HercGui createprocess error on perfectly well functioning Hercules

[Rae Egli]

I have a perfectly well functioning Hercules environment in command prompt mode.  I now wanted to add Fish's GUI (the 64 bit version).

However, as soon as I try to invoke it, I get

------------------- cut ---------------------
x:/Hercules/Prometheus.exe -fX:/Hercules/mvs/mvs.cnf -d =l dyngui

failed with rc=2 (0x00000002) The system cannot find the file specified.
------------------- cut ---------------------

I don't know why it tries to invoke Prometheus instead of Hercules.exe.  I tried to look if a 64 bit version was called Prometheus but was unable to confirm.

Any help would be greatly appreciated.

[Gerhard Postpischil]

I'm running Hercules 3.06 2009/07/11 with HercGUI 1.11.1.5265.

When I try to use the reader on 00C (ASCII EOF options) I get a
Hercules crash. I uploaded Hercules.dmp as dump00C.zip to the
file directory - please notify when it's safe to delete.

I seem to recall reading about this problem before, but a search
turned up too many hits to go through. I thought this was fixed,
or did I miss updating some files (e.g. HercRdr)?

The second problem is weirder. I use BlueZone to start six
sessions (each definition starts the next); the code used to be
single threaded, with each start triggered by completion of the
prior one. Now they start haphazardly, which makes it difficult
to get the sessions to match the terminal definitions (their
option for a specific device address, unlike TB's Vista, doesn't
work). When the assignments don't match, I power down the
machine, wait a second, and power up again (each session has a
different retry time delay), and on this restart they usually
come in in correct sequence. However, about 5% of the time, one
of the sessions won't connect, and instead I get a never ending
flood of console messages that makes it impossible to do any
work, and necessitating shutting Hercules down completely,
ending the terminal sessions, and starting from scratch:

** Win32 porting error: invalid call to 'w32_select' from
console.c(2076): mixed set(s)
console: DBG028: select: No error

Both problems have occurred on earlier releases, also.

Gerhard Postpischil
Bradford, VT

[kerravon86]

--- In hercules-390@..., Gerhard Postpischil <gerhard@...> wrote:
>
> ** Win32 porting error: invalid call to 'w32_select' from
> console.c(2076): mixed set(s)
> console: DBG028: select: No error
>
> Both problems have occurred on earlier releases, also.

Yeah, I have that problem too (this one just now):

DBG028: select: No error
porting error: invalid call to 'w32_select' from console.c(2129): mixed set(s)
DBG028: select: No error
porting error: invalid call to 'w32_select' from console.c(2129): mixed set(s)
DBG028: select: No error
porting error: invalid call to 'w32_select' from console.c(2129): mixed set(s)
DBG028: select: No error
porting error: invalid call to 'w32_select' from console.c(2129): mixed set(s)

When I go through a sequence of something along the lines
of:

"startmvs autoipl"

and let it start up properly and fully.

Run "startterm" in c:\mvs380\conf

which starts 5 terminals really quickly.

Close them all down.

startterm again.

Close them all down again.  Last one triggers off
the select errors ...

