Steel Bank Common Lisp

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv4238

Modified Files:
	version.lisp-expr 
Log Message:
1.0.15.41: sod this, revert to the old & simple tar-using contrib installation

 * Windows has occasionally *.FASL files, so *.fasl is not robust. Copy
   all the files using tar, but keep the find + chown so that contribs
   are owned by the installer, not the builder.


Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3911
retrieving revision 1.3912
diff -u -d -r1.3911 -r1.3912
--- version.lisp-expr	26 Mar 2008 19:32:35 -0000	1.3911
+++ version.lisp-expr	27 Mar 2008 17:40:30 -0000	1.3912
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.40"
+"1.0.15.41"

Update of /cvsroot/sbcl/sbcl/contrib
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv4238/contrib

Modified Files:
	asdf-module.mk 
Log Message:
1.0.15.41: sod this, revert to the old & simple tar-using contrib installation

 * Windows has occasionally *.FASL files, so *.fasl is not robust. Copy
   all the files using tar, but keep the find + chown so that contribs
   are owned by the installer, not the builder.


Index: asdf-module.mk
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/asdf-module.mk,v
retrieving revision 1.18
retrieving revision 1.19
diff -u -d -r1.18 -r1.19
--- asdf-module.mk	26 Mar 2008 19:32:35 -0000	1.18
+++ asdf-module.mk	27 Mar 2008 17:40:31 -0000	1.19
@@ -27,10 +27,8 @@
 	     "(asdf:operate (quote asdf:test-op) :$(SYSTEM))" | \
 	  $(SBCL) --eval '(load "../asdf/asdf")'
 
-# KLUDGE / FIXME: Perhaps each module should have it's own list of
-# files to install? At any rate, this is a portable (we hope) way of
-# installing all the files needed -- as long as all the files are in
-# the first level directory...
+# KLUDGE: There seems to be no portable way to tell tar to not to
+# preserve owner, so chown after installing for the current user.
 install: $(EXTRA_INSTALL_TARGETS)
-	cp -p $(SYSTEM).asd *.lisp *.fasl "$(BUILD_ROOT)$(INSTALL_DIR)"
+	tar cf - . | ( cd "$(BUILD_ROOT)$(INSTALL_DIR)" && tar xpvf - )
 	find "$(BUILD_ROOT)$(INSTALL_DIR)" -type f -exec chown `id -u`:`id -g` {} \;

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv9682

Modified Files:
	version.lisp-expr 
Log Message:
1.0.15.40: more portable contrib module installation

 * Going in despite the freeze: earlier version depended on GNU tar,
   and broke install.sh on installations where tar was not GNU.

 * Patch by Eugene Ossintsev.


Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3910
retrieving revision 1.3911
diff -u -d -r1.3910 -r1.3911
--- version.lisp-expr	26 Mar 2008 15:54:11 -0000	1.3910
+++ version.lisp-expr	26 Mar 2008 19:32:35 -0000	1.3911
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.39"
+"1.0.15.40"

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv10004

Modified Files:
	version.lisp-expr NEWS 
Log Message:
1.0.15.39: Remove subclasses of sb-posix:syscall-error

These were added in 1.0.14 (but never documented), and turn out to be
impediments to some future progress on Windows.



Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3909
retrieving revision 1.3910
diff -u -d -r1.3909 -r1.3910
--- version.lisp-expr	25 Mar 2008 21:27:33 -0000	1.3909
+++ version.lisp-expr	26 Mar 2008 15:54:11 -0000	1.3910
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.38"
+"1.0.15.39"

Index: NEWS
===================================================================
RCS file: /cvsroot/sbcl/sbcl/NEWS,v
retrieving revision 1.1322
retrieving revision 1.1323
diff -u -d -r1.1322 -r1.1323
--- NEWS	17 Mar 2008 17:13:40 -0000	1.1322
+++ NEWS	26 Mar 2008 15:54:12 -0000	1.1323
@@ -1,5 +1,7 @@
 ;;;; -*- coding: utf-8; -*-
 changes in sbcl-1.0.16 relative to 1.0.15:
+  * minor incompatible change: revert the changes to sb-posix's error
+    signaling added in 1.0.14.
   * minor incompatible change: change PROBE-FILE back to returning
     NIL whenever we can't get a truename, as was the case before 1.0.14.
   * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now

Update of /cvsroot/sbcl/sbcl/contrib/sb-posix
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv10004/contrib/sb-posix

Modified Files:
	interface.lisp 
Log Message:
1.0.15.39: Remove subclasses of sb-posix:syscall-error

These were added in 1.0.14 (but never documented), and turn out to be
impediments to some future progress on Windows.



