Conference hydra::axp-developer

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121" 14-FEB-1997 16:58:08.91
To:	nm%vbormc::"[email protected]"
CC:	HUDSON
Subj:	RE:  19903, Iona Technologies TO ASAP READING: c++ exceptions loop    

Hello John Morrow,

Thanks for your asap query on looping exceptions.

If you could provide a small example that would be ideal.

I've been having a look for problems similar, The one that came up is related
to someone using pthread calls in a C++ program.  Are you using pthread or
DECthread stuff at all in your code?

Incidentally, the other problem with destructors being called twice has been
acknowledged by the engineers who say they're "looking into it", I'll let you
know what comes of that.

nick

From:	VBORMC::"[email protected]" "John Morrow" 14-FEB-1997 18:09:49.72
To:	"[email protected] - UK Software Partner Engineering 830-4121
14-Feb-1997 1658 +0000" <[email protected]>
CC:	[email protected], [email protected]
Subj:	Re: 19903, Iona Technologies TO ASAP READING: c++ exceptions loop 

Hi Nick,

Yes, the code I'm building is C++ and multi-threaded using pthread calls.

Sometimes my code runs and gives this exception system message at the end, 
other times it cores and the core dump traces show the core occuring due to 
some pthread call. The thing is that this isn't the easiest thing to 
reproduce, probably because of the threads, in that I don't always get this 
exception every time I run the same code. I've been trying to get a small test 
program which produces this error but the code I'm building is pretty large 
and it's not easy to strip it down.

I've tried debugging the code with a runtime debugger insure++ which is 
indicating that the heap is getting corrupted, but it's not very exact about 
where this is happening.

I've tried installing the following patch:

dUNIX_OSF400-080_ECO_summary

and I've rebuilt all my code but I still have the same exception problem.



If I can get a small example which produces this error I'll pass it on.


Thanks,

John.

% ====== Internet headers and postmarks (see DECWRL::GATEWAY.DOC) ======
% Received: from mail.vbo.dec.com (mail.vbo.dec.com [16.36.208.34]) by
vbormc.vbo.dec.com (8.7.3/8.7) with ESMTP id TAA03990 for
<[email protected]>; Fri, 14 Feb 1997 19:06:48 +0100
% Received: from server21.digital.fr (server21.digital.fr [193.56.15.21]) by
mail.vbo.dec.com (8.7.3/8.7) with ESMTP id TAA01844 for
<[email protected]>; Fri, 14 Feb 1997 19:11:33 +0100 (MET)
% Received: from dublin.iona.ie (operation.dublin.iona.ie [192.122.221.5]) by
server21.digital.fr (8.7.5/8.7) with ESMTP id TAA09102 for
<[email protected]>; Fri, 14 Feb 1997 19:13:45 +0100 (MET)
% Received: from ultra (ultra [192.122.221.136]) by dublin.iona.ie
(8.7.5/jm-1.01) with ESMTP id RAA07246; Fri, 14 Feb 1997 17:48:27 GMT
% Message-Id: <[email protected]>
% X-Mailer: exmh version 1.6.6 3/24/96
% To: "[email protected] - UK Software Partner Engineering 830-4121
14-Feb-1997 1658 +0000" <[email protected]
% cc: [email protected], [email protected]
% Subject: Re: 19903, Iona Technologies TO ASAP READING: c++ exceptions loop 
% Mime-Version: 1.0
% Content-Type: text/plain; charset=us-ascii
% Date: Fri, 14 Feb 1997 17:54:45 +0000
% From: John Morrow <[email protected]>

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121" 19-FEB-1997 10:32:57.36
To:	nm%vbormc::"[email protected]"
CC:	HUDSON
Subj:	re:19903, c++ exceptions loop

Hi John,

I asked about the patch for the problem that seems similar to yours.  The
engineering group say a patch is being worked on but isn't yet finished.

I'll let you know when I hear more.  In the meantime, if you are able to
provide any more information to confirm whether what you're seeing is the same
thing, that would be useful

Regards

Nick

From:	VBORMC::"[email protected]" "John Morrow" 19-FEB-1997 11:00:37.66
To:	"[email protected] - UK Software Partner Engineering 830-4121
18-Feb-1997 1617 +0000" <[email protected]>
CC:	[email protected], [email protected]
Subj:	Re: 19903, Iona Technologies TO ASAP READING: c++ exceptions loop 

Hi Nick,

This seems to be the same problem as I'm having. The exception system message 
is certainly the same. I'm running a suite of tests against some code and 
previously, certain tests consistently gave me this exception system. Since 
dUNIX_OSF400-080_ECO_summary patch was installed on our system, I cannot get a 
consistently reproducable example of this exception system, it just pops up 
every now and then.


