Re: [patch 28/28] posix clocks: Introduce dynamic clocks

[Richard Cochran]

On Thu, Mar 03, 2011 at 06:07:59PM +0100, torbenh wrote:
> On Thu, Mar 03, 2011 at 05:01:14PM +0100, Richard Cochran wrote:
> > 
> > You are right, but I think the check should be for the capability
> > instead. Checking the file mode for RDWR seems a bit pedantic to me.
> 
> i dont see, why an fd based clock, which already has associated permissions,
> should check against the capability.
> why should the ptpd be running as root ?
> changing the permissions of /dev/ptp0 to allow ptpd to set the
> clock should be enough. 

Thinking a bit more about this, I can see three options:

1. Enfore CAP_SYS_TIME in the posix dynamic clock layer.

2. Defer the CAP_SYS_TIME check to the underlying dynamic clock. That
   puts the decision of whether a clock counts as a "system clock" to
   the author of the driver.

3. As you suggest, just use file read/write as get/set permissions.

The admin can still restrict device node ownership and read access in
any case. You could event combine these methods (1 and 3, or 1 and 2)
but I think that would only lead to user confusion.

I am not opinionated about this, but I would like to gather some
feedback before going forward. The implementation is easy in any case.

Thanks,

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