Dear Scott Wood,
In message 4A551D5A.5060300@freescale.com you wrote:
But then applying this patch would break some boards that are working now. Shirking off responsibility and have the board maintainers fix it again is IMHO not the right thing to do.
What would break? If things would no longer fit where they currently fit, that could happen on any change that increases code size --
Right, which is why I continue to resist to certain changes that increase code size.
possibly even just by changing compilers (this exact thing happened to the NAND bootstrap on some boards with very recent GCC). That's life,
But then changing the tool chain is something that is under control of the board maintainer, so he is automatically also called on fixing any problems resulting from this.
This is different from changes that other people are doing.
A quick grep shows several instances of %ll/%L in other places that may not be obvious to the board maintainer (cmd_mmc, ubi, disk/part_efi, cpu/mpc8xxx/ddr, lmb, disk/part, cmd_ide, reiserfs). Boards that use those without 64-bit printf are broken *right now*.
Agreed. It would be good to get at least error messages for such cases. It is not good that they go through unnoticed.
Best regards,
Wolfgang Denk