(Illustration by Gaich Muramatsu)
Hi, On Redhat Linux 6.1, I changed the case statement in vice-setup-rvm just after the LOG file is created to the following: case `uname -sr` in "NetBSD 1.2" ) rvmstart=0x20000000 ;; NetBSD\ 1.3* ) rvmstart=0x50000000 ;; NetBSD\ 1.4* ) rvmstart=0x50000000 ;; FreeBSD\ * ) rvmstart=0x50000000 ;; Linux\ * ) rvmstart=0x50000000 ;; *) rvmstart=0x20000000 ;; esac And rvm setup now works! Many thanks Robin Hunt -----Original Message----- From: Shafeeq Sinnamohideen [mailto:shafeeq_at_cs.cmu.edu] Sent: Thursday, March 30, 2000 8:16 PM To: codalist_at_TELEMANN.coda.cs.cmu.edu Subject: Re: problem with rdsinit On Thu, 30 Mar 2000, Jens Thiele wrote: > > all compiles fine > but vice-setup reports an error. > The error is reported by rdsinit (started from vice-setup-rvm started > from vice-setup). > relevant output from vice-setup-rvm (I added the Calling line): > > > Rdsinit will initialize data and log. > This takes a while. > Calling rdsinit -f /var/log/codalog /var/log/codadata 23068672 > 0x20000000 0x1200000 0x100000 80 32 > rvm_initialize succeeded. > Going to initialize data file to zero, could take awhile. > done. > ? ERROR: rds_zap_heap RVM_EOFFSET. > > Now I read in the mailinglist that this perhaps happens with low memory > => i tried another machine > (The same software configuration but much more memory) but the same > error occured. > btw i tried it verry often ;-) and even a reboot didn't help ;-)) The problem isn't necessarily low memory, instead that RVM tries to map its data segment at address 0x20000000 and fails because something else (like a shared library) starts less than 22mb above that address. Since RVM is not smart enough to find a free address range by itself, you have to pick one for it. 0x50000000 is the default in vice-setup-rvm for other OSs, it might be a good one to try first. If not, you can examine a few of the /proc/<pid>/map as described a few posts earlier. Good luck ShafeeqReceived on 2000-03-31 02:58:57