Index: interface.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/sb-posix/interface.lisp,v
retrieving revision 1.51
retrieving revision 1.52
diff -u -d -r1.51 -r1.52
--- interface.lisp	27 Feb 2008 17:33:45 -0000	1.51
+++ interface.lisp	26 Mar 2008 15:54:12 -0000	1.52
@@ -63,23 +63,8 @@
                (format s "System call error ~A (~A)"
                        errno (sb-int:strerror errno))))))
 
-(defvar *errno-table*
-  (let ((errno-max 0)
-        list)
-    (do-symbols (symbol (find-package "SB-POSIX"))
-      (when (get symbol 'errno)
-        (let ((errno (symbol-value symbol)))
-          (setf errno-max (max errno  errno-max))
-          (push (cons errno
-                      (eval `(define-condition ,symbol (syscall-error) ())))
-                list))))
-    (let ((table (make-array (1+ errno-max))))
-      (mapc #'(lambda (cons) (setf (elt table (car cons)) (cdr cons))) list)
-      table)))
-
 (defun syscall-error ()
-  (let ((errno (get-errno)))
-    (error (elt *errno-table* errno) :errno errno)))
+  (error 'sb-posix:syscall-error :errno (get-errno)))
 
 (defun unsupported-error (lisp-name c-name)
   (error "~S is unsupported by SBCL on this platform due to lack of ~A()."
@@ -89,60 +74,6 @@
   (warn "~S is unsupported by SBCL on this platform due to lack of ~A()."
         lisp-name c-name))
 
-;; Note that we inherit from SIMPLE-FILE-ERROR first, to get its
-;; error reporting, rather than SYSCALL-ERROR's.
-(define-condition file-syscall-error
-    (sb-impl::simple-file-error syscall-error)
-  ())
-
-(defvar *file-errno-table*
-  (let ((array (copy-seq *errno-table*)))
-    (map-into array
-              (lambda (condition-class-name)
-                (if (symbolp condition-class-name)
-                    (let ((file-condition-name
-                           (read-from-string
-                            (format nil "FILE-~A" condition-class-name))))
-                      ;; Should condition class names like FILE-ENOENT
-                      ;; and FILE-ENOTDIR be exported?  I want to say
-                      ;; "no", since we already export ENOENT, ENOTDIR
-                      ;; et al, and so the user can write handlers
-                      ;; such as
-                      ;;
-                      ;;  (handler-bind ((sb-posix:enoent ...)
-                      ;;                 (sb-posix:enotdir ...)
-                      ;;                 (file-error ...))
-                      ;;    ...)
-                      ;;
-                      ;; which will do the right thing for all our
-                      ;; FILE-SYSCALL-ERRORs, without exposing this
-                      ;; implementation detail.  (Recall that some
-                      ;; FILE-ERRORs don't strictly have to do with
-                      ;; the file system, e.g., supplying a wild
-                      ;; pathname to some functions.)  But if the
-                      ;; prevailing opinion is otherwise, uncomment
-                      ;; the following.
-                      #| (export file-condition-name) |#
-                      (eval `(define-condition ,file-condition-name
-                                 (,condition-class-name file-syscall-error)
-                               ())))
-                    condition-class-name))
-              array)
-    array))
-
-;; Note: do we have to declare SIMPLE-FILE-PERROR notinline in
-;; fd-stream.lisp?
-(sb-ext:without-package-locks
-  (defun sb-impl::simple-file-perror (note-format pathname errno)
-    (error (elt *file-errno-table* errno)
-           :pathname pathname
-           :errno errno
-           :format-control "~@<~?: ~2I~_~A~:>"
-           :format-arguments
-           (list note-format (list pathname) (sb-int:strerror errno)))))
-
-;; Note: it might prove convenient to develop a parallel set of
-;; condition classes for STREAM-ERRORs, too.
 (declaim (inline never-fails))
 (defun never-fails (&rest args)
   (declare (ignore args))

