[omniORB] HELP..!!!!client Request hangs on the server (seems to be deadloc ked) running as a task on PowerPC

Mon Aug 19 22:02:01 2002

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01C247C3.ED975D30
Content-Type: text/plain;
	charset="iso-8859-1"

Hi,

Do we have to compile the OmniOrb source with a particular version of gcc
???. I have compiled with gcc 2.95.2 (cross compiler) on RedHat Linux 6.2
for PowerPC.

However, when I run the server and try to connect from the client, I see
SIGSEGV in different places. This is happening when I integrate the servant
with my application (which is also cross compiled with the same gcc). It
happens in different places. Looks like some threading library issue or
something .

Can someone please tell me as to what is the correct version of gcc to use
on Linux/PowerPC.

thanks

-sheshi

-----Original Message-----
From: Sheshi Sankineni 
Sent: Wednesday, August 14, 2002 11:35 AM
To: 'Duncan Grisby'
Cc: 'omniorb-list@realvnc.com'
Subject: RE: [omniORB] HELP..!!!!client Request hangs on the server (seems
to be deadloc ked) running as a task on PowerPC 

Hi Duncan, 

Thanks for replying to my email. 

I am aware of why 'sync' was made a stack variable. It was my mistake that I
missed a 'delete sync' when I created it on the heap. That was the reason of
the hang. However,

I am still not sure why making it a stack variable is causing it to SIGSEV.
Do you suspect 
that that stack is getting corrupted ?? I have increated the stack size of
the thread is 2 meg. It didnt help either. Since you are familiar with the
code, can you please tell me what you think might be the cause of SIGSEGV.

Again, thanks for the reply. 

-sheshi 

-----Original Message----- 
From: Duncan Grisby [ mailto:duncan@grisby.org <mailto:duncan@grisby.org> ] 
Sent: Wednesday, August 14, 2002 6:33 AM 
To: Sheshi Sankineni 
Cc: 'omniorb-list@realvnc.com' 
Subject: Re: [omniORB] HELP..!!!!client Request hangs on the server 
(seems to be deadloc ked) running as a task on PowerPC 

On Tuesday 13 August, Sheshi Sankineni wrote: 

> This message is in MIME format. Since your mail reader does not understand

> this format, some or all of this message may not be legible. 

Please don't send HTML email to the omniORB mailing list. 

[...] 
> In the beginning I ran into SIGSEGV when trying to resolve the RootPoa. I 
> was getting SIGSEGV consistently in the omni_tracedmutex::lock()during 
> initialization of the ORB. 
> 
> What I noticed was in that method, there is a local variable sync declared

> as follows 
> 
> omni_mutex_lock sync(pd_lock); 
> 
> I changed this to a heap variable as follows 
> 
> omni_mutex_lock* sync = new omni_mutex_lock(pd_lock); // sheshi 

That's why you're getting the hang!  omni_mutex_lock has to be created 
on the stack -- it's purpose it to acquire the mutex when it is 
created, and release it when it goes out of scope. By putting it on 
the heap, you've made sure the mutex is never unlocked. 

You need to track down why the segfault is happening in the lock 
code. Randomly changing bits of code you don't understand isn't going 
to help. 

Cheers, 

Duncan. 

-- 
 -- Duncan Grisby         -- 
  -- duncan@grisby.org     -- 
   -- http://www.grisby.org <http://www.grisby.org>  -- 

------_=_NextPart_001_01C247C3.ED975D30
Content-Type: text/html;
	charset="iso-8859-1"

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<TITLE>RE: [omniORB] HELP..!!!!client Request hangs on the server (seems to be deadloc ked) running as a task on PowerPC</TITLE>

