On Thu, 18 Nov 2010, Jun Nie wrote: > >From 642b2e88fc4b6344290c88ce506fcda1bcc860a3 Mon Sep 17 00:00:00 2001 > From: Jun Nie <njun@xxxxxxxxxxx> > Date: Fri, 12 Nov 2010 19:43:49 +0800 > Subject: [PATCH] mmc: check SDIO card wake up during suspending > > It is possible that SDIO card send card interrupt to wake up > system just after SDIO func driver and SDIO card firmware commit > suspend state, while system is still in suspending process and > mmc interrupt is not release yet. > > Interrupt storm may happen if func driver does not notify firmware > to clear the wake up source. If func driver issue SDIO bus transaction > to have firmware clear the source and driver continue to handle the > wake up event, system may enter suspend before all transaction done. > In this case, SDIO card will not enter lower power mode never. But > system can not be wake up for only card insert/remove/interrupt irq are > listed in wake up source in SD spec. Data/command done irq can > not wake up system. > > This patch check SDIO card interrupt and stop system suspend if there > are card interrupt after func driver suspend. > > Signed-off-by: Jun Nie <njun@xxxxxxxxxxx> There is a structural problem with your patch. Why do you have to dig into card specific data from sdhci_suspend_host()? The fact that you need to include mmc/card.h into sdhci.c is a sign of layering violation, and all host drivers would have to be similarly modified, leading to unmaintainable code. While the race might be real, there has to be another way to fix it. Nicolas -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
