Re: bug in VxWorks memPartInfoGet (used by VxStats)

[Benjamin Franksen <franksen@mail.bessy.de>]

Hoff, Lawrence wrote:
> 
>         At SNS, we ran into a bug in memPartInfoGet, using VxWorks
> kernel 5.4.2 and VxStats. Apparently there is a bug in memPartInfoGet()
> which can corrupt the heap if another task preempts memPartInfoGet()
> and does a heap operation (e.g. malloc/free). There is an SPR covering
> this bug which seems to imply that the bug also exists in kernel
> version 5.5. It is not clear which earlier versions of VxWorks might
> also be affected, but the date on the SPR is more than 3 years ago.

Quite interesting.

Two or more years ago we ran into this same problem with VxWorks 5.2
(which we still use). Only it was worse because our IOCs hung up
completely. It turned out that the problem occurred only when the IOC
had GPIB support for a LAN-GPIB gateway running and also vxStats. I
suspected the RPC library then but now it seems that the severity of our
problem was probably only due to a lot of allocation/deallocation done
by the RPC library -- and the actual bug is in memPartInfo. Well, back
then (and until today) we disabled the VxStats memory check records in
these IOCs and all was fine (or at least seemed to be).

Ben