On Tue, Feb 15, 2022 at 12:08:26PM +0100, Jiri Slaby wrote: > On 15. 02. 22, 11:31, Andy Shevchenko wrote: > > On Tue, Feb 15, 2022 at 12:29:49PM +0200, Andy Shevchenko wrote: > > > On Tue, Feb 15, 2022 at 11:11:41AM +0100, Jiri Slaby wrote: > > > > On 15. 02. 22, 11:09, Andy Shevchenko wrote: > > > > > The pci_get_slot() increases its reference count, the caller > > > > > must decrement the reference count by calling pci_dev_put(). > > > > > > > And what about the -EINVAL case? > > > > > > What about it? The ->probe() calls ->exit() in its error path. > > > Or did I miss something? > > > > Or you mean that we call pci_dev_put() on NULL pointer? > > This is completely valid case and handled in the callee. > > No, I mixed up patches :P. This was meant as a comment to > "serial: 8250_lpss: Balance reference count for PCI DMA device" > > byt_serial_exit() isn't called when byt_serial_setup() fails with EINVAL if > I'm looking correctly. I see now, thanks for catching this! I will send a v2 soon. -- With Best Regards, Andy Shevchenko
