Hi Charles, On Sat, May 15, 2010 at 4:43 AM, Charles D. Krebs <ckrebs@xxxxxxxxxxxxxxxxxxxx> wrote: >>> Just saw this - am I right in thinking someone has written SDIO >>> support for TMIO ? Im afraid I dont appear to have the preceeding >>> emails. >> >> This is the first email regarding TMIO SDIO that I'm aware of - from >> Charles, he's sent it tothe sh ML and to me and I've added you and the mmc >> list to CC. > > To clarify - I didn't write that code. It was developed by some of the > Kernel guys over at Renesas Japan specifically for my request to support > SDIO under their BSP for 2.6.33. The code in tmio and sh_mobile_sdhi > replaces a previously closed source module for SD support on the hardware > due to licensing restrictions of the SD consortium. Apparently those issues > were resolved and now the complete SD stack is integrated into the Kernel > directly. However, the transition didn't include support for SDIO, so the > patch to tmio was provided in the interim. > > Magnus - Do you know the author for the SDIO patch? I believe that the patch was written by one of the BSP guys at Renesas. I guess I'm to blame for not posting anything better upstream, but I decided to hold until I could test this feature properly myself. There are a bunch of SDHI/tmio_mmc related patches posted for 2.6.35 already, and I guess SDIO would be targeted for 2.6.36. The SDIO patch I've seen is not exactly upstream quality, but I've been working a bit on trying to come up with a more long term solution. At this point it's a bit stalled since my Wifi SDIO cards that i've ordered are still on hold for some unknown reason. I have however tested some other SDIO cards with the SDHI/tmio_mmc drivers and they seem ok. At least the one that actually has a linux driver. The quality of the driver for that card is far from perfect, so I'd like to test with hardware that has upstream SDIO drivers (broadcomm and perhaps also atheros chipsets) before claiming that SDIO works. FYI, you don't really need any patch for SDIO support. You can do SDIO on a non-SDIO host by the automatic polling mode that the kernel provides for you. So the SDHI/tmio_mmc code should work out of the box as-is. You may want to enable the IRQ pin feature, but for that you need to patch the SDHI/tmio_MMC driver(s). I've attached my prototype SDIO patch. It's a scaled down version of the BSP patch, but still in a hackish state. / magnus
Attachment:
linux-2.6.35-pre-sdhi-sdio-rewrite-20100427b.patch
Description: Binary data
