在 2016/10/7 12:12, Jaehoon Chung 写道:
Hi Shawn,
On 09/30/2016 05:48 PM, Shawn Lin wrote:
This patch add dw_mci_runtime_suspend/resume interfaces
and expose it to dw_mci variant driver to support runtime
PM.
Signed-off-by: Shawn Lin <shawn.lin@xxxxxxxxxxxxxx>
---
drivers/mmc/host/dw_mmc.c | 60 +++++++++++++++++++++++++++++++++++++++++++++--
drivers/mmc/host/dw_mmc.h | 4 +++-
2 files changed, 61 insertions(+), 3 deletions(-)
diff --git a/drivers/mmc/host/dw_mmc.c b/drivers/mmc/host/dw_mmc.c
index 4fcbc40..54b860e 100644
--- a/drivers/mmc/host/dw_mmc.c
+++ b/drivers/mmc/host/dw_mmc.c
@@ -3266,7 +3266,7 @@ EXPORT_SYMBOL(dw_mci_remove);
-#ifdef CONFIG_PM_SLEEP
+#ifdef CONFIG_PM
/*
* TODO: we should probably disable the clock to the card in the suspend path.
*/
@@ -3324,7 +3324,63 @@ int dw_mci_resume(struct dw_mci *host)
return 0;
}
EXPORT_SYMBOL(dw_mci_resume);
-#endif /* CONFIG_PM_SLEEP */
+
+int dw_mci_runtime_suspend(struct dw_mci *host)
+{
+ printk("dw_mci_runtime_suspend\n");
Maybe I think you will remove this..if you send the patch, not RFC.
yep, it just for me to print info for testing the rpm governer.
I will remove this when sending non-RFC one.
+
+ if (host->use_dma && host->dma_ops->exit)
+ host->dma_ops->exit(host);
Can just call dw_mci_suspend()
If we deploy rpm properly, we don't need system PM at all..
To be frank, I was wanna remove dw_mci_suspend/resume pairs,
and only leave the runtime pairs here. As you could see that
runtime PM and system PM does the same thing, so personally
I realize we should help dw_mmc variant drivers migrate to use
runtime PM. That is what I did for dw_mmc-rockchip. If we are
not sure should we enable rpm for i.e., exynos, or k3, we don't
add pm_runtime_* for them for the first step and leave it to
the owners/users to make the decision. The following code should
be enough to help we achieve the first step.
static const struct dev_pm_ops dw_mci_XXXXXX_dev_pm_ops = {
SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
pm_runtime_force_resume)
SET_RUNTIME_PM_OPS(dw_mci_runtime_suspend,
dw_mci_runtime_resume,
NULL)
};
If there are some special ops for system PM like exynos, we could wrap
new dw_mci_exynos_runtime_resume there..
The same for dw_mci_resume.
What's your opinion? Seems Ulf was too busy to make comments
here. But I believe it is in the right direction.
+
+ clk_disable_unprepare(host->ciu_clk);
+
+ return 0;
+}
+EXPORT_SYMBOL(dw_mci_runtime_suspend);
+
+int dw_mci_runtime_resume(struct dw_mci *host)
+{
+ int ret = 0;
+ int i;
Can make one line.
+
+ printk("dw_mci_runtime_resume\n");
+
+ ret = clk_prepare_enable(host->ciu_clk);
+ if (ret)
+ return ret;
+
+ if (host->use_dma && host->dma_ops->init)
+ host->dma_ops->init(host);
+
+ mci_writel(host, FIFOTH, host->fifoth_val);
+ host->prev_blksz = 0;
+
+ mci_writel(host, TMOUT, 0xFFFFFFFF);
+ mci_writel(host, RINTSTS, 0xFFFFFFFF);
+ mci_writel(host, INTMASK, SDMMC_INT_CMD_DONE |
+ SDMMC_INT_DATA_OVER |
+ SDMMC_INT_TXDR | SDMMC_INT_RXDR |
+ DW_MCI_ERROR_FLAGS);
+ mci_writel(host, CTRL, SDMMC_CTRL_INT_ENABLE);
+
+ for (i = 0; i < host->num_slots; i++) {
+ struct dw_mci_slot *slot = host->slot[i];
+
+ if (!slot)
+ continue;
+
+ if (slot->mmc->pm_flags & MMC_PM_KEEP_POWER) {
+ dw_mci_set_ios(slot->mmc, &slot->mmc->ios);
+ dw_mci_setup_bus(slot, true);
+ }
+ }
+
+ dw_mci_enable_cd(host);
Some part of this function can be reused with codes in dw_mci_resume().
Best Regards,
Jaehoon Chung
+
+ return ret;
+}
+EXPORT_SYMBOL(dw_mci_runtime_resume);
+#endif /* CONFIG_PM */
static int __init dw_mci_init(void)
{
diff --git a/drivers/mmc/host/dw_mmc.h b/drivers/mmc/host/dw_mmc.h
index e8cd2de..baa7261 100644
--- a/drivers/mmc/host/dw_mmc.h
+++ b/drivers/mmc/host/dw_mmc.h
@@ -234,9 +234,11 @@
extern int dw_mci_probe(struct dw_mci *host);
extern void dw_mci_remove(struct dw_mci *host);
-#ifdef CONFIG_PM_SLEEP
+#ifdef CONFIG_PM
extern int dw_mci_suspend(struct dw_mci *host);
extern int dw_mci_resume(struct dw_mci *host);
+extern int dw_mci_runtime_suspend(struct dw_mci *host);
+extern int dw_mci_runtime_resume(struct dw_mci *host);
#endif
/**
--
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
--
Best Regards
Shawn Lin
--
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
