Error With Networking when Upgrading

[jcristy at motioncontrol.org]

Chris,
Thanks for the reply.  I have tried different stack sizes for the Rx thread (4096 to 512 * 
4096), but it seems to top out at around 9 clusters.  I can call 
rtems_bsdnet_show_mbuf_stats(); here is what i get:
in bsd_init() after the mbuf data structures are set up
************ MBUF STATISTICS ************
mbufs:4096    clusters:1024    free:1024
drops:   0       waits:   0  drains:   0
      free:4096          data:0           header:0           socket:0       
       pcb:0           rtable:0           htable:0           atable:0       
    soname:0           soopts:0           ftable:0           rights:0       
    ifaddr:0          control:0          oobdata:0       
at the beginning of fec_RxDaemon()
************ MBUF STATISTICS ************
mbufs:4096    clusters:1024    free:1024
drops:   0       waits:   0  drains:   0
      free:4095          data:1           header:0           socket:0       
       pcb:0           rtable:0           htable:0           atable:0       
    soname:0           soopts:0           ftable:0           rights:0       
    ifaddr:0          control:0          oobdata:0   
in mclalloc()
************ MBUF STATISTICS ************
mbufs:4096    clusters:1024    free:1018
drops:   0       waits:   0  drains:   1
      free:4088          data:8           header:0           socket:0       
       pcb:0           rtable:0           htable:0           atable:0       
    soname:0           soopts:0           ftable:0           rights:0       
    ifaddr:0          control:0          oobdata:0   

Also, I noticed that the mclfree->mcl_next list is intact until I call rtems_bsdnet_newproc to 
start rxDaemon.  More specifically, it is ruined at the call of rtems_task_create inside 
rtems_bsdnet_newroc.  When it breaks, the list begins to include 0's at the same point, as 
expected, that rxDaemon is unable to continue.

Thanks again,
John

On 11 Jun 2010 at 8:42, Chris Johns wrote:

Date sent:      	Fri, 11 Jun 2010 08:42:17 +1000
From:           	Chris Johns <chrisj at rtems.org>
To:             	jcristy at motioncontrol.org
Copies to:      	rtems-users at rtems.org
Subject:        	Re: Error With Networking when Upgrading

> On 11/06/10 3:33 AM, jcristy at motioncontrol.org wrote:
> >
> > I am working on upgrading from 4.7.1 to 4.9.4.  I am encountering an error: "Still Waiting for
> > mbuf cluster" in m_clalloc in rtems_glue.c.  It seems that in fec_RxDaemon it will allocate 5
> > clusters before coming to the error.  I haven't changed anything else.  Are there any things
> > that are known to cause this error that I can check for?
> >
> 
> How much memory is allocated to the stack ?
> 
> Are you able to call 'rtems_bsdnet_show_mbuf_stats()' ?
> 
> Chris


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20100611/1daf5c27/attachment-0001.html>