On Thu, 2020-09-17 at 21:45 +0300, Shay Agroskin wrote: > David Miller <da...@davemloft.net> writes: > > > From: Shay Agroskin <shay...@amazon.com> > > Date: Sun, 13 Sep 2020 11:16:34 +0300 > > > > > ENA logs are adjusted to display the full ENA representation to > > > distinct each ENA device in case of multiple interfaces. > > > Using dev_err/warn/info function family for logging provides > > > uniform > > > printing with clear distinction of the driver and device. > > > > > > This patch changes all printing in ena_com files to use dev_* > > > logging > > > messages. It also adds some log messages to make driver > > > debugging > > > easier. > > > > > > Signed-off-by: Amit Bernstein <amitb...@amazon.com> > > > Signed-off-by: Shay Agroskin <shay...@amazon.com> > > > > This device prefix is so much less useful than printing the > > actual > > networking adapter that the ena_com operations are for. > > > > So if you are going to do this, go all the way and pass the > > ena_adapter > > or the netdev down into these ena_com routines so that you can > > use > > the netdev_*() message helpers. > > > > Thank you. > > Hi David, I researched the possibility to use netdev_* functions > in this patch. Currently our driver initializes the net_device > only after calling some functions in ena_com files. > Although netdev_* log family functions can be used before > allocating a net_device struct, the print it produces in such a > case is less informative than the dev_* log print (which at least > specifies what pcie device made the print). > > I would rather change the allocation order for the net_device > struct in our driver, so that when calling ena_com device it would > always be allocated (and this way all ena_com prints could be > transformed into netdev/netif_* log family). > This change seems doable, but requires us to do some internal > testing before sending it. I could remove this whole patch, but > then our driver would be left with pr_err() functions in it which > is even less informative than dev_err(). >
allocated but unregistered netdevices also do not help much as the name of the netdev is not assigned yet. why don't use dev_info(pci_dev) macors for low level functions when netdev is not available or not allocated yet. > Can we go through with this patch, and send a future patch which > changes all dev_* functions into netif/netdev_* along with the > change in the allocation order of the net_device struct ? I know > it sounds like a procrastination attempt, but I would really > prefer not to drop the patch and leave the driver with pr_* log > prints. > > Thanks, > Shay
