Re: PHP performance on Xen domU with mulitple vcpu

>>  do {
>>    (void)gettimeofday((struct timeval *) &tv, (struct timezone *) NULL);
>>  } while (tv.tv_sec == prev_tv.tv_sec && tv.tv_usec == prev_tv.tv_usec);

> Yes, that's the difference between precision and resolution.

Yes, but that doesn't strike me as very relevant.

> While gettimeofday yields a value with a microsecond precision, the
> resolution depends on the timecounter, and some timecounters only
> provide a resolution of HZ ticks per second (usually 10ms).

Does any of this vary with CPU count?

Even if so:

(1) IIRC, I saw code to artificially increment tv_usec, to cause
consecutive calls to return different values.  I'm not sure where I saw
that, though; it may not be relevant here.

(2) Even if it changes only every 10ms, waiting for it to change 100
times should not take more than a little over a second.

(3) None of the above explains ktrace showing gettimeofday taking
multiple _seconds_ - nor two CALLs and only one RET.  That last
actually looks odd enough I suspect something is corrupting the trace.

> and using a polling loop for uniquid is what I'd expect from php.

Whether or not it's good, bad, or just ugly, it looks to me as though
it's exposed a bug.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		[email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B