<rperier>
mhhhhh.... what kind of other stuffs can do a bootloader except touch the clk tree ? pin muxing and that's it , right ?
<Graphicscore>
Are there any good RK3399 boards / boxes that are CE certified and do have an Android SDK available for download ?
<phh>
rperier: uh many things. you're still on ethernet?
<phh>
you've checked all power supplies?
<phh>
would be easier to just follow schematics ;)
<rperier>
phh: I have checked the clk tree, most of the pll , hclk_peri, clock cpu were wrong , so I have fixed it
<rperier>
but there is still something wrong
<rperier>
phh: basically, it's working when booting from NAND (with the rk bootloader) and not when boot from sdcard via uboot, same kernel, same config. The LEDs of the phy are blicking very fast, autonego is failing
<phh>
rperier: you've already went through rk bootloader sources?
<phh>
rperier: do you know if rk bootloader is built with CMD_NET and/or RK_GMAC?
<rperier>
I am talking about the RK proprietary bootloader binary blob
<rperier>
there is sources for that ?
<phh>
ah, sorry, i was on the rockchip u-boot
<phh>
I thought rk3188's bootloader was based on it, but it's not...
BenG83 has quit [Quit: Leaving]
<rperier>
no problem, yeah no it's not
<phh>
anyway, you've checked all regulators on your pmic? you've checked pin states? (not just muxing, possibly the power supply is on a gpio)
<rperier>
from the kernel point of view, everything looks good regarding regulators (I have already checked), however I have not the equipment to measure this on the board :(
<phh>
well, I'd suggest to check pin states then, I'm not too sure how
<phh>
I guess you have no schematics? perhaps some other rk3188 board with ethernet has a public schematic?
<rperier>
the only difference I see in these regulators (between the working and the failing kernel) is the regulator for cpu0
<rperier>
other regulators are the same
<rperier>
875mV vs 925 mV (for cpu0)
<phh>
well, looking at the schematics, I agree with you, except for clocks and pinmux, there can't be much things. could be ethernet-block power control, or grf
<phh>
(I know that on rk3288 pwm requires the bootloade to tinker with grf...)
<phh>
rperier: perhaps you can try to dump the grf zone
<rperier>
already tried via devmem
<rperier>
:D
<phh>
yup, without any probe that's getting really tough
<rperier>
I agree :(
Graphicscore has quit [Ping timeout: 260 seconds]
phinxy has joined #linux-rockchip
wadim_ has quit [Quit: Leaving.]
lkcl has quit [Ping timeout: 255 seconds]
cyteen has quit [Ping timeout: 255 seconds]
lkcl has joined #linux-rockchip
BenG83 has joined #linux-rockchip
BW^- has joined #linux-rockchip
<BW^->
Guys, do the RK3399 have the PXN and PAN extensions?
<BW^->
Also what about "the general things like register 7 being usable in place of clflush for rowhammer attacks"?
BluesKaj has joined #linux-rockchip
BluesKaj has left #linux-rockchip ["Konversation terminated!"]
ganbold has quit [Remote host closed the connection]
afaerber has quit [Quit: Leaving]
fullstop has left #linux-rockchip [#linux-rockchip]