[tfgbd_] Rasberry PI is an embedded computer system but not a PC.

[flyback] only problem is hyoenmadan

[oiaohm] hyoenmadan: some things don't make sense to exist in a uboot system. Like EFI varables and when the board does not have a clock providing a clock service makes no sense.

[flyback] doesn't gaytrail use a imagination technologies gpu

[tfgbd_] ReactOS only works on PCs right now

[flyback] they are such nda asssoles even m$ stopped working with them

[oiaohm] hyoenmadan: so uboot efi is always going to be the most min EFI you are going to hit.

[tfgbd_] Who, intel?

[tfgbd_] Baytrail uses an Intel HD Graphics gpu.

[flyback] oh good

[flyback] you aren't canucked like I am

[hyoenmadan] <oiaohm>: we probably can ditch some stuff, as far as the very basics required by Kernel/HAL are there running

[tfgbd_] So does Cherry trail

[hyoenmadan] <flyback>: well... Imagination uses an UEFI GOP driver in these tables

[hyoenmadan] we can use that

[hyoenmadan] although there will not be hardware acceleration

[tfgbd_] Why call it gaytrail?

[hyoenmadan] not at least until we don't support WDDM style drivers

[tfgbd_] Yeah, all the drivers are WDDM.

[oiaohm] hyoenmadan: Yes agree that reactos should be able to be made run on uboot bare bones EFI loader. Mostly by being sane and accepting that some services that are mandorary by EFI spec are pure optional because places like uboot have no reason to implement them.

[oiaohm] hyoenmadan: like it pass the basic Linux grub2 test.

[flyback] oh really?

[flyback] cause I have a tablet that is stuck on 4.4

[flyback] WAS THINKING ABOUT running linux on it even

[hyoenmadan] <oiaohm>: as i said, ROS makes extensive use of 3 things: ACPI, GIC and SMM... even without framebuffer working, we can output to serial terminal if reuired for testing with these 3 things

[hyoenmadan] would be shorted to 2 if someone implements a bridge between device trees and the WDM model

[tfgbd_] Android 4.4?

[hyoenmadan] as generally ARM socs come mostly with DTs instead ACPI trees

[hyoenmadan] which don't make sense in uboot

[tfgbd_] Just install Linux in a chroot

[oiaohm] hyoenmadan: If that SMM is System Management Mode that does not exist on arm64 or arm32.

[hyoenmadan] uboot and linux/vxworks/qnx

[oiaohm] hyoenmadan: that should shorten list down to 2.

[flyback] I gotta hack my other tablet and motorola phone to move them off android 4.4

[flyback] but the x86 tablet I might just do something else

[hyoenmadan] <oiaohm>: not sure how intel/qualcomm/broadcom managed to do it, but RPis and Qualcomm CPUs provide SMM in ARM

[hyoenmadan] with UEFI ofc

[hyoenmadan] probably Uboot in these socs can enable it too, as sounds more like something implemented in microcode mode

[hyoenmadan] but not sure about other well known socs

[hyoenmadan] is an ugly tasks

[hyoenmadan] *task

[oiaohm] hyoenmadan: at worst that could be something running in arm trustedzone. Put it this way it will not really bet implemented in SMM because the arm cpu does not have that.

[oiaohm] hyoenmadan: SMM as a cpu feature only exists on x86.

[hyoenmadan] yeh, i know

[oiaohm] hyoenmadan: https://lists.linaro.org/pipermail/boot-architecture/2013-December/000390.html trustzone.

[oiaohm] hyoenmadan: and that is a level of fun.

[tfgbd_] Some of the Android only Baytrail devices use a PowerVR gpu

[hyoenmadan] oh damn

[hyoenmadan] yeah, now all makes sense

[hyoenmadan] how the RPi3 supports SMM for Windows/ARM64

[oiaohm] hyoenmadan: if you can boot without SMM/trustzone it will make thing saner for multi OS boot from a single bootloader on ARM64

[tfgbd_] flyback: There is wine for x86 android now

[tfgbd_] But it requires 5.0

[oiaohm] hyoenmadan: yes trustzone is fairly nasty black box at times.

[hyoenmadan] <oiaohm>: thing is i don't have expertise how lack of SMM affects WDM model, driver compatibility and Kernel/HAL coding

[Mysoft] wine for non rooted x86 android?

[hyoenmadan] i only know MS didn't ditched it in its NT Port for some reason

[tfgbd_] yes

[tfgbd_] No root required

[hyoenmadan] and made it mandatory to support windows in ARM64

[stevendale] Can I rune Wine on Windows

[stevendale] o3o

[hyoenmadan] maybe, when time comes and the UEFI loader gets running, we can see what are the real implications

[hyoenmadan] of the lacking bits

[oiaohm] hyoenmadan: Lot of power management stuff historically has been done from SMM. But the thing I find super odd is there are x86 laptops with out any SMM code.

[oiaohm] hyoenmadan: and those x86 laptops without any SMM run windows fine.

[oiaohm] hyoenmadan: so its really good question how required is that SMM stuff.

[hyoenmadan] proving someone codes the mentioned DT -> WDM bridge, or the UBoot UEFI shim provides dynamic build ACPI tables from provided DTs

[hyoenmadan] *built

[hyoenmadan] ROS can't work without an hardware enumerator

[oiaohm] hyoenmadan: please note Microsoft instructions to x86 hardware vendors say you have to implement SMM as well. Even that examples exist with windows 7-10 without any on x86.

[oiaohm] hyoenmadan: so I am not exactly sure how much of a hard limit SMM is.

[oiaohm] hyoenmadan: I could be just do power management drivers different.

[hyoenmadan] <oiaohm>: as long as it doesn't affect other drivers compatibility, as we will intent to use binary IHV provided drivers in ROS ARM64 too

[hyoenmadan] when available

[tfgbd_] No ARM32?

[hyoenmadan] <oiaohm>: btw, do you have some expertise in qemu coding? i've tried to make some patches which add a virtual firewire link to work with QEMU

[oiaohm] stevendale: fragments of wine like the wined3d part can be used on Windows. Items like the loader don't.

[hyoenmadan] but i can't really make any progress with that

[oiaohm] hyoenmadan: every time I have attempt to alter qemu I have failed. Its a code base and half to get head around with all the different timing conditions.