[RFC PATCH] ARM: dts: meson8b-odroidc1: add microSD support

Martin Blumenstingl martin.blumenstingl at googlemail.com
Sun Mar 18 16:00:58 PDT 2018

Hi Linus,

On Sat, Mar 17, 2018 at 6:11 PM, Linus Lüssing <linus.luessing at c0d3.blue> wrote:
> Hi Martin,
> Sorry for the late reply.
> On Mon, Mar 12, 2018 at 10:00:17PM +0100, Martin Blumenstingl wrote:
>> >> Not sure whether this error message could potentially point to some
>> >> error in the changes in here.
>> > these errors may be a problem with the meson-mx-sdio (MMC) driver
>> > can you please show the output of /sys/kernel/debug/mmc0/ios ?
>> some users on IRC have successfully tested this patch (together with
>> the two patches from my series which fix the GPIO handling)
>> can you please re-send this patch so Kevin can queue it for the next
>> development cycle?
> Glad to hear! I also just retried it with a 4.16-rc5 kernel and
> your latest fixes and microSD still works for me, too :-).
that's great :-)

> Regarding the error message, I can't reproduce it anymore. With
> the 4.16-rc5 kernel I did not see it. Nevertheless, here's the
> output of mmc/ios:
> -----
> $ cat /sys/kernel/debug/mmc0/ios
> clock:          50000000 Hz
> actual clock:   39843750 Hz
> vdd:            21 (3.3 ~ 3.4 V)
> bus mode:       2 (push-pull)
> chip select:    0 (don't care)
> power mode:     2 (on)
> bus width:      2 (4 bits)
> timing spec:    2 (sd high-speed)
> signal voltage: 0 (3.30 V)
> driver type:    0 (driver type B)
> -----
thank you!
not sure why you got an error before, let's hope it won't happen again

> Out of curiousity, is the "actual clock" supposed to be so
> much lower than "clock"?
short answer: yes
long answer: ideally "clock" and "actual clock" should be identical.
however, due to the clock setup for the "SDIO" controller we can't get
much closer because there's no "fractional" divider and we can't
change the rate of the top-most clock (see
/sys/kernel/debug/clk/clk_summary). there is a second MMC controller
in the Meson8b SoCs (which also supports 8-bit bus width, typically
used for eMMC) which can generate better frequencies. the other
controller however is not supported (by the mainline kernel) yet

> Anyway, will resend the patch without the "RFC" now.
thank you, I'll send a Tested-by once I have time to run it on my Odroid-C1


More information about the linux-amlogic mailing list