On 2016年08月26日 16:51, Shawn Lin wrote:
corecfg_clockmultiplier indicates clock multiplier value of programmable clock generator which should be the same value of SDHCI_CAPABILITIES_1. The default value of the register, corecfg_clockmultiplier, is 0x10. But actually it is a mistake by designer as our intention was to set it to be zero which means we don't support programmable clock generator. So we have to make it to be zero on bootloader which seems work fine until now. But now we find an issue that when deploying genpd support for it, the remove callback will trigger the genpd to poweroff the power domain for sdhci-of-arasan which manage the controller, phy and corecfg_* stuff. So when we do bind/unbind the driver, we have already reinit the controller and phy, but without doing that for corecfg_*. Regarding to only the corecfg_clockmultipler is wrong, let's fix it by explicitly marking it to be zero when probing. With this change, we could do bind/unbind successfully. Reported-by: Ziyuan Xu <xzy.xu@xxxxxxxxxxxxxx> Cc: Douglas Anderson <dianders@xxxxxxxxxxxx> Signed-off-by: Shawn Lin <shawn.lin@xxxxxxxxxxxxxx>
Okay, it looks good to me. Reviewed-by: Ziyuan Xu <xzy.xu@xxxxxxxxxxxxxx> Tested-by: Ziyuan Xu <xzy.xu@xxxxxxxxxxxxxx>
--- Changes in v2: - fix some typos and build - move the configuration of corecfg_clockmultiplier after enabling aclk_emmc to avoid the off state of aclk_emmc_grf. I add a new function to make it more common for other coming users who have the same requirment for setting diff clkmul.
-- 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
