Re: [RFC PATCH v2 0/3] acpi: Introduce prepare_remove deviceoperation

[Toshi Kani]

On Fri, 2012-11-16 at 22:43 +0100, Rafael J. Wysocki wrote:
> On Thursday, November 15, 2012 11:22:47 AM Vasilis Liaskovitis wrote:
> > As discussed in https://patchwork.kernel.org/patch/1581581/
> > the driver core remove function needs to always succeed. This means we need
> > to know that the device can be successfully removed before acpi_bus_trim / 
> > acpi_bus_hot_remove_device are called. This can cause panics when OSPM-initiated
> > eject or driver unbind of memory devices fails e.g with:
> > 
> > echo 1 >/sys/bus/pci/devices/PNP0C80:XX/eject
> > echo "PNP0C80:XX" > /sys/bus/acpi/drivers/acpi_memhotplug/unbind
> > 
> > since the ACPI core goes ahead and ejects the device regardless of whether the
> > the memory is still in use or not.
> 
> So the question is, does the ACPI core have to do that and if so, then why?

The problem is that acpi_memory_devcie_remove() can fail.  However,
device_release_driver() is a void function, so it cannot report its
error.  Here are function flows for SCI, sysfs eject and unbind.

SCI & sysfs eject
===
acpi_bus_hot_remove_device()
  acpi_bus_trim()
    acpi_bus_remove()
      device_release_driver()  // Driver Core
        acpi_device_remove()
          acpi_memory_device_remove()  // ACPI Driver
  acpi_evaluate_object(handle, "_EJ0",,)  // Eject


sysfs unbind
===
driver_unbind()  // Driver Core
  device_release_driver()  // Driver Core
    acpi_device_remove()
      acpi_memory_device_remove()  // ACPI Driver
  put_device()
  bus_put()

Yasuaki's approach was to change device_release_driver() to report an
error so that acpi_bus_hot_remove_device() can fail without ejecting.
Vasilis's approach was to call ACPI driver via a new interface before
device_release_driver(), but still requires to change driver_unbind().
It looks to me that some changes to driver core is needed...

Thanks,
-Toshi

--
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/