On Mon, 2009-08-24 at 18:10 -0500, Scott Wood wrote:
Andy Fleming wrote:
On Mon, Aug 24, 2009 at 2:32 PM, Kim Phillips kim.phillips@freescale.comwrote:
if you don't have firmware installed for the PHY to come to life, this wait can be painful - let's give the option to avoid it if we want.
Signed-off-by: Kim Phillips kim.phillips@freescale.com
Acked-by: Andy Fleming afleming@freescale.com
I've always wanted this change, but never looked into how to implement it.
However, it should be noted that this change is somewhat lessened in impact by the proposal to change the timeout to be only 2 seconds, instead of ...whatever it is now.
I thought it was established that that wasn't long enough for some configurations?
It looks like 3.5 seconds was agreed upon to be safe:
http://lists.denx.de/pipermail/u-boot/2009-February/047233.html
The current timeout is set to 5 seconds.
We could set a default timeout value of 3.5 seconds, but allow boards to override this by defining something like CONFIG_SYS_PHY_AUTONEG_TIMEOUT to tweak the timeout if they wanted.
But even with 2 seconds, it's annoying to wait for N ports to be tried when you know none of them are hooked up. Or that you forgot to properly set the ipaddr, anyway.
Setting ethprime should work for the common case where you know which ports are connected.
ethrotate is annoying in general...
Agreed. For a project a while back we replaced ethact, ethprime and ethrotate with an "ethorder" variable. If defined, the interfaces in ethorder would be tried in order. So to only use 1 interface you could set ethorder to "eTSEC1", to cycle between 2 non-contiguous ports it could be set to "eTSEC1, eTSEC3", to only use PCI ports it could be set to "e1000#0, e1000#1", etc. ethorder was more flexible and a bit more straightforward in my opinion. Anyone have any interest in the patches?
Best, Peter