Update of /cvsroot/sbcl/sbcl/contrib
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv9682/contrib

Modified Files:
	asdf-module.mk 
Log Message:
1.0.15.40: more portable contrib module installation

 * Going in despite the freeze: earlier version depended on GNU tar,
   and broke install.sh on installations where tar was not GNU.

 * Patch by Eugene Ossintsev.


Index: asdf-module.mk
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/asdf-module.mk,v
retrieving revision 1.17
retrieving revision 1.18
diff -u -d -r1.17 -r1.18
--- asdf-module.mk	13 Mar 2008 10:59:59 -0000	1.17
+++ asdf-module.mk	26 Mar 2008 19:32:35 -0000	1.18
@@ -27,6 +27,10 @@
 	     "(asdf:operate (quote asdf:test-op) :$(SYSTEM))" | \
 	  $(SBCL) --eval '(load "../asdf/asdf")'
 
-
+# KLUDGE / FIXME: Perhaps each module should have it's own list of
+# files to install? At any rate, this is a portable (we hope) way of
+# installing all the files needed -- as long as all the files are in
+# the first level directory...
 install: $(EXTRA_INSTALL_TARGETS)
-	tar --exclude=CVS -cf - . | ( cd "$(BUILD_ROOT)$(INSTALL_DIR)" && tar --no-same-owner -xpvf - )
+	cp -p $(SYSTEM).asd *.lisp *.fasl "$(BUILD_ROOT)$(INSTALL_DIR)"
+	find "$(BUILD_ROOT)$(INSTALL_DIR)" -type f -exec chown `id -u`:`id -g` {} \;

Update of /cvsroot/sbcl/sbcl/contrib/sb-posix
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv30788/contrib/sb-posix

Modified Files:
	posix-tests.lisp 
Log Message:
1.0.15.38: One more fix for Win32 (oops)

* contrib/sb-posix/posix-tests.lisp: feature-out the fcntl tests on Win32.



Index: posix-tests.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/sb-posix/posix-tests.lisp,v
retrieving revision 1.37
retrieving revision 1.38
diff -u -d -r1.37 -r1.38
--- posix-tests.lisp	27 Feb 2008 17:33:45 -0000	1.37
+++ posix-tests.lisp	25 Mar 2008 21:27:33 -0000	1.38
@@ -428,6 +428,7 @@
                   sb-posix::o-nonblock)))
   t)
 
+#-win32
 (deftest fcntl.flock.1
     (locally (declare (sb-ext:muffle-conditions sb-ext:compiler-note))
       (let ((flock (make-instance 'sb-posix:flock
@@ -462,6 +463,7 @@
   42)
 
 
+#-win32
 (deftest fcntl.flock.2
     (locally (declare (sb-ext:muffle-conditions sb-ext:compiler-note))
       (let ((flock (make-instance 'sb-posix:flock

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv30788

Modified Files:
	version.lisp-expr 
Log Message:
1.0.15.38: One more fix for Win32 (oops)

* contrib/sb-posix/posix-tests.lisp: feature-out the fcntl tests on Win32.



Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3908
retrieving revision 1.3909
diff -u -d -r1.3908 -r1.3909
--- version.lisp-expr	25 Mar 2008 21:24:24 -0000	1.3908
+++ version.lisp-expr	25 Mar 2008 21:27:33 -0000	1.3909
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.37"
+"1.0.15.38"

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29555

Modified Files:
	version.lisp-expr 
Log Message:
1.0.15.37: Windows fixes.

* tools-for-build/grovel-headers.c: add a grovel for EINTR

* contrib/sb-posix/constants.lisp: feature-out the ALIEN-FLOCK structure.



Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3907
retrieving revision 1.3908
diff -u -d -r1.3907 -r1.3908
--- version.lisp-expr	17 Mar 2008 17:13:40 -0000	1.3907
+++ version.lisp-expr	25 Mar 2008 21:24:24 -0000	1.3908
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.36"
+"1.0.15.37"

Update of /cvsroot/sbcl/sbcl/contrib/sb-posix
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29555/contrib/sb-posix

Modified Files:
	constants.lisp 
Log Message:
1.0.15.37: Windows fixes.

* tools-for-build/grovel-headers.c: add a grovel for EINTR

* contrib/sb-posix/constants.lisp: feature-out the ALIEN-FLOCK structure.



Index: constants.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/sb-posix/constants.lisp,v
retrieving revision 1.41
retrieving revision 1.42
diff -u -d -r1.41 -r1.42
--- constants.lisp	27 Feb 2008 17:33:45 -0000	1.41
+++ constants.lisp	25 Mar 2008 21:24:24 -0000	1.42
@@ -362,6 +362,7 @@
  (:integer f-wrlck "F_WRLCK" nil t)
  (:integer f-unlck "F_UNLCK" nil t)
 
+ #-win32
  (:structure alien-flock
              ("struct flock"
               (short type "short" "l_type")

Update of /cvsroot/sbcl/sbcl/tools-for-build
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29555/tools-for-build

Modified Files:
	grovel-headers.c 
Log Message:
1.0.15.37: Windows fixes.

* tools-for-build/grovel-headers.c: add a grovel for EINTR

* contrib/sb-posix/constants.lisp: feature-out the ALIEN-FLOCK structure.



Index: grovel-headers.c
===================================================================
RCS file: /cvsroot/sbcl/sbcl/tools-for-build/grovel-headers.c,v
retrieving revision 1.32
retrieving revision 1.33
diff -u -d -r1.32 -r1.33
--- grovel-headers.c	31 Dec 2007 18:39:23 -0000	1.32
+++ grovel-headers.c	25 Mar 2008 21:24:24 -0000	1.33
@@ -208,6 +208,7 @@
 
     defconstant("enoent", ENOENT);
     defconstant("eexist", EEXIST);
+    defconstant("eintr", EINTR);
 
     defconstant("s-ifmt",  S_IFMT);
     defconstant("s-ifdir", S_IFDIR);

Update of /cvsroot/sbcl/sbcl/src/compiler/mips
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/mips

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/mips/c-call.lisp,v
retrieving revision 1.11
retrieving revision 1.12
diff -u -d -r1.11 -r1.12
--- c-call.lisp	1 Sep 2007 18:06:17 -0000	1.11
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.12
@@ -271,6 +271,7 @@
 
 (define-vop (alloc-number-stack-space)
   (:info amount)
+  (:result-types system-area-pointer)
   (:results (result :scs (sap-reg any-reg)))
   (:temporary (:scs (unsigned-reg) :to (:result 0)) temp)
   (:generator 0

Update of /cvsroot/sbcl/sbcl/src/compiler/x86
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/x86

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/x86/c-call.lisp,v
retrieving revision 1.40
retrieving revision 1.41
diff -u -d -r1.40 -r1.41
--- c-call.lisp	30 Nov 2007 17:46:10 -0000	1.40
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.41
@@ -279,6 +279,7 @@
 (define-vop (alloc-number-stack-space)
   (:info amount)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   (:generator 0
     (aver (location= result esp-tn))
     (unless (zerop amount)
@@ -291,6 +292,7 @@
   (:info amount)
   #!+sb-thread (:temporary (:sc unsigned-reg) temp)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   #!+sb-thread
   (:generator 0
     (aver (not (location= result esp-tn)))

Update of /cvsroot/sbcl/sbcl/src/compiler/sparc
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/sparc

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/sparc/c-call.lisp,v
retrieving revision 1.8
retrieving revision 1.9
diff -u -d -r1.8 -r1.9
--- c-call.lisp	14 Jul 2005 18:48:34 -0000	1.8
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.9
@@ -235,6 +235,7 @@
 (define-vop (alloc-number-stack-space)
   (:info amount)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   (:temporary (:scs (unsigned-reg) :to (:result 0)) temp)
   (:generator 0
     (unless (zerop amount)