21:38:07 HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C0
21:38:18 HHCTE007I 3270 device 00C0 client 127.0.0.1 connection closed
21:38:31 console: DBG022: recv: An existing connection was forcibly closed by t
21:39:21 console: DBG022: recv: An existing connection was forcibly closed by t
21:39:21 console: DBG021: send: An existing connection was forcibly closed by t
21:39:21 console: DBG021: send: An existing connection was forcibly closed by t
21:39:21 console: DBG021: send: An existing connection was forcibly closed by t
21:39:21 console: DBG021: send: An existing connection was forcibly closed by t
21:39:21 console: DBG021: send: An existing connection was forcibly closed by t
21:39:43 HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C0
21:40:05 console: DBG021: send: A request to send or receive data was disallowe
21:40:05 console: DBG021: send: A request to send or receive data was disallowe
21:40:05 console: DBG032: select failed: An operation was attempted on somethin
21:40:05 ** Win32 porting error: invalid call to 'w32_select' from console.c(21
21:40:05 console: DBG028: select: No error
21:40:05 console: DBG021: send: A request to send or receive data was disallowe
21:40:05 console: DBG021: send: A request to send or receive data was disallowe
21:40:05 ** Win32 porting error: invalid call to 'w32_select' from console.c(21
21:40:05 console: DBG028: select: No error
21:40:05 ** Win32 porting error: invalid call to 'w32_select' from console.c(21
21:40:05 console: DBG028: select: No error
21:40:05 ** Win32 porting error: invalid call to 'w32_select' from console.c(21
21:40:05 console: DBG028: select: No error
21:40:05 ** Win32 porting error: invalid call to 'w32_select' from console.c(21
21:40:05 console: DBG028: select: No error


I just tried again, more controlled, and this time
the problem only occurred at the end of the 3rd set
of startterm closures.

I previously debugged it to this point:

http://tech.groups.yahoo.com/group/hercules-390/message/58060

http://tech.groups.yahoo.com/group/hercules-390/message/58062

http://tech.groups.yahoo.com/group/hercules-390/message/58065

where I found code that looked wrong and asked for
some guidance, which I didn't get, possibly because
of all the other commotion due to other bugs I
discovered at the same time.

If I can get answers to those questions, I can possibly
debug it further.

Maybe it's time to put an exit into the code that
returns -1 for a file handle (which the caller is
meant to check, but doesn't) so that at least the
"internal error" is detected and exits instead of
giving silly messages in a loop?

BFN.  Paul.

[somitcw-2]

--- In hercules-390@...,
 "kerravon86" <kerravon86@...> wrote:
- - - snipped - - -

> When I go through a sequence of something along
>the lines of:
> "startmvs autoipl"
> and let it start up properly and fully.
> Run "startterm" in c:\mvs380\conf
> which starts 5 terminals really quickly.
> Close them all down.
> startterm again.
> Close them all down again.  Last one triggers off
> the select errors ...

- - - snipped - - -

   Hercules now supports opening and closing
TN3270 sessions as needed.  Do you need startterm?
Has anyone ever really needed startterm?

   Two early fixes for startterm:

1. Put sleep 2 between each terminal start.

2. Get rid of startterm.

   Both worked but the second was much more
satisfying and corrected more problems.

[kerravon86]

--- In hercules-390@..., "somitcw" <somitcw@...> wrote:
>
>    Hercules now supports opening and closing
> TN3270 sessions as needed.  Do you need startterm?

I do if I want to more reliably reproduce the obvious
Hercules bug where it goes into a loop.

Apart from that, I normally start 0 terminals, and
when I don't start 0, I start 1, using a modified
startterm that only starts 1.

So. Can you reproduce the Hercules bug that randomly
hits Gerhard? Maybe if you can it will help resolve
the randomly-hitting bug that is giving him grief.

BFN.  Paul.

[somitcw-2]

--- In hercules-390@...,
 Gerhard Postpischil <gerhard@...> wrote:
> I'm running Hercules 3.06 2009/07/11 with
>HercGUI 1.11.1.5265.  When I try to use the
>reader on 00C (ASCII EOF options) I get a
>Hercules crash. I uploaded Hercules.dmp as
>dump00C.zip to the file directory - please
>notify when it's safe to delete.

   I not involved in fixing it but am curious.
More details would help recreating the problem.
How was the reader defined.  Hercules control
file or devinit?  What were the parameters.
What were the HERCRDR parameters?  Was it:

HERCRDR 127.0.0.1:3505 A:\SOLITARE.TXT
Are you using 127.0.0.1 instead of localhost?
The HERCRDR documentation is wrong on that point.
The default localhost on many new systems is ::1
which is the IPV6 version of IPV4 127.0.0.1 .

