On 10/12/2018 5:18 AM, Christoph Hellwig wrote:
On Thu, Oct 11, 2018 at 04:50:00AM +0000, Sinan Kaya wrote:
The difference between pci_reset_function_locked() and
__pci_reset_function_locked() is the saving and restoring of the registers.
Unify these API by adding saverestore argument that caller passes.
Adding random boolean arguments doesn't really help the API. Either
make this another flag for reset_type or if there is a clear benefit
add an additional flags parameter with well described flags.
Good idea. This is the kind of feedback I was looking for. Here is what I
collected so far.
Alex doesn't like the API changes as it should be obvious that there are
differences between pci_reset_function_locked() and
__pci_reset_function_locked(). Sinan thinks that the difference is not that
obvious as some drivers implemented their own save/restore mechanism using
__pci_reset_function_locked(). Sinan would rather reduce the reset API flavors
and give a unified interface where user specifies what they are interested in.
Christoph asked for additional flags to be rolled into the reset_type argument.
Sinan thinks that this is a good idea.
Anybody else?
