Posts by ilmich

pcsx-rearmed contains special optimizations for arm architectures (dynamic recompiler and emulated graphics card optimized with NEON instructions). if they are not enabled it's essentially unusable. Lakka enables them, the kodi addon (used by libreelec) does not. Or at least that's what I saw. I'll try to take a look at it.

jaum20 I use estuary and my builds are optimized for this skin. personally on the boxes with ddr2 I tried pellucid for a period but estuary is the best currently.

However, try this pre-release build if you have an extra SD card.

Index of /libreelec-11/test/

there are some kodi optimizations that use smaller textures (lower quality but less stress on the cpu/gpu). Even third party skins should be better (I tried arctic fuse and it's not too bad)

remlei I imagine you also tried changing the HDMI cable. In any case, the problem is that the edid is not read correctly (so basically your TV is not correctly detected). this can happen if the HDMI cable is of poor quality, or the TV returns an EDID not correctly recognized by the driver.

Android in these cases sets the resolution fixed to 720p with audio. Linux instead at 768p without audio.
The best thing would be to understand if this is the problem (I'm pretty sure it is) because I could simulate what android does but you miss out on all the other resolutions.
So, as requested, please send a log (tutorial here https://wiki.libreelec.tv/support/log-files)

hi pocng

Quote from pocng

it's because I don't have any SDIO device information in the dmesg

that's correct, if you don't see this

mmc2: new high speed SDIO card at address 0001

at least it means that the wifi card is not active. I looked at the DTS and it seems to me to be the usual pin (90 --> gpio2 PD2) so the problem is really strange.

Quote from pocng

Thanks for rtl8822bs-aml git projet, I'll test it as soon as I get to detect SDIO

Index of /libreelec-11/test/

in this image, I have compiled the driver I was telling you about. the source is here (I forked) https://github.com/rockchip-with-…rtchip-sci-wifi

Quote from pocng

PS: for the r329q v3.1 I've finally made the merge request for boot support on the emmc and share this patched LibreElec built here

Thanks so much, I'll take a look at it as soon as I can

Quote from pocng

Hi @ilmich,

I quickly checked the GPIOs for Wlan and so far I haven't found the source of the problem. At first glance, the pins are indeed those defined in the dtb...

To help me identify the pinctrl from the gpio kernel debug file, I've made myself a correspondence table, I don't know if it can help other people but i share it here : http://sandbox02.poc.ng/tvbox/rk322x_linux_gpio.txt

Code

WIP :

* From Extract info from android DTS:
WIFI,host_wake_irq = <0x75 0x1c 0x00>; => gpio0/RK_PD4 
WIFI,poweren_gpio = <0x7c 0x1a 0x00>; => gpio2/RK_PD2   
BT,power_gpio = <0x7c 0x1d 0x00>; => gpio2/RK_PD5 
uart_rts_gpios = <0xa9 0x06 0x01>; => gpio3/RK_PA6
BT,wake_gpio = <0xa9 0x1b 0x00>; => gpio3/RK_PD3

* From dmesg
[    5.250216] [BT_RFKILL]: bluetooth_platdata_parse_dt: get property: uart_rts_gpios = 102. => GPIO3    RK_PA6    
[    5.250334] [BT_RFKILL]: bluetooth_platdata_parse_dt: get property: BT,power_gpio = 93. => GPIO2    RK_PD5 => rk322x.dtsi =??=> uart0-cts (rockchip,pins = <2 RK_PD5 1 &pcfg_pull_none>;)
[    5.250466] [BT_RFKILL]: bluetooth_platdata_parse_dt: get property: BT,wake_gpio = 123.  => GPIO3    RK_PD3    
[    5.250604] [BT_RFKILL]: bluetooth_platdata_parse_dt: get property: BT,wake_host_irq = 122.
[    4.789424] export led gpio: 39 => GPIO1    RK_PA7
/* defined in android dtb but fd650 not present */
[    4.965395] fd650_devs_get_of_pdata sda gpio:13 => GPIO0    RK_PB5
[    4.965530] fd650_devs_get_of_pdata scl gpio:116 => GPIO3    RK_PC4    

* from /sys/kernel/degub/gpio 
GPIOs 0-31, platform/pinctrl.17, gpio0:
 gpio-13  (fd650_sda           ) out hi 		GPIO0	RK_PB5 /* fd650 controle not present */


GPIOs 32-63, platform/pinctrl.17, gpio1:
 gpio-35  (spk_ctl             ) out lo 		GPIO1	RK_PA3
 gpio-39  (power-led-normal    ) out lo			GPIO1	RK_PA7	


GPIOs 64-95, platform/pinctrl.17, gpio2:
 gpio-72  (phy_led             ) out lo			GPIO2	RK_PB0	/* orange led */
 gpio-75  (power-hold          ) in  lo			GPIO2	RK_PB3
 gpio-90  (wlan_default_wlan_po) out hi			GPIO2	RK_PD2
 gpio-93  (ap6335_poweron      ) out lo			GPIO2	RK_PD3

GPIOs 96-127, platform/pinctrl.17, gpio3:
 gpio-102 (ap6335_rts          ) in  hi			GPIO3	RK_PA6
 gpio-116 (fd650_scl           ) out hi			GPIO3	RK_PC2 /* fd650 controle not present */
 gpio-117 (phy_link            ) out lo  		GPIO3	RK_PC3
 gpio-118 (otg_drv_gpio        ) out hi			GPIO3	RK_PC4
 gpio-119 (power               ) in  hi			GPIO3	RK_PC5
 gpio-121 (menu                ) in  hi			GPIO3	RK_PC7
 gpio-122 (ap6335_wake_host    ) in  hi			GPIO3	RK_PD2
 gpio-123 (ap6335_wake         ) in  hi			GPIO3	RK_PD3

Display More

To be continue ...

maybe you have already sent it, but could you please send me a dmesg? at this point the problem could be elsewhere.