Update of /cvsroot/sbcl/sbcl/src/compiler/x86-64
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/x86-64

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/x86-64/c-call.lisp,v
retrieving revision 1.23
retrieving revision 1.24
diff -u -d -r1.23 -r1.24
--- c-call.lisp	1 Jan 2008 14:45:33 -0000	1.23
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.24
@@ -269,6 +269,7 @@
 (define-vop (alloc-number-stack-space)
   (:info amount)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   (:generator 0
     (aver (location= result rsp-tn))
     (unless (zerop amount)
@@ -289,6 +290,7 @@
   (:info amount)
   #!+sb-thread (:temporary (:sc unsigned-reg) temp)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   #!+sb-thread
   (:generator 0
     (aver (not (location= result rsp-tn)))

Update of /cvsroot/sbcl/sbcl/src/compiler/ppc
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/ppc

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/ppc/c-call.lisp,v
retrieving revision 1.17
retrieving revision 1.18
diff -u -d -r1.17 -r1.18
--- c-call.lisp	23 Nov 2006 15:11:27 -0000	1.17
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.18
@@ -428,6 +428,7 @@
 (define-vop (alloc-number-stack-space)
   (:info amount)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   (:temporary (:scs (unsigned-reg) :to (:result 0)) temp)
   (:generator 0
     (unless (zerop amount)

Update of /cvsroot/sbcl/sbcl/src/compiler/alpha
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/alpha

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/alpha/c-call.lisp,v
retrieving revision 1.9
retrieving revision 1.10
diff -u -d -r1.9 -r1.10
--- c-call.lisp	20 Aug 2005 17:21:14 -0000	1.9
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.10
@@ -142,6 +142,7 @@
 (define-vop (alloc-number-stack-space)
   (:info amount)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   (:temporary (:scs (unsigned-reg) :to (:result 0)) temp)
   (:generator 0
     (unless (zerop amount)

Update of /cvsroot/sbcl/sbcl/tests
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/tests

Modified Files:
	compiler.pure.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: compiler.pure.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/tests/compiler.pure.lisp,v
retrieving revision 1.170
retrieving revision 1.171
diff -u -d -r1.170 -r1.171
--- compiler.pure.lisp	30 Nov 2007 17:46:10 -0000	1.170
+++ compiler.pure.lisp	17 Mar 2008 17:13:41 -0000	1.171
@@ -2323,3 +2323,22 @@
                (declare (fixnum y) (character x))
                (sb-sys:with-pinned-objects (x y)
                  (some-random-function))))
+
+;;; *CHECK-CONSISTENCY* and TRULY-THE
+
+(with-test (:name :bug-423)
+  (let ((sb-c::*check-consistency* t))
+    (handler-bind ((warning #'error))
+      (flet ((make-lambda (type)
+               `(lambda (x)
+                  ((lambda (z)
+                     (if (listp z)
+                         (let ((q (truly-the list z)))
+                           (length q))
+                         (if (arrayp z)
+                             (let ((q (truly-the vector z)))
+                               (length q))
+                             (error "oops"))))
+                   (the ,type x)))))
+        (compile nil (make-lambda 'list))
+        (compile nil (make-lambda 'vector))))))

Update of /cvsroot/sbcl/sbcl/src/compiler/hppa
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler/hppa

Modified Files:
	c-call.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: c-call.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/hppa/c-call.lisp,v
retrieving revision 1.6
retrieving revision 1.7
diff -u -d -r1.6 -r1.7
--- c-call.lisp	14 Jul 2005 18:48:31 -0000	1.6
+++ c-call.lisp	17 Mar 2008 17:13:41 -0000	1.7
@@ -149,6 +149,7 @@
 (define-vop (alloc-number-stack-space)
   (:info amount)
   (:results (result :scs (sap-reg any-reg)))
+  (:result-types system-area-pointer)
   (:temporary (:scs (unsigned-reg) :to (:result 0)) temp)
   (:generator 0
     (move nsp-tn result)

