This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: TCP/IP Deadlock
- From: "Giuseppe Monteleone" <info at ing-monteleone dot com>
- To: gary at mlbassoc dot com
- Cc: ecos-discuss at sourceware dot org
- Date: Tue, 29 Mar 2005 00:21:13 +0200
- Subject: Re: [ECOS] TCP/IP Deadlock
- Reply-to: "Giuseppe Monteleone" <info at ing-monteleone dot com>
I made furter analisys the system ping also with lock
The deadlock problem is in the function cyg_stdio_read in the module
stream.cxx of 'language' package the underlaying device for same obscure reason doesn't return to caller.
Now i'am investigating the entry point for that device and then set breakpoint.
Thanks for help
Giuseppe
P.S. i still use the version 2.0 and not the current
>----- Original Message -----
>From: Gary Thomas
>To: Giuseppe Monteleone
Cc: eCos Discussion
>Data: Mon, 28 Mar 2005 07:25:23 -0700
>Subject: Re: [ECOS] TCP/IP Deadlock
>On Sun, 2005-03-27 at 20:12 +0200, Giuseppe Monteleone wrote:
>> I Have some problem with TCP/IP Deadlock. The system work well
>> and the ICMP program work fine (the board respond to "ping"),
>> In my configuration i added the OpenTCP/IP stack. SO if i try to
>> open more than one web page the webserver goes down and doest
>> respond anymore.
>> In my opinion the problem is the TCP itself (IP work good because
>> the ICMP work fine).
>
>It's doubtful that there is such a major problem in the stack
>itself as it has been successfully deployed on a number of systems
>over many years. More likely the issue is with your application or
>possibly how you have eCos configured.
>
>Have you made sure that there is adequate stack for the HTTP processing
>threads?
>
>Have you attempted to analyze the state of the system when this
>happens, e.g. break in with GDB and see what threads are doing,
>what state they are in, etc?
>
>When it gets in this state, can you still ping it from other hosts?
>
>Have you enabled any debugging? There are many levels of this that
>can be applied, from tracing network packets to a full runtime trace
>(using the infra debug features)
>
>--
>------------------------------------------------------------
>Gary Thomas | Consulting for the
>MLB Associates | Embedded world
>------------------------------------------------------------
>
>
--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss