Thierry Reding wrote at Wednesday, November 16, 2011 12:14 AM:
- Thierry Reding wrote:
- Simon Glass wrote:
[...]
If you are not implementing MMC, just remove it and all the MMC code below.
I don't understand. What makes you think I don't want MMC? Both the Plutux and the Medcom boot from MMC. If you are referring to the CD being hard-coded, then that is just because I copied from Harmony. You are right though that I should probably fill the implementation. I'm not really sure about the best way to do this, though. The struct mmc passed into board_mmc_getcd() doesn't contain enough information to decide which GPIO to check. I could assume it is always SD4 since that is the only controller that is initialized, but would that be acceptable?
I was going to implement proper card-detection (it's quite trivial) but then I saw that board_mmc_getcd() is never called in the Tegra2 MMC driver so I guess that should be implemented as well. Did anybody work on that already or would you like me to prepare a patch?
I noticed that too, but adding CD support to the Tegra driver itself didn't seem right; surely it should be added to the MMC core?
Anyway, if you're looking at hacking on Tegra MMC, please do be aware of a few patches I posted re: Tegra boards and MMC changes:
tegra2: Move board_mmc_init into board files http://patchwork.ozlabs.org/patch/118143/
tegra2: Modify MMC driver to handle power and cd GPIOs http://patchwork.ozlabs.org/patch/118142/
tegra2: Add support for Ventana http://patchwork.ozlabs.org/patch/118144/
I hope these will be merged soon, but IIRC they haven't received any feedback either way from Albert.