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062

Modified Files:
	BUGS NEWS version.lisp-expr 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: BUGS
===================================================================
RCS file: /cvsroot/sbcl/sbcl/BUGS,v
retrieving revision 1.533
retrieving revision 1.534
diff -u -d -r1.533 -r1.534
--- BUGS	13 Mar 2008 15:40:28 -0000	1.533
+++ BUGS	17 Mar 2008 17:13:40 -0000	1.534
@@ -1889,26 +1889,6 @@
 2007-07-06. (We don't _have_ to check things like this, but we
 generally try to check returns in safe code, so we should here too.)
 
-423: TRULY-THE and *CHECK-CONSISTENCY*
-
- The following signals errors due to TRULY-THEs in dead code:
-
- (let ((sb-c::*check-consistency* t))
-  (handler-bind ((warning #'error))
-    (flet ((make-lambda (type)
-             `(lambda (x)
-                ((lambda (z)
-                   (if (listp z)
-                       (let ((q (truly-the list z)))
-                         (length q))
-                       (if (arrayp z)
-                           (let ((q (truly-the vector z)))
-                             (length q))
-                           (error "oops"))))
-                 (the ,type x)))))
-      (compile nil (make-lambda 'list))
-      (compile nil (make-lambda 'vector)))))
-
 424: toplevel closures and *CHECK-CONSISTENCY*
 
  The following breaks under COMPILE-FILE if *CHECK-CONSISTENCY* is true.

Index: NEWS
===================================================================
RCS file: /cvsroot/sbcl/sbcl/NEWS,v
retrieving revision 1.1321
retrieving revision 1.1322
diff -u -d -r1.1321 -r1.1322
--- NEWS	14 Mar 2008 19:44:25 -0000	1.1321
+++ NEWS	17 Mar 2008 17:13:40 -0000	1.1322
@@ -15,6 +15,7 @@
   * optimization: modular arithmetic for a particular requested width
     is implemented using a tagged representation unless a better 
     representation is available.
+  * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
   * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
     for wellformedness and returns a specialized vector. (reported by
     Francois-Rene Rideau)

Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3906
retrieving revision 1.3907
diff -u -d -r1.3906 -r1.3907
--- version.lisp-expr	17 Mar 2008 14:10:04 -0000	1.3906
+++ version.lisp-expr	17 Mar 2008 17:13:40 -0000	1.3907
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.35"
+"1.0.15.36"

Update of /cvsroot/sbcl/sbcl/contrib/sb-introspect
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/contrib/sb-introspect

Modified Files:
	test-driver.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: test-driver.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/sb-introspect/test-driver.lisp,v
retrieving revision 1.10
retrieving revision 1.11
diff -u -d -r1.10 -r1.11
--- test-driver.lisp	18 Jun 2007 16:15:54 -0000	1.10
+++ test-driver.lisp	17 Mar 2008 17:13:40 -0000	1.11
@@ -10,7 +10,7 @@
 (assert (equal (function-arglist 'cl-user::one)
                '(cl-user::a cl-user::b cl-user::c)))
 (assert (equal (function-arglist 'the)
-               '(type sb-c::value)))
+               '(sb-c::value-type sb-c::form)))
 
 (assert (equal (function-arglist #'(sb-pcl::slow-method cl-user::j (t)))
                '(sb-pcl::method-args sb-pcl::next-methods)))
@@ -81,7 +81,7 @@
 (assert (equal (function-arglist 'cl-user::one)
                '(cl-user::a cl-user::b cl-user::c)))
 (assert (equal (function-arglist 'the)
-               '(type sb-c::value)))
+               '(sb-c::value-type sb-c::form)))
 
 ;;; Check wrt. interplay of generic functions and their methods.

Update of /cvsroot/sbcl/sbcl/src/compiler
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv29062/src/compiler

Modified Files:
	ir1-translators.lisp 
Log Message:
1.0.15.36: fix bug 423

 * TRULY-THE no longer eagerly annotates the LVAR, but rather acts
   like THE in unsafe code. (Inserts a cast for which no type-check is
   generated.)

