Errors with largers and reads turned out not be cache issues, but
a bug in consistent_sync() instead. After fixing consistent_sync()
DMA transfers on 24xx MMC work fine.
host->power_pin = minfo->power_pin;
host->switch_pin = minfo->switch_pin;
host->wp_pin = minfo->wp_pin;
-
- /* FIXME: Remove once DMA works on 24xx */
- if (cpu_is_omap24xx())
- host->use_dma = 0;
- else
- host->use_dma = 1;
+ host->use_dma = 1;
host->dma_ch = -1;
host->irq = pdev->resource[1].start;