Was 000C defined like:
000C 3505 NOTEBOOK:3505 sockdev ascii trunc eof
Does your HOSTS file equate NOTEBOOK to
IP address 127.0.0.1 ?

or are you not using HERCRDR and use:
devinit 000C A:SOLITARE.TXT ascii eof

[Steve And Grace Bovy]

 

________________________________

From: hercules-390@... [mailto:hercules-390@...] On
Behalf Of somitcw
Sent: Saturday, July 11, 2009 7:23 AM
To: hercules-390@...
Subject: [hercules-390] Re: Problems





--- In hercules-390@... <mailto:hercules-390%40yahoogroups.com>
,
"kerravon86" <kerravon86@...> wrote:
- - - snipped - - -

> When I go through a sequence of something along
>the lines of:
> "startmvs autoipl"
> and let it start up properly and fully.
> Run "startterm" in c:\mvs380\conf
> which starts 5 terminals really quickly.
> Close them all down.
> startterm again.
> Close them all down again. Last one triggers off
> the select errors ...

- - - snipped - - -

Hercules now supports opening and closing
TN3270 sessions as needed. Do you need startterm?
Has anyone ever really needed startterm?

Two early fixes for startterm:

1. Put sleep 2 between each terminal start.

2. Get rid of startterm.

Both worked but the second was much more
satisfying and corrected more problems.

>> Sorry the sleep command is not available on
Most windows systems

[somitcw-2]

--- In hercules-390@...,
 "Steve And Grace Bovy" <sg.bovy@...> wrote:
- - - snipped - - -
> Sorry the sleep command is not available on
>Most windows systems

   startterm.bat was for starting multiple CygWin
c3270 sessions, not a Windows application.

For Windows, either down load something like:

http://sourceforge.net/projects/unxutils/

or use Windows tricks like:

choice /t 2 /d y

or if you prefer the sleep.exe from the
Microsoft Windows Server 2003 Resource Kit:

http://download.microsoft.com/download/8/e/c/\
8ec3a7d8-05b4-440a-a71e-ca3ee25fe057/rktools.exe

sleep      time-to-sleep-in-seconds
sleep [-m] time-to-sleep-in-milliseconds

[somitcw-2]

--- In hercules-390@...,
 "kerravon86" <kerravon86@...> wrote:

> --- In hercules-390@...,
>"somitcw" <somitcw@> wrote:
>> Hercules now supports opening and closing
>>TN3270 sessions as needed.  Do you need startterm?
> I do if I want to more reliably reproduce the
>obvious Hercules bug where it goes into a loop.
> Apart from that, I normally start 0 terminals, and
>when I don't start 0, I start 1, using a modified
>startterm that only starts 1.
> So. Can you reproduce the Hercules bug that randomly
>hits Gerhard? Maybe if you can it will help resolve
>the randomly-hitting bug that is giving him grief.
> BFN.  Paul.

   I created breakterm.bat for wc3270 and QWS3270Secure.
wc3270 started all seven terminals but the third one
was hung.  I could not unhang with reset,attn,pa1,clear,
or other keys.  QWS3270Secure also started all seven
terminals but the second one was hung.  Could not unhang.

   startterm hanging terminals is not new.  That is
why sleep was added by anyone that tried to use it

[kerravon86]

--- In hercules-390@..., "somitcw" <somitcw@...> wrote:
>
> http://sourceforge.net/projects/unxutils/

Excellent!

C:\scratch\iiii\xxx\usr\local\wbin>objdump -p sleep.exe | grep "DLL Name"
        DLL Name: WSOCK32.dll
        DLL Name: MSVCRT.dll
        DLL Name: KERNEL32.dll

> or use Windows tricks like:
>
> choice /t 2 /d y

Cool!

C:\scratch\bbbb>choice /t 2 /d y >nul:


BFN.  Paul.

