On Tue, 2021-01-12 at 20:42 +0100, Mark Kettenis wrote:
> >
> > Date: Tue, 12 Jan 2021 20:40:05 +0100 (CET)
> > From: Mark Kettenis <[email protected]>
> >
> > >
> > > From: Kurt Miller <[email protected]>
> > > Date: Tue, 12 Jan 2021 12:29:34 -0500
> > >
> > > I purchased one of the rev c boards with 8GB ram with the
> > > hopes that it could build the full ports tree. Good news
> > > it installs and boots fairly easily. Bad news is that I
> > > missed that the 8GB ram is split between the CPU and the
> > > NPU (4GB for CPU/GPU, 4GB for NPU) so only 4GB is usable
> > > for building ports.
> > >
> > > Install notes:
> > > 1) Setup serial console and use 1500000 baud
> > > 2) Boot into default os
> > > 3) Erase u-boot from eMMC:
> > > su
> > > dd if=/dev/zero of=/dev/block/mmcblk2 bs=1m count=16
> > > sync
> > > reboot
> > > 4) Power off board
> > > 5) Prepare uSD with miniroot68.fs and u-boot-rockchip.bin at sector 64
> > > 6) Install uSD
> > > 7) Switch serial console to use 115200 baud
> > > 8) Power on board
> > > 9) Install OpenBSD
> > > root on sd1 (uSD) 1g
> > > rest of the partions on sd0 (nvme)
> > >
> > > I opted to install root on the uSD so that I can test u-boot
> > > and easily revert any versions that don't work by fixing the
> > > uSD on another computer. If u-boot is on the eMMC board it
> > > prefers that over what's on the uSD.
> > Note that it is perfectly fine to have the root filesystem on eMMC and
> > U-Boot on uSD. That is my preferred configuration as it allows for
> > easy testing of U-Boot updates since you can simply swap out uSD cards.
> >
> > >
> > > rkiic1 at mainbus0
> > > iic1 at rkiic1
> > > "haoyu,hym8563" at iic1 addr 0x51 not configured
> > So that's your RTC. Judging from the data sheet, this is a PCF8563
> > clone. Can you try the attached diff?
> >
> > >
> > > rkiic3 at mainbus0
> > > iic3 at rkiic3
> > > "rockchip,rk809" at iic3 addr 0x20 not configured
> > So this board uses a different PMIC than the "classic" RK3399 boards.
> > Judging from the datasheet there are some significant differences, so
> > this isn't a matter of just adding a compatible string to the
> > rkpmic(4) driver. But the hardware is probably similar enough to
> > extend that driver instead of adding a new one.
> >
> > This would probably get rid of:
> >
> > >
> > > cpu0: regulator not implemented
> > and provide proper DVFS support on this board which will make the CPU
> > run at higher clock speeds.
> And here is the diff I promised:
>
Thanks. It worked as expected. Now I see:
rkiic1 at mainbus0
iic1 at rkiic1
pcxrtc0 at iic1 addr 0x51: battery ok
and it got rid of these lines as well:
WARNING: bad clock chip time
WARNING: CHECK AND RESET THE DATE!
okay kurt@
> Index: dev/i2c/pcf8563.c
> ===================================================================
> RCS file: /cvs/src/sys/dev/i2c/pcf8563.c,v
> retrieving revision 1.2
> diff -u -p -r1.2 pcf8563.c
> --- dev/i2c/pcf8563.c 6 Oct 2019 15:45:37 -0000 1.2
> +++ dev/i2c/pcf8563.c 12 Jan 2021 19:41:54 -0000
> @@ -44,8 +44,6 @@
> * PCF8563 Real-Time Clock
> */
>
> -#define PCF8563_ADDR 0x51 /* Fixed I2C Slave Address */
> -
> #define PCF8563_CONTROL1 0x00
> #define PCF8563_CONTROL2 0x01
> #define PCF8563_SECONDS 0x02
> @@ -106,8 +104,8 @@ pcxrtc_match(struct device *parent, void
> {
> struct i2c_attach_args *ia = arg;
>
> - if (strcmp(ia->ia_name, "nxp,pcf8563") == 0 &&
> - ia->ia_addr == PCF8563_ADDR)
> + if (strcmp(ia->ia_name, "nxp,pcf8563") == 0 ||
> + strcmp(ia->ia_name, "haoyu,hym8563") == 0)
> return (1);
>
> return (0);
>
