kernel/linux.git/drivers/nvmem/core.c, branch v4.14.263

nvmem: Fix shift-out-of-bound (UBSAN) with byte size cells

2021-10-20T08:42:04+00:00

commit 5d388fa01fa6eb310ac023a363a6cb216d9d8fe9 upstream. If a cell has 'nbits' equal to a multiple of BITS_PER_BYTE the logic *p &= GENMASK((cell->nbits%BITS_PER_BYTE) - 1, 0); will become undefined behavior because nbits modulo BITS_PER_BYTE is 0, and we subtract one from that making a large number that is then shifted more than the number of bits that fit into an unsigned long. UBSAN reports this problem: UBSAN: shift-out-of-bounds in drivers/nvmem/core.c:1386:8 shift exponent 64 is too large for 64-bit type 'unsigned long' CPU: 6 PID: 7 Comm: kworker/u16:0 Not tainted 5.15.0-rc3+ #9 Hardware name: Google Lazor (rev3+) with KB Backlight (DT) Workqueue: events_unbound deferred_probe_work_func Call trace: dump_backtrace+0x0/0x170 show_stack+0x24/0x30 dump_stack_lvl+0x64/0x7c dump_stack+0x18/0x38 ubsan_epilogue+0x10/0x54 __ubsan_handle_shift_out_of_bounds+0x180/0x194 __nvmem_cell_read+0x1ec/0x21c nvmem_cell_read+0x58/0x94 nvmem_cell_read_variable_common+0x4c/0xb0 nvmem_cell_read_variable_le_u32+0x40/0x100 a6xx_gpu_init+0x170/0x2f4 adreno_bind+0x174/0x284 component_bind_all+0xf0/0x264 msm_drm_bind+0x1d8/0x7a0 try_to_bring_up_master+0x164/0x1ac __component_add+0xbc/0x13c component_add+0x20/0x2c dp_display_probe+0x340/0x384 platform_probe+0xc0/0x100 really_probe+0x110/0x304 __driver_probe_device+0xb8/0x120 driver_probe_device+0x4c/0xfc __device_attach_driver+0xb0/0x128 bus_for_each_drv+0x90/0xdc __device_attach+0xc8/0x174 device_initial_probe+0x20/0x2c bus_probe_device+0x40/0xa4 deferred_probe_work_func+0x7c/0xb8 process_one_work+0x128/0x21c process_scheduled_works+0x40/0x54 worker_thread+0x1ec/0x2a8 kthread+0x138/0x158 ret_from_fork+0x10/0x20 Fix it by making sure there are any bits to mask out. Fixes: 69aba7948cbe ("nvmem: Add a simple NVMEM framework for consumers") Cc: Douglas Anderson Cc: stable@vger.kernel.org Signed-off-by: Stephen Boyd Signed-off-by: Srinivas Kandagatla Link: https://lore.kernel.org/r/20211013124511.18726-1-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman Signed-off-by: Greg Kroah-Hartman

nvmem: core: return error code instead of NULL from nvmem_device_get

2019-11-20T16:59:56+00:00

[ Upstream commit ca6ac25cecf0e740d7cc8e03e0ebbf8acbeca3df ] nvmem_device_get() should return ERR_PTR() on error or valid pointer on success, but one of the code path seems to return NULL, so fix it. Reported-by: Niklas Cassel Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman Signed-off-by: Sasha Levin

nvmem: Use the same permissions for eeprom as for nvmem

2019-09-19T07:08:08+00:00

commit e70d8b287301eb6d7c7761c6171c56af62110ea3 upstream. The compatibility "eeprom" attribute is currently root-only no matter what the configuration says. The "nvmem" attribute does respect the setting of the root_only configuration bit, so do the same for "eeprom". Signed-off-by: Jean Delvare Fixes: b6c217ab9be6 ("nvmem: Add backwards compatibility support for older EEPROM drivers.") Reviewed-by: Bartosz Golaszewski Cc: Andrew Lunn Cc: Srinivas Kandagatla Cc: Arnd Bergmann Link: https://lore.kernel.org/r/20190728184255.563332e6@endymion Signed-off-by: Greg Kroah-Hartman

