Re: sparse warning, or why does jifies_to_msecs() return an int?

[David Mosberger <davidm@napali.hpl.hp.com>]

>>>>> On Fri, 14 Jan 2005 20:29:07 -0800 (PST), Linus Torvalds <torvalds@osdl.org> said:

  >> Is there are a good reason to constrain the return value to 4
  >> billion msecs?  If so, what's the proper way to shut up sparse?

  Linus> There's no good way to shut up sparse, I think. The fact is,
  Linus> we _are_ losing bits, but it doesn't matter much in this
  Linus> case.

Yes, you're right.

  Linus> I think "jiffies_to_msecs(MAX_JIFFY_OFFSET)" is fundamentally
  Linus> a suspect operation (since the ranges are different for the
  Linus> two types), and I think that the sparse warnign is correct,
  Linus> but it's one of those "doing the wrong thing is not always
  Linus> wrogn enough to matter".

How about something along the lines of the attached?  The test in
msecs_to_jiffies is non-sensical for HZ>=1000 because

 (a) jiffies_to_msecs(x) <= x, if HZ >= 1000, and
 (b) MAX_UINT <= MAX_ULONG

	--david

===== include/linux/jiffies.h 1.11 vs edited =====
--- 1.11/include/linux/jiffies.h	2005-01-04 18:48:02 -08:00
+++ edited/include/linux/jiffies.h	2005-01-14 21:52:46 -08:00
@@ -276,8 +276,10 @@
 
 static inline unsigned long msecs_to_jiffies(const unsigned int m)
 {
+#if HZ < 1000
 	if (m > jiffies_to_msecs(MAX_JIFFY_OFFSET))
 		return MAX_JIFFY_OFFSET;
+#endif
 #if HZ <= 1000 && !(1000 % HZ)
 	return (m + (1000 / HZ) - 1) / (1000 / HZ);
 #elif HZ > 1000 && !(HZ % 1000)