Message ID | 1459181919-13825-1-git-send-email-colin.king@canonical.com (mailing list archive) |
---|---|
State | Changes Requested, archived |
Headers | show |
On Mon, Mar 28, 2016 at 05:18:39PM +0100, Colin King wrote: > From: Colin Ian King <colin.king@canonical.com> > > intel_menlow_memory_remove sanity checks to see if device is null, however, > this check is performed after we have already passed device into a call > to acpi_driver_data. If device is null, then acpi_driver_data will produce > a null pointer dereference on device. The correct action is to sanity check > device, then assign cdev, then check if cdev is null. > Hrm, looking at this locally, that all makes sense. Taking a step back however, I notice that intel_menlow_memory_remove is an ops function pointer inside the acpi_driver structure itself, which is called from acpi_device_remove() (and probe) (drivers/acpi/bus.c). This already verifies acpi_driver is not NULL and can't get acpi_driver if acpi_device is NULL. So unless there is some other use case for this callback I'm unaware of (certainly possible) it appears to be totally redundant to do this checking here. +Rafael - is there a best practices for these acpi callbacks with respect to input validation? > Signed-off-by: Colin Ian King <colin.king@canonical.com> > --- > drivers/platform/x86/intel_menlow.c | 8 ++++++-- > 1 file changed, 6 insertions(+), 2 deletions(-) > > diff --git a/drivers/platform/x86/intel_menlow.c b/drivers/platform/x86/intel_menlow.c > index 0a919d8..185a1bd 100644 > --- a/drivers/platform/x86/intel_menlow.c > +++ b/drivers/platform/x86/intel_menlow.c > @@ -196,9 +196,13 @@ static int intel_menlow_memory_add(struct acpi_device *device) > > static int intel_menlow_memory_remove(struct acpi_device *device) > { > - struct thermal_cooling_device *cdev = acpi_driver_data(device); > + struct thermal_cooling_device *cdev; > + > + if (!device) > + return -EINVAL; > > - if (!device || !cdev) > + cdev = acpi_driver_data(device); > + if (!cdev) > return -EINVAL; > > sysfs_remove_link(&device->dev.kobj, "thermal_cooling"); > -- > 2.7.4 > >
On Monday, March 28, 2016 11:18:05 AM Darren Hart wrote: > On Mon, Mar 28, 2016 at 05:18:39PM +0100, Colin King wrote: > > From: Colin Ian King <colin.king@canonical.com> > > > > intel_menlow_memory_remove sanity checks to see if device is null, however, > > this check is performed after we have already passed device into a call > > to acpi_driver_data. If device is null, then acpi_driver_data will produce > > a null pointer dereference on device. The correct action is to sanity check > > device, then assign cdev, then check if cdev is null. > > > > Hrm, looking at this locally, that all makes sense. > > Taking a step back however, I notice that intel_menlow_memory_remove is an ops > function pointer inside the acpi_driver structure itself, which is called from > acpi_device_remove() (and probe) (drivers/acpi/bus.c). This already verifies > acpi_driver is not NULL and can't get acpi_driver if acpi_device is NULL. So > unless there is some other use case for this callback I'm unaware of (certainly > possible) it appears to be totally redundant to do this checking here. > > +Rafael - is there a best practices for these acpi callbacks with respect to > input validation? No best practices I'm aware of, but if the core does this checks anyway before calling this, they are clearly not necessary here. Thanks, Rafael -- To unsubscribe from this list: send the line "unsubscribe platform-driver-x86" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Mar 29, 2016 at 03:13:43PM +0200, Rafael Wysocki wrote: > On Monday, March 28, 2016 11:18:05 AM Darren Hart wrote: > > On Mon, Mar 28, 2016 at 05:18:39PM +0100, Colin King wrote: > > > From: Colin Ian King <colin.king@canonical.com> > > > > > > intel_menlow_memory_remove sanity checks to see if device is null, however, > > > this check is performed after we have already passed device into a call > > > to acpi_driver_data. If device is null, then acpi_driver_data will produce > > > a null pointer dereference on device. The correct action is to sanity check > > > device, then assign cdev, then check if cdev is null. > > > > > > > Hrm, looking at this locally, that all makes sense. > > > > Taking a step back however, I notice that intel_menlow_memory_remove is an ops > > function pointer inside the acpi_driver structure itself, which is called from > > acpi_device_remove() (and probe) (drivers/acpi/bus.c). This already verifies > > acpi_driver is not NULL and can't get acpi_driver if acpi_device is NULL. So > > unless there is some other use case for this callback I'm unaware of (certainly > > possible) it appears to be totally redundant to do this checking here. > > > > +Rafael - is there a best practices for these acpi callbacks with respect to > > input validation? > > No best practices I'm aware of, but if the core does this checks anyway before > calling this, they are clearly not necessary here. My position as well. Colin, would you care to respin these 2?
diff --git a/drivers/platform/x86/intel_menlow.c b/drivers/platform/x86/intel_menlow.c index 0a919d8..185a1bd 100644 --- a/drivers/platform/x86/intel_menlow.c +++ b/drivers/platform/x86/intel_menlow.c @@ -196,9 +196,13 @@ static int intel_menlow_memory_add(struct acpi_device *device) static int intel_menlow_memory_remove(struct acpi_device *device) { - struct thermal_cooling_device *cdev = acpi_driver_data(device); + struct thermal_cooling_device *cdev; + + if (!device) + return -EINVAL; - if (!device || !cdev) + cdev = acpi_driver_data(device); + if (!cdev) return -EINVAL; sysfs_remove_link(&device->dev.kobj, "thermal_cooling");