[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-kernel
Subject:    Re: [NEW PATCH] timers, sysctl, signal (only to get POSIX timers and clocks)
From:       "Albert D. Cahalan" <acahalan () cs ! uml ! edu>
Date:       1999-08-07 1:15:09
[Download RAW message or body]


Robert de Vries patches:

+       KERN_CLK_TCK=41,        /* int: Ticks per second (clock_t) */

There are several HZ values that currently happen to be the same.
We have the HZ as seen in /proc, the HZ as seen in the clocks system
call, and the HZ as actually used in the kernel for jiffies. Linus
has in the past objected to exposing the real HZ because the real HZ
may differ from that seen in /proc. I think you can satisfy that
objection by making a distinction.

+       KERN_LINUX_COUNTER_HZ=77 /* int: Frequency of free running counter */

This is a CPU hardware counter? I never would have guessed.
CPU_COUNTER_HZ and CPU_TSC_HZ are more descriptive.

+int good_timespec(const struct timespec *ts)
+{
+       if (ts == NULL) return 0;
+       if (ts->tv_sec < 0) return 0;
+       if ((ts->tv_nsec < 0) || (ts->tv_nsec >= NSEC_PER_SEC)) return 0;
+       return 1;
+}

You wouldn't have to do so much IN THE KERNEL (hint, hint) if you
used nice simple 64-bit values. Ask yourself "Why am I doing this?".
The kernel API is something we are stuck with long-term.

+       /* check whether the time lies in the past */
+       if ((val->tv_sec < ts.tv_sec) || 
+           ((val->tv_sec == ts.tv_sec) &&
+            (val->tv_nsec <= ts.tv_nsec))) {
+               /* expire immediately */
+               val->tv_sec = 0;
+               val->tv_nsec = 0;
+       }
+       else {
+               val->tv_sec -= ts.tv_sec;
+               val->tv_nsec -= ts.tv_nsec;
+               if (val->tv_nsec < 0) {
+                       val->tv_nsec += NSEC_PER_SEC;
+                       val->tv_sec--;
+               }
+       }


Hmmm...

if(val<=ts) val=0;
else val -= ts;

The structs become a permanent part of the kernel API, even on 64-bit
hardware that would be able to directly work with larger values.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

[prev in list] [next in list] [prev in thread] [next in thread]