Anyway, one particular test which used to give me this exception system now 
just hangs. I've tried running this test using a runtime debugger "insight" 
and it causes this test to core dump saying....


	DECthreads Last Chance handler: thread 1 exiting on status exception 
0x177db005
	Exception: Invalid memory address (dce / thd)
	IOT/Abort trap (core dumped)


I've run it a good few times and examined the core files. The most common 
trace is this....



Thread 0x2 terminated at PC 0x3ff8053eab0 by signal IOT
(ladebug) t
>0  0x3ff8053eab0 in nxm_thread_kill(0x3ffc102fe10, 0x3ffc102e338, 0x6, 
0x3ffc102e31
	0, 0x2, 0x14030cd00) DebugInformationStrippedFromFile19:???

 #1  0x3ff8101edbc in pthread_kill(0x3ffc1030590, 0x14030cd00, 0x2130455, 
0x3ff8100bd
	e0, 0x0, 0x140319e78) DebugInformationStrippedFromFile99:???

 #2  0x3ff810253fc in UnknownProcedure10FromFile103(0x2130455, 0x3ff8100bde0, 
0x0, 0x
	140319e78, 0x3ff810bdb14, 0x3ffc102ec50)
DebugInformationStrippedFromFile103:?
??

 #3  0x3ff810bdb10 in /usr/shlib/libc.so

 #4  0x3ff81107f30 in raise(0x3ff810bdb14, 0x3ffc102ec50, 0x3ff81107f34, 
0x3ff8102532
	8, 0x3ff8111ea8c, 0x0) DebugInformationStrippedFromFile441:???

 #5  0x3ff8111ea88 in abort(0x140319438, 0x0, 0x0, 0x0, 0xffffffff00000000, 
0x177db00
	5) DebugInformationStrippedFromFile329:???

 #6  0x3ff810155c4 in errAbort(0x0, 0x0, 0xffffffff00000000, 0x177db005, 
0x3ff81015e5
	c, 0x3ffc102e2a0) DebugInformationStrippedFromFile90:???

 #7  0x3ff81015e58 in UnknownProcedure2FromFile91(0x3ff807b3728, 0x0, 
0x1403196f8, 0x
	3ff81016324, 0x1403199d8, 0x3ff807b3488)
DebugInformationStrippedFromFile91:??
?

 #8  0x3ff807b220c in UnknownProcedure4FromFile1(0x3ff807b3624, 0x0, 
0x1403196f8, 0x3
	ff810241a0, 0x140319a48, 0x3ff81016324)
DebugInformationStrippedFromFile1:???

 #9  0x3ff807b3620 in UnknownProcedure16FromFile1(0x0, 0x140319e78, 
0x140319e78, 0x14
	0318f18, 0xabadabad00beed00, 0x140305b0f)
DebugInformationStrippedFromFile1:??
?

 #10 0x3ff807b3664 in exc_unwind(0x140319e78, 0x140318f18, 0xabadabad00beed00, 
0x1403
	05b0f, 0x3ff807b254c, 0x140305ad0) DebugInformationStrippedFromFile1:???

 #11 0x3ff807b2548 in exc_raise_exception(0xffffffff, 0x140319a20, 
0x140308760, 0x0,
	0x1201025b0, 0x140308860) DebugInformationStrippedFromFile1:???

 #12 0x3ff81016324 in pthread_exc_raise_np(0x0, 0x140307ab0, 0x3ff81023f88, 
0x120104e
	1c, 0x14030cd00, 0x140308720) DebugInformationStrippedFromFile91:???

 #13 0x3ff810241a0 in thdBase(0x0, 0x0, 0x0, 0x0, 0x45586732, 0x3) 
DebugInformationSt
	rippedFromFile101:???

(ladebug)

.....or sometimes #12 is a call to __sigtramp and #11 is 
exc_raise_signal_exception.
I hope this is not a problem with my runtime debugger, but the trace does show 
exc_raise_signal_exception and thdBase which seem to agree with the 
diagnostics in your last message.


