On Wed, Mar 01, 2023 at 09:09:57AM -0700, Jeffrey Hugo wrote: > On 2/28/2023 4:52 AM, Stanislaw Gruszka wrote: > > On Mon, Feb 06, 2023 at 08:41:40AM -0700, Jeffrey Hugo wrote: > > > + mhi_cntl = kzalloc(sizeof(*mhi_cntl), GFP_KERNEL); > > [snip] > > > + mhi_cntl->irq = kmalloc(sizeof(*mhi_cntl->irq), GFP_KERNEL); > > > > I recommend usage of devm_kzalloc(), devm_kmalloc() for those > > to simplify error and exit paths. > > When this was written, I didn't want to pass the struct device to the > mhi_controller just for the purpose of using devm_*. Today, I'm thinking > that is not the end of the world, and devm has advantages. Will change. If already available &pci_dev->dev can not be used in devm_ due to different life times of pci_dev->dev and mhi_cntl, I don't think change would be justifiable and kmalloc/kzalloc should stay. Regards Stanislaw
