Theodore Y. Ts'o writes:
> Date: Fri, 30 Jun 2000 13:46:09 -0700 (PDT)
> From: Andre Hedrick <[EMAIL PROTECTED]>
>
> > It can already do it with a fast system, HZ=1000 and ingos patch.
>
> You can't do that and expect sub-systems that use HZ as a reference base.
> Since I have not followed the thread, completely, I may ge out of context.
> Regardless if I am, a global assuption of the value of HZ is used by more
> than just me.
>
> What user programs are depending on HZ? We really should have a way
> of letting portable programs find out what the HZ value, so that
> it's possible to change it in the future. The POSIX interface to
> this is sysconf(_SC_CLK_TCK), but that currently returns a
> hard-coded value. It should probably find out this information via
> a sysctl argument, so that user-mode program don't have to be
> recompiled if/when the HZ value changes. (There are some really
> good arguments why it should perhaps be raised to 1024, like it is
> on other platforms such as the Alpha. After all, modern i386's are
> as fast or faster than Alpha's were back when the Alpha port decided
> to use a HZ value of 1024.)
We've had this debate before. Linus said that there are two versions
of HZ: one for the kernel and one for user-space. The user-space one
should not change. All time values reported to user-space should be
absolute units, or jiffie values should be convert to the user-space
HZ units.
Then the kernel can run the timer tick at whatever rate it likes.
Regards,
Richard....
Permanent: [EMAIL PROTECTED]
Current: [EMAIL PROTECTED]
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
Please read the FAQ at http://www.tux.org/lkml/
