General PTY and CONSOLE question--SOLVED

[Gene Smith]

Gene Smith wrote, On 09/23/2008 10:29 AM:
> Joel Sherrill wrote, On 09/23/2008 09:39 AM:
>> Chris Johns wrote:
>>> Joel Sherrill wrote:
>>>   
>>>> Sounds like the socket close code has no idea that it shouldn't
>>>> close it a second time.  Somehow we get deep in the bowels
>>>> for a closed file descriptor.  I wonder if simply calling close()
>>>> twice on a socket fd would reproduce the problem.
>>>>     
>>> I think this is bug 785.
>>>
>>> (https://www.rtems.org/bugzilla/show_bug.cgi?id=785)
>>>   
>> Gene.. can you confirm this fixes your situation?  If so,
>> I would like to apply this patch to all open branches and
>> the head.
> 
> Joel,Chris,
> This may indeed be a bug but I don't think it is what I am seeing 
> exactly. First, this assumes you have already blocked in accept(). In my 
> situation I have not yet blocked in accept. I go into accept() and then 
> return when I see that the SS_CANTRECEIVEMORE flag is set. This flag is 
> set when the code chain beginning with fclose(stdin) occurs 
> (improperly?) with no console driver installed after calling 
> rtems_panic() in another thread.
> 
> Also, the patches are against 4.6 and 4.7. Not sure if they are OK with 
> 4.8.0 which I have. Can you send me an updated patch or, even better, 
> the complete files with the patch applied and I will try them? You can 
> send to gd.smth at gmail dot com.  (Yes, smth is right).
> 
> -gene

I tried the patch for 4.7 against my 4.8.0 and only rtems_glue.c worked, 
the other two produced .rej files. One is short and I can patch manually 
  (rtems_bsdnet_internal.h). The other is quite long and hard to patch 
manually (rtems_syscall.c).