Hi Joe,
On 26 June 2018 at 16:31, Joe Hershberger joe.hershberger@ni.com wrote:
On Tue, Jun 26, 2018 at 6:18 PM, Simon Glass sjg@chromium.org wrote:
Hi Joe,
On 26 June 2018 at 14:19, Joe Hershberger joe.hershberger@ni.com wrote:
With systemd stable interface names, eth0 will almost never exist. Instead of using that name in the sandbox.dts, use index 2 (the first interface after "lo"). Enumerate the interfaces on the system to choose a valid interace name.
Signed-off-by: Joe Hershberger joe.hershberger@ni.com
arch/sandbox/cpu/eth-raw-os.c | 28 ++++++++++++++++++++++++++++ arch/sandbox/dts/sandbox.dts | 2 +- arch/sandbox/include/asm/eth-raw-os.h | 4 +++- drivers/net/sandbox-raw.c | 8 +++++++- 4 files changed, 39 insertions(+), 3 deletions(-)
This seems brittle too. A name seems better to me. I might be missing something though. Can we ask the OS for a name, etc?
This asks for the equivalent of "eth0", but asks the OS to tell us the actual name. I don't know that we can possibly choose the correct network interface to use across any machine configuration. Choosing the first one, whatever it's name is, seems like a pretty reliable way to get a good interface in most cases.
OK well let's see how it works out.
Reviewed-by: Simon Glass sjg@chromium.org
Regards, Simon