More conservative defaults in GENCGC

[sbcl.git] / src / compiler / ppc / backend-parms.lisp

diff --git a/src/compiler/ppc/backend-parms.lisp b/src/compiler/ppc/backend-parms.lisp
index 9f061d6..e20c1c2 100644 (file)
--- a/src/compiler/ppc/backend-parms.lisp
+++ b/src/compiler/ppc/backend-parms.lisp
@@ -3,5 +3,24 @@
 (def!constant +backend-fasl-file-implementation+ :ppc)
 (setf *backend-register-save-penalty* 3)
 (setf *backend-byte-order* :big-endian)
-(setf *backend-page-size* 4096)
 
+(eval-when (:compile-toplevel :load-toplevel :execute)
+  ;; On Linux, the ABI specifies the page size to be 4k-64k, use the
+  ;; maximum of that range. FIXME: it'd be great if somebody would
+  ;; find out whether using exact multiples of the page size actually
+  ;; matters in the few places where that's done, or whether we could
+  ;; just use 4k everywhere.
+  (setf *backend-page-bytes* #!+linux 65536 #!-linux 4096))
+
+;;; The size in bytes of GENCGC cards, i.e. the granularity at which
+;;; writes to old generations are logged.  With mprotect-based write
+;;; barriers, this must be a multiple of the OS page size.
+(def!constant gencgc-card-bytes *backend-page-bytes*)
+;;; The minimum size of new allocation regions.  While it doesn't
+;;; currently make a lot of sense to have a card size lower than
+;;; the alloc granularity, it will, once we are smarter about finding
+;;; the start of objects.
+(def!constant gencgc-alloc-granularity 0)
+;;; The minimum size at which we release address ranges to the OS.
+;;; This must be a multiple of the OS page size.
+(def!constant gencgc-release-granularity *backend-page-bytes*)