On Fri, Feb 12, 2016 at 07:37:10PM -0500, Stefan Berger wrote: > Jason Gunthorpe <jguntho...@obsidianresearch.com> wrote on 02/12/2016 > 07:04:30 PM: > > > > This is a hold over from before the struct device conversion. > > > > - All prints should be using &chip->dev, which is the Linux > > standard. This changes prints to use tpm0 as the device name, > > not the PnP/etc ID. > > - The few places involving sysfs/modules that really do need the > > parent just use chip->dev.parent instead > > - We no longer need to get_device(pdev) in any places since it is no > > longer used by any of the code. The kref on the parent is held > > by the device core during device_add and dropped in device_del > That is exactly what was needed for the vtpm driver and now you're > removing it. Is that still going to work after this change? Or do we > need to re-add it as get/put_device(chip->dev.parent) ?
That code was not correct, the get_device side has racy lack-of-locking problems and it serves no purpose for the tpm core or any existing driver. I already fixed this once in commit ba0ef85479c46a 'tpm: Fix initialization of the cdev' - that solves the racing of get_device, and grabs the correct device kref, but I forgot to delete the broken residual get_device. Sigh. It is unfortunate that bogus code sent you down this rabbit hole. My bad :( I'll send you something else that might work for vtpm... Jason