 * ALLOC-NUMBER-STACK-SPACE and ALLOC-ALIEN-STACK-SPACE were missing
   :RESULT-TYPES, which was masked by the old TRULY-THE
   implementation. (Tested on x86, x86-64, PPC, but all backends
   updated -- hopefully correctly.)

 * Docstrings for THE and TRULY-THE.

 FWIW, this patch seems to make PPC+GENCGC build happy once again: but
 I don't know if that means the problems there were/are related to bug
 423 -- or if the exact memory layout just happens to change subtly so
 that whatever corruption occurs, just happens to occur in a place
 where the GC doesn't see it anymore. (Or if the exact point at which
 GC runs is now just slightly different so that the problem pointers
 are not live anymore, or...)


Index: ir1-translators.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/compiler/ir1-translators.lisp,v
retrieving revision 1.85
retrieving revision 1.86
diff -u -d -r1.85 -r1.86
--- ir1-translators.lisp	4 Feb 2008 20:49:27 -0000	1.85
+++ ir1-translators.lisp	17 Mar 2008 17:13:40 -0000	1.86
@@ -860,26 +860,34 @@
 
 ;;; Assert that FORM evaluates to the specified type (which may be a
 ;;; VALUES type). TYPE may be a type specifier or (as a hack) a CTYPE.
-(def-ir1-translator the ((type value) start next result)
-  (the-in-policy type value (lexenv-policy *lexenv*) start next result))
+(def-ir1-translator the ((value-type form) start next result)
+  #!+sb-doc
+  "Specifies that the values returned by FORM conform to the VALUE-TYPE.
+
+CLHS specifies that the consequences are undefined if any result is
+not of the declared type, but SBCL treats declarations as assertions
+as long as SAFETY is at least 2, in which case incorrect type
+information will result in a runtime type-error instead of leading to
+eg. heap corruption. This is however expressly non-portable: use
+CHECK-TYPE instead of THE to catch type-errors at runtime. THE is best
+considered an optimization tool to inform the compiler about types it
+is unable to derive from other declared types."
+  (the-in-policy value-type form (lexenv-policy *lexenv*) start next result))
 
 ;;; This is like the THE special form, except that it believes
 ;;; whatever you tell it. It will never generate a type check, but
 ;;; will cause a warning if the compiler can prove the assertion is
 ;;; wrong.
-(def-ir1-translator truly-the ((type value) start next result)
+(def-ir1-translator truly-the ((value-type form) start next result)
   #!+sb-doc
-  ""
-  #-nil
-  (let ((type (coerce-to-values (compiler-values-specifier-type type)))
-        (old (when result (find-uses result))))
-    (ir1-convert start next result value)
-    (when result
-      (do-uses (use result)
-        (unless (memq use old)
-          (derive-node-type use type)))))
-  #+nil
-  (the-in-policy type value '((type-check . 0)) start cont))
+  "Specifies that the values returned by FORM conform to the
+VALUE-TYPE, and causes the compiler to trust this information
+unconditionally.
+
+Consequences are undefined if any result is not of the declared type
+-- typical symptoms including memory corruptions. Use with great
+care."
+  (the-in-policy value-type form '((type-check . 0)) start next result))
 
 ;;;; SETQ

Update of /cvsroot/sbcl/sbcl/src/runtime
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv16384/src/runtime

Modified Files:
	gc-common.c gencgc.c 
Log Message:
1.0.15.35: fix bitrotted GC sanity checks

 * One missing #ifdef.

 * Better lossage messages for for some cases.


