Hello Liu Jinsong, This is a semi-automatic email about new static checker warnings. The patch 259f201cb7ea: "xen/acpi: ACPI memory hotplug" from Jan 24, 2013, leads to the following Smatch complaint: drivers/xen/xen-acpi-memhotplug.c:198 acpi_memory_get_device() error: we previously assumed 'device' could be null (see line 171) drivers/xen/xen-acpi-memhotplug.c 170 171 if (!acpi_bus_get_device(handle, &device) && device) ^^^^^^ New check. Btw, checking device is unnecessary. if (acpi_bus_get_device(handle, &device) == 0) goto end; A successful "Get Device" means that "device" is non-NULL; that's built into the name. Anyway, if acpi_bus_get_device() fails either something else will fail or we will Oops in the call to acpi_driver_data(). 172 goto end; 173 174 status = acpi_get_parent(handle, &phandle); 175 if (ACPI_FAILURE(status)) { 176 pr_warn(PREFIX "Cannot find acpi parent\n"); 177 return -EINVAL; 178 } 179 180 /* Get the parent device */ 181 result = acpi_bus_get_device(phandle, &pdevice); 182 if (result) { 183 pr_warn(PREFIX "Cannot get acpi bus device\n"); 184 return -EINVAL; 185 } 186 187 /* 188 * Now add the notified device. This creates the acpi_device 189 * and invokes .add function 190 */ 191 result = acpi_bus_scan(handle); 192 if (result) { 193 pr_warn(PREFIX "Cannot add acpi bus\n"); 194 return -EINVAL; 195 } 196 197 end: 198 *mem_device = acpi_driver_data(device); ^^^^^^ Dereference. 199 if (!(*mem_device)) { 200 pr_err(PREFIX "Driver data not found\n"); regards, dan carpenter
Dan Carpenter wrote:> Hello Liu Jinsong, > > This is a semi-automatic email about new static checker warnings. > > The patch 259f201cb7ea: "xen/acpi: ACPI memory hotplug" from Jan 24, > 2013, leads to the following Smatch complaint: > > drivers/xen/xen-acpi-memhotplug.c:198 acpi_memory_get_device() > error: we previously assumed 'device' could be null (see line 171) > > drivers/xen/xen-acpi-memhotplug.c > 170 > 171 if (!acpi_bus_get_device(handle, &device) && device) > ^^^^^^ > New check. > > Btw, checking device is unnecessary. > > if (acpi_bus_get_device(handle, &device) == 0) > goto end; > > A successful "Get Device" means that "device" is non-NULL; that's > built into the name. Anyway, if acpi_bus_get_device() fails either > something else will fail or we will Oops in the call to > acpi_driver_data(). > > 172 goto end; > 173 > 174 status = acpi_get_parent(handle, &phandle); > 175 if (ACPI_FAILURE(status)) { > 176 pr_warn(PREFIX "Cannot find acpi parent\n"); > 177 return -EINVAL; > 178 } > 179 > 180 /* Get the parent device */ > 181 result = acpi_bus_get_device(phandle, &pdevice); > 182 if (result) { > 183 pr_warn(PREFIX "Cannot get acpi bus device\n"); > 184 return -EINVAL; > 185 } > 186 > 187 /* > 188 * Now add the notified device. This creates the acpi_device > 189 * and invokes .add function > 190 */ > 191 result = acpi_bus_scan(handle); > 192 if (result) { > 193 pr_warn(PREFIX "Cannot add acpi bus\n"); > 194 return -EINVAL; > 195 } > 196 > 197 end: > 198 *mem_device = acpi_driver_data(device); > ^^^^^^ > Dereference. > > 199 if (!(*mem_device)) { > 200 pr_err(PREFIX "Driver data not found\n"); > > regards, > dan carpenterThanks Dan, updated, will send out minutes later. Regards, Jinsong
Reasonably Related Threads
- xen/acpi: ACPI memory hotplug
- [PATCH] linux: fix processor handling in presence of external control
- [RFC PATCH 6/7] dt-bindings: arm: Add virtio transport for SCMI
- [PATCH 1/4] drm: retrieve EDID via ACPI _DDC method
- [PATCH] drm/nouveau: don't return freed object from nouveau_handle_create