Re: [PATCH] hpet: factor timer allocate from open

[Clemens Ladisch <clemens@ladisch.de>]

Magnus Lynch wrote:
> The current implementation of the /dev/hpet driver couples opening the
> device with allocating one of the (scarce) timers (aka comparators).
> This is a limitation in that the main counter may be valuable to
> applications seeking a high-resolution timer who have no use for the
> interrupt generating functionality of the comparators.
> 
> This patch alters the open semantics so that when the device is
> opened, no timer is allocated. Operations that depend on a timer being
> in context implicitly attempt allocating a timer, to maintain backward
> compatibility.

Many thanks; I had planned to do something like this, but never found
the time.

> I also have changes for adding IOCTLs to get the main counter
> value--which is more straightforward than mmaping and reading the
> registers directly, and possibly necessary if mmap is ifdef'd
> away--but have separated that logically into another patch.

This driver was written before the high-resolution timer API was
available, so the only actual use case, besides backwards compatibility,
is the ability to read the timer register directly without a syscall.

A ioctl to read the main counter would just duplicate clock_gettime(),
but I cannot see any benefit over that.

> diff --git a/drivers/char/hpet.c b/drivers/char/hpet.c

Your mailer killed tabs and wrapped lines.

> @@ -438,6 +469,17 @@ hpet_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
>  {
>        struct hpet_dev *devp;
> 
> +       switch (cmd) {
> +       case HPET_INFO:
> +               break;
> +       default:
> +               {
> +                       int r = hpet_alloc_timer(file);

Applications might want to use HPET_INFO to find out which timer they
got, so I think the driver cannot avoid allocating a timer in this case.


Regards,
Clemens