Index: gc-common.c
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/runtime/gc-common.c,v
retrieving revision 1.51
retrieving revision 1.52
diff -u -d -r1.51 -r1.52
--- gc-common.c	19 Feb 2008 10:18:07 -0000	1.51
+++ gc-common.c	17 Mar 2008 14:10:04 -0000	1.52
@@ -137,7 +137,9 @@
 
         lispobj object = *object_ptr;
 #ifdef LISP_FEATURE_GENCGC
-        gc_assert(!forwarding_pointer_p(object_ptr));
+        if (forwarding_pointer_p(object_ptr))
+            lose("unexpect forwarding pointer in scavenge: %p, start=%p, n=%l\n",
+                 object_ptr, start, n_words);
 #endif
         if (is_lisp_pointer(object)) {
             if (from_space_p(object)) {

Index: gencgc.c
===================================================================
RCS file: /cvsroot/sbcl/sbcl/src/runtime/gencgc.c,v
retrieving revision 1.125
retrieving revision 1.126
diff -u -d -r1.125 -r1.126
--- gencgc.c	19 Feb 2008 16:06:34 -0000	1.125
+++ gencgc.c	17 Mar 2008 14:10:05 -0000	1.126
@@ -1571,6 +1571,8 @@
     if (!check_code_fixups)
         return;
 
+    FSHOW((stderr, "/sniffing code: %p, %lu\n", code, displacement));
+
     ncode_words = fixnum_value(code->code_size);
     nheader_words = HeaderValue(*(lispobj *)code);
     nwords = ncode_words + nheader_words;
@@ -3502,15 +3504,14 @@
 #ifdef LUTEX_WIDETAG
                 case LUTEX_WIDETAG:
 #endif
+#ifdef NO_TLS_VALUE_MARKER_WIDETAG
+                case NO_TLS_VALUE_MARKER_WIDETAG:
+#endif
                     count = (sizetab[widetag_of(*start)])(start);
                     break;
 
                 default:
-                    FSHOW((stderr,
-                           "/Unhandled widetag 0x%x at 0x%x\n",
-                           widetag_of(*start), start));
-                    fflush(stderr);
-                    gc_abort();
+                    lose("Unhandled widetag 0x%x at 0x%x\n", widetag_of(*start), start);
                 }
             }
         }
@@ -4445,8 +4446,7 @@
 
     if (verify_after_free_heap) {
         /* Check whether purify has left any bad pointers. */
-        if (gencgc_verbose)
-            SHOW("checking after free_heap\n");
+        FSHOW((stderr, "checking after free_heap\n"));
         verify_gc();
     }
 }

Update of /cvsroot/sbcl/sbcl
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv16384

Modified Files:
	version.lisp-expr 
Log Message:
1.0.15.35: fix bitrotted GC sanity checks

 * One missing #ifdef.

 * Better lossage messages for for some cases.


Index: version.lisp-expr
===================================================================
RCS file: /cvsroot/sbcl/sbcl/version.lisp-expr,v
retrieving revision 1.3905
retrieving revision 1.3906
diff -u -d -r1.3905 -r1.3906
--- version.lisp-expr	14 Mar 2008 20:41:25 -0000	1.3905
+++ version.lisp-expr	17 Mar 2008 14:10:04 -0000	1.3906
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"1.0.15.34"
+"1.0.15.35"

Update of /cvsroot/sbcl/sbcl/contrib/sb-bsd-sockets
In directory sc8-pr-cvs8.sourceforge.net:/tmp/cvs-serv26928/contrib/sb-bsd-sockets

Modified Files:
	inet.lisp 
Log Message:
1.0.15.34: tweak the PCL cache improvement from 1.0.15.12

 * Thanks to Paul Khuong for noting that we weren't actually dropping
   50% of the entries, but somewhat less. Now (assuming our RANDOM is
   good) we really should average 50%.


Index: inet.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/sb-bsd-sockets/inet.lisp,v
retrieving revision 1.11
retrieving revision 1.12
diff -u -d -r1.11 -r1.12
--- inet.lisp	14 Mar 2008 19:44:25 -0000	1.11
+++ inet.lisp	14 Mar 2008 20:41:26 -0000	1.12
@@ -25,7 +25,7 @@
 \"127.0.0.1\". Signals an error if the string is malformed."
   (declare (type string dotted-quads))
   (labels ((oops ()
-             (error "~S is not a string designating an IP address." 
+             (error "~S is not a string designating an IP address."
                     dotted-quads))
            (check (x)
              (if (typep x '(unsigned-byte 8))