pr1255 (was Re: MCF5329EVB BSP)

Till Straumann strauman at slac.stanford.edu
Fri Oct 5 20:28:43 CDT 2007


Feng, Kate wrote:
> Joel Sherrill wrote:
>
>   
>> I wanted time to analyze the problems you were reporting and
>> deal with them at a leisurely pace.  That's one of the reasons I
>> went ahead and cut a 4.8 branch without adding them.
>>     
>
> Can you describe the problems that were reported to this list
> a little bit more ? Or was there a PR related to this problem ?
>   
PR1255

However, it is unlikely that this causes a problem with NFS; in particular
if you don't have two threads accessing the same file there shouldn't be a
1255 related problem.

IMFS is another beast and I believe that bad things happen if multiple
threads concurrently do things like accessing a file and deleting it
or delete the same file at the same time.

T.
> It's very strange.  This afternoon, while a file was being written
> via NFS, it crashed in the middle of the process. I roughly traced
> it down from <_Thread_Handler>
> to  <taskEntry>....... This is the first time I saw this message.
> However, it is the first time I had the NFS via a different network
> setup, too. I did not have enough time to trace the bug further
> because I thought I had introduced a new bug in my application
> program without knowing.
>   
>
> Regards,
> Kate
> -----Original Message-----
> From: rtems-users-bounces+feng1=bnl.gov at rtems.org on behalf of Till Straumann
> Sent: Fri 10/5/2007 5:34 PM
> To: Joel Sherrill
> Cc: Kirspel, Kevin; 'rtems-users at rtems.org'
> Subject: Re: MCF5329EVB BSP
>  
> Joel Sherrill wrote:
>   
>> Till Straumann wrote:
>>     
>>> Joel Sherrill wrote:
>>>       
>>>> Till Straumann wrote:
>>>>         
>>>>> Just a reminder that the filesystem framework currently is not 
>>>>> thread-safe!
>>>>>
>>>>>           
>>>> Your outstanding patches are in the queue for 4.9.
>>>>         
>>> IIRC my patches don't cover making the FS framework thread-safe.
>>>       
>> I wanted time to analyze the problems you were reporting and
>> deal with them at a leisurely pace.  That's one of the reasons I
>> went ahead and cut a 4.8 branch without adding them.
>>
>> The other reason was that that it hasn't been a major problem
>> for all these years so one more release cycle wasn't a big deal.
>>     
> Sure - my concerns are that as more filesystems become available
> there will be more users and more potential for problems to show.
>
> Note that NFS is special: it it essentially stateless and therefore
> insensitive to race conditions (but could still be that subtle issues
> in the FS framework code could cause problems).
>
> T.
>   
>> --joel
>>     
>>> T
>>>       
>>>> --joel
>>>>         
>>>>> T.
>>>>> Joel Sherrill wrote:
>>>>>           
>>>>>> Kirspel, Kevin wrote:
>>>>>>  
>>>>>>             
>>>>>>> I have almost completed a port to RTEMS of the MCF5329EVB 
>>>>>>> evaluation board from Freescale. With this BSP I had to port the 
>>>>>>> YAFFS2 file system to RTEMS. I would like to submit the BSP and 
>>>>>>> YAFFS2 port to RTEMS. The YAFFS2 license is GPL. I don't know if 
>>>>>>> that is OK with RTEMS. I would like to do some robustness testing 
>>>>>>> on the YAFFS2 port. Does RTEMS have a test plan/procedure to 
>>>>>>> validate new file systems?
>>>>>>>
>>>>>>>     
>>>>>>>               
>>>>>> Congratulations on the BSP and port. Is the license OK on the BSP? 
>>>>>> Other email
>>>>>> deals with the YAFFS2 license issue.
>>>>>>
>>>>>> There are POSIX tests which exercise files through the C library 
>>>>>> IO calls and
>>>>>> should exercise all the functionality available through "normal" 
>>>>>> APIs. You
>>>>>> would just have to arrange to have that mounted and the paths 
>>>>>> point to it.
>>>>>>
>>>>>> --joel
>>>>>>  
>>>>>>             
>>>>>>> Kevin Kirspel
>>>>>>>
>>>>>>> Opti Medical Systems
>>>>>>>
>>>>>>> Electrical Engineer
>>>>>>>
>>>>>>> 235 Hembree Park Drive
>>>>>>>
>>>>>>> Roswell GA 30076
>>>>>>>
>>>>>>> (ph) 770-510-4444 x4568
>>>>>>>
>>>>>>> (fax) 770-510-4445
>>>>>>>
>>>>>>> ------------------------------------------------------------------------ 
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> rtems-users mailing list
>>>>>>> rtems-users at rtems.com
>>>>>>> http://rtems.rtems.org/mailman/listinfo/rtems-users
>>>>>>>       
>>>>>>>               
>>>>>> _______________________________________________
>>>>>> rtems-users mailing list
>>>>>> rtems-users at rtems.com
>>>>>> http://rtems.rtems.org/mailman/listinfo/rtems-users
>>>>>>   
>>>>>>             
>
>
> _______________________________________________
> rtems-users mailing list
> rtems-users at rtems.com
> http://rtems.rtems.org/mailman/listinfo/rtems-users
>
>   




More information about the rtems-users mailing list