[Openmcl-devel] bug(?) loading iolib under certain optimization settings
gb at clozure.com
Fri Jan 17 11:45:00 PST 2014
This looks like ticket 1131 (Mhttp://trac.clozure.com/ccl/ticket/1131>),
That was supposed to have been fixed in r15973
(http://trac.clozure.com/ccl/changeset/15973>), though for some reason
the commit message didn't update the ticket status.
Uprrade to a more recent version and try again.
On Fri, 17 Jan 2014, Jared C. Davis wrote:
> I seem to have hit a bug with either CCL or IOLIB. I'm not sure which.
> I hit this while trying to get (ql:quickload "iolib") to work within an ACL2
> image (on CCL). Oddly, the command worked fine when run from ordinary CCL,
> just not from within ACL2. I eventually figured out that the problem seems
> to stem from the optimization settings that ACL2 sets up. Here's a simple
> script to provoke it (run this in a new directory):
> rm -rf inst quicklisp.lisp
> curl -O http://beta.quicklisp.org/quicklisp.lisp
> ccl <<EOF
> (load "quicklisp.lisp")
> (quicklisp-quickstart:install :path "./inst")
> ccl <<EOF
> (proclaim '(optimize (speed 3) (safety 0)))
> (load "inst/setup.lisp")
> (ql:quickload "iolib")
> exit 0
> When I run this, I end up seeing the following error:
> ;Compiling "[...]/inst/dists/quicklisp/software/iolib-0.7.3/src/sockets/common.lisp"...
> > Error: bad regspec: NIL
> > While executing: CCL::%HARD-REGSPEC-VALUE, in process listener(1).
> By commenting things out, it seems that the error is caused while compiling(?)
> the following function in common.lisp:
> (defun map-ipv4-vector-to-ipv6 (addr)
> (declare (type ipv4-array addr))
> (let ((ipv6addr (make-array 8
> :element-type 'ub16
> :initial-element 0)))
> ;; setting the IPv4 marker
> (setf (aref ipv6addr 5) #xFFFF)
> ;; setting the first two bytes
> (setf (aref ipv6addr 6) (+ (ash (aref addr 0) 8)
> (aref addr 1)))
> ;; setting the last two bytes
> (setf (aref ipv6addr 7) (+ (ash (aref addr 2) 8)
> (aref addr 3)))
> (values ipv6addr)))
> If I simply comment out the :element-type 'ub16, the problem goes away and I
> appear to be able to load the file successfully.
> Anyway, at this point I think I'd better hand this over to people who know
> about CCL and/or Quicklisp and/or IOLIB. I'm running CCL from a recent SVN
> trunk on x86-64 linux. If you have any trouble recreating this, please let me
> know and I can provide more information about my setup.
> Jared C. Davis <jared at cs.utexas.edu>
> 11410 Windermere Meadows
> Austin, TX 78759
> Openmcl-devel mailing list
> Openmcl-devel at clozure.com
More information about the Openmcl-devel