<META content="MSHTML 5.00.3211.1700" name=GENERATOR></HEAD>
<BODY>
<DIV>Hi,</DIV>
<DIV>&nbsp;</DIV>
<DIV>Do we 
have to compile the OmniOrb source with a particular version of gcc ???. I have 
compiled with gcc 2.95.2 (cross compiler) on RedHat Linux 6.2 for 
PowerPC.</DIV>
<DIV>&nbsp;</DIV>
<DIV>However, when I run the server and try to connect from 
the client, I see SIGSEGV in different places. This is happening when I 
integrate the servant with my application (which is also cross compiled with the 
same gcc). It happens in different places. Looks like some threading library 
issue or something .</DIV>
<DIV>&nbsp;</DIV>
<DIV>Can 
someone please tell me as to what is the correct version of gcc to use on 
Linux/PowerPC.</DIV>
<DIV>&nbsp;</DIV>
<DIV>thanks</DIV>
<DIV>&nbsp;</DIV>
<DIV>-sheshi</DIV>
<BLOCKQUOTE style="MARGIN-RIGHT: 0px">
 <DIV align=left class=OutlookMessageHeader dir=ltr>-----Original Message----- From: Sheshi Sankineni 
 Sent: Wednesday, August 14, 2002 11:35 AM To: 'Duncan 
 Grisby' Cc: 'omniorb-list@realvnc.com' Subject: RE: 
 [omniORB] HELP..!!!!client Request hangs on the server (seems to be deadloc 
 ked) running as a task on PowerPC </DIV> 
 Hi Duncan, 
 Thanks for replying to my email. 
 I am aware of why 'sync' was made a stack variable. It was my 
 mistake that I missed a 'delete sync' when I created it on the heap. That was 
 the reason of the hang. However,
 I am still not sure why making it a stack variable is causing 
 it to SIGSEV. Do you suspect that that stack is 
 getting corrupted ?? I have increated the stack size of the thread is 2 meg. 
 It didnt help either. Since you are familiar with the code, can you please 
 tell me what you think might be the cause of SIGSEGV.
 Again, thanks for the reply. 
 -sheshi 
 -----Original Message----- From: 
 Duncan Grisby [<A 
 href="mailto:duncan@grisby.org">mailto:duncan@grisby.org</A>] Sent: Wednesday, August 14, 2002 6:33 AM To: 
 Sheshi Sankineni Cc: 'omniorb-list@realvnc.com' 
 Subject: Re: [omniORB] HELP..!!!!client Request hangs on the 
 server (seems to be deadloc ked) running as a task on 
 PowerPC 
 On Tuesday 13 August, Sheshi Sankineni wrote: 
 &gt; This message is in MIME format. Since your mail reader 
 does not understand &gt; this format, some or all of 
 this message may not be legible. 
 Please don't send HTML email to the omniORB mailing 
 list. 
 [...] &gt; In the beginning I ran into 
 SIGSEGV when trying to resolve the RootPoa. I &gt; was 
 getting SIGSEGV consistently in the omni_tracedmutex::lock()during 
 &gt; initialization of the ORB. &gt; 
 &gt; What I noticed was in that method, there is a 
 local variable sync declared &gt; as follows 
 &gt; &gt; omni_mutex_lock 
 sync(pd_lock); &gt; &gt; I 
 changed this to a heap variable as follows &gt; 
 &gt; omni_mutex_lock* sync = new 
 omni_mutex_lock(pd_lock); // sheshi 
 That's why you're getting the hang!&nbsp; omni_mutex_lock has 
 to be created on the stack -- it's purpose it to 
 acquire the mutex when it is created, and release it 
 when it goes out of scope. By putting it on the heap, 
 you've made sure the mutex is never unlocked. 
 You need to track down why the segfault is happening in the 
 lock code. Randomly changing bits of code you don't 
 understand isn't going to help. 
 Cheers, 
 Duncan. 
 -- &nbsp;-- Duncan 
 Grisby&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; -- &nbsp; -- duncan@grisby.org&nbsp;&nbsp;&nbsp;&nbsp; -- &nbsp;&nbsp; -- <A href="http://www.grisby.org" 
 target=_blank>http://www.grisby.org</A> -- 
</BLOCKQUOTE></BODY></HTML>

------_=_NextPart_001_01C247C3.ED975D30--