From mikael.roming@saabgroup.com  Mon Mar 26 08:15:04 2007
From: Mikael Roming <mikael.roming@saabgroup.com>
Date: Mon, 26 Mar 2007 14:14:53 -0000
Content-Type: text/plain ; charset="us-ascii"
To: report@adacore.com
CC: Ture Andersen <ture.andersen@saabgroup.com>, Curt Strm <curt.strom@saabsystems.se>, 
 Per Sandberg <per.sandberg@saabgroup.com>
Message-Id: <20070326121454.E1213165010@province.act-europe.fr>
Subject: [G326-009] #466 AIX debugger useless

Priority: Blocking

We are using GPS 4.0.2 on Linux with remote compilation and 
debugging on AIX using GNAT 6.0.1.

You must be very lucky and have a small program to be able to debug in this environment. The bigger program you have the greater problems you have. In most cases when we try to debug it just won't work so there is no other description to give than just that. When we are lucky we often run into the problems below.

Really bad things:

1. It's very rare that you can set a breakpoint by clicking in the source code.
2. Debugger crash after reaching a breakpoint and executing step.
3. Debugger crash when pressing interrupt more than once in a debug session.
4. When trying to catch exceptions they are not caught.
5. Very often when you want to look at a variable GPS mixes this up and sends a command to the debugger to display the type

Annoying:

1. You can't see all processes on the remote machine when attaching to a process, you want to see not just the processes running as root but processes running under other users as well.
 
Problems were simmilar with GPS 4.0.1 and GNAT 5.0.4



From report-internal@adacore.com  Mon Mar 26 08:15:24 2007
To: mikael.roming@saabgroup.com
From: report@adacore.com
Cc: Ture Andersen <ture.andersen@saabgroup.com>, report@adacore.com, Curt Strm <curt.strom@saabsystems.se>, Per Sandberg <per.sandberg@saabgroup.com>
Message-Id: <20070326121518.E6EC448CC46@nile.gnat.com>
Date: Mon, 26 Mar 2007 08:15:18 -0400 (EDT)
Subject: Re: [G326-009] #466 AIX debugger useless

[Automatic message from GNAT report ticket mechanism]

Dear Customer,

Your report has been automatically assigned ticket number [G326-009].
Please be sure to put this number in the subject line of any
subsequent followup message. Messages from us on this report will
quote this ticket number.

Thank you for using GNAT.

-- The GNAT support team




From charlet@ACT-Europe.FR  Mon Mar 26 08:21:36 2007
Date: Mon, 26 Mar 2007 14:21:26 +0200
From: Arnaud Charlet <charlet@adacore.com>
To: Mikael Roming <mikael.roming@saabgroup.com>
Cc: report@adacore.com, Ture Andersen <ture.andersen@saabgroup.com>,
	Curt =?iso-8859-1?Q?Str=F6m?= <curt.strom@saabsystems.se>,
	Per Sandberg <per.sandberg@saabgroup.com>
Message-ID: <20070326122126.GA3702@adacore.com>
References: <20070326121454.E1213165010@province.act-europe.fr>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <20070326121454.E1213165010@province.act-europe.fr>
Subject: Re: [G326-009] #466 AIX debugger useless

Mikael,

I'm sorry to hear that you are having troubles with GPS in remote mode.

We will try to investigate and will likely need more input from you
(e.g. configuration of your network, kind of remote access, GPS logs
corresponding to your debug sessions with GPS crashing, ...).

In the mean time, could you confirm that using gdb directly on your AIX
machines allow you to move forward and do most of the debugging, or
are there also issues occurring with gdb itself on AIX ? In order
to investigate all these issues, it's important to identify which issue
is related to the GPS remote and which is due to gdb.

