Re: [PATCH] Add getcpu

[Florian Weimer <fweimer at redhat dot com>]

* H. J. Lu:

> +@node Process Information
> +@section Process Information

I'm not sure if this is the right place.  This is specifically not a
process attribute, after all.  Perhaps it will fit better into the
discussion of scheduling-related functionality?

> +@deftypefun int getcpu (unsigned int *cpu, unsigned int *node)
> +@standards{Linux, <sched.h>}
> +@safety{@prelim{}@mtsafe{}@assafe{}@acsafe{}}

Can we unwind through vsyscalls?  If not, the function is not AC-safe.

> +The @code{getcpu} function identifies the processor and node on which
> +the calling thread or process is currently running and writes them into
> +the integers pointed to by the @var{cpu} and @var{node} arguments.  The
> +processor is a unique small integer identifying a CPU.  The node is a

“small nonnegative integer”

> +unique small identifier identifying a NUMA node.  When either @var{cpu}

“small nonnegative integer” (not “identifier”)

> +or @var{node} is @code{NULL}, nothing is written to the respective
> +pointer.

Should we add a caveat that due to CPU hotplugging, the values might not
be so small after a while?

> +The return value is @code{0} on success and @code{-1} on failure.  The
> +following @code{errno} error condition is defined for this function:
> +
> +@table @code
> +@item EFAULT
> +Arguments point outside the calling process's address space.
> +@end table
> +@end deftypefun

I don't think the vsyscall can detect the EFAULT error condition
reliably.  Should we really document it?

What about ENOSYS?  Is this system call really supported on all
architectures?

Thanks,
Florian