You also said:

>	  If you avoid using DCE and DECthreads exceptions, then
>          you can safely use C++ exceptions in multi-threaded
>          programs so long as the handling of an exception does
>          not cause the destruction of a running thread. The DEC
>          C++ Class Library task package has a similar restriction
>          pertaining to return from such a constructor. [1445]
>
>
>The customer's interpretation of this is that you cannot mix c++ exceptions
>and pthread exceptions, but you can still use pthread exceptions by
>themselves.  True?

My code uses C++ exceptions and is multi-threaded. How do I avoid using DCE 
and DECthreads exceptions, I know that for some other systems you can get 
information about the status of a thread routine from a thread exception or 
from a status value which the routine returnes to the thread. Is there 
something similar on Digital UNIX?

(By the way, what's the difference between DCE, DECthreads and pthreads, does 
DECthreads contain all pthread routines plus other thread routines?)

Thanks,

John.

From:	VBORMC::"[email protected]" "John Morrow" 20-FEB-1997 12:45:15.94
To:	"[email protected] - UK Software Partner Engineering 830-4121
19-Feb-1997 1116 +0000" <[email protected]>
CC:	
Subj:	Re: 19903, c++ exceptions loop 

Hi Nick,

I'm using Digital UNIX v4.0A but have a copy of v4.0B. It says there are some 
thread patches. Is it worth installing v4.0b?

John

% ====== Internet headers and postmarks (see DECWRL::GATEWAY.DOC) ======
% Received: from mail.vbo.dec.com (mail.vbo.dec.com [16.36.208.34]) by
vbormc.vbo.dec.com (8.7.3/8.7) with ESMTP id NAA11495 for
<[email protected]>; Thu, 20 Feb 1997 13:42:05 +0100
% Received: from server21.digital.fr (server21.digital.fr [193.56.15.21]) by
mail.vbo.dec.com (8.7.3/8.7) with ESMTP id NAA07479 for
<[email protected]>; Thu, 20 Feb 1997 13:46:50 +0100 (MET)
% Received: from dublin.iona.ie (operation.dublin.iona.ie [192.122.221.5]) by
server21.digital.fr (8.7.5/8.7) with ESMTP id NAA22874 for
<[email protected]>; Thu, 20 Feb 1997 13:49:05 +0100 (MET)
% Received: from ultra (ultra [192.122.221.136]) by dublin.iona.ie
(8.7.5/jm-1.01) with ESMTP id MAA05450 for <[email protected]>; Thu, 20
Feb 1997 12:23:36 GMT
% Message-Id: <[email protected]>
% X-Mailer: exmh version 1.6.6 3/24/96
% To: "[email protected] - UK Software Partner Engineering 830-4121
19-Feb-1997 1116 +0000" <[email protected]
% Subject: Re: 19903, c++ exceptions loop 
% Mime-Version: 1.0
% Content-Type: text/plain; charset=us-ascii
% Date: Thu, 20 Feb 1997 12:29:52 +0000
% From: John Morrow <[email protected]>

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121" 20-FEB-1997 13:45:45.75
To:	nm%VBORMC::"[email protected]"
CC:	HUDSON
Subj:	Re: 19903, c++ exceptions loop 

Hi John

My thread-expert colleague tells me that the patches he knows of relating to
threads are to do with performance improvements rather than fixing crashing
programs.  So I don't think it's particularly likely that 4.0b will fix your
problem.  If you do install it though, let me know whether it makes any
difference....

nick

From:	WATNOW::"[email protected]"  2-APR-1997 12:56:29.43
To:	[email protected]
CC:	[email protected], [email protected]
Subj:	Re: 19903

Hi,

I was wondering if you could give me a date when the patch for the c++ 
exceptions loop problem will be available.

Thanks,

John.

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121"  2-APR-1997 14:28:30.39
To:	nm%vbormc::"[email protected]"
CC:	HUDSON,nm%vbormc::"[email protected]"
Subj:	Re: 19903

Hello John

Thanks for your reminder.  I've asked the engineering group for an update

nick

From:	WATNOW::"[email protected]"  4-APR-1997 09:47:04.49
To:	[email protected]
CC:	[email protected], [email protected]
Subj:	Re: 19903

Hi Nick,

I'm going to be out of the office for the next two weeks until the 21st and I 
won't be reading my email. My boss, Aileen Cunningham, will be looking after 
my work while I'm away so if your sending me any news on the exceptions patch 
will you also cc [email protected] on the email. 

