On Tue, Sep 27, 2005 at 05:55:00PM -0400, Camm Maguire wrote: > Greetings! I've looked at the build log, and know where the error > is. I suspect some problem in the fault address recovery code of the > building kernel. GCL has two garbage collection modes, the > conventional, and "SGC" or stratefied garbage collection, which marks > pages read only, waits for segfaults triggered by write attempts, > recovers the fault address and marks the page read-write, and then > continues. Only the read-write pages need be garbage collected. > > One can test this by intalling GCL, running it with 'gcl', typeing > (fmakunbound 'si::sgc-on) and then (si::save-system "/tmp/foo"), then > move /tmp/foo to /usr/lib/gcl-2.6.7/unixport/saved_gcl (if that is the > current path as I seem to recall). If this is not the case, then I > can only imagine that 2.9.3 is taking up a bit more memory which is > exceeding the bounds only on 64bit.
[EMAIL PROTECTED]:~# GCL (GNU Common Lisp) 2.6.7 CLtL1 Sep 22 2005 14:31:27 Source License: LGPL(gcl,gmp), GPL(unexec,bfd) Binary License: GPL due to GPL'ed components: (READLINE BFD UNEXEC) Modifications of this banner must retain notice of a compatible license Dedicated to the memory of W. Schelter Use (help) to get some basic information on how to use GCL. >(fmakunbound 'si::sgc-on) SYSTEM:SGC-ON >(si::save-system "/tmp/foo") [EMAIL PROTECTED]:~# ls -l /usr/lib/gcl-2.6.7/unixport/saved_gcl /tmp/foo -rwxr-xr-x 1 root root 9377808 Sep 28 00:13 /tmp/foo -rwxr-xr-x 1 root root 9361360 Sep 22 14:39 /usr/lib/gcl-2.6.7/unixport/saved_gcl I could move the file to the /usr/lib/gcl-2.6.7/unixport/saved_gcl, but I'm not sure what you're expecting me to do next? Try building acl2 again? If I understand you correct, you're suspecting the conventional garbage collection sometimes has a problem and you want to see if this new binary doesn't have it? Since this only seems to happen sometimes, this seems to be a race condition to me, and it would be nice if what I was testing actually has a chance of triggering it. Is it only a certain thing that triggers it, or should I do the full build? I hope it's just a single test I can run a few 100 times, instead of the full build since that seems to take quiet some time. Kurt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
