On Thu, 2004-08-19 at 09:40, Dave Ellis wrote:
Jon Loeliger wrote:
I'd like to get your opinion on a proposed change to a few files that handle some aspects of the various enetaddr fields as found in asm-ppc/u-boot.h, common/cmd_bdinfo.c and lib_ppc/board.c.
In particular, I'd like to propose a shift from having these fields be present when certain boards are #defined to having these fields be present when CONFIG_ETH1ADDR symbols are defined.
I like the idea of using a common symbol, but I would prefer a new one, something like CONFIG_HAS_ETH1, so I can have bi_enet1addr in the kernel interface without putting a default value for it in the environment.
BTW, I am willing to make the change so that the code uses the symbols:
CONFIG_HAS_ETH CONFIG_HAS_ETH1 CONFIG_HAS_ETH2 and CONFIG_HAS_ETH3
as suggested. I like it.
However, now I need answers to the following question: Do you want me to retrofit code into all the Config files to #define CONFIG_HAS_ETHx where it currently also has CONFIG_ETHxADDR defined, or where the code has a board name even though a CONIG_ETHxADDR is not defined too?
Happy to do this, just realize that to be backwards compatible with existing config files, I'll have to change many config files. I can not test them all. I can test the 4 I have in front of me.
FYI, I am also willing to remove the #ifdef conditionality from the bd_t structure around these ETH addr fields as well, but with the caveat that it changes other people's bd_t structures and potentially messes up their Linux interfaces. Again, I can't test all that either...
Thanks, jdl