Skip to contentSkip to conversation

acpi: fix for memory leak in i2c_acpi_install_space_handler

PATCH In review Vamshi K Sthambamkadi 2 participants 2 messages Nov 28, 2019 – Nov 29, 2019 ACPI

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 acpi: fix for memory leak in i2c_acpi_install_space_handler

kmemleak reported backtrace:

    [<bbee0454>] kmem_cache_alloc_trace+0x128/0x260
    [<6677f215>] i2c_acpi_install_space_handler+0x4b/0xe0
    [<1180f4fc>] i2c_register_adapter+0x186/0x400
    [<6083baf7>] i2c_add_adapter+0x4e/0x70
    [<a3ddf966>] intel_gmbus_setup+0x1a2/0x2c0 [i915]
    [<84cb69ae>] i915_driver_probe+0x8d8/0x13a0 [i915]
    [<81911d4b>] i915_pci_probe+0x48/0x160 [i915]
    [<4b159af1>] pci_device_probe+0xdc/0x160
    [<b3c64704>] really_probe+0x1ee/0x450
    [<bc029f5a>] driver_probe_device+0x142/0x1b0
    [<d8829d20>] device_driver_attach+0x49/0x50
    [<de71f045>] __driver_attach+0xc9/0x150
    [<df33ac83>] bus_for_each_dev+0x56/0xa0
    [<80089bba>] driver_attach+0x19/0x20
    [<cc73f583>] bus_add_driver+0x177/0x220
    [<7b29d8c7>] driver_register+0x56/0xf0

In i2c_acpi_remove_space_handler function, leak occurs whenever "data" parameter is initialized to 0 in call to acpi_bus_get_private_data(). This is because the parameter validity check in acpi_bus_get_private_data (condition->if(!*data)) returns EINVAL and as a consequence, memory is never freed in i2c_acpi_remove_space_handler function. Fix/Correct the parameter validity check in acpi_bus_get_private_data() as that of similarly done in acpi_get_data_full().

---
 drivers/acpi/bus.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/acpi/bus.c +1 -1
index 48bc96d..5400267 100644
@@ -153,7 +153,7 @@ int acpi_bus_get_private_data(acpi_handle handle, void **data)
 {
 	acpi_status status;
 
-	if (!*data)
+	if (!data)
 		return -EINVAL;
 
 	status = acpi_get_data(handle, acpi_bus_private_data_handler, data);
Signature · 1 line
2.7.4
#1 lore raw
Quoted 36 lines · CET Vamshi K Sthambamkadi

kmemleak reported backtrace:

    [<bbee0454>] kmem_cache_alloc_trace+0x128/0x260
    [<6677f215>] i2c_acpi_install_space_handler+0x4b/0xe0
    [<1180f4fc>] i2c_register_adapter+0x186/0x400
    [<6083baf7>] i2c_add_adapter+0x4e/0x70
    [<a3ddf966>] intel_gmbus_setup+0x1a2/0x2c0 [i915]
    [<84cb69ae>] i915_driver_probe+0x8d8/0x13a0 [i915]
    [<81911d4b>] i915_pci_probe+0x48/0x160 [i915]
    [<4b159af1>] pci_device_probe+0xdc/0x160
    [<b3c64704>] really_probe+0x1ee/0x450
    [<bc029f5a>] driver_probe_device+0x142/0x1b0
    [<d8829d20>] device_driver_attach+0x49/0x50
    [<de71f045>] __driver_attach+0xc9/0x150
    [<df33ac83>] bus_for_each_dev+0x56/0xa0
    [<80089bba>] driver_attach+0x19/0x20
    [<cc73f583>] bus_add_driver+0x177/0x220
    [<7b29d8c7>] driver_register+0x56/0xf0

In i2c_acpi_remove_space_handler function, leak occurs whenever "data" parameter is initialized to 0 in call to acpi_bus_get_private_data(). This is because the parameter validity check in acpi_bus_get_private_data (condition->if(!*data)) returns EINVAL and as a consequence, memory is never freed in i2c_acpi_remove_space_handler function. Fix/Correct the parameter validity check in acpi_bus_get_private_data() as that of similarly done in acpi_get_data_full().

---
 drivers/acpi/bus.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/acpi/bus.c +0 -0
index 48bc96d..5400267 100644

drivers/acpi/bus.c int acpi_bus_get_private_data(acpi_handle handle, void **data) · line 159

Quoted 8 lines
drivers/acpi/bus.c +1 -1
@@ -153,7 +153,7 @@ int acpi_bus_get_private_data(acpi_handle handle, void **data)
 {
 	acpi_status status;
 
-	if (!*data)
+	if (!data)
 		return -EINVAL;
 
 	status = acpi_get_data(handle, acpi_bus_private_data_handler, data);

Applying as a stable-candidate fix for 5.5 with rewritten subject and changelog.

Thanks!