On Tue, 19 Oct 2010, Lin Ming wrote:
> On Tue, Oct 19, 2010 at 12:49 AM, Len Brown <lenb@xxxxxxxxxx> wrote:
> > On Mon, 18 Oct 2010, Stephen Hemminger wrote:
> >
> >> Found by running make namespacecheck on linux-next
> >>
> >> Signed-off-by: Stephen Hemminger <shemminger@xxxxxxxxxx>
> >>
> >> --- a/drivers/acpi/osl.c 2010-10-06 13:42:58.722974685 +0900
> > ...
> >> /*
> >> - * Allocate the memory for a spinlock and initialize it.
> >> - */
> >> -acpi_status acpi_os_create_lock(acpi_spinlock * handle)
> >> -{
> >> - spin_lock_init(*handle);
> >> -
> >> - return AE_OK;
> >> -}
> >
> > ACPICA declares acpi_os_create_lock in acpiosxf.h,
> > but never calls it. Bob, can we do away with it?
>
> In ACPICA code (AcpiUtMutexInitialize),
>
> Status = AcpiOsCreateLock (&AcpiGbl_GpeLock);
> if (ACPI_FAILURE (Status))
> {
> return_ACPI_STATUS (Status);
> }
>
> Status = AcpiOsCreateLock (&AcpiGbl_HardwareLock);
> if (ACPI_FAILURE (Status))
> {
> return_ACPI_STATUS (Status);
> }
>
> In Linux code (acpi_ut_mutex_initialize)
>
> spin_lock_init(acpi_gbl_gpe_lock);
> spin_lock_init(acpi_gbl_hardware_lock);
>
> This is another divergence between ACPICA and Linux code.
Lin-Ming,
Thanks for the explanation.
I've applied Stephen's patch, since it is correct.
If you later want to send me patches to reduce the divergence
between ACPICA and Linux/ACPICA, I'll be happy to accept them.
thanks,
Len Brown, Intel Open Source Technology Center
