| Message ID | 1418244929-8254-1-git-send-email-dianders@chromium.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Am Mittwoch, 10. Dezember 2014, 12:55:29 schrieb Doug Anderson: > It seems that ever since (536f6b9 mmc: dw_mmc: Reset DMA before > enabling IDMAC) landed upstream that SD cards have been very unhappy > on rk3288-evb. They were a little unhappy before that change, but > after that change they're REALLY unhappy. > > It turns out that the above fix happens to fix a corruption when > reading card information during probe time. Without the fix we didn't > detect that high speed SD cards could actually support high speed. > With the fix we suddenly detect that they're high speed and we try to > use them at 50MHz. That doesn't work so well on EVB with the default > drive strength (maybe because there are two physical SD card slots > hooked up to the same pin?). > > Fix the problem by bumping up the drive strength of the sdmmc lines. > > Signed-off-by: Doug Anderson <dianders@chromium.org> > Fixes: 536f6b91d21b ("mmc: dw_mmc: Reset DMA before enabling IDMAC") applied to my 3.19 dts-fixes branch (after a slight modification of the subject: please use "ARM: dts: rockchip:") Heiko
diff --git a/arch/arm/boot/dts/rk3288-evb.dtsi b/arch/arm/boot/dts/rk3288-evb.dtsi index 3e067dd..6194d67 100644 --- a/arch/arm/boot/dts/rk3288-evb.dtsi +++ b/arch/arm/boot/dts/rk3288-evb.dtsi @@ -155,6 +155,15 @@ }; &pinctrl { + pcfg_pull_none_drv_8ma: pcfg-pull-none-drv-8ma { + drive-strength = <8>; + }; + + pcfg_pull_up_drv_8ma: pcfg-pull-up-drv-8ma { + bias-pull-up; + drive-strength = <8>; + }; + backlight { bl_en: bl-en { rockchip,pins = <7 2 RK_FUNC_GPIO &pcfg_pull_none>; @@ -173,6 +182,27 @@ }; }; + sdmmc { + /* + * Default drive strength isn't enough to achieve even + * high-speed mode on EVB board so bump up to 8ma. + */ + sdmmc_bus4: sdmmc-bus4 { + rockchip,pins = <6 16 RK_FUNC_1 &pcfg_pull_up_drv_8ma>, + <6 17 RK_FUNC_1 &pcfg_pull_up_drv_8ma>, + <6 18 RK_FUNC_1 &pcfg_pull_up_drv_8ma>, + <6 19 RK_FUNC_1 &pcfg_pull_up_drv_8ma>; + }; + + sdmmc_clk: sdmmc-clk { + rockchip,pins = <6 20 RK_FUNC_1 &pcfg_pull_none_drv_8ma>; + }; + + sdmmc_cmd: sdmmc-cmd { + rockchip,pins = <6 21 RK_FUNC_1 &pcfg_pull_up_drv_8ma>; + }; + }; + usb { host_vbus_drv: host-vbus-drv { rockchip,pins = <0 14 RK_FUNC_GPIO &pcfg_pull_none>;
It seems that ever since (536f6b9 mmc: dw_mmc: Reset DMA before enabling IDMAC) landed upstream that SD cards have been very unhappy on rk3288-evb. They were a little unhappy before that change, but after that change they're REALLY unhappy. It turns out that the above fix happens to fix a corruption when reading card information during probe time. Without the fix we didn't detect that high speed SD cards could actually support high speed. With the fix we suddenly detect that they're high speed and we try to use them at 50MHz. That doesn't work so well on EVB with the default drive strength (maybe because there are two physical SD card slots hooked up to the same pin?). Fix the problem by bumping up the drive strength of the sdmmc lines. Signed-off-by: Doug Anderson <dianders@chromium.org> Fixes: 536f6b91d21b ("mmc: dw_mmc: Reset DMA before enabling IDMAC") --- arch/arm/boot/dts/rk3288-evb.dtsi | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)