Cyril Chemparathy <[email protected]> writes: > This series takes a new approach to the ioremap() conundrum posed by the > introduction of the tnetv107x architecture. > > Unlike earlier approaches, no new init time interfaces have been added in this > implementation. Instead, this series converts the contents of > davinci_soc_info > to use physical instead of virtual addresses. These physical addresses are > then ioremap()ed as necessary in each individual module (psc, timer, etc.). > > As a pleasant side-effect, this series also gets rid of a bunch of > IO_ADDRESS() > uses. :-)
As another pleasant side effect, you made my day. :) This is something that's been on my TODO for some time and I've been ignoring it. Thanks a ton for tackling it. So as a next step, lets break out the stuff I mentioned as ready to merge into a separate series, and get that applied and in davinci-next for 2.6.35. Then, fixup some of the very minor issues I pointed out and get some more testing on your ioremap series for the existing platforms as well... > These changes were tested only on the tnetv107x evm platform. I tried a boot test on dm355 and da8xx and neither booted, but didn't do any more digging, because I want to take off for the weekend on a happy note. :) Next step is to track the boot problems on dm* and da* and then get this merged. If we take care of it next week, I'd like to queue this up for 2.6.35. Thanks again Cyril, and nice work! Kevin _______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