Regards,

Arno



From report@gnat.com  Mon Mar 26 08:21:38 2007
From: report@gnat.com
To: file@gnat.com
Message-Id: <20070326122137.21A0B48CFAF@nile.gnat.com>
Date: Mon, 26 Mar 2007 08:21:37 -0400 (EDT)
Subject: [bugtool:G326-009] Assigned to charlet





From charlet@gnat.com  Mon Mar 26 08:21:47 2007
From: charlet@gnat.com
To: file@gnat.com
Message-Id: <20070326122146.9032748CE47@nile.gnat.com>
Date: Mon, 26 Mar 2007 08:21:46 -0400 (EDT)
Subject: [bugtool:G326-009] attributes modified

Reassigned from: charlet to: lambourg




From charlet@gnat.com  Mon Mar 26 08:22:03 2007
From: charlet@gnat.com
To: file@gnat.com
Message-Id: <20070326122202.C79FF48CE47@nile.gnat.com>
Date: Mon, 26 Mar 2007 08:22:02 -0400 (EDT)
Subject: [bugtool:G326-009] attributes modified

Description changed from: AIX debugger useless
  to: GPS: troubles with remote debugging on AIX




From lambourg@adacore.com  Wed Mar 28 08:53:16 2007
Message-ID: <460A65C0.4050003@adacore.com>
Date: Wed, 28 Mar 2007 14:55:28 +0200
From: Jerome Lambourg <lambourg@adacore.com>
MIME-Version: 1.0
To: Mikael Roming <mikael.roming@saabgroup.com>
CC:  report@adacore.com, Ture Andersen <ture.andersen@saabgroup.com>, 
 =?ISO-8859-1?Q?Curt_Str=F6m?= <curt.strom@saabsystems.se>, 
 Per Sandberg <per.sandberg@saabgroup.com>
References: <20070326121454.E1213165010@province.act-europe.fr>
In-Reply-To: <20070326121454.E1213165010@province.act-europe.fr>
Content-Type: text/plain; charset=ISO-8859-1
Subject: Re: [G326-009] #466 AIX debugger useless

Hello  Mikael,

I come back to you to better understand the underlying problems you have

Mikael Roming wrote:
> Priority: Blocking
>
> We are using GPS 4.0.2 on Linux with remote compilation and 
> debugging on AIX using GNAT 6.0.1.
>
> You must be very lucky and have a small program to be able to debug in this environment. The bigger program you have the greater problems you have. In most cases when we try to debug it just won't work so there is no other description to give than just that. When we are lucky we often run into the problems below.
>
> Really bad things:
>
> 1. It's very rare that you can set a breakpoint by clicking in the source code.
>   
Have you tried setting the breakpoint manually (e.g. in the debugger
console window, type "break file:line". Does this work ?
> 2. Debugger crash after reaching a breakpoint and executing step.
>   
> 3. Debugger crash when pressing interrupt more than once in a debug session.
>   
> 4. When trying to catch exceptions they are not caught.
> 5. Very often when you want to look at a variable GPS mixes this up and sends a command to the debugger to display the type
>
>   
In all the above cases, do you have any log or error message shown ? A
reproducer ?

> Annoying:
>
> 1. You can't see all processes on the remote machine when attaching to a process, you want to see not just the processes running as root but processes running under other users as well.
>   
What is your remote configuration ? Are you logged as root on this machine ?
More generally, we need to know:
* your precise remote configuration: what remote tool are you using
(rsh, ssh, telnet), are you using a specific account
* the exact error messages you are seeing (if any).
* have you tried with a nativegdb command line session ? Do you see the
same errors in this case ?
>  
> Problems were simmilar with GPS 4.0.1 and GNAT 5.0.4
>   
OK.



From lambourg@gnat.com  Thu Mar 29 08:54:07 2007
From: lambourg@gnat.com
To: file@gnat.com
Message-Id: <20070329125406.C237548CC12@nile.gnat.com>
Date: Thu, 29 Mar 2007 08:54:06 -0400 (EDT)
Subject: [bugtool:G326-009] attributes modified

Moved from: open to: hold
Why: 



From lambourg@adacore.com  Thu Apr  5 11:02:00 2007
Message-ID: <46150FF1.4010209@adacore.com>
Date: Thu, 05 Apr 2007 17:04:17 +0200
From: Jerome Lambourg <lambourg@adacore.com>
MIME-Version: 1.0
To: Roming Mikael <mikael.roming@saabgroup.com>,  report <report@adacore.com>
Subject: Re: [G326-009] #466 AIX debugger useless
References: <6F2815FEE7F896458CDE99423F396992B2144C@corpappl002.corp.saab.se>
In-Reply-To: <6F2815FEE7F896458CDE99423F396992B2144C@corpappl002.corp.saab.se>
Content-Type: text/plain; charset=ISO-8859-1

Hello Mikael,

Just a procedural note: you should always copy report@adacore.com when
sending us e-mail (a reply-all is a good way of achieving this): this
ensures that the discussion is properly tracked on our side, and also
ensures that the issue is properly handled in case of vacations.

