CVE-2022-48975: gpiolib: fix memory leak in gpiochip_setup_dev()

In the Linux kernel, the following vulnerability has been resolved:

gpiolib: fix memory leak in gpiochipsetupdev()

Here is a backtrace report about memory leak detected in gpiochipsetupdev():

unreferenced object 0xffff88810b406400 (size 512): comm "python3", pid 1682, jiffies 4295346908 (age 24.090s) backtrace: kmalloctrace deviceadd deviceprivateinit at drivers/base/core.c:3361 (inlined by) deviceadd at drivers/base/core.c:3411 cdevdeviceadd gpiolibcdevregister gpiochipsetupdev gpiochipadddatawithkey

gcdevregister() & gcdevunregister() would call deviceadd() & devicedel() (no matter CONFIGGPIOCDEV is enabled or not) to register/unregister device.

However, if deviceadd() succeeds, some resource (like struct deviceprivate allocated by deviceprivateinit()) is not released by devicedel().

Therefore, after deviceadd() succeeds by gcdevregister(), it needs to call putdevice() to release resource in the error handle path.

Here we move forward the register of release function, and let it release every piece of resource by putdevice() instead of kfree().

While at it, fix another subtle issue, i.e. when gc->ngpio is equal to 0, we still call kcalloc() and, in case of further error, kfree() on the ZEROPTR pointer, which is not NULL. It's not a bug per se, but rather waste of the resources and potentially wrong expectation about contents of the gdev->descs variable.