Re: [RFC patch 0/4] TSC calibration improvements

[Willy Tarreau <w@1wt.eu>]

On Sat, Sep 06, 2008 at 02:10:32PM -0700, Linus Torvalds wrote:

> The fact is, the code that Ingo added was totally bogus. The real bug was 
> that he did a totally bogus "--expect" in the argument to that last call. 

BTW, I hate to see state-changing instructions inside an if condition.
I've been bitten several times while debugging. You try to temporarily
comment out the if statement for a test and you end up with different
code. Same for printf. Examples of dangerous usages :

      i = 0;
      for (x = 0; x < 100; x++) {
         update_var(&i);
         if (debug && i--)
             printf("Hey I'm here\n");
      }
      return i;

You can bet that the if will go away before production. Variant with
similar effects :

      i = 0;
      for (x = 0; x < 100; x++) {
         update_var(&i);
         printf("Hey I'm here : %d\n", --i);
      }
      return i;

Since it costs nothing (except one tab and one LF) to put the instruction
out of the condition, I prefer to see them extracted :

      i = 0;
      for (x = 0; x < 100; x++) {
         update_var(&i);
         i--;
         if (debug)
             printf("Hey I'm here\n");
      }
      return i;
> 

Willy