nvmem: core: fix read buffer in place

2019-06-15T09:54:54+00:00

[ Upstream commit 2fe518fecb3a4727393be286db9804cd82ee2d91 ] When the bit_offset in the cell is zero, the pointer to the msb will not be properly initialized (ie, will still be pointing to the first byte in the buffer). This being the case, if there are bits to clear in the msb, those will be left untouched while the mask will incorrectly clear bit positions on the first byte. This commit also makes sure that any byte unused in the cell is cleared. Signed-off-by: Jorge Ramirez-Ortiz Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman Signed-off-by: Sasha Levin

nvmem: Don't let a NULL cell_id for nvmem_cell_get() crash us

2018-08-24T11:09:14+00:00

[ Upstream commit 87ed1405ef09d29a14df43295f7b6a93b63bfe6e ] In commit ca04d9d3e1b1 ("phy: qcom-qusb2: New driver for QUSB2 PHY on Qcom chips") you can see a call like: devm_nvmem_cell_get(dev, NULL); Note that the cell ID passed to the function is NULL. This is because the qcom-qusb2 driver is expected to work only on systems where the PHY node is hooked up via device-tree and is nameless. This works OK for the most part. The first thing nvmem_cell_get() does is to call of_nvmem_cell_get() and there it's documented that a NULL name is fine. The problem happens when the call to of_nvmem_cell_get() returns -EINVAL. In such a case we'll fall back to nvmem_cell_get_from_list() and eventually might (if nvmem_cells isn't an empty list) crash with something that looks like: strcmp nvmem_find_cell __nvmem_device_get nvmem_cell_get_from_list nvmem_cell_get devm_nvmem_cell_get qusb2_phy_probe There are several different ways we could fix this problem: One could argue that perhaps the qcom-qusb2 driver should be changed to use of_nvmem_cell_get() which is allowed to have a NULL name. In that case, we'd need to add a patche to introduce devm_of_nvmem_cell_get() since the qcom-qusb2 driver is using devm managed resources. One could also argue that perhaps we could just add a name to qcom-qusb2. That would be OK but I believe it effectively changes the device tree bindings, so maybe it's a no-go. In this patch I have chosen to fix the problem by simply not crashing when a NULL cell_id is passed to nvmem_cell_get(). NOTE: that for the qcom-qusb2 driver the "nvmem-cells" property is defined to be optional and thus it's expected to be a common case that we would hit this crash and this is more than just a theoretical fix. Fixes: ca04d9d3e1b1 ("phy: qcom-qusb2: New driver for QUSB2 PHY on Qcom chips") Signed-off-by: Douglas Anderson Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

nvmem: properly handle returned value nvmem_reg_read

2018-08-03T05:50:34+00:00

[ Upstream commit 50808bfcc14b854775a9f1d0abe3dac2babcf5c3 ] Function nvmem_reg_read can return a non zero value indicating an error. This returned value must be read and error propagated to nvmem_cell_prepare_write_buffer. Silence the following gcc warning (W=1): drivers/nvmem/core.c:1093:9: warning: variable 'rc' set but not used [-Wunused-but-set-variable] Signed-off-by: Mathieu Malaterre Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

nvmem: add missing of_node_put() in of_nvmem_cell_get()

2017-09-18T14:12:26+00:00

of_get_next_parent() increments the refcount of the returned node. It should be put when done. Signed-off-by: Masahiro Yamada Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman

nvmem: core: return EFBIG on out-of-range write

2017-09-18T14:12:26+00:00

When writing data that exceeds the nvmem size to a nvmem sysfs file using the sh redirection operator >, the shell hangs, trying to write the out-of-range bytes endlessly. Fix the problem by returning EFBIG described in man 2 write. Similar change was done for binary sysfs files on commit 0936896056365349afa867c16e9f9100a6707cbf Signed-off-by: Guy Shapiro Cc: linux-api@vger.kernel.org Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman

nvmem: core: remove unneeded NULL check

2017-08-28T15:33:23+00:00

"p" is the list iterator so it can't be NULL. Static checkers complain about this unnecessary check because we dereference the list iterator to get the next item in the list so we'd be in trouble if it really was NULL. I have removed the check. Signed-off-by: Dan Carpenter Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman

nvmem: core: Add nvmem_cell_read_u32

2017-08-28T15:33:23+00:00

This function does a quick and easy read of an u32 value without any kind of resource management code on the consumer side. Signed-off-by: Leonard Crestez Reviewed-by: Shawn Guo Signed-off-by: Srinivas Kandagatla Signed-off-by: Greg Kroah-Hartman