Messages in this thread | | | Date | Mon, 27 Jan 2014 19:34:40 +0400 | Subject | Re: [PATCH 3/3] net: via-rhine: add OF bus binding | From | Alexey Charkov <> |
| |
2014/1/27 Ben Hutchings <ben@decadent.org.uk>: > On Mon, 2014-01-27 at 15:51 +0400, Alexey Charkov wrote: >> This should make the driver usable with VIA/WonderMedia ARM-based >> Systems-on-Chip integrated Rhine III adapters. Note that these >> are always in MMIO mode, and don't have any known EEPROM. > [...] >> --- a/drivers/net/ethernet/via/Kconfig >> +++ b/drivers/net/ethernet/via/Kconfig >> @@ -19,7 +19,7 @@ if NET_VENDOR_VIA >> >> config VIA_RHINE >> tristate "VIA Rhine support" >> - depends on PCI >> + depends on (PCI || USE_OF) >> select CRC32 >> select MII >> ---help--- > > This seems like the right thing to do, but it means you need to add > #ifdef CONFIG_PCI and #ifdef CONFIG_USE_OF around the driver structures > and related functions.
Frankly, I would like to avoid that if possible (as pointed out in the cover email), as I believe we would get a cleaner driver without #ifdef. This is also the way it was done in via-velocity, and it works just fine.
> You should compile-test in configurations that have just one of those > dependencies enabled.
This has been compile-tested and runtime-tested in OF-only configuration on WM8950, and Roger also tested it in PCI-only configuration, so it seems to work fine.
> [...] >> --- a/drivers/net/ethernet/via/via-rhine.c >> +++ b/drivers/net/ethernet/via/via-rhine.c > [...] >> @@ -847,7 +856,8 @@ static void rhine_hw_init(struct net_device *dev, long pioaddr) >> msleep(5); >> >> /* Reload EEPROM controlled bytes cleared by soft reset */ >> - rhine_reload_eeprom(pioaddr, dev); >> + if (!strncmp(dev->dev.parent->bus->name, "pci", 3)) >> + rhine_reload_eeprom(pioaddr, dev); > [...] > > Ew. I think you should use dev_is_pci(), although you might also need > to guard that with #ifdef CONFIG_PCI.
Oh, cool. Didn't realize it existed :) Will adjust, thanks.
I believe the #ifdef is not strictly required, though, as we include the PCI header anyway (and the macro expands to just a simple test). Any specific concerns why we should do that, apart from the +3.8% module size increase?
Thanks, Alexey
| |