Re: Re: Re: Re: HARDINGE HNC RETROFIT

Hi Mark

EMCSH is a "wish" like Tcl/Tk connector into the EMC. To a system usage
monitor like Top it would appear the be the owner of those EMC processes
that use wish. This would include tkemc, the backplotter, and any of the
tkemc scripts. I know that the original backplotter, the one I wrote,
can consume great amounts of RAM. If that was running with the chips
plot or the dome test it would probably account for a good chunk of
memory.

I haven't tried leaving that up overnight but I know that it can take a
significant bit of time freeing up memory allocated to it. If one were
in a big hurry this might appear to be a near dead system, particularly
after Linux had done it's periodic housekeeping and had reasigned
priorities to nonrealtime things like emcsh.

I seem to remember a report laying about someplace about Tcl/Tk and
memory leaks. It satisfied me but then I don't remember (memory leak)
much about it.

My first question when asked about some "goofy" behavior on the part of
an EMC is where'd you get it. If it is a download, did you run the
checksums and did they come out with the known value. That satisfied, my
next question is how long has it been running without interruption. My
first suggestion if it has been up for quite a while is to shut down the
EMC and restart and see if the behavior persists. The problem here would
not be related to a EMCSH problem but is related to the SHMEM processes.
In the 2+ rtlinux kernels this is a memory manager named mbuff. If you
find that a message pops up regarding mbuff being busy, you can clean it
out with a little program called mbuff_dealloc. This program is created
by running make in the directory where the mbuff module source resides.
The memory structure that you will be clearing out is named emcmotStruct.

Hope this helps

Ray

>    From: "Mark" <markatswat@...>
> Subject: Re: Re: Re: HARDINGE HNC RETROFIT
>
>
> Speaking of EMC, I was playing with it and left it running overnight
> and noticed something very disturbing, it seems to have a rather bad
> memory leak, as the next morning (this happened several times) I got
> up and the hard drive was swapping like mad and performance was down
> the tubes. Looked at the processes and EMCSH was using 100+ MB of ram
> on my 128 MB machine. Anyone have any clues???
>
> Mark