GKN%OAK.SAINET.MFENET@LLL-MFE.ARPA (01/31/86)
Date: Thu, 30-JAN-1986 19:14 EST
To: info-vax@sri-kl.arpa
Message-ID: <[OAK.SAINET.MFENET].8C581940.008E9D0E.GKN>
US-Mail: Science Applications; P.O. Box 2501; Oak Ridge, TN 37831
Telephone: (615) 482-9031
X-VMS-Mail-To: ARPA%"info-vax@sri-kl.arpa"
From: Arpa%"dean @A.PSY.CMU.EDU" 29-JAN-1986 06:06
Subj: EXE$ALONONPAGED and EXE$DEANONPAGED
Date: Monday, 27 Jan 86 15:09:25 EST
The problem I suspect is a synch probelm with allocation and
dealloction of memory. ie. I am in the middle of a dealloction, the
system interrupts at high IPL and is able to allocate the peice I
just deaalocated. I then finish deallocating. When the kernel mode
ast later dealloctes a previously deallocted piece, POW.
The Fiche doesn't mention any need to raise IPL when allocating or
deallocting though.
Anyone have any thoughts/experience/comments ?
Randall W. Dean
Well, the only thing I can think of is that you ALWAYS RAISE IPL to
synchronize access, never lower it. For EXE$ALONONPAGED, this means
that your current IPL must be .leq. IPL$_MAILBOX (11) when you call
it. If you are at a higher IPL (eg - device interrupt level), then
you must create a fork process at a lower IPL in order for it to raise
IPL to allocate memory.
Regards,
gerard
[sorry if you get the top half of this - my fingers slipped]