On 12/14/2011 10:23 PM, Zdenek Styblik wrote:
> On Thu, Dec 15, 2011 at 12:28 AM, Corey Minyard<tcminy...@gmail.com> wrote:
>> On 12/14/2011 06:26 AM, Zdenek Styblik wrote:
>>> Hello,
>>>
>>> as you may have noticed, there was thread about ipmitool getting
>>> stucked. I dug into this issue and found the problem.
>>>
>>> Line to blame is:
>>> ~~~ 'src/plugins/open/open.c' ~~~
>>> if (select(intf->fd+1,&rset, NULL, NULL, NULL)< 0) {
>>> lperror(LOG_ERR, "I/O Error");
>>> return NULL;
>>> }
>>> ~~~ 'src/plugins/open/open.c' ~~~
>>>
>>> Using select() here is good/wise. However select() is never going to
>>> time-out, because time-out is set to NULL. I believe I can patch it in
>>> the very same way, at least for time-out option, as LAN/LAN-PLUS is
>>> patched by Harshad Prabhu. Yet, I'm wondering whether it is a good
>>> idea to do so. Whether there is some reason why this, open driver -
>>> KCS interface, should never ever time-out.
>> That's fine, I suppose, but the driver is designed to always return a
>> response, a timeout error if nothing comes back. So a timeout here is
>> really unnecessary.
>>
>> -corey
>>
> Designed? Yes. Implemented? No. Unless I've missed something, KCS gets stuck
> and never comes back, because of select() I've pasted.
>
Can you reproduce? That is a driver bug if so, and I'd prefer to fix it.
-corey
------------------------------------------------------------------------------
10 Tips for Better Server Consolidation
Server virtualization is being driven by many needs.
But none more important than the need to reduce IT complexity
while improving strategic productivity. Learn More!
http://www.accelacomm.com/jaw/sdnl/114/51507609/
_______________________________________________
Ipmitool-devel mailing list
Ipmitool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ipmitool-devel
