Update 22.02.2019 Feature-complete Debian build - SDIO WiFi/BT / PCIe / 3D / 2D

[MisterA]

I tested my pcie sata controller based on a Marvel 9215 chipset, but it fails to initialize successfully.
The card itself gets extremely hot as well.  

You can see the output of lspci in the attachment.

[festor89]

I might suggest replacing libmali.so on the build with this blob:
https://github.com/rockchip-linux/libmal...x11-gbm.so

I haven't test it yet, but *in theory* that blob should support both gbm and x11. Hopefully that would allow you to run XORG applications, such as web browsers.

If you want wayland instead of X then you could try this blob:
https://github.com/rockchip-linux/libmal...and-gbm.so

Again, I haven't tested either one, so please report back if you have success

Just download the blob and save it as libmali.so and overwrite the current symlink with this. If it works out then I'll include it in the next release.

Unfortunately, the swap file did not give anything. Chromium still does not start

[rosseba]

I tested my pcie sata controller based on a Marvel 9215 chipset, but it fails to initialize successfully.

The same for me with the RockPro64 PCIe Dual SATA card (https://www.pine64.org/?product=rockpro6...rface-card):

Code:

# dmesg | grep -i pci
                  PCI I/O : 0xffffffbffee00000 - 0xffffffbfffe00000   (    16 MB)
[    0.127759] PCI/MSI: /interrupt-controller@fee00000/interrupt-controller@fee20000 domain created
[    0.220531] of_get_named_gpiod_flags: parsed 'gpio' property of node '/vcc3v3-pcie-regulator[0]' - status (0)
[    0.220577] reg-fixed-voltage vcc3v3-pcie-regulator: Looking up vin-supply from device tree
[    0.220623] vcc3v3_pcie: supplied by dc_12v
[    0.220674] vcc3v3_pcie: 3300 mV
[    0.220827] reg-fixed-voltage vcc3v3-pcie-regulator: vcc3v3_pcie supplying 3300000uV
[    0.280807] PCI: CLS 0 bytes, default 64
[    0.414935] ehci-pci: EHCI PCI platform driver
[    0.897651] vcc3v3_pcie: disabling
[    1.837381] phy phy-pcie-phy.9: Looking up phy-supply from device tree
[    1.837397] phy phy-pcie-phy.9: Looking up phy-supply property in node /pcie-phy failed
[    1.839113] rockchip-pcie f8000000.pcie: GPIO lookup for consumer ep
[    1.839128] rockchip-pcie f8000000.pcie: using device tree for GPIO lookup
[    1.839185] of_get_named_gpiod_flags: parsed 'ep-gpios' property of node '/pcie@f8000000[0]' - status (0)
[    1.839472] rockchip-pcie f8000000.pcie: Looking up vpcie3v3-supply from device tree
[    1.839673] rockchip-pcie f8000000.pcie: Looking up vpcie1v8-supply from device tree
[    1.839684] rockchip-pcie f8000000.pcie: Looking up vpcie1v8-supply property in node /pcie@f8000000 failed
[    1.839721] rockchip-pcie f8000000.pcie: no vpcie1v8 regulator found
[    1.839727] rockchip-pcie f8000000.pcie: Looking up vpcie0v9-supply from device tree
[    1.839735] rockchip-pcie f8000000.pcie: Looking up vpcie0v9-supply property in node /pcie@f8000000 failed
[    1.839749] rockchip-pcie f8000000.pcie: no vpcie0v9 regulator found
[    1.839827] rockchip-pcie f8000000.pcie: invalid power supply
[    2.339880] rockchip-pcie f8000000.pcie: PCIe link training gen1 timeout!
[    2.340080] rockchip-pcie: probe of f8000000.pcie failed with error -110

[jja2000]

I tested my pcie sata controller based on a Marvel 9215 chipset, but it fails to initialize successfully.

The same for me with the RockPro64 PCIe Dual SATA card (https://www.pine64.org/?product=rockpro6...rface-card):

Code:

# dmesg | grep -i pci
                  PCI I/O : 0xffffffbffee00000 - 0xffffffbfffe00000   (    16 MB)
[    0.127759] PCI/MSI: /interrupt-controller@fee00000/interrupt-controller@fee20000 domain created
[    0.220531] of_get_named_gpiod_flags: parsed 'gpio' property of node '/vcc3v3-pcie-regulator[0]' - status (0)
[    0.220577] reg-fixed-voltage vcc3v3-pcie-regulator: Looking up vin-supply from device tree
[    0.220623] vcc3v3_pcie: supplied by dc_12v
[    0.220674] vcc3v3_pcie: 3300 mV
[    0.220827] reg-fixed-voltage vcc3v3-pcie-regulator: vcc3v3_pcie supplying 3300000uV
[    0.280807] PCI: CLS 0 bytes, default 64
[    0.414935] ehci-pci: EHCI PCI platform driver
[    0.897651] vcc3v3_pcie: disabling
[    1.837381] phy phy-pcie-phy.9: Looking up phy-supply from device tree
[    1.837397] phy phy-pcie-phy.9: Looking up phy-supply property in node /pcie-phy failed
[    1.839113] rockchip-pcie f8000000.pcie: GPIO lookup for consumer ep
[    1.839128] rockchip-pcie f8000000.pcie: using device tree for GPIO lookup
[    1.839185] of_get_named_gpiod_flags: parsed 'ep-gpios' property of node '/pcie@f8000000[0]' - status (0)
[    1.839472] rockchip-pcie f8000000.pcie: Looking up vpcie3v3-supply from device tree
[    1.839673] rockchip-pcie f8000000.pcie: Looking up vpcie1v8-supply from device tree
[    1.839684] rockchip-pcie f8000000.pcie: Looking up vpcie1v8-supply property in node /pcie@f8000000 failed
[    1.839721] rockchip-pcie f8000000.pcie: no vpcie1v8 regulator found
[    1.839727] rockchip-pcie f8000000.pcie: Looking up vpcie0v9-supply from device tree
[    1.839735] rockchip-pcie f8000000.pcie: Looking up vpcie0v9-supply property in node /pcie@f8000000 failed
[    1.839749] rockchip-pcie f8000000.pcie: no vpcie0v9 regulator found
[    1.839827] rockchip-pcie f8000000.pcie: invalid power supply
[    2.339880] rockchip-pcie f8000000.pcie: PCIe link training gen1 timeout!
[    2.340080] rockchip-pcie: probe of f8000000.pcie failed with error -110

I have exactly the same problem with Ayufan's builds, armbian's builds and this build on a myriad of kernels. Yet a lot of people seem to have SATA or just pcie in general working. How is it that nobody higher up at Pine64 or even Ayufan has taken a look at this? I get that people have stuff to do and all that, but I've spent way too much money on a product that's basically an expensive paperweight to me.

For the record, this also happens when there's no pcie card inserted

I'm not happy with this...

https://forum.pine64.org/showthread.php?...&pid=41632
https://forum.pine64.org/showthread.php?...3#pid41513
https://forum.pine64.org/showthread.php?...3#pid42243

[Mentaluproar]

It shows that this images is running in 32 bit mode. Why is that?

[NoDiskNoFun]

@mrfixit: First off: Thank you for your work!!!
I wanted to ask you how you get the SDIO WIFI module to work. I am running Arch Linux now and i tried to boot your kernel but it did not work (seems like some kind of firmware was not found). I also tried to enable all SDIO related options in kernel-config and built it against Mainline 4.19. Building did work but WIFI cannot be detected. I am asking you to either share your trick to get Linux to detect SDIO WIFI or alternatively, as open source programming generally would suggest, to share the complete patchset and/or config with the community.

[Mrfixit2001]

There is an updated release coming (I expect next week sometime) which will include additional kernel drivers and patches, including some for PCIe That should take care of the remaining issues being experienced.

I've currently only provided arm32 images (kernel is 64-bit with a 32-bit userspace), but I can work on providing an arm64 image in my next release as well if there's a need for it.

At some point after that I'll work on a desktop image as well so you don't have to worry about the hassle of installing all the packages yourselves.

As for how to get SDIO, WIFI, and BT working in the kernel, I haven't published my kernel source itself yet. But I did write this patch for Batocera today which should outline the minimum requirements to get these features enabled. Hasn't been tested yet, as my kernel has an enormous amount of backports from mainline in it so I had to write this patch specifically for the "stock" rockchip kernel. Basically I'm saying I cannot guarantee how it will function without all the driver updates I've back-ported, but I think it's sufficient to enable everything.
https://github.com/batocera-linux/batoce...147d0ef192

[tblindba]

There is an updated release coming (I expect next week sometime) which will include additional kernel drivers and patches, including some for PCIe That should take care of the remaining issues being experienced.

I've currently only provided arm32 images (kernel is 64-bit with a 32-bit userspace), but I can work on providing an arm64 image in my next release as well if there's a need for it.

At some point after that I'll work on a desktop image as well so you don't have to worry about the hassle of installing all the packages yourselves.

As for how to get SDIO, WIFI, and BT working in the kernel, I haven't published my kernel source itself yet. But I did write this patch for Batocera today which should outline the minimum requirements to get these features enabled. Hasn't been tested yet, as my kernel has an enormous amount of backports from mainline in it so I had to write this patch specifically for the "stock" rockchip kernel. Basically I'm saying I cannot guarantee how it will function without all the driver updates I've back-ported, but I think it's sufficient to enable everything.
https://github.com/batocera-linux/batoce...147d0ef192

I'm pretty excited about the work you are doing. A functional desktop environment with GPU support would make this board amazing. I put the board down after a while due to the software situation, but this build really reinvigorated my interest.

Great job so far!

[fosf0r]

Still dying for kernel source tree.
(Doesn't that violate the GPL?)

[NoDiskNoFun]

Still dying for kernel source tree.
(Doesn't that violate the GPL?)

Me too