[Sbcl-commits] CVS: sbcl/contrib/sb-sprof sb-sprof.lisp,1.25,1.26

[Nikodemus S. <de...@us...>]

Update of /cvsroot/sbcl/sbcl/contrib/sb-sprof
In directory fdv4jf1.ch3.sourceforge.com:/tmp/cvs-serv8381/contrib/sb-sprof

Modified Files:
	sb-sprof.lisp 
Log Message:
1.0.28.56: more robust stack frame pointer checking in SB-SPROF

  Foreign code might not have a frame pointer like we expect. Use
  CONTROL-STACK-POINTER-VALID-P to check it.

  Patch by Bart Botta.


Index: sb-sprof.lisp
===================================================================
RCS file: /cvsroot/sbcl/sbcl/contrib/sb-sprof/sb-sprof.lisp,v
retrieving revision 1.25
retrieving revision 1.26
diff -u -d -r1.25 -r1.26
--- sb-sprof.lisp	13 Dec 2008 10:52:07 -0000	1.25
+++ sb-sprof.lisp	17 May 2009 18:01:01 -0000	1.26
@@ -562,14 +562,11 @@
               (with-alien ((scp (* os-context-t) :local scp))
                 (let* ((pc-ptr (sb-vm:context-pc scp))
                        (fp (sb-vm::context-register scp #.sb-vm::ebp-offset)))
-                  ;; For some reason completely bogus small values for the
-                  ;; frame pointer are returned every now and then, leading
-                  ;; to segfaults. Try to avoid these cases.
-                  ;;
-                  ;; FIXME: Do a more thorough sanity check on ebp, or figure
-                  ;; out why this is happening.
-                  ;; -- JES, 2005-01-11
-                  (when (< fp 4096)
+                  ;; foreign code might not have a useful frame
+                  ;; pointer in ebp/rbp, so make sure it looks
+                  ;; reasonable before walking the stack
+                  (unless (sb-di::control-stack-pointer-valid-p (sb-sys:int-sap fp))
+                    (record samples pc-ptr)
                     (return-from sigprof-handler nil))
                   (incf (samples-trace-count samples))
                   (pushnew self (samples-sampled-threads samples))