Concerning the technical response to your problem, I'm looking at it and
will answer shortly.

Regards,
- Jerome

Roming Mikael wrote:
> Ups, I did miss your last line there, no we do not have a nice little reproducer and you do not really se anything but the debugger hanging/crashing. The way to reproduce this I guess is to try to debug one of our programs, but I do not think that I'm be able to give you our complete application, becase of both commersial and military secrecy. A way to solve this could be that someone from Adacore vistited us instead.
>
> In the case where the debugger does a ptype instead of a print you can see that this error is in GPS that sends the debugger the ptype command despite the fact that you have choosen to print a variable.
>
> Your complier is really good so it is a real shame that the debugger doesn't work
>  
> Regards
>
>   

From lambourg@adacore.com  Thu Apr  5 11:02:13 2007
Message-ID: <46150FFD.5070602@adacore.com>
Date: Thu, 05 Apr 2007 17:04:29 +0200
From: Jerome Lambourg <lambourg@adacore.com>
MIME-Version: 1.0
To: Bug Reminder <file@adacore.com>
Subject: [Fwd: Re: [G326-009] #466 AIX debugger useless]
Content-Type: multipart/mixed; boundary="------------090705020403050309020406"
X-Act-Attachment: /gnat.dev/ada/gnatbugs/hold/G326-009/Re: [G326-009] #466 AIX 
 debugger useless (message/rfc822)

This is a multi-part message in MIME format.
--------------090705020403050309020406
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit



--------------090705020403050309020406--

From lambourg@adacore.com  Thu Apr  5 11:02:25 2007
Message-ID: <46151005.6060405@adacore.com>
Date: Thu, 05 Apr 2007 17:04:37 +0200
From: Jerome Lambourg <lambourg@adacore.com>
MIME-Version: 1.0
To: Bug Reminder <file@adacore.com>
Subject: [Fwd: Re: [G326-009] #466 AIX debugger useless]
Content-Type: multipart/mixed; boundary="------------090905090807040802050406"
X-Act-Attachment: /gnat.dev/ada/gnatbugs/hold/G326-009/Re: [G326-009] #466 AIX 
 debugger useless_1 (message/rfc822)

This is a multi-part message in MIME format.
--------------090905090807040802050406
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit



--------------090905090807040802050406--

From lambourg@adacore.com  Tue Apr 10 09:05:33 2007
Message-ID: <461B8B5E.7080501@adacore.com>
Date: Tue, 10 Apr 2007 15:04:30 +0200
From: Jerome Lambourg <lambourg@adacore.com>
MIME-Version: 1.0
To: Roming Mikael <mikael.roming@saabgroup.com>
CC: Report <report@adacore.com>
Subject: Re: [G326-009] #466 AIX debugger useless
References: <6F2815FEE7F896458CDE99423F396992B2144C@corpappl002.corp.saab.se>
In-Reply-To: <6F2815FEE7F896458CDE99423F396992B2144C@corpappl002.corp.saab.se>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Hello Mikael

Roming Mikael wrote :
> In the case where the debugger does a ptype instead of a print you can see that this error is in GPS that sends the debugger the ptype command despite the fact that you have choosen to print a variable.

I think we figured out what happens here: in order to display a 
variable, GPS needs to use both print and ptype : the first one is to 
retrieve the value itself while the second one is needed because GPS 
will use it to correctly format the output.

Now the bug itself: the fact that you are seeing the ptype result 
instead of the print result probably means that a timeout occurs 
somewhere, discarding the print output and retrieving the ptype output 
instead.

In fact, we just discovered an incorrectly set timeout value in the GPS 
remote functionality, that is now corrected in the coming GPS 4.1.1.

This version is not released yet, but we can provide it to you if you 
want to try it.