Quote from pocng

SDIO WIFI : not detected => wifi is sdio 02e7/9086 / smartchip sci S9082H (on armbian forum i can see : its a clone of some wifi rtl chips , its closse source , no linux support)

I found this driver

GitHub - ChalesYu/rtl8822bs-aml at smartchip_sci_s9012p

unoffical driver for linux (Support rtl8189es rtl8723bs rtl8822bs) and SCI(smartchip)WiFi(S9082C S9012P) - GitHub - ChalesYu/rtl8822bs-aml at…

github.com

I don't like it much but I can try.

Quote from chewitt

That might not be a simple exercise though, and his time is limited.

I'm starting to work on libreelec 12, but unfortunately, it's true that I have little time available.

pocng , do you have a link to the Android firmware of your box (alternatively also the first ~100mb of a dump of the internal memory) ? I looked at the logs and you should have a Broadcom BCM4330 wifi chip which is supported by the mainline kernel. If it's not detected, the correct gpio that enables it probably needs to be added, but I need the android dtb.

guerrero2694 as pocng said, try sending a photo of the board

The resizing process appears to have been interrupted because the partition is 7GB but the filesystem has not been completely resized. as they told you, post the resize log because it may have had problems completing. however you can also try the manual resize directly from the multitool

try this one

https://test.libreelec.tv/13.0/Rockchip/rock-4c-plus/

deleonkikko has tested it and everything seems to work except the 2k port. Unfortunately, I don't have time to work on it, but it should still be usable.

Quote from pocng

for max-frequency I based it on a DTS with the same emmc (a samsung iot dev platform) and on another box that uses the same eemc component (an Amlogic Phicomm N1 box) => max-frequency = <0x2faf080>.

thank you, try again

Index of /libreelec-11/test/

diff https://github.com/ilmich/LibreEL…ef46e8da6d0f435

Quote from ilmich

pocng try this new build

Index of /libreelec-11/test/

Fingers crossed

I modified the emmc node like this

+&emmc {
+	clock-frequency = <50000000>; ---> added
+	clock-freq-min-max = <400000 50000000>; ----> added
+	mmc-hs200-1_8v; ----> added
+	/delete-property/ pinctrl-names; 
+	/delete-property/ pinctrl-0;
+	/delete-property/ default-sample-phase;
+	status = "okay";
+};

pocng try this new build

Index of /libreelec-11/test/

Fingers crossed

jock2 is the greatest expert on these boxes (and beyond). Without him, these builds wouldn't exist because he taught me a lot. So the options he used are 100% the best.

I'll try to make some changes and send you a build to test.

For the options you can use in the dts unfortunately u-boot legacy is not well documented so the only way is to look at the driver

u-boot/drivers/mmc/rockchip_dw_mmc.c at next-dev · rockchip-linux/u-boot

U-Boot tree for pending commits. Contribute to rockchip-linux/u-boot development by creating an account on GitHub.

github.com

However, from what I understand, the emmc that your box has is of a particular type and requires some precautions (I think the option that makes it work is the max-frequency option)

Hi pocng looks promising. I'll try to make the changes to the emmc node of dts or if you have time, feel free to send me a pull request on github

Quote from pocng

PS : in my previous post when i say "multiboot" = U-Boot form multitool => https://github.com/paolosabatino/…tepnand-patches)

So you're saying that
- libreelec flashed in emmc
- replaced the bootloader with the one of multitool
does boot work?

If so, I'll try to check the jock2 patches and understand what needs to be done to fix your problem.

hi pocng are you sure you have an mmc and not a NAND?

When you start the multitool it should alert you to the difference. This is because libreelec 11 does not work with NAND so the behavior you are complaining about is normal.

Quote from pocng

Which parameters did you modify in U-Boot?

I disabled NAND because since it doesn't work, there's no point in having it enabled in u-boot

pocng try this test image

Index of /libreelec-11/test/

let me know

hi pocng it's strange because in both cases the emmc memory is detected by the miniloader

Boot1 Release Time: May 13 2019 17:02:59, version: 2.56
ChipType = 0xc, 380
mmc2:cmd19,100
SdmmcInit=2 0
BootCapSize=1000
UserCapSize=7456MB

I'll try to make some changes to the bootloader and send you a test image.

Hi,

I don't have this board, but with the help of deleonkikkoI produced this test build.

rock-4c-plus

If anyone can do more testing (especially with the 2k display port connector) before officially releasing it and report the problems to me that would be great.