[kerravon86]

--- In hercules-390@..., "somitcw" <somitcw@...> wrote:

>
> --- In hercules-390@...,
>  "kerravon86" <kerravon86@> wrote:
> > --- In hercules-390@...,
> >"somitcw" <somitcw@> wrote:
> >> Hercules now supports opening and closing
> >>TN3270 sessions as needed.  Do you need startterm?
> > I do if I want to more reliably reproduce the
> >obvious Hercules bug where it goes into a loop.
> > Apart from that, I normally start 0 terminals, and
> >when I don't start 0, I start 1, using a modified
> >startterm that only starts 1.
> > So. Can you reproduce the Hercules bug that randomly
> >hits Gerhard? Maybe if you can it will help resolve
> >the randomly-hitting bug that is giving him grief.
> > BFN.  Paul.
>
>    I created breakterm.bat for wc3270 and QWS3270Secure.
> wc3270 started all seven terminals but the third one
> was hung.  I could not unhang with reset,attn,pa1,clear,
> or other keys.  QWS3270Secure also started all seven
> terminals but the second one was hung.  Could not unhang.
>
>    startterm hanging terminals is not new.  That is
> why sleep was added by anyone that tried to use it

But that is a possibly separate bug in Hercules.

Are you able to get the select error in a loop after
you have the hung terminal?

BFN.  Paul.

[somitcw-2]

--- In hercules-390@...,
 "kerravon86" <kerravon86@...> wrote:
- - - snipped - - -
>> startterm hanging terminals is not new.  That is
>>why sleep was added by anyone that tried to use it
> But that is a possibly separate bug in Hercules.
> Are you able to get the select error in a loop
>after you have the hung terminal?
> BFN.  Paul.

Vista Home Premium
Pentium DUO Core 2
Hercules pre-built on Jun 5 2009 at 17:12:56

   No error messages.  Here is an attempt to connect
six QWS3270Secure sessions from my break.bat :

HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C0
HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C1
HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C2
HHCTE009I Client 127.0.0.1 connected to 3270 device 0:01C0
HHCTE009I Client 127.0.0.1 connected to 3270 device 0:01C1

Note: Hercules does not show the hung terminal as being
connected.

Then I tried seven wc3270 connections.  The first hung
and the others showed:

Connect to 127.0.0.1, port 3270: No connection could be
made because the target machine actively refused it.
[Press <Enter>]

P.S. The rktools sleep.exe was the one I liked best.
It allowed seconds and milli-seconds instead of
seconds, minutes, hours, and days.  Like you, I also
used >nul for: choice /t 1 /d y >nul

[fish-8]

 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Gerhard Postpischil wrote:

> I'm running Hercules 3.06 2009/07/11 with HercGUI 1.11.1.5265.

What Windows flavor? 32-bit Windows XP? 64-bit Vista? Etc...


> When I try to use the reader on 00C (ASCII EOF options) I get
> a Hercules crash. I uploaded Hercules.dmp as dump00C.zip to
> the file directory - please notify when it's safe to delete.

You can delete it.

It's unfortunately completely worthless without the associated
debugging symbols to go along with it.  :(

And this is not something YOU can provide, Gerhard.

Rather, Ivan is supposed to be providing them with his daily
snapshots.

(Ivan? Can this get fixed? Thanks!)

Normally debugging symbols (*.pdb files) are not ever needed for
anything. You DON'T need them to run Hercules. The only time you DO
need them is whenever you're debugging a crash dump. When that's the
case it's *critical* to have them. Without them crash dumps are
unfortunately completely 100% worthless. :(

Also, since we're on the subject, IMO the Hercules Windows
installation package should include a recent version of the
DbgHelp.dll redistributable as well. I've often found that when
analyzing crash dumps the version of DbgHelp.dll that was on the
user's system (that was used to *create* the crash dump with) was too
old to provide a meaningful dump.

It'd be nice if Hercules could distribute a relatively recent version
of DbgHelp.dll along with its binaries. That way whenever someone
does encounter Hercules crash, we can be reasonably assured the dump
will be good.

