On 05/21/16 14:05, Conrad Meyer wrote:
On Sat, May 21, 2016 at 10:52 AM, Pedro F. Giffuni <p...@freebsd.org> wrote:
Author: pfg
Date: Sat May 21 17:52:44 2016
New Revision: 300377
URL: https://svnweb.freebsd.org/changeset/base/300377

Log:
  ndis(4): Avoid overflow.

  This is a long standing problem: our random() function returns an
  unsigned integer but the rand provided by ndis(4) returns an int.
  Scale it down.

  MFC after:    2 weeks

Modified:
  head/sys/compat/ndis/subr_ntoskrnl.c

Modified: head/sys/compat/ndis/subr_ntoskrnl.c
==============================================================================
--- head/sys/compat/ndis/subr_ntoskrnl.c        Sat May 21 17:38:43 2016        
(r300376)
+++ head/sys/compat/ndis/subr_ntoskrnl.c        Sat May 21 17:52:44 2016        
(r300377)
@@ -3189,7 +3189,7 @@ static int
 rand(void)
 {

-       return (random());
+       return (random() / 2 + 1);
 }

 static void



Won't this still return a negative integer in many cases?

random(9) returns u_long, whereas this rand() routine returns 'int'.

Even on architectures where long is the same size as ordinary
integers, the range of possible results of the 'random() / 2 + 1'
expression, before implicit cast to signed, is [1, 2^31] (inclusive).

According to:
sys/libkern/random.c

The result is uniform on [0, 2^31 - 1].


2^31 is not representable by typical signed 32-bit integers, so this
will wrap to INT_MIN.  Also, I'm not sure why zero is excluded from
the range.


It is not a good reason but the zero is sometimes inconvenient: if
the value is going to be used as a multiplier in some calculation
it will basically kill the random component.

On architectures where long is larger than ordinary integers, this
expression has no hope of fitting in the non-negative range of a
signed integer.

Why not instead:

return ((u_int)random() / 2);


TBH, I have seen the same conversion formula over and over and I
just repeated it, so I am glad you are asking the question. Hopefully
some else has a better answer? ;).

Pedro.
Pedro.

Best,
Conrad

_______________________________________________
svn-src-all@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

Reply via email to