Thanks in advance,

John.

Had a call from Chris Coggrave @OLO (dtn 851-1085, mobile 0831-656696).

He is involved with SI project at Kelloggs; they want to use Iona's product
on Unix but Iona is holding up delivery of the unix v4 product until the
problem referred to in this note is resolved.

I've said to Chris that we don't know for sure that this problem is the
same as that in the CMA notesfile #1487, but that we are waiting for patch
to be able to check, as symptoms are similar.

Since my last two requests for info in the CMA notesfile have been ignored
I said best formal way to follow this up is for Chris to log call on
Iona's behalf at the CSC so they can follow up the IPMT.  I will also
attempt to contact engineering group to get update.

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121" 15-APR-1997 10:51:05.86
To:	mts$"olo::chris coggrave"
CC:	kernel::birkinshaw,HUDSON
Subj:	patch for Iona info

Hi Chris

I spoke with the DECthreads product manager (Jean Fullerton) yesterday, and she
subsequently entered the following note.

If you need to have more definite idea of the patch release date (which has not
yet been provided, I guess because they don't know), I suggest you contact her
directly so that you can make your concerns known to her

In the meantime I'll monitor the specified web page

Regards
nick


           <<< CLT::DISK$CLT_LIBRARY3:[NOTES$LIBRARY]CMA.NOTE;1 >>>
                           -< DECthreads Conference >-
================================================================================
Note 1487.6   Patch for non-native handling of Signal -> exception        6 of 6
PTHRED::J_FULLERTON "Jean Fullerton (ZKO)"           11 lines  14-APR-1997 11:48
                       -< UNIX V4.0 BL5 patch kit info >-
--------------------------------------------------------------------------------
> So, any progress?

The patch for this problem is included in the UNIX V4.0 BL5 patch kit.
It should be available soon.  Please check web page 
  http://www.zk3.dec.com/dupatchwww/patch-schedule/kits.html
for its status and 
  http://www.zk3.dec.com/dupatchwww/
for information on how to obtain the patch kit.

Regards,
Jean

had call from micky balladelli in VBO wanting status (Iona gave him my
name).

Pointed him at notes in cma file

his mail address is [email protected]

From:	WATNOW::"[email protected]" 22-APR-1997 10:42:26.71
To:	[email protected]
CC:	
Subj:	Re: 19903

Hi Nick,

Can you let me know when you expect the exceptions patch will be available?

Thanks in advance,

John.

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121" 22-APR-1997 10:55:21.94
To:	nm%WATNOW::"[email protected]"
CC:	HUDSON
Subj:	Re: 19903

Hi John

I'm afraid I have no date at the moment.  Looking at the current patch status,
they say that the kit has been built, is being tested, and that a minor fault
was identified and fixed on 19/apr.  The last comment is that verification of
the latest kit started on 21/apr.  So I suppose it depends on whether they find
any other problems.  If not, then I would hope that it should be released
fairly soon.  Sorry I can't be more specific, but I will keep checking the
status and as soon I have news I'll let you know.

nick

From:	DEC:.REO.REOVTX::HUDSON       "[email protected] - UK Software
Partner Engineering 830-4121" 13-MAY-1997 09:29:01.88
To:	WATNOW::"[email protected]"
CC:	HUDSON
Subj:	Re: 19903

Hi John

The Unix patch that contains patches for DECthreads has become available.

There are different versions depending on what variant of V4 you have (e.g.
V4.0, V4.0a, V4.0b).

Can you let me know what version of Unix you need the patch for (maybe all?)
and where I can send it?  Unfortunately I don't believe it's available on the
public Digital web pages, but I can put it on a DAT tape and send it to you

Regards
Nick

From:	WATNOW::"[email protected]" 14-MAY-1997 09:51:52.66
To:	[email protected]
CC:	
Subj:	Re: 19903 

Hi Nick,

We actually got the patch a week or two ago, for 4.0b, but I've been up to my 
ears
with other stuff and I haven't had a chance to test it yet. 

When I get a chance to test it I'll let you know.

Many thanks,

John.