RTAI 3.3 test4: shm problems [solved]
mantegazza at aero.polimi.it
Mon Jan 16 20:34:53 CET 2006
ralf eckhardt wrote:
> On Monday 16 January 2006 14:26, Paolo Mantegazza wrote:
>>Paolo Mantegazza wrote:
>>>>[rtai_free() seems to fail]
>>>Yep, your fix seems to work.
>>That's OK. Take into account the such a fix has been simple but the bug
>>has been there for quite some time. Likely none was developing with such
>>large chunks of memory or repeated applications were related to multi
>>registrations of named obejct of a smaller size.
> As I am just now porting some applications from brave old rtai-2.4.11
> to a today's version i must admit that i have provoked this symptom by
> recklessly exploiting the limits of these functions during preparatory tests;o)
> In the real world i would allocate only some 50M, or so; there, this
> issue would have slipped through, too.
> However, the same test with similar amounts of shm on another machine
> yields that this bug did not show up on old rtai-2.4.11.
Well it is due to the new hashed registration, while previously it was
on simple linear arrays. In any case you could have recovered quickly
the old 2.4.11 by simply uncommenting a macro, as it is still there. In
fact the first thing I did was to reset the old way and see it worked so
that it was easy to track the bug.
More information about the Rtai