usb: pd: Fix memory leak in usbpd_create()

In both the usbpd_create() failure path and usbpd_destroy(),
some associated allocations, namely the string allocated from
dev_set_name() as well as the resource-mananaged dual_role_instance
structure were not getting freed. These were detected by kmemleak.

unreferenced object 0xffffffe158c37c80 (size 128):
  comm "kworker/6:1", pid 198, jiffies 4294938493 (age 469.090s)
  hex dump (first 32 bytes):
    75 73 62 70 64 30 00 6b 6b 6b 6b 6b 6b 6b 6b 6b  usbpd0.kkkkkkkkk
    6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b  kkkkkkkkkkkkkkkk
  backtrace:
    [<ffffff894d897f38>] kmemleak_alloc+0x60/0xc0
    [<ffffff894ca3ca98>] __kmalloc_track_caller+0x220/0x3bc
    [<ffffff894cc12d00>] kvasprintf+0x80/0x10c
    [<ffffff894cc12dfc>] kvasprintf_const+0x70/0xe4
    [<ffffff894d8892d8>] kobject_set_name_vargs+0x5c/0xe8
    [<ffffff894cf3afdc>] dev_set_name+0x84/0xb0
    [<ffffff894d16aa6c>] usbpd_create+0x84/0x700
    [<ffffff894d172d14>] pdphy_probe+0x21c/0x2ec
    [<ffffff894cf42d88>] platform_drv_probe+0x5c/0xb0
    [<ffffff894cf40208>] driver_probe_device+0x2e8/0x41c
    [<ffffff894cf40c80>] __device_attach_driver+0xcc/0x144
    [<ffffff894cf3e374>] bus_for_each_drv+0x80/0xc8
    [<ffffff894cf40464>] __device_attach+0xa4/0x154
    [<ffffff894cf40534>] device_initial_probe+0x20/0x2c
    [<ffffff894cf3e518>] bus_probe_device+0x34/0x8c
    [<ffffff894cf40a1c>] deferred_probe_work_func+0xcc/0x16c

unreferenced object 0xffffffe158c00580 (size 128):
  comm "kworker/6:1", pid 198, jiffies 4294938493 (age 469.090s)
  hex dump (first 32 bytes):
    78 5d c9 58 e1 ff ff ff 78 5d c9 58 e1 ff ff ff  x].X....x].X....
    24 ef 0c 4d 89 ff ff ff 80 b0 3a 59 e1 ff ff ff  $..M......:Y....
  backtrace:
    [<ffffff894d897f38>] kmemleak_alloc+0x60/0xc0
    [<ffffff894ca3ca98>] __kmalloc_track_caller+0x220/0x3bc
    [<ffffff894cf44198>] devres_alloc_node+0x34/0x70
    [<ffffff894d0cedec>] devm_dual_role_instance_register+0x40/0x178
    [<ffffff894d16af6c>] usbpd_create+0x584/0x700
    [<ffffff894d172d14>] pdphy_probe+0x21c/0x2ec
    [<ffffff894cf42d88>] platform_drv_probe+0x5c/0xb0
    [<ffffff894cf40208>] driver_probe_device+0x2e8/0x41c
    [<ffffff894cf40c80>] __device_attach_driver+0xcc/0x144
    [<ffffff894cf3e374>] bus_for_each_drv+0x80/0xc8
    [<ffffff894cf40464>] __device_attach+0xa4/0x154
    [<ffffff894cf40534>] device_initial_probe+0x20/0x2c
    [<ffffff894cf3e518>] bus_probe_device+0x34/0x8c
    [<ffffff894cf40a1c>] deferred_probe_work_func+0xcc/0x16c
    [<ffffff894c8d1c64>] process_one_work+0x1c0/0x3d4
    [<ffffff894c8d19a4>] process_scheduled_works+0x28/0x40

Both of these can be fixed by making sure to use put_device()
instead of simply kfree(pd), as that will make sure to call
additional cleanup routines for the associated objects.

Change-Id: Id755ddef66308c7d10217b2b8ef56c47421188a5
Signed-off-by: Jack Pham <jackp@codeaurora.org>