Mike Frysinger wrote:
On Tuesday, March 01, 2011 03:00:47 Michal Simek wrote:
If both functions should return 0 then any code should check it and all others drivers should be fixed.
i agree, but that doesnt mean new code should knowingly be left broken
I agree that make no sense do not fix it right now.
the init func is a special beast -- it returns the # of devices registered, not "1 is success".
Where is it written?
atm, only the mailing list. i have a local patch that i havent gotten around to pushing up yet for the README.drivers.eth.
ep93xx_eth.c returns also 1. Anyway if is number of registered devices, "1" should means one registered device. If zero means one registered device then please point me to that documentation.
the change hasnt been ported to all drivers yet. but new drivers should be doing it as i described.
How does it look like phy lib u-boot support?
also, you should change the "hang()" to "return 0" in the init func.
Are you sure return 0 which should mean success. Anything different from 0 seems to me relevant.
you too. int bfin_EMAC_initialize(bd_t *bis) { struct eth_device *dev; dev = malloc(sizeof(*dev)); if (dev == NULL) hang();
I maintain emaclite driver and none tell me this that's why the process is so slow. I believe if you release that documentation, which you are talking about, then others will clean/test their drivers.
Michal