> Your complier is really good so it is a real shame that the debugger doesn't work

Have you tried the command-line gdb to see if all the problems you are 
expriencing are related to GPS,  or if they are also present in gdb ?

Regards,
- Jerome

From mikael.roming@saabgroup.com  Mon Apr 16 08:38:00 2007
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Subject: RE: [G326-009] #466 AIX debugger useless
Date: Mon, 16 Apr 2007 14:37:45 +0200
Message-ID: <6F2815FEE7F896458CDE99423F396992B21450@corpappl002.corp.saab.se>
References: <6F2815FEE7F896458CDE99423F396992B2144C@corpappl002.corp.saab.se> 
 <461B8B5E.7080501@adacore.com>
From: "Roming Mikael" <mikael.roming@saabgroup.com>
To: "Jerome Lambourg" <lambourg@adacore.com>
CC: <report@adacore.com>

Hello Jerome

I haven't succeded to produce a simple testcase for you yet but I will
try more.

Nice that you have fixed one problem.

Conecrning the other problems we have tried gdb in textmode but not very
much.
It is more a feeling than anything else that gdb is the one that
acctually causes most of the problems
Let me get back to you in a couple of days.

Regards

Mikael Roming

 =20

-----Original Message-----
From: Jerome Lambourg [mailto:lambourg@adacore.com]=20
Sent: den 10 april 2007 15:05
To: Roming Mikael
Cc: Report
Subject: Re: [G326-009] #466 AIX debugger useless

Hello Mikael

Roming Mikael wrote :
> In the case where the debugger does a ptype instead of a print you can
see that this error is in GPS that sends the debugger the ptype command
despite the fact that you have choosen to print a variable.

I think we figured out what happens here: in order to display a
variable, GPS needs to use both print and ptype : the first one is to
retrieve the value itself while the second one is needed because GPS
will use it to correctly format the output.

Now the bug itself: the fact that you are seeing the ptype result
instead of the print result probably means that a timeout occurs
somewhere, discarding the print output and retrieving the ptype output
instead.

In fact, we just discovered an incorrectly set timeout value in the GPS
remote functionality, that is now corrected in the coming GPS 4.1.1.

This version is not released yet, but we can provide it to you if you
want to try it.

> Your complier is really good so it is a real shame that the debugger=20
> doesn't work

Have you tried the command-line gdb to see if all the problems you are
expriencing are related to GPS,  or if they are also present in gdb ?

Regards,
- Jerome

From lambourg@adacore.com  Mon Apr 16 08:41:15 2007
Message-ID: <46236EE0.1070300@adacore.com>
Date: Mon, 16 Apr 2007 14:41:04 +0200
From: Jerome Lambourg <lambourg@adacore.com>
MIME-Version: 1.0
To: Roming Mikael <mikael.roming@saabgroup.com>
CC:  report@adacore.com
Subject: Re: [G326-009] #466 AIX debugger useless
References: <6F2815FEE7F896458CDE99423F396992B2144C@corpappl002.corp.saab.se> 
 <461B8B5E.7080501@adacore.com> <6F2815FEE7F896458CDE99423F396992B21450@corpappl002.corp.saab.se>
In-Reply-To: <6F2815FEE7F896458CDE99423F396992B21450@corpappl002.corp.saab.se>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Hello Mikael,

Roming Mikael wrote :
> I haven't succeded to produce a simple testcase for you yet but I will
> try more.
> 
> Nice that you have fixed one problem.
> 
> Conecrning the other problems we have tried gdb in textmode but not very
> much.
> It is more a feeling than anything else that gdb is the one that
> acctually causes most of the problems
> Let me get back to you in a couple of days.

OK, I'll wait for more input from you !

Regards,
- Jerome

From lambourg@gnat.com  Wed Apr 18 04:58:16 2007
From: lambourg@gnat.com
To: file@gnat.com
Subject: [bugtool:G326-009] GPS: troubles with remote debugging on AIX - moved to hold
Message-ID: <20070418085815.AC46048CE18@nile.gnat.com>
Date: Wed, 18 Apr 2007 04:58:15 -0400 (EDT)

Moved from: open to: hold
Why: 