But of course no matter how "good" the dump it, it's worthless
without the .PDB files.

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBSllIvEj11/TE7j4qEQKACgCg5SIq6fzZ9w1kc9Sx+IWTr7LGwSQAoL7e
kSFonzy+d8wDoLixjRduEOn0
=n1R4
-----END PGP SIGNATURE-----

[kerravon86]

--- In hercules-390@..., "somitcw" <somitcw@...> wrote:

>
> --- In hercules-390@...,
>  "kerravon86" <kerravon86@> wrote:
> - - - snipped - - -
> >> startterm hanging terminals is not new.  That is
> >>why sleep was added by anyone that tried to use it
> > But that is a possibly separate bug in Hercules.
> > Are you able to get the select error in a loop
> >after you have the hung terminal?
> > BFN.  Paul.
>
> Vista Home Premium
> Pentium DUO Core 2
> Hercules pre-built on Jun 5 2009 at 17:12:56
>
>    No error messages.  Here is an attempt to connect
> six QWS3270Secure sessions from my break.bat :
>
> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C0
> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C1
> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C2
> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:01C0
> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:01C1
>
> Note: Hercules does not show the hung terminal as being
> connected.
>
> Then I tried seven wc3270 connections.  The first hung
> and the others showed:
>
> Connect to 127.0.0.1, port 3270: No connection could be
> made because the target machine actively refused it.
> [Press <Enter>]

So, you have so far been unable to reproduce the
problem of the w32select going into a loop?

Did you try closing all those terminals and then
starting up a new batch?

If that doesn't do the trick, does c3270 show the
problem instead?

BFN.  Paul.

[kerravon86]

--- In hercules-390@..., "Fish" <fish@...> wrote:
>
> > When I try to use the reader on 00C (ASCII EOF options) I get
> > a Hercules crash. I uploaded Hercules.dmp as dump00C.zip to
> > the file directory - please notify when it's safe to delete.
>
> You can delete it.
>
> It's unfortunately completely worthless without the associated
> debugging symbols to go along with it.  :(

Can't that be recreated from the source code
from SVN?

BFN.  Paul.

[Gerhard Postpischil]

somitcw wrote:
>    I not involved in fixing it but am curious.
> More details would help recreating the problem.
> How was the reader defined.  Hercules control
> file or devinit?  What were the parameters.
> What were the HERCRDR parameters?  Was it:

The config file has 000C NOTEBOOK:3505 sockdev ascii trunc eof

The command issued by HercGUI was devinit 000C A:solitare.txt
ASCII EOF

NOTEBOOK on a ping shows as 192.168.0.20

FWIW, the machine has a wireless PCMCIA card that's disabled.
Only the ethernet port is enabled. Running Win XP w/SP3 and all
current fixes.


Gerhard Postpischil
Bradford, VT

[Gerhard Postpischil]

Fish wrote:
> What Windows flavor? 32-bit Windows XP? 64-bit Vista? Etc...

XP w/SP 3 and all current fixes.

> You can delete it.

Done.

> But of course no matter how "good" the dump it, it's worthless
> without the .PDB files.

Most of us have been around long enough to read mainframe dumps
without looking at the manuals or the listings <g>  How long
will it take for the PC platforms???




Gerhard Postpischil
Bradford, VT

[Binyamin Dissen]

On Sun, 12 Jul 2009 02:33:27 -0400 Gerhard Postpischil <gerhard@...>
wrote:

:>Fish wrote:
:>> What Windows flavor? 32-bit Windows XP? 64-bit Vista? Etc...

:>XP w/SP 3 and all current fixes.

:>> You can delete it.

:>Done.

:>> But of course no matter how "good" the dump it, it's worthless
:>> without the .PDB files.

Don't understand that as well. One would think that the PDB (whatever it is)
might make it easier. But required??

