[vorbis-dev] Re: libogg 1.1 test failure on alpha

[Christian Weisgerber]

Bill Nottingham <notting at redhat.com> wrote:

> The oggpackB_{look,read} functions appear to not be right on 64-bit.

I finally sat down and looked at this.

Both functions fail to mask off the upper bits.  There's an implicit
assumption that all arithmetic is & 0xffffffff, which holds on
32-bit machines, but fails on 64-bit ones.

(BTW, I don't understand the change from 1.16 to 1.17 in bitwise.c.
These expressions are exactly equivalent, aren't they?)

<p>--- src/bitwise.c.orig	2003-12-07 01:48:17.000000000 +0100
+++ src/bitwise.c	2003-12-07 01:52:58.000000000 +0100
@@ -251,7 +251,8 @@ long oggpack_look(oggpack_buffer *b,int 
 /* Read in bits without advancing the bitptr; bits <= 32 */
 long oggpackB_look(oggpack_buffer *b,int bits){
   unsigned long ret;
-  int m=32-bits;
+  unsigned long m=mask[bits];
+  int s=32-bits;
 
   bits+=b->endbit;
 
@@ -272,7 +273,7 @@ long oggpackB_look(oggpack_buffer *b,int
       }
     }
   }
-  return (ret>>(m>>1))>>((m+1)>>1);
+  return ((ret>>(s>>1))>>((s+1)>>1)&m);
 }
 
 long oggpack_look1(oggpack_buffer *b){
@@ -347,7 +348,8 @@ long oggpack_read(oggpack_buffer *b,int 
 /* bits <= 32 */
 long oggpackB_read(oggpack_buffer *b,int bits){
   unsigned long ret;
-  long m=32-bits;
+  unsigned long m=mask[bits];
+  long s=32-bits;
   
   bits+=b->endbit;
 
@@ -369,7 +371,7 @@ long oggpackB_read(oggpack_buffer *b,int
       }
     }
   }
-  ret=(ret>>(m>>1))>>((m+1)>>1);
+  ret=((ret>>(s>>1))>>((s+1)>>1)&m);
   
  overflow:
 

-- 
Christian "naddy" Weisgerber                          naddy at mips.inka.de
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.