Merge branch 'devlink_register-last'
Leon Romanovsky says: ==================== Move devlink_register to be last devlink command This is second version of patch series https://lore.kernel.org/netdev/cover.1628599239.git.leonro@nvidia.com/ The main change is addition of delayed notification logic that will allowed us to delete devlink_params_publish API (future series will remove it completely) and conversion of all drivers to have devlink_register being last commend. The series itself is pretty straightforward, except liquidio driver which performs initializations in various workqueues without proper locks. That driver doesn't hole device_lock and it is clearly broken for any parallel driver core flows (modprobe + devlink + PCI reset will 100% crash it). In order to annotate devlink_register() will lockdep of holding device_lock, I added workaround in this driver. Thanks ---------------------- From previous cover letter: Hi Dave and Jakub, This series prepares code to remove devlink_reload_enable/_disable API and in order to do, we move all devlink_register() calls to be right before devlink_reload_enable(). The best place for such a call should be right before exiting from the probe(). This is done because devlink_register() opens devlink netlink to the users and gives them a venue to issue commands before initialization is finished. 1. Some drivers were aware of such "functionality" and tried to protect themselves with extra locks, state machines and devlink_reload_enable(). Let's assume that it worked for them, but I'm personally skeptical about it. 2. Some drivers copied that pattern, but without locks and state machines. That protected them from reload flows, but not from any _set_ routines. 3. And all other drivers simply didn't understand the implications of early devlink_register() and can be seen as "broken". ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
Please register or sign in to comment