022 at 09:44 AM, Mark Waterbury wrote:
So I am confused ... Surprise! It would seem "most" userids don't have the OPTION record, much less the ECMODE parm, but when it is specified, ECMODE is set ON at logon. Hence, if you need ECMODE on, it would be best to have it set in USER DIRECT?? ? |
Mark A. Stevens