Hi Heiko, Shawn
On 2018-11-01 10:44, Heiko Stübner wrote:
Hi Shawn, Hal,
Am Donnerstag, 1. November 2018, 02:20:03 CET schrieb Shawn Lin:
On 2018/11/1 0:07, Hal Emmerich wrote:
> Hello everyone,
>
> When booting a veyron-speedy device with the rk3288 and kernel 4.19 I
> get the following errors regarding the mmc and invalid clk rates:
>
>
> [ 0.000000] NR_IRQS: 16, nr_irqs: 16, preallocated irqs: 16
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] rockchip_mmc_get_phase: invalid clk rate
> [ 0.000000] arch_timer: cp15 timer(s) running at 24.00MHz (phys).
Sorry for late!
The log should be be improved as the mmc_phase clock my be
temporarily orphaned before re-parenting to its parents when
all things settled down. But we are definitely in trouble when
seeing these log if mmc driver starts doing tuning.
Did you card boot normally in HS200/UHS mode? If yes, no things
wrong happened.
I can see these messages as well, but only during initial bootup, when
the
clock-tree gets assembled, like the logs above indicate as well and
mmc devices work fine for me so far.
So it really looks like simply the phase-clocks still being orphans at
this
point and we could just make the log a bit less talkative.
Heiko
About 1 boot out of 10 or 20 boot hangs due to the mmc.
The card has an issue reading a block during boot, so the
mmc block system calls for a __mmc_hw_reset(), which tries to flush the
cache but
the command to clear the cache hangs indefinitely as at this point the
mmc is in a
bad state for an unknown reason. Possibly the unknown reason is why the
block read
failed in the first place.
So theres likely a deeper problem with talking to the mmc.
I reported that bug to the mmc list which you can see here:
https://www.spinics.net/lists/linux-mmc/msg51815.html
So, boot doesn't reliably go smoothly. Not sure if this clock issue is
to blame though.
I have added debug information to the "invalid clk rate" errors to print
their parents found
using getParent() so I don't think they are orphans, as the parent clock
gets printed as expected.
Thanks,
Hal
_______________________________________________
Linux-rockchip mailing list
Linux-rockchip@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/linux-rockchip
