On Thu, Feb 24, 2011 at 11:51:28PM +0800, Shawn Guo wrote:

On Thu, Feb 24, 2011 at 04:18:05PM +0800, Shawn Guo wrote:

quoted

Hi Wolfram,

On Wed, Feb 23, 2011 at 02:51:53PM +0100, Wolfram Sang wrote:

quoted

Take #3, changes:

* also intercept calls to SDHCI_SIGNAL_ENABLE (needed on mx25)
* remove unconditional BROKEN_CARD_DETECTION (leftover)
* improved kernel-doc about unused GPIO
* added tags from Eric

Tested now by me and Marc on mx35, Eric on mx25/35/51. Arnaud, did you have a
chance to retest on mx51? What about the FSL guys? :)

I'm testing it on mx25 3ds and mx51 babbage. Both card-detect and
write-protect are working on mx25 3ds, but write-protect has some
problem on babbage, and I need some time to figure it out.

I just figured out why wp_gpio does not work on mx51.

+       if (cpu_is_mx25() || cpu_is_mx35()) {
+               /* Fix errata ENGcm07207 present on i.MX25 and i.MX35 */
                host->quirks |= SDHCI_QUIRK_NO_MULTIBLOCK;
+               /* write_protect can't be routed to controller, use gpio */
+               sdhci_esdhc_ops.get_ro = esdhc_pltfm_get_ro;
+       }

Would it make sense to do the same for mx51?  On babbage board,
SD1_WP is routed to controller, but SD2_WP is not, so we have to use
gpio for SD2 write_protect.

We should probably have esdhc_pltfm_get_ro whenever platform provides
a wp_gpio, and fall on controller SD_WP only when platform does not
provide the wp_gpio.

-- 
Regards,
Shawn