Hmm, unless jiffies don't start at zero but somewhere near the wraparound in 2.6. That would also explain why 2.4 and 2.6 show different behaviour.
I remember a discussion on lkml where exactly this was mentioned: upon system startup, jiffies should be set some time before the wrap around in order to trigger these problems.
But I don't remember if this really has been implemented.
Also note that on hammer long is 64bit and int is 32bit, so the type mismatch "int timeout" vs. "long jiffies" certainly leads to problems once jiffies are large enougth.
Yes, certainly.
Gerd
CU Michael.
-- Info: To unsubscribe send a mail to [EMAIL PROTECTED] with "unsubscribe linux-dvb" as subject.