:>Most of us have been around long enough to read mainframe dumps
:>without looking at the manuals or the listings <g>  How long
:>will it take for the PC platforms???

As soon as someone build a SADUMP for it.

--
Binyamin Dissen <bdissen@...>
http://www.dissensoftware.com

Director, Dissen Software, Bar & Grill - Israel


Should you use the mailblocks package and expect a response from me,
you should preauthorize the dissensoftware.com domain.

I very rarely bother responding to challenge/response systems,
especially those from irresponsible companies.

[somitcw-2]

--- In hercules-390@...,
 "kerravon86" <kerravon86@...> wrote:

> --- In hercules-390@..., "somitcw" <somitcw@> wrote:
>> --- In hercules-390@...,
>>  "kerravon86" <kerravon86@> wrote:
>> - - - snipped - - -
>>>> startterm hanging terminals is not new.  That is
>>>>why sleep was added by anyone that tried to use it
>>> But that is a possibly separate bug in Hercules.
>>> Are you able to get the select error in a loop
>>>after you have the hung terminal?
>>> BFN.  Paul.
>> Vista Home Premium
>> Pentium DUO Core 2
>> Hercules pre-built on Jun 5 2009 at 17:12:56
>> No error messages.  Here is an attempt to connect
>> six QWS3270Secure sessions from my break.bat :
>> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C0
>> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C1
>> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:00C2
>> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:01C0
>> HHCTE009I Client 127.0.0.1 connected to 3270 device 0:01C1
>> Note: Hercules does not show the hung terminal
>>as being connected.
>> Then I tried seven wc3270 connections.  The first
>>hung and the others showed:
>> Connect to 127.0.0.1, port 3270: No connection
>>could be made because the target machine actively
>>refused it.
>> [Press <Enter>]
> So, you have so far been unable to reproduce the
>problem of the w32select going into a loop?

   I don't see any w32select message.  Do I need
some type of tracing turned on to see errors?

> Did you try closing all those terminals and then
>starting up a new batch?

   No, I just kept starting a new batch.
My system has 35 CRTs defined so there is it takes
awhile to run out.

> If that doesn't do the trick, does c3270 show the
>problem instead?
> BFN.  Paul.

   I don't have CygWin installed so use the wc3270
Windows version of c3270.  I also use QWS3270Secure.
There are a few others I can try but the startterm
hung terminal problem is not new.  Switching TN3270
products may display slightly different symptoms but
has never been reported to fix Hercules.

   I do have one additional symptom.  With QWS3270Secure
having one hung terminal then wc3270 having it's first
terminal hung and others with the "target machine
actively refused it" error message, quit froze Hercules.
MVS was already down with the normal quiesce wait state.
When I closed the first wc3270 ( the hung wc3270 ),
then Hercules ended properly from the earlier quit.

[somitcw-2]

--- In hercules-390@...,
 Gerhard Postpischil <gerhard@...> wrote:
> somitcw wrote:
>> I not involved in fixing it but am curious.
>>More details would help recreating the problem.
>>How was the reader defined.  Hercules control
>>file or devinit?  What were the parameters.
>>What were the HERCRDR parameters?  Was it:
> The config file has 000C NOTEBOOK:3505 sockdev
>ascii trunc eof

   000C starts life as as socket reader.

> The command issued by HercGUI was
> devinit 000C A:solitare.txt ASCII EOF

   000C then becomes a file reader.

> NOTEBOOK on a ping shows as 192.168.0.20

   If 192.168.0.20 is the PC that Hercules is
running on and matches what HERCRDR is pointing
to, it should work?  A DOS Prompt command
netstat -a -n
may show if 192.168.0.20:3505 is listening?

> FWIW, the machine has a wireless PCMCIA card
>that's disabled.  Only the ethernet port is
>enabled. Running Win XP w/SP3 and all
>current fixes.
> Gerhard Postpischil
> Bradford, VT

   No HERCRDR command?  Is that because 000C
was not a socket reader at the time?